diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,36917 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.655574043261232,
+  "eval_steps": 250,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0033277870216306157,
+      "grad_norm": 99.72433471679688,
+      "learning_rate": 5e-06,
+      "loss": 2.2579,
+      "num_input_tokens_seen": 62524,
+      "step": 1
+    },
+    {
+      "epoch": 0.0033277870216306157,
+      "loss": 2.4882845878601074,
+      "loss_ce": 0.4775424301624298,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.177734375,
+      "loss_xval": 2.015625,
+      "num_input_tokens_seen": 62524,
+      "step": 1
+    },
+    {
+      "epoch": 0.0066555740432612314,
+      "grad_norm": 43.31109619140625,
+      "learning_rate": 5e-06,
+      "loss": 1.8354,
+      "num_input_tokens_seen": 124728,
+      "step": 2
+    },
+    {
+      "epoch": 0.0066555740432612314,
+      "loss": 1.5098202228546143,
+      "loss_ce": 0.0459531769156456,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0849609375,
+      "loss_xval": 1.4609375,
+      "num_input_tokens_seen": 124728,
+      "step": 2
+    },
+    {
+      "epoch": 0.009983361064891847,
+      "grad_norm": 40.23197555541992,
+      "learning_rate": 5e-06,
+      "loss": 1.8923,
+      "num_input_tokens_seen": 187516,
+      "step": 3
+    },
+    {
+      "epoch": 0.009983361064891847,
+      "loss": 1.9805556535720825,
+      "loss_ce": 0.46248918771743774,
+      "loss_iou": 0.4375,
+      "loss_num": 0.12890625,
+      "loss_xval": 1.515625,
+      "num_input_tokens_seen": 187516,
+      "step": 3
+    },
+    {
+      "epoch": 0.013311148086522463,
+      "grad_norm": 41.29404067993164,
+      "learning_rate": 5e-06,
+      "loss": 1.8433,
+      "num_input_tokens_seen": 249256,
+      "step": 4
+    },
+    {
+      "epoch": 0.013311148086522463,
+      "loss": 1.760237216949463,
+      "loss_ce": 0.29978805780410767,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.10205078125,
+      "loss_xval": 1.4609375,
+      "num_input_tokens_seen": 249256,
+      "step": 4
+    },
+    {
+      "epoch": 0.016638935108153077,
+      "grad_norm": 45.62328338623047,
+      "learning_rate": 5e-06,
+      "loss": 1.9853,
+      "num_input_tokens_seen": 312896,
+      "step": 5
+    },
+    {
+      "epoch": 0.016638935108153077,
+      "loss": 1.6446335315704346,
+      "loss_ce": 0.36777815222740173,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.1005859375,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 312896,
+      "step": 5
+    },
+    {
+      "epoch": 0.019966722129783693,
+      "grad_norm": 26.73392677307129,
+      "learning_rate": 5e-06,
+      "loss": 1.5267,
+      "num_input_tokens_seen": 373800,
+      "step": 6
+    },
+    {
+      "epoch": 0.019966722129783693,
+      "loss": 1.481225848197937,
+      "loss_ce": 0.186303973197937,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0771484375,
+      "loss_xval": 1.296875,
+      "num_input_tokens_seen": 373800,
+      "step": 6
+    },
+    {
+      "epoch": 0.02329450915141431,
+      "grad_norm": 28.28041648864746,
+      "learning_rate": 5e-06,
+      "loss": 1.8401,
+      "num_input_tokens_seen": 435532,
+      "step": 7
+    },
+    {
+      "epoch": 0.02329450915141431,
+      "loss": 1.7152678966522217,
+      "loss_ce": 0.2172209620475769,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.158203125,
+      "loss_xval": 1.5,
+      "num_input_tokens_seen": 435532,
+      "step": 7
+    },
+    {
+      "epoch": 0.026622296173044926,
+      "grad_norm": 93.13752746582031,
+      "learning_rate": 5e-06,
+      "loss": 1.7189,
+      "num_input_tokens_seen": 498240,
+      "step": 8
+    },
+    {
+      "epoch": 0.026622296173044926,
+      "loss": 1.7084490060806274,
+      "loss_ce": 0.12007011473178864,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.11865234375,
+      "loss_xval": 1.5859375,
+      "num_input_tokens_seen": 498240,
+      "step": 8
+    },
+    {
+      "epoch": 0.029950083194675542,
+      "grad_norm": 55.754974365234375,
+      "learning_rate": 5e-06,
+      "loss": 1.9588,
+      "num_input_tokens_seen": 560944,
+      "step": 9
+    },
+    {
+      "epoch": 0.029950083194675542,
+      "loss": 1.9288438558578491,
+      "loss_ce": 0.43592390418052673,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.1474609375,
+      "loss_xval": 1.4921875,
+      "num_input_tokens_seen": 560944,
+      "step": 9
+    },
+    {
+      "epoch": 0.033277870216306155,
+      "grad_norm": 24.77700424194336,
+      "learning_rate": 5e-06,
+      "loss": 1.7834,
+      "num_input_tokens_seen": 623276,
+      "step": 10
+    },
+    {
+      "epoch": 0.033277870216306155,
+      "loss": 1.2600505352020264,
+      "loss_ce": 0.17655442655086517,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.1494140625,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 623276,
+      "step": 10
+    },
+    {
+      "epoch": 0.036605657237936774,
+      "grad_norm": 53.40562438964844,
+      "learning_rate": 5e-06,
+      "loss": 1.9654,
+      "num_input_tokens_seen": 686344,
+      "step": 11
+    },
+    {
+      "epoch": 0.036605657237936774,
+      "loss": 1.5656721591949463,
+      "loss_ce": 0.15307454764842987,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.15234375,
+      "loss_xval": 1.4140625,
+      "num_input_tokens_seen": 686344,
+      "step": 11
+    },
+    {
+      "epoch": 0.03993344425956739,
+      "grad_norm": 185.1483612060547,
+      "learning_rate": 5e-06,
+      "loss": 1.5234,
+      "num_input_tokens_seen": 748796,
+      "step": 12
+    },
+    {
+      "epoch": 0.03993344425956739,
+      "loss": 0.9671105146408081,
+      "loss_ce": 0.1262902468442917,
+      "loss_iou": 0.0,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.83984375,
+      "num_input_tokens_seen": 748796,
+      "step": 12
+    },
+    {
+      "epoch": 0.04326123128119801,
+      "grad_norm": 41.721710205078125,
+      "learning_rate": 5e-06,
+      "loss": 2.2144,
+      "num_input_tokens_seen": 811248,
+      "step": 13
+    },
+    {
+      "epoch": 0.04326123128119801,
+      "loss": 1.7121745347976685,
+      "loss_ce": 0.382096529006958,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.126953125,
+      "loss_xval": 1.328125,
+      "num_input_tokens_seen": 811248,
+      "step": 13
+    },
+    {
+      "epoch": 0.04658901830282862,
+      "grad_norm": 26.450149536132812,
+      "learning_rate": 5e-06,
+      "loss": 2.0257,
+      "num_input_tokens_seen": 874336,
+      "step": 14
+    },
+    {
+      "epoch": 0.04658901830282862,
+      "loss": 2.167698383331299,
+      "loss_ce": 0.07883133739233017,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.15625,
+      "loss_xval": 2.09375,
+      "num_input_tokens_seen": 874336,
+      "step": 14
+    },
+    {
+      "epoch": 0.04991680532445923,
+      "grad_norm": 28.79014015197754,
+      "learning_rate": 5e-06,
+      "loss": 1.5748,
+      "num_input_tokens_seen": 934104,
+      "step": 15
+    },
+    {
+      "epoch": 0.04991680532445923,
+      "loss": 1.746118426322937,
+      "loss_ce": 0.3081301152706146,
+      "loss_iou": 0.34375,
+      "loss_num": 0.150390625,
+      "loss_xval": 1.4375,
+      "num_input_tokens_seen": 934104,
+      "step": 15
+    },
+    {
+      "epoch": 0.05324459234608985,
+      "grad_norm": 35.701480865478516,
+      "learning_rate": 5e-06,
+      "loss": 1.6555,
+      "num_input_tokens_seen": 995740,
+      "step": 16
+    },
+    {
+      "epoch": 0.05324459234608985,
+      "loss": 1.4585609436035156,
+      "loss_ce": 0.164127379655838,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.1318359375,
+      "loss_xval": 1.296875,
+      "num_input_tokens_seen": 995740,
+      "step": 16
+    },
+    {
+      "epoch": 0.056572379367720464,
+      "grad_norm": 31.904708862304688,
+      "learning_rate": 5e-06,
+      "loss": 1.4605,
+      "num_input_tokens_seen": 1057552,
+      "step": 17
+    },
+    {
+      "epoch": 0.056572379367720464,
+      "loss": 1.500415563583374,
+      "loss_ce": 0.4437748193740845,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.10693359375,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 1057552,
+      "step": 17
+    },
+    {
+      "epoch": 0.059900166389351084,
+      "grad_norm": 26.225719451904297,
+      "learning_rate": 5e-06,
+      "loss": 1.8686,
+      "num_input_tokens_seen": 1117304,
+      "step": 18
+    },
+    {
+      "epoch": 0.059900166389351084,
+      "loss": 2.0437541007995605,
+      "loss_ce": 0.25664472579956055,
+      "loss_iou": 0.546875,
+      "loss_num": 0.1376953125,
+      "loss_xval": 1.7890625,
+      "num_input_tokens_seen": 1117304,
+      "step": 18
+    },
+    {
+      "epoch": 0.0632279534109817,
+      "grad_norm": 107.4391098022461,
+      "learning_rate": 5e-06,
+      "loss": 1.8643,
+      "num_input_tokens_seen": 1180152,
+      "step": 19
+    },
+    {
+      "epoch": 0.0632279534109817,
+      "loss": 1.5854601860046387,
+      "loss_ce": 0.12549912929534912,
+      "loss_iou": 0.359375,
+      "loss_num": 0.1484375,
+      "loss_xval": 1.4609375,
+      "num_input_tokens_seen": 1180152,
+      "step": 19
+    },
+    {
+      "epoch": 0.06655574043261231,
+      "grad_norm": 31.948070526123047,
+      "learning_rate": 5e-06,
+      "loss": 1.6704,
+      "num_input_tokens_seen": 1240584,
+      "step": 20
+    },
+    {
+      "epoch": 0.06655574043261231,
+      "loss": 1.782106876373291,
+      "loss_ce": 0.6380637884140015,
+      "loss_iou": 0.203125,
+      "loss_num": 0.1474609375,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 1240584,
+      "step": 20
+    },
+    {
+      "epoch": 0.06988352745424292,
+      "grad_norm": 39.34145736694336,
+      "learning_rate": 5e-06,
+      "loss": 1.8904,
+      "num_input_tokens_seen": 1301508,
+      "step": 21
+    },
+    {
+      "epoch": 0.06988352745424292,
+      "loss": 1.6883811950683594,
+      "loss_ce": 0.6380882263183594,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.111328125,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 1301508,
+      "step": 21
+    },
+    {
+      "epoch": 0.07321131447587355,
+      "grad_norm": 28.05674934387207,
+      "learning_rate": 5e-06,
+      "loss": 2.0444,
+      "num_input_tokens_seen": 1363652,
+      "step": 22
+    },
+    {
+      "epoch": 0.07321131447587355,
+      "loss": 2.08927845954895,
+      "loss_ce": 0.1547081470489502,
+      "loss_iou": 0.5625,
+      "loss_num": 0.162109375,
+      "loss_xval": 1.9375,
+      "num_input_tokens_seen": 1363652,
+      "step": 22
+    },
+    {
+      "epoch": 0.07653910149750416,
+      "grad_norm": 34.66334915161133,
+      "learning_rate": 5e-06,
+      "loss": 2.0099,
+      "num_input_tokens_seen": 1427244,
+      "step": 23
+    },
+    {
+      "epoch": 0.07653910149750416,
+      "loss": 2.1946048736572266,
+      "loss_ce": 0.20241737365722656,
+      "loss_iou": 0.640625,
+      "loss_num": 0.142578125,
+      "loss_xval": 1.9921875,
+      "num_input_tokens_seen": 1427244,
+      "step": 23
+    },
+    {
+      "epoch": 0.07986688851913477,
+      "grad_norm": 16.318044662475586,
+      "learning_rate": 5e-06,
+      "loss": 1.6373,
+      "num_input_tokens_seen": 1488232,
+      "step": 24
+    },
+    {
+      "epoch": 0.07986688851913477,
+      "loss": 1.2525696754455566,
+      "loss_ce": 0.21619272232055664,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.1064453125,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 1488232,
+      "step": 24
+    },
+    {
+      "epoch": 0.08319467554076539,
+      "grad_norm": 22.026212692260742,
+      "learning_rate": 5e-06,
+      "loss": 1.8012,
+      "num_input_tokens_seen": 1549668,
+      "step": 25
+    },
+    {
+      "epoch": 0.08319467554076539,
+      "loss": 1.860095500946045,
+      "loss_ce": 0.44945091009140015,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.1591796875,
+      "loss_xval": 1.4140625,
+      "num_input_tokens_seen": 1549668,
+      "step": 25
+    },
+    {
+      "epoch": 0.08652246256239601,
+      "grad_norm": 26.607149124145508,
+      "learning_rate": 5e-06,
+      "loss": 1.898,
+      "num_input_tokens_seen": 1612748,
+      "step": 26
+    },
+    {
+      "epoch": 0.08652246256239601,
+      "loss": 1.7494425773620605,
+      "loss_ce": 0.19084876775741577,
+      "loss_iou": 0.4375,
+      "loss_num": 0.13671875,
+      "loss_xval": 1.5625,
+      "num_input_tokens_seen": 1612748,
+      "step": 26
+    },
+    {
+      "epoch": 0.08985024958402663,
+      "grad_norm": 100.14049530029297,
+      "learning_rate": 5e-06,
+      "loss": 1.9263,
+      "num_input_tokens_seen": 1675456,
+      "step": 27
+    },
+    {
+      "epoch": 0.08985024958402663,
+      "loss": 1.7160944938659668,
+      "loss_ce": 0.20193445682525635,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.099609375,
+      "loss_xval": 1.515625,
+      "num_input_tokens_seen": 1675456,
+      "step": 27
+    },
+    {
+      "epoch": 0.09317803660565724,
+      "grad_norm": 31.630752563476562,
+      "learning_rate": 5e-06,
+      "loss": 2.0261,
+      "num_input_tokens_seen": 1736840,
+      "step": 28
+    },
+    {
+      "epoch": 0.09317803660565724,
+      "loss": 1.7389755249023438,
+      "loss_ce": 0.17842870950698853,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.1416015625,
+      "loss_xval": 1.5625,
+      "num_input_tokens_seen": 1736840,
+      "step": 28
+    },
+    {
+      "epoch": 0.09650582362728785,
+      "grad_norm": 23.318147659301758,
+      "learning_rate": 5e-06,
+      "loss": 1.7689,
+      "num_input_tokens_seen": 1799508,
+      "step": 29
+    },
+    {
+      "epoch": 0.09650582362728785,
+      "loss": 1.8038142919540405,
+      "loss_ce": 0.46494707465171814,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.12451171875,
+      "loss_xval": 1.3359375,
+      "num_input_tokens_seen": 1799508,
+      "step": 29
+    },
+    {
+      "epoch": 0.09983361064891846,
+      "grad_norm": 20.128877639770508,
+      "learning_rate": 5e-06,
+      "loss": 2.0295,
+      "num_input_tokens_seen": 1862340,
+      "step": 30
+    },
+    {
+      "epoch": 0.09983361064891846,
+      "loss": 2.387416362762451,
+      "loss_ce": 0.5177874565124512,
+      "loss_iou": 0.53125,
+      "loss_num": 0.162109375,
+      "loss_xval": 1.8671875,
+      "num_input_tokens_seen": 1862340,
+      "step": 30
+    },
+    {
+      "epoch": 0.10316139767054909,
+      "grad_norm": 102.55489349365234,
+      "learning_rate": 5e-06,
+      "loss": 1.695,
+      "num_input_tokens_seen": 1926308,
+      "step": 31
+    },
+    {
+      "epoch": 0.10316139767054909,
+      "loss": 1.4592796564102173,
+      "loss_ce": 0.12383048981428146,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0888671875,
+      "loss_xval": 1.3359375,
+      "num_input_tokens_seen": 1926308,
+      "step": 31
+    },
+    {
+      "epoch": 0.1064891846921797,
+      "grad_norm": 24.122255325317383,
+      "learning_rate": 5e-06,
+      "loss": 1.5715,
+      "num_input_tokens_seen": 1987296,
+      "step": 32
+    },
+    {
+      "epoch": 0.1064891846921797,
+      "loss": 1.2685531377792358,
+      "loss_ce": 0.11279135942459106,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.1572265625,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 1987296,
+      "step": 32
+    },
+    {
+      "epoch": 0.10981697171381032,
+      "grad_norm": 29.521120071411133,
+      "learning_rate": 5e-06,
+      "loss": 1.9234,
+      "num_input_tokens_seen": 2049684,
+      "step": 33
+    },
+    {
+      "epoch": 0.10981697171381032,
+      "loss": 2.4738051891326904,
+      "loss_ce": 0.3732193112373352,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.171875,
+      "loss_xval": 2.09375,
+      "num_input_tokens_seen": 2049684,
+      "step": 33
+    },
+    {
+      "epoch": 0.11314475873544093,
+      "grad_norm": 33.83547592163086,
+      "learning_rate": 5e-06,
+      "loss": 2.2031,
+      "num_input_tokens_seen": 2112524,
+      "step": 34
+    },
+    {
+      "epoch": 0.11314475873544093,
+      "loss": 2.4939706325531006,
+      "loss_ce": 0.40705662965774536,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.162109375,
+      "loss_xval": 2.09375,
+      "num_input_tokens_seen": 2112524,
+      "step": 34
+    },
+    {
+      "epoch": 0.11647254575707154,
+      "grad_norm": 28.041799545288086,
+      "learning_rate": 5e-06,
+      "loss": 1.9234,
+      "num_input_tokens_seen": 2174764,
+      "step": 35
+    },
+    {
+      "epoch": 0.11647254575707154,
+      "loss": 2.1027140617370605,
+      "loss_ce": 0.321463942527771,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.1953125,
+      "loss_xval": 1.78125,
+      "num_input_tokens_seen": 2174764,
+      "step": 35
+    },
+    {
+      "epoch": 0.11980033277870217,
+      "grad_norm": 16.05351448059082,
+      "learning_rate": 5e-06,
+      "loss": 1.6234,
+      "num_input_tokens_seen": 2236936,
+      "step": 36
+    },
+    {
+      "epoch": 0.11980033277870217,
+      "loss": 1.9183712005615234,
+      "loss_ce": 0.2826289236545563,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.11376953125,
+      "loss_xval": 1.6328125,
+      "num_input_tokens_seen": 2236936,
+      "step": 36
+    },
+    {
+      "epoch": 0.12312811980033278,
+      "grad_norm": 25.214027404785156,
+      "learning_rate": 5e-06,
+      "loss": 1.7429,
+      "num_input_tokens_seen": 2297968,
+      "step": 37
+    },
+    {
+      "epoch": 0.12312811980033278,
+      "loss": 1.8933910131454468,
+      "loss_ce": 0.352375328540802,
+      "loss_iou": 0.453125,
+      "loss_num": 0.126953125,
+      "loss_xval": 1.5390625,
+      "num_input_tokens_seen": 2297968,
+      "step": 37
+    },
+    {
+      "epoch": 0.1264559068219634,
+      "grad_norm": 41.4306526184082,
+      "learning_rate": 5e-06,
+      "loss": 1.9087,
+      "num_input_tokens_seen": 2359480,
+      "step": 38
+    },
+    {
+      "epoch": 0.1264559068219634,
+      "loss": 1.9372410774230957,
+      "loss_ce": 0.3669285476207733,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.134765625,
+      "loss_xval": 1.5703125,
+      "num_input_tokens_seen": 2359480,
+      "step": 38
+    },
+    {
+      "epoch": 0.129783693843594,
+      "grad_norm": 87.06895446777344,
+      "learning_rate": 5e-06,
+      "loss": 1.4617,
+      "num_input_tokens_seen": 2422388,
+      "step": 39
+    },
+    {
+      "epoch": 0.129783693843594,
+      "loss": 1.5843279361724854,
+      "loss_ce": 0.26157402992248535,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.10107421875,
+      "loss_xval": 1.3203125,
+      "num_input_tokens_seen": 2422388,
+      "step": 39
+    },
+    {
+      "epoch": 0.13311148086522462,
+      "grad_norm": 22.859128952026367,
+      "learning_rate": 5e-06,
+      "loss": 1.7171,
+      "num_input_tokens_seen": 2485508,
+      "step": 40
+    },
+    {
+      "epoch": 0.13311148086522462,
+      "loss": 1.6260865926742554,
+      "loss_ce": 0.1578247845172882,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.142578125,
+      "loss_xval": 1.46875,
+      "num_input_tokens_seen": 2485508,
+      "step": 40
+    },
+    {
+      "epoch": 0.13643926788685523,
+      "grad_norm": 29.294437408447266,
+      "learning_rate": 5e-06,
+      "loss": 1.6604,
+      "num_input_tokens_seen": 2547740,
+      "step": 41
+    },
+    {
+      "epoch": 0.13643926788685523,
+      "loss": 1.5736039876937866,
+      "loss_ce": 0.11828167736530304,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.1376953125,
+      "loss_xval": 1.453125,
+      "num_input_tokens_seen": 2547740,
+      "step": 41
+    },
+    {
+      "epoch": 0.13976705490848584,
+      "grad_norm": 22.002099990844727,
+      "learning_rate": 5e-06,
+      "loss": 1.9157,
+      "num_input_tokens_seen": 2610656,
+      "step": 42
+    },
+    {
+      "epoch": 0.13976705490848584,
+      "loss": 2.026005744934082,
+      "loss_ce": 0.26331043243408203,
+      "loss_iou": 0.515625,
+      "loss_num": 0.146484375,
+      "loss_xval": 1.765625,
+      "num_input_tokens_seen": 2610656,
+      "step": 42
+    },
+    {
+      "epoch": 0.14309484193011648,
+      "grad_norm": 42.256507873535156,
+      "learning_rate": 5e-06,
+      "loss": 1.9787,
+      "num_input_tokens_seen": 2671692,
+      "step": 43
+    },
+    {
+      "epoch": 0.14309484193011648,
+      "loss": 1.8853733539581299,
+      "loss_ce": 0.2652561664581299,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.12158203125,
+      "loss_xval": 1.6171875,
+      "num_input_tokens_seen": 2671692,
+      "step": 43
+    },
+    {
+      "epoch": 0.1464226289517471,
+      "grad_norm": 23.313539505004883,
+      "learning_rate": 5e-06,
+      "loss": 1.97,
+      "num_input_tokens_seen": 2734876,
+      "step": 44
+    },
+    {
+      "epoch": 0.1464226289517471,
+      "loss": 2.3043510913848877,
+      "loss_ce": 0.4186089336872101,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.1435546875,
+      "loss_xval": 1.8828125,
+      "num_input_tokens_seen": 2734876,
+      "step": 44
+    },
+    {
+      "epoch": 0.1497504159733777,
+      "grad_norm": 19.345558166503906,
+      "learning_rate": 5e-06,
+      "loss": 1.9084,
+      "num_input_tokens_seen": 2798924,
+      "step": 45
+    },
+    {
+      "epoch": 0.1497504159733777,
+      "loss": 1.7593090534210205,
+      "loss_ce": 0.1675121784210205,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.1171875,
+      "loss_xval": 1.59375,
+      "num_input_tokens_seen": 2798924,
+      "step": 45
+    },
+    {
+      "epoch": 0.15307820299500832,
+      "grad_norm": 56.2034912109375,
+      "learning_rate": 5e-06,
+      "loss": 1.5545,
+      "num_input_tokens_seen": 2862396,
+      "step": 46
+    },
+    {
+      "epoch": 0.15307820299500832,
+      "loss": 1.1661244630813599,
+      "loss_ce": 0.09337049722671509,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.091796875,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 2862396,
+      "step": 46
+    },
+    {
+      "epoch": 0.15640599001663893,
+      "grad_norm": 15.956643104553223,
+      "learning_rate": 5e-06,
+      "loss": 1.6658,
+      "num_input_tokens_seen": 2926024,
+      "step": 47
+    },
+    {
+      "epoch": 0.15640599001663893,
+      "loss": 1.5099748373031616,
+      "loss_ce": 0.054896753281354904,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.08984375,
+      "loss_xval": 1.453125,
+      "num_input_tokens_seen": 2926024,
+      "step": 47
+    },
+    {
+      "epoch": 0.15973377703826955,
+      "grad_norm": 25.402143478393555,
+      "learning_rate": 5e-06,
+      "loss": 1.8769,
+      "num_input_tokens_seen": 2989532,
+      "step": 48
+    },
+    {
+      "epoch": 0.15973377703826955,
+      "loss": 1.6056416034698486,
+      "loss_ce": 0.17790716886520386,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.10888671875,
+      "loss_xval": 1.4296875,
+      "num_input_tokens_seen": 2989532,
+      "step": 48
+    },
+    {
+      "epoch": 0.16306156405990016,
+      "grad_norm": 26.91611671447754,
+      "learning_rate": 5e-06,
+      "loss": 1.8648,
+      "num_input_tokens_seen": 3052660,
+      "step": 49
+    },
+    {
+      "epoch": 0.16306156405990016,
+      "loss": 1.767176866531372,
+      "loss_ce": 0.14803630113601685,
+      "loss_iou": 0.515625,
+      "loss_num": 0.1181640625,
+      "loss_xval": 1.6171875,
+      "num_input_tokens_seen": 3052660,
+      "step": 49
+    },
+    {
+      "epoch": 0.16638935108153077,
+      "grad_norm": 17.735488891601562,
+      "learning_rate": 5e-06,
+      "loss": 1.733,
+      "num_input_tokens_seen": 3114048,
+      "step": 50
+    },
+    {
+      "epoch": 0.16638935108153077,
+      "loss": 1.8018484115600586,
+      "loss_ce": 0.201262429356575,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.171875,
+      "loss_xval": 1.6015625,
+      "num_input_tokens_seen": 3114048,
+      "step": 50
+    },
+    {
+      "epoch": 0.16971713810316139,
+      "grad_norm": 23.010356903076172,
+      "learning_rate": 5e-06,
+      "loss": 1.7252,
+      "num_input_tokens_seen": 3177028,
+      "step": 51
+    },
+    {
+      "epoch": 0.16971713810316139,
+      "loss": 1.9260194301605225,
+      "loss_ce": 0.23461312055587769,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.1220703125,
+      "loss_xval": 1.6875,
+      "num_input_tokens_seen": 3177028,
+      "step": 51
+    },
+    {
+      "epoch": 0.17304492512479203,
+      "grad_norm": 50.80724334716797,
+      "learning_rate": 5e-06,
+      "loss": 1.5932,
+      "num_input_tokens_seen": 3240048,
+      "step": 52
+    },
+    {
+      "epoch": 0.17304492512479203,
+      "loss": 1.2517362833023071,
+      "loss_ce": 0.07107216864824295,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.1142578125,
+      "loss_xval": 1.1796875,
+      "num_input_tokens_seen": 3240048,
+      "step": 52
+    },
+    {
+      "epoch": 0.17637271214642264,
+      "grad_norm": 19.88364601135254,
+      "learning_rate": 5e-06,
+      "loss": 1.7949,
+      "num_input_tokens_seen": 3302084,
+      "step": 53
+    },
+    {
+      "epoch": 0.17637271214642264,
+      "loss": 1.9920969009399414,
+      "loss_ce": 0.1786203682422638,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.1337890625,
+      "loss_xval": 1.8125,
+      "num_input_tokens_seen": 3302084,
+      "step": 53
+    },
+    {
+      "epoch": 0.17970049916805325,
+      "grad_norm": 15.960768699645996,
+      "learning_rate": 5e-06,
+      "loss": 1.8404,
+      "num_input_tokens_seen": 3364032,
+      "step": 54
+    },
+    {
+      "epoch": 0.17970049916805325,
+      "loss": 1.923379898071289,
+      "loss_ce": 0.3872470259666443,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.125,
+      "loss_xval": 1.5390625,
+      "num_input_tokens_seen": 3364032,
+      "step": 54
+    },
+    {
+      "epoch": 0.18302828618968386,
+      "grad_norm": 34.17069625854492,
+      "learning_rate": 5e-06,
+      "loss": 1.6571,
+      "num_input_tokens_seen": 3426944,
+      "step": 55
+    },
+    {
+      "epoch": 0.18302828618968386,
+      "loss": 1.422314167022705,
+      "loss_ce": 0.04633765295147896,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.126953125,
+      "loss_xval": 1.375,
+      "num_input_tokens_seen": 3426944,
+      "step": 55
+    },
+    {
+      "epoch": 0.18635607321131448,
+      "grad_norm": 14.88204288482666,
+      "learning_rate": 5e-06,
+      "loss": 1.7171,
+      "num_input_tokens_seen": 3488980,
+      "step": 56
+    },
+    {
+      "epoch": 0.18635607321131448,
+      "loss": 1.6802122592926025,
+      "loss_ce": 0.12698963284492493,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.1240234375,
+      "loss_xval": 1.5546875,
+      "num_input_tokens_seen": 3488980,
+      "step": 56
+    },
+    {
+      "epoch": 0.1896838602329451,
+      "grad_norm": 25.315814971923828,
+      "learning_rate": 5e-06,
+      "loss": 1.5849,
+      "num_input_tokens_seen": 3552192,
+      "step": 57
+    },
+    {
+      "epoch": 0.1896838602329451,
+      "loss": 1.5455249547958374,
+      "loss_ce": 0.09777102619409561,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.10498046875,
+      "loss_xval": 1.4453125,
+      "num_input_tokens_seen": 3552192,
+      "step": 57
+    },
+    {
+      "epoch": 0.1930116472545757,
+      "grad_norm": 52.65762710571289,
+      "learning_rate": 5e-06,
+      "loss": 1.7184,
+      "num_input_tokens_seen": 3614784,
+      "step": 58
+    },
+    {
+      "epoch": 0.1930116472545757,
+      "loss": 1.7533233165740967,
+      "loss_ce": 0.1888701170682907,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.1279296875,
+      "loss_xval": 1.5625,
+      "num_input_tokens_seen": 3614784,
+      "step": 58
+    },
+    {
+      "epoch": 0.19633943427620631,
+      "grad_norm": 10.436413764953613,
+      "learning_rate": 5e-06,
+      "loss": 1.5083,
+      "num_input_tokens_seen": 3676144,
+      "step": 59
+    },
+    {
+      "epoch": 0.19633943427620631,
+      "loss": 1.4122167825698853,
+      "loss_ce": 0.03331056609749794,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.1416015625,
+      "loss_xval": 1.375,
+      "num_input_tokens_seen": 3676144,
+      "step": 59
+    },
+    {
+      "epoch": 0.19966722129783693,
+      "grad_norm": 23.417434692382812,
+      "learning_rate": 5e-06,
+      "loss": 1.513,
+      "num_input_tokens_seen": 3740152,
+      "step": 60
+    },
+    {
+      "epoch": 0.19966722129783693,
+      "loss": 1.52932870388031,
+      "loss_ce": 0.15628191828727722,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0966796875,
+      "loss_xval": 1.375,
+      "num_input_tokens_seen": 3740152,
+      "step": 60
+    },
+    {
+      "epoch": 0.20299500831946754,
+      "grad_norm": 57.756534576416016,
+      "learning_rate": 5e-06,
+      "loss": 1.9057,
+      "num_input_tokens_seen": 3803536,
+      "step": 61
+    },
+    {
+      "epoch": 0.20299500831946754,
+      "loss": 2.161196708679199,
+      "loss_ce": 0.07525897026062012,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.1640625,
+      "loss_xval": 2.09375,
+      "num_input_tokens_seen": 3803536,
+      "step": 61
+    },
+    {
+      "epoch": 0.20632279534109818,
+      "grad_norm": 34.947166442871094,
+      "learning_rate": 5e-06,
+      "loss": 1.5023,
+      "num_input_tokens_seen": 3866352,
+      "step": 62
+    },
+    {
+      "epoch": 0.20632279534109818,
+      "loss": 1.5223082304000854,
+      "loss_ce": 0.03939810022711754,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.11865234375,
+      "loss_xval": 1.484375,
+      "num_input_tokens_seen": 3866352,
+      "step": 62
+    },
+    {
+      "epoch": 0.2096505823627288,
+      "grad_norm": 27.8328800201416,
+      "learning_rate": 5e-06,
+      "loss": 1.6887,
+      "num_input_tokens_seen": 3929720,
+      "step": 63
+    },
+    {
+      "epoch": 0.2096505823627288,
+      "loss": 1.6723523139953613,
+      "loss_ce": 0.13377803564071655,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.1337890625,
+      "loss_xval": 1.5390625,
+      "num_input_tokens_seen": 3929720,
+      "step": 63
+    },
+    {
+      "epoch": 0.2129783693843594,
+      "grad_norm": 28.632810592651367,
+      "learning_rate": 5e-06,
+      "loss": 1.8225,
+      "num_input_tokens_seen": 3993200,
+      "step": 64
+    },
+    {
+      "epoch": 0.2129783693843594,
+      "loss": 2.061736583709717,
+      "loss_ce": 0.2531428337097168,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.1328125,
+      "loss_xval": 1.8125,
+      "num_input_tokens_seen": 3993200,
+      "step": 64
+    },
+    {
+      "epoch": 0.21630615640599002,
+      "grad_norm": 26.3468074798584,
+      "learning_rate": 5e-06,
+      "loss": 1.5015,
+      "num_input_tokens_seen": 4055156,
+      "step": 65
+    },
+    {
+      "epoch": 0.21630615640599002,
+      "loss": 1.671527624130249,
+      "loss_ce": 0.03480884060263634,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.1396484375,
+      "loss_xval": 1.640625,
+      "num_input_tokens_seen": 4055156,
+      "step": 65
+    },
+    {
+      "epoch": 0.21963394342762063,
+      "grad_norm": 28.5969295501709,
+      "learning_rate": 5e-06,
+      "loss": 1.621,
+      "num_input_tokens_seen": 4117564,
+      "step": 66
+    },
+    {
+      "epoch": 0.21963394342762063,
+      "loss": 1.391996145248413,
+      "loss_ce": 0.03555082529783249,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.08642578125,
+      "loss_xval": 1.359375,
+      "num_input_tokens_seen": 4117564,
+      "step": 66
+    },
+    {
+      "epoch": 0.22296173044925124,
+      "grad_norm": 13.81509780883789,
+      "learning_rate": 5e-06,
+      "loss": 1.4978,
+      "num_input_tokens_seen": 4181136,
+      "step": 67
+    },
+    {
+      "epoch": 0.22296173044925124,
+      "loss": 1.482597827911377,
+      "loss_ce": 0.14080099761486053,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.09375,
+      "loss_xval": 1.34375,
+      "num_input_tokens_seen": 4181136,
+      "step": 67
+    },
+    {
+      "epoch": 0.22628951747088186,
+      "grad_norm": 22.297697067260742,
+      "learning_rate": 5e-06,
+      "loss": 1.5529,
+      "num_input_tokens_seen": 4244384,
+      "step": 68
+    },
+    {
+      "epoch": 0.22628951747088186,
+      "loss": 1.800987720489502,
+      "loss_ce": 0.0724719688296318,
+      "loss_iou": 0.5,
+      "loss_num": 0.1455078125,
+      "loss_xval": 1.7265625,
+      "num_input_tokens_seen": 4244384,
+      "step": 68
+    },
+    {
+      "epoch": 0.22961730449251247,
+      "grad_norm": 17.73756217956543,
+      "learning_rate": 5e-06,
+      "loss": 1.6361,
+      "num_input_tokens_seen": 4306480,
+      "step": 69
+    },
+    {
+      "epoch": 0.22961730449251247,
+      "loss": 1.7959282398223877,
+      "loss_ce": 0.25491267442703247,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.1376953125,
+      "loss_xval": 1.5390625,
+      "num_input_tokens_seen": 4306480,
+      "step": 69
+    },
+    {
+      "epoch": 0.23294509151414308,
+      "grad_norm": 28.785377502441406,
+      "learning_rate": 5e-06,
+      "loss": 1.6575,
+      "num_input_tokens_seen": 4368812,
+      "step": 70
+    },
+    {
+      "epoch": 0.23294509151414308,
+      "loss": 1.6283104419708252,
+      "loss_ce": 0.05555645376443863,
+      "loss_iou": 0.4375,
+      "loss_num": 0.1396484375,
+      "loss_xval": 1.5703125,
+      "num_input_tokens_seen": 4368812,
+      "step": 70
+    },
+    {
+      "epoch": 0.23627287853577372,
+      "grad_norm": 15.95328140258789,
+      "learning_rate": 5e-06,
+      "loss": 1.923,
+      "num_input_tokens_seen": 4432212,
+      "step": 71
+    },
+    {
+      "epoch": 0.23627287853577372,
+      "loss": 2.040701389312744,
+      "loss_ce": 0.4274199604988098,
+      "loss_iou": 0.53125,
+      "loss_num": 0.10888671875,
+      "loss_xval": 1.609375,
+      "num_input_tokens_seen": 4432212,
+      "step": 71
+    },
+    {
+      "epoch": 0.23960066555740434,
+      "grad_norm": 15.339340209960938,
+      "learning_rate": 5e-06,
+      "loss": 1.6562,
+      "num_input_tokens_seen": 4495800,
+      "step": 72
+    },
+    {
+      "epoch": 0.23960066555740434,
+      "loss": 1.5833992958068848,
+      "loss_ce": 0.07705167680978775,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.1416015625,
+      "loss_xval": 1.5078125,
+      "num_input_tokens_seen": 4495800,
+      "step": 72
+    },
+    {
+      "epoch": 0.24292845257903495,
+      "grad_norm": 64.54541778564453,
+      "learning_rate": 5e-06,
+      "loss": 1.9285,
+      "num_input_tokens_seen": 4558428,
+      "step": 73
+    },
+    {
+      "epoch": 0.24292845257903495,
+      "loss": 1.8927369117736816,
+      "loss_ce": 0.13980722427368164,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.12890625,
+      "loss_xval": 1.75,
+      "num_input_tokens_seen": 4558428,
+      "step": 73
+    },
+    {
+      "epoch": 0.24625623960066556,
+      "grad_norm": 32.903106689453125,
+      "learning_rate": 5e-06,
+      "loss": 2.1965,
+      "num_input_tokens_seen": 4622312,
+      "step": 74
+    },
+    {
+      "epoch": 0.24625623960066556,
+      "loss": 1.6801223754882812,
+      "loss_ce": 0.08515171706676483,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.12109375,
+      "loss_xval": 1.59375,
+      "num_input_tokens_seen": 4622312,
+      "step": 74
+    },
+    {
+      "epoch": 0.24958402662229617,
+      "grad_norm": 18.268884658813477,
+      "learning_rate": 5e-06,
+      "loss": 1.696,
+      "num_input_tokens_seen": 4685228,
+      "step": 75
+    },
+    {
+      "epoch": 0.24958402662229617,
+      "loss": 2.0499777793884277,
+      "loss_ce": 0.2775166630744934,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.1572265625,
+      "loss_xval": 1.7734375,
+      "num_input_tokens_seen": 4685228,
+      "step": 75
+    },
+    {
+      "epoch": 0.2529118136439268,
+      "grad_norm": 15.21312141418457,
+      "learning_rate": 5e-06,
+      "loss": 1.3746,
+      "num_input_tokens_seen": 4746040,
+      "step": 76
+    },
+    {
+      "epoch": 0.2529118136439268,
+      "loss": 1.2260797023773193,
+      "loss_ce": 0.06128474697470665,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.1337890625,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 4746040,
+      "step": 76
+    },
+    {
+      "epoch": 0.2562396006655574,
+      "grad_norm": 24.16434097290039,
+      "learning_rate": 5e-06,
+      "loss": 1.6829,
+      "num_input_tokens_seen": 4809160,
+      "step": 77
+    },
+    {
+      "epoch": 0.2562396006655574,
+      "loss": 1.6074845790863037,
+      "loss_ce": 0.13531659543514252,
+      "loss_iou": 0.46875,
+      "loss_num": 0.1064453125,
+      "loss_xval": 1.46875,
+      "num_input_tokens_seen": 4809160,
+      "step": 77
+    },
+    {
+      "epoch": 0.259567387687188,
+      "grad_norm": 36.885498046875,
+      "learning_rate": 5e-06,
+      "loss": 1.6743,
+      "num_input_tokens_seen": 4872948,
+      "step": 78
+    },
+    {
+      "epoch": 0.259567387687188,
+      "loss": 1.5818986892700195,
+      "loss_ce": 0.03892991691827774,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.130859375,
+      "loss_xval": 1.546875,
+      "num_input_tokens_seen": 4872948,
+      "step": 78
+    },
+    {
+      "epoch": 0.2628951747088186,
+      "grad_norm": 12.761034965515137,
+      "learning_rate": 5e-06,
+      "loss": 1.3517,
+      "num_input_tokens_seen": 4934300,
+      "step": 79
+    },
+    {
+      "epoch": 0.2628951747088186,
+      "loss": 1.0315775871276855,
+      "loss_ce": 0.014487742446362972,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0888671875,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 4934300,
+      "step": 79
+    },
+    {
+      "epoch": 0.26622296173044924,
+      "grad_norm": 13.58120059967041,
+      "learning_rate": 5e-06,
+      "loss": 1.6854,
+      "num_input_tokens_seen": 4997784,
+      "step": 80
+    },
+    {
+      "epoch": 0.26622296173044924,
+      "loss": 1.634024739265442,
+      "loss_ce": 0.04027477279305458,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.111328125,
+      "loss_xval": 1.59375,
+      "num_input_tokens_seen": 4997784,
+      "step": 80
+    },
+    {
+      "epoch": 0.26955074875207985,
+      "grad_norm": 24.133459091186523,
+      "learning_rate": 5e-06,
+      "loss": 1.4397,
+      "num_input_tokens_seen": 5061392,
+      "step": 81
+    },
+    {
+      "epoch": 0.26955074875207985,
+      "loss": 1.667598009109497,
+      "loss_ce": 0.2418166995048523,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.1318359375,
+      "loss_xval": 1.421875,
+      "num_input_tokens_seen": 5061392,
+      "step": 81
+    },
+    {
+      "epoch": 0.27287853577371046,
+      "grad_norm": 15.399471282958984,
+      "learning_rate": 5e-06,
+      "loss": 1.3346,
+      "num_input_tokens_seen": 5124660,
+      "step": 82
+    },
+    {
+      "epoch": 0.27287853577371046,
+      "loss": 1.3418378829956055,
+      "loss_ce": 0.09183788299560547,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.08984375,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 5124660,
+      "step": 82
+    },
+    {
+      "epoch": 0.2762063227953411,
+      "grad_norm": 19.83667755126953,
+      "learning_rate": 5e-06,
+      "loss": 1.6332,
+      "num_input_tokens_seen": 5186724,
+      "step": 83
+    },
+    {
+      "epoch": 0.2762063227953411,
+      "loss": 1.8349742889404297,
+      "loss_ce": 0.2241344451904297,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.1435546875,
+      "loss_xval": 1.609375,
+      "num_input_tokens_seen": 5186724,
+      "step": 83
+    },
+    {
+      "epoch": 0.2795341098169717,
+      "grad_norm": 15.304369926452637,
+      "learning_rate": 5e-06,
+      "loss": 1.4917,
+      "num_input_tokens_seen": 5248960,
+      "step": 84
+    },
+    {
+      "epoch": 0.2795341098169717,
+      "loss": 1.3170514106750488,
+      "loss_ce": 0.10513728857040405,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.11669921875,
+      "loss_xval": 1.2109375,
+      "num_input_tokens_seen": 5248960,
+      "step": 84
+    },
+    {
+      "epoch": 0.28286189683860236,
+      "grad_norm": 30.498498916625977,
+      "learning_rate": 5e-06,
+      "loss": 1.7189,
+      "num_input_tokens_seen": 5312544,
+      "step": 85
+    },
+    {
+      "epoch": 0.28286189683860236,
+      "loss": 1.7351880073547363,
+      "loss_ce": 0.04475831985473633,
+      "loss_iou": 0.5,
+      "loss_num": 0.1376953125,
+      "loss_xval": 1.6875,
+      "num_input_tokens_seen": 5312544,
+      "step": 85
+    },
+    {
+      "epoch": 0.28618968386023297,
+      "grad_norm": 30.64577865600586,
+      "learning_rate": 5e-06,
+      "loss": 1.959,
+      "num_input_tokens_seen": 5375812,
+      "step": 86
+    },
+    {
+      "epoch": 0.28618968386023297,
+      "loss": 2.172987461090088,
+      "loss_ce": 0.3106827139854431,
+      "loss_iou": 0.578125,
+      "loss_num": 0.140625,
+      "loss_xval": 1.859375,
+      "num_input_tokens_seen": 5375812,
+      "step": 86
+    },
+    {
+      "epoch": 0.2895174708818636,
+      "grad_norm": 23.82610321044922,
+      "learning_rate": 5e-06,
+      "loss": 1.6742,
+      "num_input_tokens_seen": 5438272,
+      "step": 87
+    },
+    {
+      "epoch": 0.2895174708818636,
+      "loss": 1.6248981952667236,
+      "loss_ce": 0.05556230992078781,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.1572265625,
+      "loss_xval": 1.5703125,
+      "num_input_tokens_seen": 5438272,
+      "step": 87
+    },
+    {
+      "epoch": 0.2928452579034942,
+      "grad_norm": 14.44840145111084,
+      "learning_rate": 5e-06,
+      "loss": 1.7714,
+      "num_input_tokens_seen": 5498964,
+      "step": 88
+    },
+    {
+      "epoch": 0.2928452579034942,
+      "loss": 1.6302241086959839,
+      "loss_ce": 0.05844667926430702,
+      "loss_iou": 0.359375,
+      "loss_num": 0.1708984375,
+      "loss_xval": 1.5703125,
+      "num_input_tokens_seen": 5498964,
+      "step": 88
+    },
+    {
+      "epoch": 0.2961730449251248,
+      "grad_norm": 26.83960723876953,
+      "learning_rate": 5e-06,
+      "loss": 1.6978,
+      "num_input_tokens_seen": 5561644,
+      "step": 89
+    },
+    {
+      "epoch": 0.2961730449251248,
+      "loss": 1.8088901042938232,
+      "loss_ce": 0.06719096004962921,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.15234375,
+      "loss_xval": 1.7421875,
+      "num_input_tokens_seen": 5561644,
+      "step": 89
+    },
+    {
+      "epoch": 0.2995008319467554,
+      "grad_norm": 11.925620079040527,
+      "learning_rate": 5e-06,
+      "loss": 1.6314,
+      "num_input_tokens_seen": 5624348,
+      "step": 90
+    },
+    {
+      "epoch": 0.2995008319467554,
+      "loss": 1.4735398292541504,
+      "loss_ce": 0.1751999855041504,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.09619140625,
+      "loss_xval": 1.296875,
+      "num_input_tokens_seen": 5624348,
+      "step": 90
+    },
+    {
+      "epoch": 0.30282861896838603,
+      "grad_norm": 11.06351089477539,
+      "learning_rate": 5e-06,
+      "loss": 1.2644,
+      "num_input_tokens_seen": 5686716,
+      "step": 91
+    },
+    {
+      "epoch": 0.30282861896838603,
+      "loss": 1.3133293390274048,
+      "loss_ce": 0.03647388890385628,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.10888671875,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 5686716,
+      "step": 91
+    },
+    {
+      "epoch": 0.30615640599001664,
+      "grad_norm": 18.150497436523438,
+      "learning_rate": 5e-06,
+      "loss": 1.4589,
+      "num_input_tokens_seen": 5750560,
+      "step": 92
+    },
+    {
+      "epoch": 0.30615640599001664,
+      "loss": 1.482915997505188,
+      "loss_ce": 0.026373039931058884,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.09326171875,
+      "loss_xval": 1.453125,
+      "num_input_tokens_seen": 5750560,
+      "step": 92
+    },
+    {
+      "epoch": 0.30948419301164726,
+      "grad_norm": 31.141006469726562,
+      "learning_rate": 5e-06,
+      "loss": 1.662,
+      "num_input_tokens_seen": 5813996,
+      "step": 93
+    },
+    {
+      "epoch": 0.30948419301164726,
+      "loss": 1.7749927043914795,
+      "loss_ce": 0.032805174589157104,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.119140625,
+      "loss_xval": 1.7421875,
+      "num_input_tokens_seen": 5813996,
+      "step": 93
+    },
+    {
+      "epoch": 0.31281198003327787,
+      "grad_norm": 20.244047164916992,
+      "learning_rate": 5e-06,
+      "loss": 1.3877,
+      "num_input_tokens_seen": 5876956,
+      "step": 94
+    },
+    {
+      "epoch": 0.31281198003327787,
+      "loss": 1.3969483375549316,
+      "loss_ce": 0.058081258088350296,
+      "loss_iou": 0.421875,
+      "loss_num": 0.09912109375,
+      "loss_xval": 1.3359375,
+      "num_input_tokens_seen": 5876956,
+      "step": 94
+    },
+    {
+      "epoch": 0.3161397670549085,
+      "grad_norm": 10.707723617553711,
+      "learning_rate": 5e-06,
+      "loss": 1.552,
+      "num_input_tokens_seen": 5939564,
+      "step": 95
+    },
+    {
+      "epoch": 0.3161397670549085,
+      "loss": 1.7702038288116455,
+      "loss_ce": 0.031434379518032074,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.1328125,
+      "loss_xval": 1.7421875,
+      "num_input_tokens_seen": 5939564,
+      "step": 95
+    },
+    {
+      "epoch": 0.3194675540765391,
+      "grad_norm": 15.120357513427734,
+      "learning_rate": 5e-06,
+      "loss": 1.3944,
+      "num_input_tokens_seen": 6000348,
+      "step": 96
+    },
+    {
+      "epoch": 0.3194675540765391,
+      "loss": 1.3216090202331543,
+      "loss_ce": 0.1492457389831543,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.130859375,
+      "loss_xval": 1.171875,
+      "num_input_tokens_seen": 6000348,
+      "step": 96
+    },
+    {
+      "epoch": 0.3227953410981697,
+      "grad_norm": 27.774568557739258,
+      "learning_rate": 5e-06,
+      "loss": 1.6287,
+      "num_input_tokens_seen": 6063144,
+      "step": 97
+    },
+    {
+      "epoch": 0.3227953410981697,
+      "loss": 1.5362926721572876,
+      "loss_ce": 0.0611950121819973,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.134765625,
+      "loss_xval": 1.4765625,
+      "num_input_tokens_seen": 6063144,
+      "step": 97
+    },
+    {
+      "epoch": 0.3261231281198003,
+      "grad_norm": 23.571321487426758,
+      "learning_rate": 5e-06,
+      "loss": 1.501,
+      "num_input_tokens_seen": 6125216,
+      "step": 98
+    },
+    {
+      "epoch": 0.3261231281198003,
+      "loss": 1.4960654973983765,
+      "loss_ce": 0.02975688874721527,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.1328125,
+      "loss_xval": 1.46875,
+      "num_input_tokens_seen": 6125216,
+      "step": 98
+    },
+    {
+      "epoch": 0.32945091514143093,
+      "grad_norm": 21.455690383911133,
+      "learning_rate": 5e-06,
+      "loss": 1.2035,
+      "num_input_tokens_seen": 6186436,
+      "step": 99
+    },
+    {
+      "epoch": 0.32945091514143093,
+      "loss": 1.533522367477417,
+      "loss_ce": 0.024733252823352814,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.10888671875,
+      "loss_xval": 1.5078125,
+      "num_input_tokens_seen": 6186436,
+      "step": 99
+    },
+    {
+      "epoch": 0.33277870216306155,
+      "grad_norm": 19.567928314208984,
+      "learning_rate": 5e-06,
+      "loss": 1.6169,
+      "num_input_tokens_seen": 6248292,
+      "step": 100
+    },
+    {
+      "epoch": 0.33277870216306155,
+      "loss": 1.5605666637420654,
+      "loss_ce": 0.15138691663742065,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.1552734375,
+      "loss_xval": 1.40625,
+      "num_input_tokens_seen": 6248292,
+      "step": 100
+    },
+    {
+      "epoch": 0.33610648918469216,
+      "grad_norm": 132.5515899658203,
+      "learning_rate": 5e-06,
+      "loss": 1.564,
+      "num_input_tokens_seen": 6312152,
+      "step": 101
+    },
+    {
+      "epoch": 0.33610648918469216,
+      "loss": 1.7496979236602783,
+      "loss_ce": 0.02167057991027832,
+      "loss_iou": 0.484375,
+      "loss_num": 0.1513671875,
+      "loss_xval": 1.7265625,
+      "num_input_tokens_seen": 6312152,
+      "step": 101
+    },
+    {
+      "epoch": 0.33943427620632277,
+      "grad_norm": 20.127302169799805,
+      "learning_rate": 5e-06,
+      "loss": 1.6141,
+      "num_input_tokens_seen": 6374832,
+      "step": 102
+    },
+    {
+      "epoch": 0.33943427620632277,
+      "loss": 1.5366733074188232,
+      "loss_ce": 0.15434911847114563,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.09912109375,
+      "loss_xval": 1.3828125,
+      "num_input_tokens_seen": 6374832,
+      "step": 102
+    },
+    {
+      "epoch": 0.3427620632279534,
+      "grad_norm": 26.46497917175293,
+      "learning_rate": 5e-06,
+      "loss": 1.7083,
+      "num_input_tokens_seen": 6439248,
+      "step": 103
+    },
+    {
+      "epoch": 0.3427620632279534,
+      "loss": 1.5998561382293701,
+      "loss_ce": 0.11841069906949997,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.11767578125,
+      "loss_xval": 1.484375,
+      "num_input_tokens_seen": 6439248,
+      "step": 103
+    },
+    {
+      "epoch": 0.34608985024958405,
+      "grad_norm": 22.090496063232422,
+      "learning_rate": 5e-06,
+      "loss": 1.5894,
+      "num_input_tokens_seen": 6501596,
+      "step": 104
+    },
+    {
+      "epoch": 0.34608985024958405,
+      "loss": 1.4435570240020752,
+      "loss_ce": 0.05586168169975281,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.09912109375,
+      "loss_xval": 1.390625,
+      "num_input_tokens_seen": 6501596,
+      "step": 104
+    },
+    {
+      "epoch": 0.34941763727121466,
+      "grad_norm": 14.779273986816406,
+      "learning_rate": 5e-06,
+      "loss": 1.5203,
+      "num_input_tokens_seen": 6562808,
+      "step": 105
+    },
+    {
+      "epoch": 0.34941763727121466,
+      "loss": 1.5112072229385376,
+      "loss_ce": 0.05271116644144058,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.1416015625,
+      "loss_xval": 1.4609375,
+      "num_input_tokens_seen": 6562808,
+      "step": 105
+    },
+    {
+      "epoch": 0.3527454242928453,
+      "grad_norm": 21.532625198364258,
+      "learning_rate": 5e-06,
+      "loss": 1.6474,
+      "num_input_tokens_seen": 6626268,
+      "step": 106
+    },
+    {
+      "epoch": 0.3527454242928453,
+      "loss": 1.6102540493011475,
+      "loss_ce": 0.03603534772992134,
+      "loss_iou": 0.5,
+      "loss_num": 0.115234375,
+      "loss_xval": 1.578125,
+      "num_input_tokens_seen": 6626268,
+      "step": 106
+    },
+    {
+      "epoch": 0.3560732113144759,
+      "grad_norm": 20.450246810913086,
+      "learning_rate": 5e-06,
+      "loss": 1.6417,
+      "num_input_tokens_seen": 6688448,
+      "step": 107
+    },
+    {
+      "epoch": 0.3560732113144759,
+      "loss": 1.5604519844055176,
+      "loss_ce": 0.10928011685609818,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.11083984375,
+      "loss_xval": 1.453125,
+      "num_input_tokens_seen": 6688448,
+      "step": 107
+    },
+    {
+      "epoch": 0.3594009983361065,
+      "grad_norm": 11.704852104187012,
+      "learning_rate": 5e-06,
+      "loss": 1.4292,
+      "num_input_tokens_seen": 6751076,
+      "step": 108
+    },
+    {
+      "epoch": 0.3594009983361065,
+      "loss": 1.302997350692749,
+      "loss_ce": 0.06276305764913559,
+      "loss_iou": 0.375,
+      "loss_num": 0.0986328125,
+      "loss_xval": 1.2421875,
+      "num_input_tokens_seen": 6751076,
+      "step": 108
+    },
+    {
+      "epoch": 0.3627287853577371,
+      "grad_norm": 26.655656814575195,
+      "learning_rate": 5e-06,
+      "loss": 1.5021,
+      "num_input_tokens_seen": 6814604,
+      "step": 109
+    },
+    {
+      "epoch": 0.3627287853577371,
+      "loss": 1.4590511322021484,
+      "loss_ce": 0.03766445070505142,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.126953125,
+      "loss_xval": 1.421875,
+      "num_input_tokens_seen": 6814604,
+      "step": 109
+    },
+    {
+      "epoch": 0.36605657237936773,
+      "grad_norm": 32.32645797729492,
+      "learning_rate": 5e-06,
+      "loss": 1.293,
+      "num_input_tokens_seen": 6877660,
+      "step": 110
+    },
+    {
+      "epoch": 0.36605657237936773,
+      "loss": 1.2947750091552734,
+      "loss_ce": 0.05161091312766075,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.1025390625,
+      "loss_xval": 1.2421875,
+      "num_input_tokens_seen": 6877660,
+      "step": 110
+    },
+    {
+      "epoch": 0.36938435940099834,
+      "grad_norm": 19.443389892578125,
+      "learning_rate": 5e-06,
+      "loss": 1.4704,
+      "num_input_tokens_seen": 6940376,
+      "step": 111
+    },
+    {
+      "epoch": 0.36938435940099834,
+      "loss": 1.6598196029663086,
+      "loss_ce": 0.016264785081148148,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.15234375,
+      "loss_xval": 1.640625,
+      "num_input_tokens_seen": 6940376,
+      "step": 111
+    },
+    {
+      "epoch": 0.37271214642262895,
+      "grad_norm": 17.155519485473633,
+      "learning_rate": 5e-06,
+      "loss": 1.468,
+      "num_input_tokens_seen": 7002968,
+      "step": 112
+    },
+    {
+      "epoch": 0.37271214642262895,
+      "loss": 1.4055347442626953,
+      "loss_ce": 0.047136228531599045,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.1416015625,
+      "loss_xval": 1.359375,
+      "num_input_tokens_seen": 7002968,
+      "step": 112
+    },
+    {
+      "epoch": 0.37603993344425957,
+      "grad_norm": 233.18862915039062,
+      "learning_rate": 5e-06,
+      "loss": 1.7817,
+      "num_input_tokens_seen": 7067164,
+      "step": 113
+    },
+    {
+      "epoch": 0.37603993344425957,
+      "loss": 1.9062466621398926,
+      "loss_ce": 0.01269207801669836,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.142578125,
+      "loss_xval": 1.890625,
+      "num_input_tokens_seen": 7067164,
+      "step": 113
+    },
+    {
+      "epoch": 0.3793677204658902,
+      "grad_norm": 35.83794403076172,
+      "learning_rate": 5e-06,
+      "loss": 1.7392,
+      "num_input_tokens_seen": 7130484,
+      "step": 114
+    },
+    {
+      "epoch": 0.3793677204658902,
+      "loss": 1.760888934135437,
+      "loss_ce": 0.05971705913543701,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.1416015625,
+      "loss_xval": 1.703125,
+      "num_input_tokens_seen": 7130484,
+      "step": 114
+    },
+    {
+      "epoch": 0.3826955074875208,
+      "grad_norm": 17.227394104003906,
+      "learning_rate": 5e-06,
+      "loss": 1.4545,
+      "num_input_tokens_seen": 7192824,
+      "step": 115
+    },
+    {
+      "epoch": 0.3826955074875208,
+      "loss": 1.311098337173462,
+      "loss_ce": 0.025453729555010796,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.10498046875,
+      "loss_xval": 1.2890625,
+      "num_input_tokens_seen": 7192824,
+      "step": 115
+    },
+    {
+      "epoch": 0.3860232945091514,
+      "grad_norm": 20.637577056884766,
+      "learning_rate": 5e-06,
+      "loss": 1.4128,
+      "num_input_tokens_seen": 7254948,
+      "step": 116
+    },
+    {
+      "epoch": 0.3860232945091514,
+      "loss": 1.4620569944381714,
+      "loss_ce": 0.043355897068977356,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.1328125,
+      "loss_xval": 1.421875,
+      "num_input_tokens_seen": 7254948,
+      "step": 116
+    },
+    {
+      "epoch": 0.389351081530782,
+      "grad_norm": 32.69279479980469,
+      "learning_rate": 5e-06,
+      "loss": 1.7562,
+      "num_input_tokens_seen": 7318796,
+      "step": 117
+    },
+    {
+      "epoch": 0.389351081530782,
+      "loss": 1.807417392730713,
+      "loss_ce": 0.037886105477809906,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.15625,
+      "loss_xval": 1.765625,
+      "num_input_tokens_seen": 7318796,
+      "step": 117
+    },
+    {
+      "epoch": 0.39267886855241263,
+      "grad_norm": 11.888574600219727,
+      "learning_rate": 5e-06,
+      "loss": 1.3916,
+      "num_input_tokens_seen": 7379464,
+      "step": 118
+    },
+    {
+      "epoch": 0.39267886855241263,
+      "loss": 1.490173101425171,
+      "loss_ce": 0.04559309780597687,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.12353515625,
+      "loss_xval": 1.4453125,
+      "num_input_tokens_seen": 7379464,
+      "step": 118
+    },
+    {
+      "epoch": 0.39600665557404324,
+      "grad_norm": 21.6405029296875,
+      "learning_rate": 5e-06,
+      "loss": 1.2959,
+      "num_input_tokens_seen": 7441932,
+      "step": 119
+    },
+    {
+      "epoch": 0.39600665557404324,
+      "loss": 1.2409827709197998,
+      "loss_ce": 0.017350006848573685,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.12353515625,
+      "loss_xval": 1.2265625,
+      "num_input_tokens_seen": 7441932,
+      "step": 119
+    },
+    {
+      "epoch": 0.39933444259567386,
+      "grad_norm": 196.78163146972656,
+      "learning_rate": 5e-06,
+      "loss": 1.7212,
+      "num_input_tokens_seen": 7505232,
+      "step": 120
+    },
+    {
+      "epoch": 0.39933444259567386,
+      "loss": 1.5443594455718994,
+      "loss_ce": 0.028734492138028145,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.107421875,
+      "loss_xval": 1.515625,
+      "num_input_tokens_seen": 7505232,
+      "step": 120
+    },
+    {
+      "epoch": 0.40266222961730447,
+      "grad_norm": 118.65660858154297,
+      "learning_rate": 5e-06,
+      "loss": 1.3824,
+      "num_input_tokens_seen": 7568056,
+      "step": 121
+    },
+    {
+      "epoch": 0.40266222961730447,
+      "loss": 1.408996343612671,
+      "loss_ce": 0.02081269398331642,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.130859375,
+      "loss_xval": 1.390625,
+      "num_input_tokens_seen": 7568056,
+      "step": 121
+    },
+    {
+      "epoch": 0.4059900166389351,
+      "grad_norm": 19.64975929260254,
+      "learning_rate": 5e-06,
+      "loss": 1.5991,
+      "num_input_tokens_seen": 7632352,
+      "step": 122
+    },
+    {
+      "epoch": 0.4059900166389351,
+      "loss": 1.42240571975708,
+      "loss_ce": 0.056683123111724854,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0869140625,
+      "loss_xval": 1.3671875,
+      "num_input_tokens_seen": 7632352,
+      "step": 122
+    },
+    {
+      "epoch": 0.40931780366056575,
+      "grad_norm": 47.080657958984375,
+      "learning_rate": 5e-06,
+      "loss": 1.4264,
+      "num_input_tokens_seen": 7695112,
+      "step": 123
+    },
+    {
+      "epoch": 0.40931780366056575,
+      "loss": 1.4007999897003174,
+      "loss_ce": 0.023846877738833427,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.10546875,
+      "loss_xval": 1.375,
+      "num_input_tokens_seen": 7695112,
+      "step": 123
+    },
+    {
+      "epoch": 0.41264559068219636,
+      "grad_norm": 16.035123825073242,
+      "learning_rate": 5e-06,
+      "loss": 1.5467,
+      "num_input_tokens_seen": 7758548,
+      "step": 124
+    },
+    {
+      "epoch": 0.41264559068219636,
+      "loss": 1.532633900642395,
+      "loss_ce": 0.059977658092975616,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.115234375,
+      "loss_xval": 1.46875,
+      "num_input_tokens_seen": 7758548,
+      "step": 124
+    },
+    {
+      "epoch": 0.415973377703827,
+      "grad_norm": 15.68152904510498,
+      "learning_rate": 5e-06,
+      "loss": 1.406,
+      "num_input_tokens_seen": 7818644,
+      "step": 125
+    },
+    {
+      "epoch": 0.415973377703827,
+      "loss": 1.4248697757720947,
+      "loss_ce": 0.009464464150369167,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.1279296875,
+      "loss_xval": 1.4140625,
+      "num_input_tokens_seen": 7818644,
+      "step": 125
+    },
+    {
+      "epoch": 0.4193011647254576,
+      "grad_norm": 23.207124710083008,
+      "learning_rate": 5e-06,
+      "loss": 1.5815,
+      "num_input_tokens_seen": 7881772,
+      "step": 126
+    },
+    {
+      "epoch": 0.4193011647254576,
+      "loss": 1.6510601043701172,
+      "loss_ce": 0.010435021482408047,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.14453125,
+      "loss_xval": 1.640625,
+      "num_input_tokens_seen": 7881772,
+      "step": 126
+    },
+    {
+      "epoch": 0.4226289517470882,
+      "grad_norm": 17.7788028717041,
+      "learning_rate": 5e-06,
+      "loss": 1.3724,
+      "num_input_tokens_seen": 7943924,
+      "step": 127
+    },
+    {
+      "epoch": 0.4226289517470882,
+      "loss": 1.4064964056015015,
+      "loss_ce": 0.027834344655275345,
+      "loss_iou": 0.359375,
+      "loss_num": 0.1318359375,
+      "loss_xval": 1.375,
+      "num_input_tokens_seen": 7943924,
+      "step": 127
+    },
+    {
+      "epoch": 0.4259567387687188,
+      "grad_norm": 22.62128448486328,
+      "learning_rate": 5e-06,
+      "loss": 1.6814,
+      "num_input_tokens_seen": 8007756,
+      "step": 128
+    },
+    {
+      "epoch": 0.4259567387687188,
+      "loss": 1.7109909057617188,
+      "loss_ce": 0.05376438423991203,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.1455078125,
+      "loss_xval": 1.65625,
+      "num_input_tokens_seen": 8007756,
+      "step": 128
+    },
+    {
+      "epoch": 0.4292845257903494,
+      "grad_norm": 17.13036346435547,
+      "learning_rate": 5e-06,
+      "loss": 1.5,
+      "num_input_tokens_seen": 8071624,
+      "step": 129
+    },
+    {
+      "epoch": 0.4292845257903494,
+      "loss": 1.517619252204895,
+      "loss_ce": 0.04252162203192711,
+      "loss_iou": 0.46875,
+      "loss_num": 0.10693359375,
+      "loss_xval": 1.4765625,
+      "num_input_tokens_seen": 8071624,
+      "step": 129
+    },
+    {
+      "epoch": 0.43261231281198004,
+      "grad_norm": 28.38715934753418,
+      "learning_rate": 5e-06,
+      "loss": 1.6712,
+      "num_input_tokens_seen": 8135204,
+      "step": 130
+    },
+    {
+      "epoch": 0.43261231281198004,
+      "loss": 1.6535531282424927,
+      "loss_ce": 0.016834398731589317,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.123046875,
+      "loss_xval": 1.640625,
+      "num_input_tokens_seen": 8135204,
+      "step": 130
+    },
+    {
+      "epoch": 0.43594009983361065,
+      "grad_norm": 21.765769958496094,
+      "learning_rate": 5e-06,
+      "loss": 1.2592,
+      "num_input_tokens_seen": 8198328,
+      "step": 131
+    },
+    {
+      "epoch": 0.43594009983361065,
+      "loss": 1.0689215660095215,
+      "loss_ce": 0.04450753331184387,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.08203125,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 8198328,
+      "step": 131
+    },
+    {
+      "epoch": 0.43926788685524126,
+      "grad_norm": 12.410709381103516,
+      "learning_rate": 5e-06,
+      "loss": 1.2945,
+      "num_input_tokens_seen": 8260932,
+      "step": 132
+    },
+    {
+      "epoch": 0.43926788685524126,
+      "loss": 1.2807693481445312,
+      "loss_ce": 0.014656049199402332,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.12353515625,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 8260932,
+      "step": 132
+    },
+    {
+      "epoch": 0.4425956738768719,
+      "grad_norm": 24.125675201416016,
+      "learning_rate": 5e-06,
+      "loss": 1.4355,
+      "num_input_tokens_seen": 8324024,
+      "step": 133
+    },
+    {
+      "epoch": 0.4425956738768719,
+      "loss": 1.2214634418487549,
+      "loss_ce": 0.035916537046432495,
+      "loss_iou": 0.28125,
+      "loss_num": 0.12451171875,
+      "loss_xval": 1.1875,
+      "num_input_tokens_seen": 8324024,
+      "step": 133
+    },
+    {
+      "epoch": 0.4459234608985025,
+      "grad_norm": 249.4659423828125,
+      "learning_rate": 5e-06,
+      "loss": 1.2783,
+      "num_input_tokens_seen": 8386732,
+      "step": 134
+    },
+    {
+      "epoch": 0.4459234608985025,
+      "loss": 1.165820837020874,
+      "loss_ce": 0.022754406556487083,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.08935546875,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 8386732,
+      "step": 134
+    },
+    {
+      "epoch": 0.4492512479201331,
+      "grad_norm": 15.685258865356445,
+      "learning_rate": 5e-06,
+      "loss": 1.6974,
+      "num_input_tokens_seen": 8449704,
+      "step": 135
+    },
+    {
+      "epoch": 0.4492512479201331,
+      "loss": 1.7366528511047363,
+      "loss_ce": 0.01448477990925312,
+      "loss_iou": 0.515625,
+      "loss_num": 0.1376953125,
+      "loss_xval": 1.71875,
+      "num_input_tokens_seen": 8449704,
+      "step": 135
+    },
+    {
+      "epoch": 0.4525790349417637,
+      "grad_norm": 15.560267448425293,
+      "learning_rate": 5e-06,
+      "loss": 1.3367,
+      "num_input_tokens_seen": 8511980,
+      "step": 136
+    },
+    {
+      "epoch": 0.4525790349417637,
+      "loss": 1.5319541692733765,
+      "loss_ce": 0.06418071687221527,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.091796875,
+      "loss_xval": 1.46875,
+      "num_input_tokens_seen": 8511980,
+      "step": 136
+    },
+    {
+      "epoch": 0.4559068219633943,
+      "grad_norm": 31.106922149658203,
+      "learning_rate": 5e-06,
+      "loss": 1.3191,
+      "num_input_tokens_seen": 8575108,
+      "step": 137
+    },
+    {
+      "epoch": 0.4559068219633943,
+      "loss": 1.1032752990722656,
+      "loss_ce": 0.01050189882516861,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.10888671875,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 8575108,
+      "step": 137
+    },
+    {
+      "epoch": 0.45923460898502494,
+      "grad_norm": 19.03159523010254,
+      "learning_rate": 5e-06,
+      "loss": 1.51,
+      "num_input_tokens_seen": 8637260,
+      "step": 138
+    },
+    {
+      "epoch": 0.45923460898502494,
+      "loss": 1.4713654518127441,
+      "loss_ce": 0.05046709254384041,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.1171875,
+      "loss_xval": 1.421875,
+      "num_input_tokens_seen": 8637260,
+      "step": 138
+    },
+    {
+      "epoch": 0.46256239600665555,
+      "grad_norm": 18.39104652404785,
+      "learning_rate": 5e-06,
+      "loss": 1.5178,
+      "num_input_tokens_seen": 8699912,
+      "step": 139
+    },
+    {
+      "epoch": 0.46256239600665555,
+      "loss": 1.5592514276504517,
+      "loss_ce": 0.04411466047167778,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.1044921875,
+      "loss_xval": 1.515625,
+      "num_input_tokens_seen": 8699912,
+      "step": 139
+    },
+    {
+      "epoch": 0.46589018302828616,
+      "grad_norm": 16.95204734802246,
+      "learning_rate": 5e-06,
+      "loss": 1.4614,
+      "num_input_tokens_seen": 8763456,
+      "step": 140
+    },
+    {
+      "epoch": 0.46589018302828616,
+      "loss": 1.5307507514953613,
+      "loss_ce": 0.026844505220651627,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.1328125,
+      "loss_xval": 1.5,
+      "num_input_tokens_seen": 8763456,
+      "step": 140
+    },
+    {
+      "epoch": 0.46921797004991683,
+      "grad_norm": 13.439885139465332,
+      "learning_rate": 5e-06,
+      "loss": 1.2669,
+      "num_input_tokens_seen": 8823008,
+      "step": 141
+    },
+    {
+      "epoch": 0.46921797004991683,
+      "loss": 1.1225981712341309,
+      "loss_ce": 0.03861372545361519,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.125,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 8823008,
+      "step": 141
+    },
+    {
+      "epoch": 0.47254575707154745,
+      "grad_norm": 13.823068618774414,
+      "learning_rate": 5e-06,
+      "loss": 1.3294,
+      "num_input_tokens_seen": 8885756,
+      "step": 142
+    },
+    {
+      "epoch": 0.47254575707154745,
+      "loss": 1.3653852939605713,
+      "loss_ce": 0.011625496670603752,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.1240234375,
+      "loss_xval": 1.3515625,
+      "num_input_tokens_seen": 8885756,
+      "step": 142
+    },
+    {
+      "epoch": 0.47587354409317806,
+      "grad_norm": 46.61298751831055,
+      "learning_rate": 5e-06,
+      "loss": 1.6217,
+      "num_input_tokens_seen": 8949676,
+      "step": 143
+    },
+    {
+      "epoch": 0.47587354409317806,
+      "loss": 1.6690086126327515,
+      "loss_ce": 0.08209459483623505,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.1396484375,
+      "loss_xval": 1.5859375,
+      "num_input_tokens_seen": 8949676,
+      "step": 143
+    },
+    {
+      "epoch": 0.47920133111480867,
+      "grad_norm": 12.167203903198242,
+      "learning_rate": 5e-06,
+      "loss": 1.1975,
+      "num_input_tokens_seen": 9012212,
+      "step": 144
+    },
+    {
+      "epoch": 0.47920133111480867,
+      "loss": 1.0818878412246704,
+      "loss_ce": 0.06528623402118683,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.08251953125,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 9012212,
+      "step": 144
+    },
+    {
+      "epoch": 0.4825291181364393,
+      "grad_norm": 19.003299713134766,
+      "learning_rate": 5e-06,
+      "loss": 1.3821,
+      "num_input_tokens_seen": 9075056,
+      "step": 145
+    },
+    {
+      "epoch": 0.4825291181364393,
+      "loss": 1.3460979461669922,
+      "loss_ce": 0.01919359713792801,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.103515625,
+      "loss_xval": 1.328125,
+      "num_input_tokens_seen": 9075056,
+      "step": 145
+    },
+    {
+      "epoch": 0.4858569051580699,
+      "grad_norm": 30.158510208129883,
+      "learning_rate": 5e-06,
+      "loss": 1.3432,
+      "num_input_tokens_seen": 9138200,
+      "step": 146
+    },
+    {
+      "epoch": 0.4858569051580699,
+      "loss": 1.3490657806396484,
+      "loss_ce": 0.0026301806792616844,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.09130859375,
+      "loss_xval": 1.34375,
+      "num_input_tokens_seen": 9138200,
+      "step": 146
+    },
+    {
+      "epoch": 0.4891846921797005,
+      "grad_norm": 13.28848648071289,
+      "learning_rate": 5e-06,
+      "loss": 1.2648,
+      "num_input_tokens_seen": 9201716,
+      "step": 147
+    },
+    {
+      "epoch": 0.4891846921797005,
+      "loss": 1.31808340549469,
+      "loss_ce": 0.007536512799561024,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.1015625,
+      "loss_xval": 1.3125,
+      "num_input_tokens_seen": 9201716,
+      "step": 147
+    },
+    {
+      "epoch": 0.4925124792013311,
+      "grad_norm": 22.02879524230957,
+      "learning_rate": 5e-06,
+      "loss": 1.2787,
+      "num_input_tokens_seen": 9264876,
+      "step": 148
+    },
+    {
+      "epoch": 0.4925124792013311,
+      "loss": 1.3372578620910645,
+      "loss_ce": 0.014503922313451767,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.10888671875,
+      "loss_xval": 1.3203125,
+      "num_input_tokens_seen": 9264876,
+      "step": 148
+    },
+    {
+      "epoch": 0.49584026622296173,
+      "grad_norm": 15.431622505187988,
+      "learning_rate": 5e-06,
+      "loss": 1.4477,
+      "num_input_tokens_seen": 9327456,
+      "step": 149
+    },
+    {
+      "epoch": 0.49584026622296173,
+      "loss": 1.287103295326233,
+      "loss_ce": 0.03173219412565231,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.09619140625,
+      "loss_xval": 1.2578125,
+      "num_input_tokens_seen": 9327456,
+      "step": 149
+    },
+    {
+      "epoch": 0.49916805324459235,
+      "grad_norm": 14.655500411987305,
+      "learning_rate": 5e-06,
+      "loss": 1.5951,
+      "num_input_tokens_seen": 9391828,
+      "step": 150
+    },
+    {
+      "epoch": 0.49916805324459235,
+      "loss": 1.5570836067199707,
+      "loss_ce": 0.03169288486242294,
+      "loss_iou": 0.46875,
+      "loss_num": 0.11767578125,
+      "loss_xval": 1.5234375,
+      "num_input_tokens_seen": 9391828,
+      "step": 150
+    },
+    {
+      "epoch": 0.502495840266223,
+      "grad_norm": 12.9921236038208,
+      "learning_rate": 5e-06,
+      "loss": 1.1374,
+      "num_input_tokens_seen": 9453992,
+      "step": 151
+    },
+    {
+      "epoch": 0.502495840266223,
+      "loss": 1.1639986038208008,
+      "loss_ce": 0.00481888884678483,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.12890625,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 9453992,
+      "step": 151
+    },
+    {
+      "epoch": 0.5058236272878536,
+      "grad_norm": 12.746012687683105,
+      "learning_rate": 5e-06,
+      "loss": 1.4932,
+      "num_input_tokens_seen": 9515028,
+      "step": 152
+    },
+    {
+      "epoch": 0.5058236272878536,
+      "loss": 1.3379508256912231,
+      "loss_ce": 0.004943011794239283,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0927734375,
+      "loss_xval": 1.3359375,
+      "num_input_tokens_seen": 9515028,
+      "step": 152
+    },
+    {
+      "epoch": 0.5091514143094842,
+      "grad_norm": 12.805912017822266,
+      "learning_rate": 5e-06,
+      "loss": 1.3026,
+      "num_input_tokens_seen": 9576904,
+      "step": 153
+    },
+    {
+      "epoch": 0.5091514143094842,
+      "loss": 1.1774930953979492,
+      "loss_ce": 0.010500917211174965,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.1162109375,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 9576904,
+      "step": 153
+    },
+    {
+      "epoch": 0.5124792013311148,
+      "grad_norm": 11.221924781799316,
+      "learning_rate": 5e-06,
+      "loss": 1.4352,
+      "num_input_tokens_seen": 9640896,
+      "step": 154
+    },
+    {
+      "epoch": 0.5124792013311148,
+      "loss": 1.4379901885986328,
+      "loss_ce": 0.014650269411504269,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.09521484375,
+      "loss_xval": 1.421875,
+      "num_input_tokens_seen": 9640896,
+      "step": 154
+    },
+    {
+      "epoch": 0.5158069883527454,
+      "grad_norm": 29.179161071777344,
+      "learning_rate": 5e-06,
+      "loss": 1.4006,
+      "num_input_tokens_seen": 9704416,
+      "step": 155
+    },
+    {
+      "epoch": 0.5158069883527454,
+      "loss": 1.2948763370513916,
+      "loss_ce": 0.02729811705648899,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.109375,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 9704416,
+      "step": 155
+    },
+    {
+      "epoch": 0.519134775374376,
+      "grad_norm": 30.093252182006836,
+      "learning_rate": 5e-06,
+      "loss": 1.5561,
+      "num_input_tokens_seen": 9767272,
+      "step": 156
+    },
+    {
+      "epoch": 0.519134775374376,
+      "loss": 1.4946575164794922,
+      "loss_ce": 0.012723930180072784,
+      "loss_iou": 0.46875,
+      "loss_num": 0.1083984375,
+      "loss_xval": 1.484375,
+      "num_input_tokens_seen": 9767272,
+      "step": 156
+    },
+    {
+      "epoch": 0.5224625623960066,
+      "grad_norm": 14.34788990020752,
+      "learning_rate": 5e-06,
+      "loss": 1.4241,
+      "num_input_tokens_seen": 9829412,
+      "step": 157
+    },
+    {
+      "epoch": 0.5224625623960066,
+      "loss": 1.4533286094665527,
+      "loss_ce": 0.002156792674213648,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.109375,
+      "loss_xval": 1.453125,
+      "num_input_tokens_seen": 9829412,
+      "step": 157
+    },
+    {
+      "epoch": 0.5257903494176372,
+      "grad_norm": 45.826786041259766,
+      "learning_rate": 5e-06,
+      "loss": 1.1328,
+      "num_input_tokens_seen": 9891184,
+      "step": 158
+    },
+    {
+      "epoch": 0.5257903494176372,
+      "loss": 1.2606561183929443,
+      "loss_ce": 0.008825048804283142,
+      "loss_iou": 0.4375,
+      "loss_num": 0.07568359375,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 9891184,
+      "step": 158
+    },
+    {
+      "epoch": 0.5291181364392679,
+      "grad_norm": 14.152484893798828,
+      "learning_rate": 5e-06,
+      "loss": 1.2858,
+      "num_input_tokens_seen": 9952284,
+      "step": 159
+    },
+    {
+      "epoch": 0.5291181364392679,
+      "loss": 1.2074310779571533,
+      "loss_ce": 0.024569693952798843,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.1044921875,
+      "loss_xval": 1.1796875,
+      "num_input_tokens_seen": 9952284,
+      "step": 159
+    },
+    {
+      "epoch": 0.5324459234608985,
+      "grad_norm": 37.07005310058594,
+      "learning_rate": 5e-06,
+      "loss": 1.5076,
+      "num_input_tokens_seen": 10014964,
+      "step": 160
+    },
+    {
+      "epoch": 0.5324459234608985,
+      "loss": 1.5249152183532715,
+      "loss_ce": 0.008313634432852268,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.1298828125,
+      "loss_xval": 1.515625,
+      "num_input_tokens_seen": 10014964,
+      "step": 160
+    },
+    {
+      "epoch": 0.5357737104825291,
+      "grad_norm": 17.514577865600586,
+      "learning_rate": 5e-06,
+      "loss": 1.2458,
+      "num_input_tokens_seen": 10078884,
+      "step": 161
+    },
+    {
+      "epoch": 0.5357737104825291,
+      "loss": 0.9992653727531433,
+      "loss_ce": 0.006589556112885475,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.08740234375,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 10078884,
+      "step": 161
+    },
+    {
+      "epoch": 0.5391014975041597,
+      "grad_norm": 19.246084213256836,
+      "learning_rate": 5e-06,
+      "loss": 1.3381,
+      "num_input_tokens_seen": 10141460,
+      "step": 162
+    },
+    {
+      "epoch": 0.5391014975041597,
+      "loss": 1.0187444686889648,
+      "loss_ce": 0.012152590788900852,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.10302734375,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 10141460,
+      "step": 162
+    },
+    {
+      "epoch": 0.5424292845257903,
+      "grad_norm": 14.021342277526855,
+      "learning_rate": 5e-06,
+      "loss": 1.2232,
+      "num_input_tokens_seen": 10203772,
+      "step": 163
+    },
+    {
+      "epoch": 0.5424292845257903,
+      "loss": 1.3046138286590576,
+      "loss_ce": 0.004809187725186348,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0986328125,
+      "loss_xval": 1.296875,
+      "num_input_tokens_seen": 10203772,
+      "step": 163
+    },
+    {
+      "epoch": 0.5457570715474209,
+      "grad_norm": 20.620309829711914,
+      "learning_rate": 5e-06,
+      "loss": 1.3921,
+      "num_input_tokens_seen": 10266640,
+      "step": 164
+    },
+    {
+      "epoch": 0.5457570715474209,
+      "loss": 1.3793249130249023,
+      "loss_ce": 0.012625731527805328,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.08154296875,
+      "loss_xval": 1.3671875,
+      "num_input_tokens_seen": 10266640,
+      "step": 164
+    },
+    {
+      "epoch": 0.5490848585690515,
+      "grad_norm": 27.303794860839844,
+      "learning_rate": 5e-06,
+      "loss": 1.3677,
+      "num_input_tokens_seen": 10330220,
+      "step": 165
+    },
+    {
+      "epoch": 0.5490848585690515,
+      "loss": 1.2398979663848877,
+      "loss_ce": 0.002593299839645624,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0810546875,
+      "loss_xval": 1.234375,
+      "num_input_tokens_seen": 10330220,
+      "step": 165
+    },
+    {
+      "epoch": 0.5524126455906821,
+      "grad_norm": 15.321734428405762,
+      "learning_rate": 5e-06,
+      "loss": 1.3985,
+      "num_input_tokens_seen": 10392312,
+      "step": 166
+    },
+    {
+      "epoch": 0.5524126455906821,
+      "loss": 1.1850658655166626,
+      "loss_ce": 0.030891068279743195,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.119140625,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 10392312,
+      "step": 166
+    },
+    {
+      "epoch": 0.5557404326123128,
+      "grad_norm": 48.359825134277344,
+      "learning_rate": 5e-06,
+      "loss": 1.1362,
+      "num_input_tokens_seen": 10455024,
+      "step": 167
+    },
+    {
+      "epoch": 0.5557404326123128,
+      "loss": 1.3550812005996704,
+      "loss_ce": 0.0035187224857509136,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.1025390625,
+      "loss_xval": 1.3515625,
+      "num_input_tokens_seen": 10455024,
+      "step": 167
+    },
+    {
+      "epoch": 0.5590682196339434,
+      "grad_norm": 31.070018768310547,
+      "learning_rate": 5e-06,
+      "loss": 1.0637,
+      "num_input_tokens_seen": 10516804,
+      "step": 168
+    },
+    {
+      "epoch": 0.5590682196339434,
+      "loss": 0.8580918312072754,
+      "loss_ce": 0.006285225972533226,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.099609375,
+      "loss_xval": 0.8515625,
+      "num_input_tokens_seen": 10516804,
+      "step": 168
+    },
+    {
+      "epoch": 0.562396006655574,
+      "grad_norm": 13.717340469360352,
+      "learning_rate": 5e-06,
+      "loss": 1.3105,
+      "num_input_tokens_seen": 10579816,
+      "step": 169
+    },
+    {
+      "epoch": 0.562396006655574,
+      "loss": 1.291137933731079,
+      "loss_ce": 0.017700420692563057,
+      "loss_iou": 0.40625,
+      "loss_num": 0.09228515625,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 10579816,
+      "step": 169
+    },
+    {
+      "epoch": 0.5657237936772047,
+      "grad_norm": 14.410500526428223,
+      "learning_rate": 5e-06,
+      "loss": 1.1444,
+      "num_input_tokens_seen": 10640804,
+      "step": 170
+    },
+    {
+      "epoch": 0.5657237936772047,
+      "loss": 1.0607528686523438,
+      "loss_ce": 0.012901253998279572,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.1064453125,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 10640804,
+      "step": 170
+    },
+    {
+      "epoch": 0.5690515806988353,
+      "grad_norm": 21.776578903198242,
+      "learning_rate": 5e-06,
+      "loss": 1.5343,
+      "num_input_tokens_seen": 10705984,
+      "step": 171
+    },
+    {
+      "epoch": 0.5690515806988353,
+      "loss": 1.5333166122436523,
+      "loss_ce": 0.013785396702587605,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0869140625,
+      "loss_xval": 1.515625,
+      "num_input_tokens_seen": 10705984,
+      "step": 171
+    },
+    {
+      "epoch": 0.5723793677204659,
+      "grad_norm": 20.108116149902344,
+      "learning_rate": 5e-06,
+      "loss": 1.4882,
+      "num_input_tokens_seen": 10768956,
+      "step": 172
+    },
+    {
+      "epoch": 0.5723793677204659,
+      "loss": 1.473107099533081,
+      "loss_ce": 0.015587646514177322,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0986328125,
+      "loss_xval": 1.4609375,
+      "num_input_tokens_seen": 10768956,
+      "step": 172
+    },
+    {
+      "epoch": 0.5757071547420965,
+      "grad_norm": 60.56779479980469,
+      "learning_rate": 5e-06,
+      "loss": 1.6353,
+      "num_input_tokens_seen": 10832704,
+      "step": 173
+    },
+    {
+      "epoch": 0.5757071547420965,
+      "loss": 1.7748465538024902,
+      "loss_ce": 0.0072684986516833305,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.14453125,
+      "loss_xval": 1.765625,
+      "num_input_tokens_seen": 10832704,
+      "step": 173
+    },
+    {
+      "epoch": 0.5790349417637272,
+      "grad_norm": 19.885292053222656,
+      "learning_rate": 5e-06,
+      "loss": 1.3888,
+      "num_input_tokens_seen": 10895876,
+      "step": 174
+    },
+    {
+      "epoch": 0.5790349417637272,
+      "loss": 1.2018749713897705,
+      "loss_ce": 0.0016796982381492853,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.09423828125,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 10895876,
+      "step": 174
+    },
+    {
+      "epoch": 0.5823627287853578,
+      "grad_norm": 32.432376861572266,
+      "learning_rate": 5e-06,
+      "loss": 1.2947,
+      "num_input_tokens_seen": 10957804,
+      "step": 175
+    },
+    {
+      "epoch": 0.5823627287853578,
+      "loss": 1.3988152742385864,
+      "loss_ce": 0.003551559057086706,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.1083984375,
+      "loss_xval": 1.3984375,
+      "num_input_tokens_seen": 10957804,
+      "step": 175
+    },
+    {
+      "epoch": 0.5856905158069884,
+      "grad_norm": 81.50908660888672,
+      "learning_rate": 5e-06,
+      "loss": 1.1717,
+      "num_input_tokens_seen": 11017420,
+      "step": 176
+    },
+    {
+      "epoch": 0.5856905158069884,
+      "loss": 1.09592866897583,
+      "loss_ce": 0.021709948778152466,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.138671875,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 11017420,
+      "step": 176
+    },
+    {
+      "epoch": 0.589018302828619,
+      "grad_norm": 14.417983055114746,
+      "learning_rate": 5e-06,
+      "loss": 1.5833,
+      "num_input_tokens_seen": 11080940,
+      "step": 177
+    },
+    {
+      "epoch": 0.589018302828619,
+      "loss": 1.7335617542266846,
+      "loss_ce": 0.020182903856039047,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.1337890625,
+      "loss_xval": 1.7109375,
+      "num_input_tokens_seen": 11080940,
+      "step": 177
+    },
+    {
+      "epoch": 0.5923460898502496,
+      "grad_norm": 67.00255584716797,
+      "learning_rate": 5e-06,
+      "loss": 1.3181,
+      "num_input_tokens_seen": 11144980,
+      "step": 178
+    },
+    {
+      "epoch": 0.5923460898502496,
+      "loss": 1.448054313659668,
+      "loss_ce": 0.03496834263205528,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.09716796875,
+      "loss_xval": 1.4140625,
+      "num_input_tokens_seen": 11144980,
+      "step": 178
+    },
+    {
+      "epoch": 0.5956738768718802,
+      "grad_norm": 17.97549057006836,
+      "learning_rate": 5e-06,
+      "loss": 1.3473,
+      "num_input_tokens_seen": 11208712,
+      "step": 179
+    },
+    {
+      "epoch": 0.5956738768718802,
+      "loss": 1.2413945198059082,
+      "loss_ce": 0.0031132774893194437,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.08447265625,
+      "loss_xval": 1.234375,
+      "num_input_tokens_seen": 11208712,
+      "step": 179
+    },
+    {
+      "epoch": 0.5990016638935108,
+      "grad_norm": 23.58061408996582,
+      "learning_rate": 5e-06,
+      "loss": 1.3682,
+      "num_input_tokens_seen": 11271028,
+      "step": 180
+    },
+    {
+      "epoch": 0.5990016638935108,
+      "loss": 1.1757802963256836,
+      "loss_ce": 0.0268545038998127,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.09814453125,
+      "loss_xval": 1.1484375,
+      "num_input_tokens_seen": 11271028,
+      "step": 180
+    },
+    {
+      "epoch": 0.6023294509151415,
+      "grad_norm": 15.865267753601074,
+      "learning_rate": 5e-06,
+      "loss": 1.1611,
+      "num_input_tokens_seen": 11332560,
+      "step": 181
+    },
+    {
+      "epoch": 0.6023294509151415,
+      "loss": 1.332716941833496,
+      "loss_ce": 0.01826382614672184,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.1123046875,
+      "loss_xval": 1.3125,
+      "num_input_tokens_seen": 11332560,
+      "step": 181
+    },
+    {
+      "epoch": 0.6056572379367721,
+      "grad_norm": 14.636528968811035,
+      "learning_rate": 5e-06,
+      "loss": 1.4119,
+      "num_input_tokens_seen": 11394592,
+      "step": 182
+    },
+    {
+      "epoch": 0.6056572379367721,
+      "loss": 1.3619801998138428,
+      "loss_ce": 0.004070098511874676,
+      "loss_iou": 0.421875,
+      "loss_num": 0.1025390625,
+      "loss_xval": 1.359375,
+      "num_input_tokens_seen": 11394592,
+      "step": 182
+    },
+    {
+      "epoch": 0.6089850249584027,
+      "grad_norm": 14.351191520690918,
+      "learning_rate": 5e-06,
+      "loss": 1.3834,
+      "num_input_tokens_seen": 11457912,
+      "step": 183
+    },
+    {
+      "epoch": 0.6089850249584027,
+      "loss": 1.633519172668457,
+      "loss_ce": 0.019261367619037628,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.10302734375,
+      "loss_xval": 1.6171875,
+      "num_input_tokens_seen": 11457912,
+      "step": 183
+    },
+    {
+      "epoch": 0.6123128119800333,
+      "grad_norm": 16.637283325195312,
+      "learning_rate": 5e-06,
+      "loss": 1.23,
+      "num_input_tokens_seen": 11521492,
+      "step": 184
+    },
+    {
+      "epoch": 0.6123128119800333,
+      "loss": 1.2156696319580078,
+      "loss_ce": 0.05990792065858841,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.09716796875,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 11521492,
+      "step": 184
+    },
+    {
+      "epoch": 0.6156405990016639,
+      "grad_norm": 34.948036193847656,
+      "learning_rate": 5e-06,
+      "loss": 1.5363,
+      "num_input_tokens_seen": 11584208,
+      "step": 185
+    },
+    {
+      "epoch": 0.6156405990016639,
+      "loss": 1.6890095472335815,
+      "loss_ce": 0.034224435687065125,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.09619140625,
+      "loss_xval": 1.65625,
+      "num_input_tokens_seen": 11584208,
+      "step": 185
+    },
+    {
+      "epoch": 0.6189683860232945,
+      "grad_norm": 14.760610580444336,
+      "learning_rate": 5e-06,
+      "loss": 1.129,
+      "num_input_tokens_seen": 11643560,
+      "step": 186
+    },
+    {
+      "epoch": 0.6189683860232945,
+      "loss": 1.1673153638839722,
+      "loss_ce": 0.0008114665979519486,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.1240234375,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 11643560,
+      "step": 186
+    },
+    {
+      "epoch": 0.6222961730449251,
+      "grad_norm": 16.94452667236328,
+      "learning_rate": 5e-06,
+      "loss": 1.2258,
+      "num_input_tokens_seen": 11705768,
+      "step": 187
+    },
+    {
+      "epoch": 0.6222961730449251,
+      "loss": 1.3460886478424072,
+      "loss_ce": 0.0033152345567941666,
+      "loss_iou": 0.4375,
+      "loss_num": 0.09326171875,
+      "loss_xval": 1.34375,
+      "num_input_tokens_seen": 11705768,
+      "step": 187
+    },
+    {
+      "epoch": 0.6256239600665557,
+      "grad_norm": 11.506011009216309,
+      "learning_rate": 5e-06,
+      "loss": 1.3508,
+      "num_input_tokens_seen": 11766764,
+      "step": 188
+    },
+    {
+      "epoch": 0.6256239600665557,
+      "loss": 1.2646206617355347,
+      "loss_ce": 0.04477202519774437,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.10888671875,
+      "loss_xval": 1.21875,
+      "num_input_tokens_seen": 11766764,
+      "step": 188
+    },
+    {
+      "epoch": 0.6289517470881864,
+      "grad_norm": 13.971025466918945,
+      "learning_rate": 5e-06,
+      "loss": 1.395,
+      "num_input_tokens_seen": 11829848,
+      "step": 189
+    },
+    {
+      "epoch": 0.6289517470881864,
+      "loss": 1.7523114681243896,
+      "loss_ce": 0.011588791385293007,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.1328125,
+      "loss_xval": 1.7421875,
+      "num_input_tokens_seen": 11829848,
+      "step": 189
+    },
+    {
+      "epoch": 0.632279534109817,
+      "grad_norm": 34.798828125,
+      "learning_rate": 5e-06,
+      "loss": 1.3202,
+      "num_input_tokens_seen": 11893144,
+      "step": 190
+    },
+    {
+      "epoch": 0.632279534109817,
+      "loss": 1.4185304641723633,
+      "loss_ce": 0.0669679343700409,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.123046875,
+      "loss_xval": 1.3515625,
+      "num_input_tokens_seen": 11893144,
+      "step": 190
+    },
+    {
+      "epoch": 0.6356073211314476,
+      "grad_norm": 23.807941436767578,
+      "learning_rate": 5e-06,
+      "loss": 1.3981,
+      "num_input_tokens_seen": 11956764,
+      "step": 191
+    },
+    {
+      "epoch": 0.6356073211314476,
+      "loss": 1.4876669645309448,
+      "loss_ce": 0.012813407927751541,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.10888671875,
+      "loss_xval": 1.4765625,
+      "num_input_tokens_seen": 11956764,
+      "step": 191
+    },
+    {
+      "epoch": 0.6389351081530782,
+      "grad_norm": 10.772053718566895,
+      "learning_rate": 5e-06,
+      "loss": 1.3734,
+      "num_input_tokens_seen": 12020672,
+      "step": 192
+    },
+    {
+      "epoch": 0.6389351081530782,
+      "loss": 1.2527387142181396,
+      "loss_ce": 0.014945641160011292,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.08642578125,
+      "loss_xval": 1.234375,
+      "num_input_tokens_seen": 12020672,
+      "step": 192
+    },
+    {
+      "epoch": 0.6422628951747088,
+      "grad_norm": 16.477436065673828,
+      "learning_rate": 5e-06,
+      "loss": 1.3591,
+      "num_input_tokens_seen": 12083392,
+      "step": 193
+    },
+    {
+      "epoch": 0.6422628951747088,
+      "loss": 1.431098461151123,
+      "loss_ce": 0.002875822363421321,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.12451171875,
+      "loss_xval": 1.4296875,
+      "num_input_tokens_seen": 12083392,
+      "step": 193
+    },
+    {
+      "epoch": 0.6455906821963394,
+      "grad_norm": 12.675455093383789,
+      "learning_rate": 5e-06,
+      "loss": 1.3098,
+      "num_input_tokens_seen": 12146444,
+      "step": 194
+    },
+    {
+      "epoch": 0.6455906821963394,
+      "loss": 1.375436782836914,
+      "loss_ce": 0.015329442918300629,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.115234375,
+      "loss_xval": 1.359375,
+      "num_input_tokens_seen": 12146444,
+      "step": 194
+    },
+    {
+      "epoch": 0.64891846921797,
+      "grad_norm": 15.562612533569336,
+      "learning_rate": 5e-06,
+      "loss": 1.0385,
+      "num_input_tokens_seen": 12208136,
+      "step": 195
+    },
+    {
+      "epoch": 0.64891846921797,
+      "loss": 1.0402494668960571,
+      "loss_ce": 0.0009428322082385421,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.08203125,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 12208136,
+      "step": 195
+    },
+    {
+      "epoch": 0.6522462562396006,
+      "grad_norm": 28.349838256835938,
+      "learning_rate": 5e-06,
+      "loss": 1.197,
+      "num_input_tokens_seen": 12270244,
+      "step": 196
+    },
+    {
+      "epoch": 0.6522462562396006,
+      "loss": 1.0950136184692383,
+      "loss_ce": 0.006146472413092852,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.07958984375,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 12270244,
+      "step": 196
+    },
+    {
+      "epoch": 0.6555740432612313,
+      "grad_norm": 16.476137161254883,
+      "learning_rate": 5e-06,
+      "loss": 1.3331,
+      "num_input_tokens_seen": 12333024,
+      "step": 197
+    },
+    {
+      "epoch": 0.6555740432612313,
+      "loss": 1.3048908710479736,
+      "loss_ce": 0.005574405658990145,
+      "loss_iou": 0.34375,
+      "loss_num": 0.1220703125,
+      "loss_xval": 1.296875,
+      "num_input_tokens_seen": 12333024,
+      "step": 197
+    },
+    {
+      "epoch": 0.6589018302828619,
+      "grad_norm": 13.58584976196289,
+      "learning_rate": 5e-06,
+      "loss": 1.2588,
+      "num_input_tokens_seen": 12396340,
+      "step": 198
+    },
+    {
+      "epoch": 0.6589018302828619,
+      "loss": 1.4304707050323486,
+      "loss_ce": 0.005665954202413559,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.115234375,
+      "loss_xval": 1.421875,
+      "num_input_tokens_seen": 12396340,
+      "step": 198
+    },
+    {
+      "epoch": 0.6622296173044925,
+      "grad_norm": 38.477630615234375,
+      "learning_rate": 5e-06,
+      "loss": 1.3283,
+      "num_input_tokens_seen": 12459396,
+      "step": 199
+    },
+    {
+      "epoch": 0.6622296173044925,
+      "loss": 1.2707278728485107,
+      "loss_ce": 0.000708332285284996,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.11865234375,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 12459396,
+      "step": 199
+    },
+    {
+      "epoch": 0.6655574043261231,
+      "grad_norm": 30.41405487060547,
+      "learning_rate": 5e-06,
+      "loss": 1.1544,
+      "num_input_tokens_seen": 12522324,
+      "step": 200
+    },
+    {
+      "epoch": 0.6655574043261231,
+      "loss": 0.9297307133674622,
+      "loss_ce": 0.007001231890171766,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.10009765625,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 12522324,
+      "step": 200
+    },
+    {
+      "epoch": 0.6688851913477537,
+      "grad_norm": 32.16323471069336,
+      "learning_rate": 5e-06,
+      "loss": 1.4879,
+      "num_input_tokens_seen": 12585900,
+      "step": 201
+    },
+    {
+      "epoch": 0.6688851913477537,
+      "loss": 1.5877773761749268,
+      "loss_ce": 0.02771873027086258,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.140625,
+      "loss_xval": 1.5625,
+      "num_input_tokens_seen": 12585900,
+      "step": 201
+    },
+    {
+      "epoch": 0.6722129783693843,
+      "grad_norm": 19.184961318969727,
+      "learning_rate": 5e-06,
+      "loss": 1.3649,
+      "num_input_tokens_seen": 12647776,
+      "step": 202
+    },
+    {
+      "epoch": 0.6722129783693843,
+      "loss": 1.634108066558838,
+      "loss_ce": 0.015455802902579308,
+      "loss_iou": 0.515625,
+      "loss_num": 0.11767578125,
+      "loss_xval": 1.6171875,
+      "num_input_tokens_seen": 12647776,
+      "step": 202
+    },
+    {
+      "epoch": 0.6755407653910149,
+      "grad_norm": 13.572768211364746,
+      "learning_rate": 5e-06,
+      "loss": 1.0858,
+      "num_input_tokens_seen": 12708056,
+      "step": 203
+    },
+    {
+      "epoch": 0.6755407653910149,
+      "loss": 0.7612197995185852,
+      "loss_ce": 0.04417877644300461,
+      "loss_iou": 0.10546875,
+      "loss_num": 0.10107421875,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 12708056,
+      "step": 203
+    },
+    {
+      "epoch": 0.6788685524126455,
+      "grad_norm": 15.890493392944336,
+      "learning_rate": 5e-06,
+      "loss": 1.139,
+      "num_input_tokens_seen": 12770736,
+      "step": 204
+    },
+    {
+      "epoch": 0.6788685524126455,
+      "loss": 1.2692952156066895,
+      "loss_ce": 0.0007404821808449924,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0908203125,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 12770736,
+      "step": 204
+    },
+    {
+      "epoch": 0.6821963394342762,
+      "grad_norm": 25.105133056640625,
+      "learning_rate": 5e-06,
+      "loss": 1.4867,
+      "num_input_tokens_seen": 12834956,
+      "step": 205
+    },
+    {
+      "epoch": 0.6821963394342762,
+      "loss": 1.2058188915252686,
+      "loss_ce": 0.002937908982858062,
+      "loss_iou": 0.421875,
+      "loss_num": 0.072265625,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 12834956,
+      "step": 205
+    },
+    {
+      "epoch": 0.6855241264559068,
+      "grad_norm": 57.6992073059082,
+      "learning_rate": 5e-06,
+      "loss": 1.2759,
+      "num_input_tokens_seen": 12898988,
+      "step": 206
+    },
+    {
+      "epoch": 0.6855241264559068,
+      "loss": 1.0813333988189697,
+      "loss_ce": 0.03372599184513092,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.09619140625,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 12898988,
+      "step": 206
+    },
+    {
+      "epoch": 0.6888519134775375,
+      "grad_norm": 17.233978271484375,
+      "learning_rate": 5e-06,
+      "loss": 1.1929,
+      "num_input_tokens_seen": 12962164,
+      "step": 207
+    },
+    {
+      "epoch": 0.6888519134775375,
+      "loss": 1.2135835886001587,
+      "loss_ce": 0.01924763433635235,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0625,
+      "loss_xval": 1.1953125,
+      "num_input_tokens_seen": 12962164,
+      "step": 207
+    },
+    {
+      "epoch": 0.6921797004991681,
+      "grad_norm": 16.188570022583008,
+      "learning_rate": 5e-06,
+      "loss": 1.3785,
+      "num_input_tokens_seen": 13024784,
+      "step": 208
+    },
+    {
+      "epoch": 0.6921797004991681,
+      "loss": 1.215705156326294,
+      "loss_ce": 0.005255852825939655,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.09326171875,
+      "loss_xval": 1.2109375,
+      "num_input_tokens_seen": 13024784,
+      "step": 208
+    },
+    {
+      "epoch": 0.6955074875207987,
+      "grad_norm": 23.61080551147461,
+      "learning_rate": 5e-06,
+      "loss": 1.3336,
+      "num_input_tokens_seen": 13087760,
+      "step": 209
+    },
+    {
+      "epoch": 0.6955074875207987,
+      "loss": 1.5246670246124268,
+      "loss_ce": 0.002206052653491497,
+      "loss_iou": 0.484375,
+      "loss_num": 0.1103515625,
+      "loss_xval": 1.5234375,
+      "num_input_tokens_seen": 13087760,
+      "step": 209
+    },
+    {
+      "epoch": 0.6988352745424293,
+      "grad_norm": 13.288503646850586,
+      "learning_rate": 5e-06,
+      "loss": 1.1365,
+      "num_input_tokens_seen": 13150884,
+      "step": 210
+    },
+    {
+      "epoch": 0.6988352745424293,
+      "loss": 1.2659443616867065,
+      "loss_ce": 0.018141645938158035,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0888671875,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 13150884,
+      "step": 210
+    },
+    {
+      "epoch": 0.7021630615640599,
+      "grad_norm": 30.597015380859375,
+      "learning_rate": 5e-06,
+      "loss": 1.4653,
+      "num_input_tokens_seen": 13214276,
+      "step": 211
+    },
+    {
+      "epoch": 0.7021630615640599,
+      "loss": 1.5506691932678223,
+      "loss_ce": 0.01111849956214428,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.1201171875,
+      "loss_xval": 1.5390625,
+      "num_input_tokens_seen": 13214276,
+      "step": 211
+    },
+    {
+      "epoch": 0.7054908485856906,
+      "grad_norm": 23.87419319152832,
+      "learning_rate": 5e-06,
+      "loss": 1.1347,
+      "num_input_tokens_seen": 13276424,
+      "step": 212
+    },
+    {
+      "epoch": 0.7054908485856906,
+      "loss": 1.1488089561462402,
+      "loss_ce": 0.0018362791743129492,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.08447265625,
+      "loss_xval": 1.1484375,
+      "num_input_tokens_seen": 13276424,
+      "step": 212
+    },
+    {
+      "epoch": 0.7088186356073212,
+      "grad_norm": 16.2000732421875,
+      "learning_rate": 5e-06,
+      "loss": 1.2645,
+      "num_input_tokens_seen": 13339488,
+      "step": 213
+    },
+    {
+      "epoch": 0.7088186356073212,
+      "loss": 1.0974924564361572,
+      "loss_ce": 0.0013010293478146195,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0849609375,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 13339488,
+      "step": 213
+    },
+    {
+      "epoch": 0.7121464226289518,
+      "grad_norm": 34.518123626708984,
+      "learning_rate": 5e-06,
+      "loss": 1.0251,
+      "num_input_tokens_seen": 13402452,
+      "step": 214
+    },
+    {
+      "epoch": 0.7121464226289518,
+      "loss": 1.091152548789978,
+      "loss_ce": 0.0013087954139336944,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.083984375,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 13402452,
+      "step": 214
+    },
+    {
+      "epoch": 0.7154742096505824,
+      "grad_norm": 13.655023574829102,
+      "learning_rate": 5e-06,
+      "loss": 1.157,
+      "num_input_tokens_seen": 13465432,
+      "step": 215
+    },
+    {
+      "epoch": 0.7154742096505824,
+      "loss": 1.149668574333191,
+      "loss_ce": 0.00928770937025547,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.09228515625,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 13465432,
+      "step": 215
+    },
+    {
+      "epoch": 0.718801996672213,
+      "grad_norm": 25.365924835205078,
+      "learning_rate": 5e-06,
+      "loss": 1.125,
+      "num_input_tokens_seen": 13528332,
+      "step": 216
+    },
+    {
+      "epoch": 0.718801996672213,
+      "loss": 1.1836438179016113,
+      "loss_ce": 0.13774539530277252,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0908203125,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 13528332,
+      "step": 216
+    },
+    {
+      "epoch": 0.7221297836938436,
+      "grad_norm": 27.475000381469727,
+      "learning_rate": 5e-06,
+      "loss": 1.4337,
+      "num_input_tokens_seen": 13590456,
+      "step": 217
+    },
+    {
+      "epoch": 0.7221297836938436,
+      "loss": 1.3032087087631226,
+      "loss_ce": 0.01927315816283226,
+      "loss_iou": 0.40625,
+      "loss_num": 0.09375,
+      "loss_xval": 1.28125,
+      "num_input_tokens_seen": 13590456,
+      "step": 217
+    },
+    {
+      "epoch": 0.7254575707154742,
+      "grad_norm": 26.22903823852539,
+      "learning_rate": 5e-06,
+      "loss": 1.1997,
+      "num_input_tokens_seen": 13653520,
+      "step": 218
+    },
+    {
+      "epoch": 0.7254575707154742,
+      "loss": 1.0637105703353882,
+      "loss_ce": 0.0012105784844607115,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0888671875,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 13653520,
+      "step": 218
+    },
+    {
+      "epoch": 0.7287853577371048,
+      "grad_norm": 21.157546997070312,
+      "learning_rate": 5e-06,
+      "loss": 1.2113,
+      "num_input_tokens_seen": 13717416,
+      "step": 219
+    },
+    {
+      "epoch": 0.7287853577371048,
+      "loss": 1.2450811862945557,
+      "loss_ce": 0.006799900438636541,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.08447265625,
+      "loss_xval": 1.234375,
+      "num_input_tokens_seen": 13717416,
+      "step": 219
+    },
+    {
+      "epoch": 0.7321131447587355,
+      "grad_norm": 16.073261260986328,
+      "learning_rate": 5e-06,
+      "loss": 1.3171,
+      "num_input_tokens_seen": 13780356,
+      "step": 220
+    },
+    {
+      "epoch": 0.7321131447587355,
+      "loss": 1.1784981489181519,
+      "loss_ce": 0.005646565929055214,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.07470703125,
+      "loss_xval": 1.171875,
+      "num_input_tokens_seen": 13780356,
+      "step": 220
+    },
+    {
+      "epoch": 0.7354409317803661,
+      "grad_norm": 9.423686027526855,
+      "learning_rate": 5e-06,
+      "loss": 1.4231,
+      "num_input_tokens_seen": 13843264,
+      "step": 221
+    },
+    {
+      "epoch": 0.7354409317803661,
+      "loss": 1.6099144220352173,
+      "loss_ce": 0.0015159165486693382,
+      "loss_iou": 0.484375,
+      "loss_num": 0.1279296875,
+      "loss_xval": 1.609375,
+      "num_input_tokens_seen": 13843264,
+      "step": 221
+    },
+    {
+      "epoch": 0.7387687188019967,
+      "grad_norm": 13.743108749389648,
+      "learning_rate": 5e-06,
+      "loss": 1.1927,
+      "num_input_tokens_seen": 13905668,
+      "step": 222
+    },
+    {
+      "epoch": 0.7387687188019967,
+      "loss": 1.1869020462036133,
+      "loss_ce": 0.001599297858774662,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.07666015625,
+      "loss_xval": 1.1875,
+      "num_input_tokens_seen": 13905668,
+      "step": 222
+    },
+    {
+      "epoch": 0.7420965058236273,
+      "grad_norm": 31.587129592895508,
+      "learning_rate": 5e-06,
+      "loss": 1.3488,
+      "num_input_tokens_seen": 13968100,
+      "step": 223
+    },
+    {
+      "epoch": 0.7420965058236273,
+      "loss": 1.445157527923584,
+      "loss_ce": 0.01302863098680973,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.10400390625,
+      "loss_xval": 1.4296875,
+      "num_input_tokens_seen": 13968100,
+      "step": 223
+    },
+    {
+      "epoch": 0.7454242928452579,
+      "grad_norm": 11.37446403503418,
+      "learning_rate": 5e-06,
+      "loss": 1.1052,
+      "num_input_tokens_seen": 14029952,
+      "step": 224
+    },
+    {
+      "epoch": 0.7454242928452579,
+      "loss": 1.2456074953079224,
+      "loss_ce": 0.0007344639161601663,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.08251953125,
+      "loss_xval": 1.2421875,
+      "num_input_tokens_seen": 14029952,
+      "step": 224
+    },
+    {
+      "epoch": 0.7487520798668885,
+      "grad_norm": 16.84494972229004,
+      "learning_rate": 5e-06,
+      "loss": 1.301,
+      "num_input_tokens_seen": 14093280,
+      "step": 225
+    },
+    {
+      "epoch": 0.7487520798668885,
+      "loss": 1.3011748790740967,
+      "loss_ce": 0.015042103826999664,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.12451171875,
+      "loss_xval": 1.2890625,
+      "num_input_tokens_seen": 14093280,
+      "step": 225
+    },
+    {
+      "epoch": 0.7520798668885191,
+      "grad_norm": 9.384833335876465,
+      "learning_rate": 5e-06,
+      "loss": 1.2195,
+      "num_input_tokens_seen": 14157284,
+      "step": 226
+    },
+    {
+      "epoch": 0.7520798668885191,
+      "loss": 1.217372179031372,
+      "loss_ce": 0.0010636431397870183,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.11083984375,
+      "loss_xval": 1.21875,
+      "num_input_tokens_seen": 14157284,
+      "step": 226
+    },
+    {
+      "epoch": 0.7554076539101497,
+      "grad_norm": 15.191215515136719,
+      "learning_rate": 5e-06,
+      "loss": 1.1685,
+      "num_input_tokens_seen": 14219260,
+      "step": 227
+    },
+    {
+      "epoch": 0.7554076539101497,
+      "loss": 1.1222490072250366,
+      "loss_ce": 0.015803655609488487,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.07373046875,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 14219260,
+      "step": 227
+    },
+    {
+      "epoch": 0.7587354409317804,
+      "grad_norm": 24.131282806396484,
+      "learning_rate": 5e-06,
+      "loss": 1.1008,
+      "num_input_tokens_seen": 14281308,
+      "step": 228
+    },
+    {
+      "epoch": 0.7587354409317804,
+      "loss": 1.2230792045593262,
+      "loss_ce": 0.0006670955335721374,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.09912109375,
+      "loss_xval": 1.21875,
+      "num_input_tokens_seen": 14281308,
+      "step": 228
+    },
+    {
+      "epoch": 0.762063227953411,
+      "grad_norm": 14.669888496398926,
+      "learning_rate": 5e-06,
+      "loss": 1.1266,
+      "num_input_tokens_seen": 14343964,
+      "step": 229
+    },
+    {
+      "epoch": 0.762063227953411,
+      "loss": 1.3784270286560059,
+      "loss_ce": 0.008309826254844666,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.08935546875,
+      "loss_xval": 1.3671875,
+      "num_input_tokens_seen": 14343964,
+      "step": 229
+    },
+    {
+      "epoch": 0.7653910149750416,
+      "grad_norm": 18.83928871154785,
+      "learning_rate": 5e-06,
+      "loss": 1.3428,
+      "num_input_tokens_seen": 14407128,
+      "step": 230
+    },
+    {
+      "epoch": 0.7653910149750416,
+      "loss": 1.1617732048034668,
+      "loss_ce": 0.0025936225429177284,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0869140625,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 14407128,
+      "step": 230
+    },
+    {
+      "epoch": 0.7687188019966722,
+      "grad_norm": 14.189291000366211,
+      "learning_rate": 5e-06,
+      "loss": 0.9483,
+      "num_input_tokens_seen": 14470136,
+      "step": 231
+    },
+    {
+      "epoch": 0.7687188019966722,
+      "loss": 0.9246655702590942,
+      "loss_ce": 0.01060311309993267,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 14470136,
+      "step": 231
+    },
+    {
+      "epoch": 0.7720465890183028,
+      "grad_norm": 14.045916557312012,
+      "learning_rate": 5e-06,
+      "loss": 0.8905,
+      "num_input_tokens_seen": 14532512,
+      "step": 232
+    },
+    {
+      "epoch": 0.7720465890183028,
+      "loss": 0.65977543592453,
+      "loss_ce": 0.014023483730852604,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 14532512,
+      "step": 232
+    },
+    {
+      "epoch": 0.7753743760399334,
+      "grad_norm": 20.161911010742188,
+      "learning_rate": 5e-06,
+      "loss": 1.3524,
+      "num_input_tokens_seen": 14594204,
+      "step": 233
+    },
+    {
+      "epoch": 0.7753743760399334,
+      "loss": 1.3588159084320068,
+      "loss_ce": 0.006765137426555157,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.083984375,
+      "loss_xval": 1.3515625,
+      "num_input_tokens_seen": 14594204,
+      "step": 233
+    },
+    {
+      "epoch": 0.778702163061564,
+      "grad_norm": 12.568875312805176,
+      "learning_rate": 5e-06,
+      "loss": 1.1444,
+      "num_input_tokens_seen": 14656860,
+      "step": 234
+    },
+    {
+      "epoch": 0.778702163061564,
+      "loss": 0.9393011331558228,
+      "loss_ce": 0.014252342283725739,
+      "loss_iou": 0.296875,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.92578125,
+      "num_input_tokens_seen": 14656860,
+      "step": 234
+    },
+    {
+      "epoch": 0.7820299500831946,
+      "grad_norm": 21.507532119750977,
+      "learning_rate": 5e-06,
+      "loss": 1.3511,
+      "num_input_tokens_seen": 14720468,
+      "step": 235
+    },
+    {
+      "epoch": 0.7820299500831946,
+      "loss": 1.420245885848999,
+      "loss_ce": 0.011066140606999397,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0888671875,
+      "loss_xval": 1.40625,
+      "num_input_tokens_seen": 14720468,
+      "step": 235
+    },
+    {
+      "epoch": 0.7853577371048253,
+      "grad_norm": 33.57356643676758,
+      "learning_rate": 5e-06,
+      "loss": 1.4398,
+      "num_input_tokens_seen": 14784184,
+      "step": 236
+    },
+    {
+      "epoch": 0.7853577371048253,
+      "loss": 1.4989573955535889,
+      "loss_ce": 0.0023754474241286516,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.115234375,
+      "loss_xval": 1.5,
+      "num_input_tokens_seen": 14784184,
+      "step": 236
+    },
+    {
+      "epoch": 0.7886855241264559,
+      "grad_norm": 13.441980361938477,
+      "learning_rate": 5e-06,
+      "loss": 0.9306,
+      "num_input_tokens_seen": 14846184,
+      "step": 237
+    },
+    {
+      "epoch": 0.7886855241264559,
+      "loss": 0.9496654272079468,
+      "loss_ce": 0.007770880591124296,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 14846184,
+      "step": 237
+    },
+    {
+      "epoch": 0.7920133111480865,
+      "grad_norm": 23.40178871154785,
+      "learning_rate": 5e-06,
+      "loss": 1.1604,
+      "num_input_tokens_seen": 14908792,
+      "step": 238
+    },
+    {
+      "epoch": 0.7920133111480865,
+      "loss": 1.1228361129760742,
+      "loss_ce": 0.0007658317917957902,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.09814453125,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 14908792,
+      "step": 238
+    },
+    {
+      "epoch": 0.7953410981697171,
+      "grad_norm": 29.678503036499023,
+      "learning_rate": 5e-06,
+      "loss": 1.4377,
+      "num_input_tokens_seen": 14971520,
+      "step": 239
+    },
+    {
+      "epoch": 0.7953410981697171,
+      "loss": 1.4047870635986328,
+      "loss_ce": 0.006837797816842794,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.07763671875,
+      "loss_xval": 1.3984375,
+      "num_input_tokens_seen": 14971520,
+      "step": 239
+    },
+    {
+      "epoch": 0.7986688851913477,
+      "grad_norm": 20.842504501342773,
+      "learning_rate": 5e-06,
+      "loss": 1.1844,
+      "num_input_tokens_seen": 15033176,
+      "step": 240
+    },
+    {
+      "epoch": 0.7986688851913477,
+      "loss": 1.3855491876602173,
+      "loss_ce": 0.02422105148434639,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.10498046875,
+      "loss_xval": 1.359375,
+      "num_input_tokens_seen": 15033176,
+      "step": 240
+    },
+    {
+      "epoch": 0.8019966722129783,
+      "grad_norm": 9.68075942993164,
+      "learning_rate": 5e-06,
+      "loss": 1.3038,
+      "num_input_tokens_seen": 15094324,
+      "step": 241
+    },
+    {
+      "epoch": 0.8019966722129783,
+      "loss": 1.2901484966278076,
+      "loss_ce": 0.0098751625046134,
+      "loss_iou": 0.359375,
+      "loss_num": 0.1123046875,
+      "loss_xval": 1.28125,
+      "num_input_tokens_seen": 15094324,
+      "step": 241
+    },
+    {
+      "epoch": 0.8053244592346089,
+      "grad_norm": 14.69822883605957,
+      "learning_rate": 5e-06,
+      "loss": 1.2158,
+      "num_input_tokens_seen": 15156516,
+      "step": 242
+    },
+    {
+      "epoch": 0.8053244592346089,
+      "loss": 1.2247101068496704,
+      "loss_ce": 0.02549135871231556,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.1064453125,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 15156516,
+      "step": 242
+    },
+    {
+      "epoch": 0.8086522462562395,
+      "grad_norm": 16.155757904052734,
+      "learning_rate": 5e-06,
+      "loss": 1.2242,
+      "num_input_tokens_seen": 15218404,
+      "step": 243
+    },
+    {
+      "epoch": 0.8086522462562395,
+      "loss": 1.1564106941223145,
+      "loss_ce": 0.015785671770572662,
+      "loss_iou": 0.375,
+      "loss_num": 0.07763671875,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 15218404,
+      "step": 243
+    },
+    {
+      "epoch": 0.8119800332778702,
+      "grad_norm": 13.781820297241211,
+      "learning_rate": 5e-06,
+      "loss": 1.1725,
+      "num_input_tokens_seen": 15282228,
+      "step": 244
+    },
+    {
+      "epoch": 0.8119800332778702,
+      "loss": 1.1227807998657227,
+      "loss_ce": 0.017311980947852135,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0673828125,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 15282228,
+      "step": 244
+    },
+    {
+      "epoch": 0.8153078202995009,
+      "grad_norm": 15.639060020446777,
+      "learning_rate": 5e-06,
+      "loss": 1.1997,
+      "num_input_tokens_seen": 15345232,
+      "step": 245
+    },
+    {
+      "epoch": 0.8153078202995009,
+      "loss": 1.175334095954895,
+      "loss_ce": 0.0005293773720040917,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.07177734375,
+      "loss_xval": 1.171875,
+      "num_input_tokens_seen": 15345232,
+      "step": 245
+    },
+    {
+      "epoch": 0.8186356073211315,
+      "grad_norm": 20.45536994934082,
+      "learning_rate": 5e-06,
+      "loss": 1.2539,
+      "num_input_tokens_seen": 15408764,
+      "step": 246
+    },
+    {
+      "epoch": 0.8186356073211315,
+      "loss": 1.2810962200164795,
+      "loss_ce": 0.00033442690619267523,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.076171875,
+      "loss_xval": 1.28125,
+      "num_input_tokens_seen": 15408764,
+      "step": 246
+    },
+    {
+      "epoch": 0.8219633943427621,
+      "grad_norm": 12.981587409973145,
+      "learning_rate": 5e-06,
+      "loss": 1.0421,
+      "num_input_tokens_seen": 15471008,
+      "step": 247
+    },
+    {
+      "epoch": 0.8219633943427621,
+      "loss": 1.138602375984192,
+      "loss_ce": 0.00383676472119987,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.072265625,
+      "loss_xval": 1.1328125,
+      "num_input_tokens_seen": 15471008,
+      "step": 247
+    },
+    {
+      "epoch": 0.8252911813643927,
+      "grad_norm": 51.6541633605957,
+      "learning_rate": 5e-06,
+      "loss": 0.8726,
+      "num_input_tokens_seen": 15533376,
+      "step": 248
+    },
+    {
+      "epoch": 0.8252911813643927,
+      "loss": 0.745193362236023,
+      "loss_ce": 0.01130670215934515,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.07666015625,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 15533376,
+      "step": 248
+    },
+    {
+      "epoch": 0.8286189683860233,
+      "grad_norm": 20.74786949157715,
+      "learning_rate": 5e-06,
+      "loss": 1.274,
+      "num_input_tokens_seen": 15596260,
+      "step": 249
+    },
+    {
+      "epoch": 0.8286189683860233,
+      "loss": 0.9992030262947083,
+      "loss_ce": 0.005550671834498644,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.109375,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 15596260,
+      "step": 249
+    },
+    {
+      "epoch": 0.831946755407654,
+      "grad_norm": 60.06270980834961,
+      "learning_rate": 5e-06,
+      "loss": 1.4534,
+      "num_input_tokens_seen": 15658848,
+      "step": 250
+    },
+    {
+      "epoch": 0.831946755407654,
+      "eval_seeclick_CIoU": 0.12590423226356506,
+      "eval_seeclick_GIoU": 0.17364877462387085,
+      "eval_seeclick_IoU": 0.24328559637069702,
+      "eval_seeclick_MAE_all": 0.18153271079063416,
+      "eval_seeclick_MAE_h": 0.07490194030106068,
+      "eval_seeclick_MAE_w": 0.10936548560857773,
+      "eval_seeclick_MAE_x_boxes": 0.31923606991767883,
+      "eval_seeclick_MAE_y_boxes": 0.09831257537007332,
+      "eval_seeclick_NUM_probability": 0.9999359548091888,
+      "eval_seeclick_inside_bbox": 0.30520834028720856,
+      "eval_seeclick_loss": 2.6519570350646973,
+      "eval_seeclick_loss_ce": 0.03889298997819424,
+      "eval_seeclick_loss_iou": 0.8387451171875,
+      "eval_seeclick_loss_num": 0.1761322021484375,
+      "eval_seeclick_loss_xval": 2.558349609375,
+      "eval_seeclick_runtime": 70.0545,
+      "eval_seeclick_samples_per_second": 0.671,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 15658848,
+      "step": 250
+    },
+    {
+      "epoch": 0.831946755407654,
+      "eval_icons_CIoU": 0.07106863334774971,
+      "eval_icons_GIoU": 0.19157247245311737,
+      "eval_icons_IoU": 0.21291274577379227,
+      "eval_icons_MAE_all": 0.1373068317770958,
+      "eval_icons_MAE_h": 0.07063675299286842,
+      "eval_icons_MAE_w": 0.13146401941776276,
+      "eval_icons_MAE_x_boxes": 0.1353582739830017,
+      "eval_icons_MAE_y_boxes": 0.05130962934345007,
+      "eval_icons_NUM_probability": 0.9999799430370331,
+      "eval_icons_inside_bbox": 0.2986111119389534,
+      "eval_icons_loss": 2.3059701919555664,
+      "eval_icons_loss_ce": 2.8694971661025193e-05,
+      "eval_icons_loss_iou": 0.806396484375,
+      "eval_icons_loss_num": 0.14170265197753906,
+      "eval_icons_loss_xval": 2.322265625,
+      "eval_icons_runtime": 66.8297,
+      "eval_icons_samples_per_second": 0.748,
+      "eval_icons_steps_per_second": 0.03,
+      "num_input_tokens_seen": 15658848,
+      "step": 250
+    },
+    {
+      "epoch": 0.831946755407654,
+      "eval_screenspot_CIoU": 0.023540629694859188,
+      "eval_screenspot_GIoU": 0.07850407063961029,
+      "eval_screenspot_IoU": 0.1775170018275579,
+      "eval_screenspot_MAE_all": 0.22443277140458426,
+      "eval_screenspot_MAE_h": 0.09910715123017629,
+      "eval_screenspot_MAE_w": 0.17734555900096893,
+      "eval_screenspot_MAE_x_boxes": 0.31622066100438434,
+      "eval_screenspot_MAE_y_boxes": 0.13959191491206488,
+      "eval_screenspot_NUM_probability": 0.9998429814974467,
+      "eval_screenspot_inside_bbox": 0.2912500003973643,
+      "eval_screenspot_loss": 2.9710144996643066,
+      "eval_screenspot_loss_ce": 0.0023049935698509216,
+      "eval_screenspot_loss_iou": 0.92724609375,
+      "eval_screenspot_loss_num": 0.23094685872395834,
+      "eval_screenspot_loss_xval": 3.009765625,
+      "eval_screenspot_runtime": 119.5082,
+      "eval_screenspot_samples_per_second": 0.745,
+      "eval_screenspot_steps_per_second": 0.025,
+      "num_input_tokens_seen": 15658848,
+      "step": 250
+    },
+    {
+      "epoch": 0.831946755407654,
+      "eval_compot_CIoU": -0.05822424963116646,
+      "eval_compot_GIoU": 0.039091480895876884,
+      "eval_compot_IoU": 0.10423949733376503,
+      "eval_compot_MAE_all": 0.23665092885494232,
+      "eval_compot_MAE_h": 0.08639108017086983,
+      "eval_compot_MAE_w": 0.25060585141181946,
+      "eval_compot_MAE_x_boxes": 0.2275974601507187,
+      "eval_compot_MAE_y_boxes": 0.14669033139944077,
+      "eval_compot_NUM_probability": 0.9997861981391907,
+      "eval_compot_inside_bbox": 0.2204861119389534,
+      "eval_compot_loss": 3.053682804107666,
+      "eval_compot_loss_ce": 0.0055244737304747105,
+      "eval_compot_loss_iou": 0.95751953125,
+      "eval_compot_loss_num": 0.2315673828125,
+      "eval_compot_loss_xval": 3.07421875,
+      "eval_compot_runtime": 67.9773,
+      "eval_compot_samples_per_second": 0.736,
+      "eval_compot_steps_per_second": 0.029,
+      "num_input_tokens_seen": 15658848,
+      "step": 250
+    },
+    {
+      "epoch": 0.831946755407654,
+      "eval_custom_ui_MAE_all": 0.11324165761470795,
+      "eval_custom_ui_MAE_x": 0.11285695433616638,
+      "eval_custom_ui_MAE_y": 0.11362636089324951,
+      "eval_custom_ui_NUM_probability": 0.9999939203262329,
+      "eval_custom_ui_loss": 0.5810784697532654,
+      "eval_custom_ui_loss_ce": 0.01076856441795826,
+      "eval_custom_ui_loss_num": 0.1113433837890625,
+      "eval_custom_ui_loss_xval": 0.556640625,
+      "eval_custom_ui_runtime": 57.5498,
+      "eval_custom_ui_samples_per_second": 0.869,
+      "eval_custom_ui_steps_per_second": 0.035,
+      "num_input_tokens_seen": 15658848,
+      "step": 250
+    },
+    {
+      "epoch": 0.831946755407654,
+      "loss": 0.5382449626922607,
+      "loss_ce": 0.010657099075615406,
+      "loss_iou": 0.0,
+      "loss_num": 0.10546875,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 15658848,
+      "step": 250
+    },
+    {
+      "epoch": 0.8352745424292846,
+      "grad_norm": 42.612083435058594,
+      "learning_rate": 5e-06,
+      "loss": 1.1362,
+      "num_input_tokens_seen": 15721148,
+      "step": 251
+    },
+    {
+      "epoch": 0.8352745424292846,
+      "loss": 1.1814727783203125,
+      "loss_ce": 0.0027619153261184692,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.09423828125,
+      "loss_xval": 1.1796875,
+      "num_input_tokens_seen": 15721148,
+      "step": 251
+    },
+    {
+      "epoch": 0.8386023294509152,
+      "grad_norm": 13.122926712036133,
+      "learning_rate": 5e-06,
+      "loss": 1.0092,
+      "num_input_tokens_seen": 15783360,
+      "step": 252
+    },
+    {
+      "epoch": 0.8386023294509152,
+      "loss": 1.0700435638427734,
+      "loss_ce": 0.0011958942050114274,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.08056640625,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 15783360,
+      "step": 252
+    },
+    {
+      "epoch": 0.8419301164725458,
+      "grad_norm": 21.410629272460938,
+      "learning_rate": 5e-06,
+      "loss": 1.1255,
+      "num_input_tokens_seen": 15845752,
+      "step": 253
+    },
+    {
+      "epoch": 0.8419301164725458,
+      "loss": 1.3312301635742188,
+      "loss_ce": 0.004081668332219124,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.09814453125,
+      "loss_xval": 1.328125,
+      "num_input_tokens_seen": 15845752,
+      "step": 253
+    },
+    {
+      "epoch": 0.8452579034941764,
+      "grad_norm": 15.538365364074707,
+      "learning_rate": 5e-06,
+      "loss": 1.2166,
+      "num_input_tokens_seen": 15908960,
+      "step": 254
+    },
+    {
+      "epoch": 0.8452579034941764,
+      "loss": 1.152634859085083,
+      "loss_ce": 0.010545015335083008,
+      "loss_iou": 0.375,
+      "loss_num": 0.07861328125,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 15908960,
+      "step": 254
+    },
+    {
+      "epoch": 0.848585690515807,
+      "grad_norm": 40.248104095458984,
+      "learning_rate": 5e-06,
+      "loss": 0.985,
+      "num_input_tokens_seen": 15968648,
+      "step": 255
+    },
+    {
+      "epoch": 0.848585690515807,
+      "loss": 0.7882208824157715,
+      "loss_ce": 0.02186344563961029,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.0810546875,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 15968648,
+      "step": 255
+    },
+    {
+      "epoch": 0.8519134775374376,
+      "grad_norm": 28.604537963867188,
+      "learning_rate": 5e-06,
+      "loss": 1.1575,
+      "num_input_tokens_seen": 16030956,
+      "step": 256
+    },
+    {
+      "epoch": 0.8519134775374376,
+      "loss": 1.093517780303955,
+      "loss_ce": 0.017590083181858063,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.1005859375,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 16030956,
+      "step": 256
+    },
+    {
+      "epoch": 0.8552412645590682,
+      "grad_norm": 19.028539657592773,
+      "learning_rate": 5e-06,
+      "loss": 1.1595,
+      "num_input_tokens_seen": 16092152,
+      "step": 257
+    },
+    {
+      "epoch": 0.8552412645590682,
+      "loss": 1.3261454105377197,
+      "loss_ce": 0.008274221792817116,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.11181640625,
+      "loss_xval": 1.3203125,
+      "num_input_tokens_seen": 16092152,
+      "step": 257
+    },
+    {
+      "epoch": 0.8585690515806988,
+      "grad_norm": 11.434488296508789,
+      "learning_rate": 5e-06,
+      "loss": 1.0794,
+      "num_input_tokens_seen": 16155380,
+      "step": 258
+    },
+    {
+      "epoch": 0.8585690515806988,
+      "loss": 1.0630097389221191,
+      "loss_ce": 0.01808791421353817,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 16155380,
+      "step": 258
+    },
+    {
+      "epoch": 0.8618968386023295,
+      "grad_norm": 8.151037216186523,
+      "learning_rate": 5e-06,
+      "loss": 0.9681,
+      "num_input_tokens_seen": 16216940,
+      "step": 259
+    },
+    {
+      "epoch": 0.8618968386023295,
+      "loss": 0.9589591026306152,
+      "loss_ce": 0.0010733783710747957,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.08837890625,
+      "loss_xval": 0.95703125,
+      "num_input_tokens_seen": 16216940,
+      "step": 259
+    },
+    {
+      "epoch": 0.8652246256239601,
+      "grad_norm": 19.071046829223633,
+      "learning_rate": 5e-06,
+      "loss": 1.4472,
+      "num_input_tokens_seen": 16280368,
+      "step": 260
+    },
+    {
+      "epoch": 0.8652246256239601,
+      "loss": 1.5674210786819458,
+      "loss_ce": 0.09525313228368759,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.107421875,
+      "loss_xval": 1.46875,
+      "num_input_tokens_seen": 16280368,
+      "step": 260
+    },
+    {
+      "epoch": 0.8685524126455907,
+      "grad_norm": 19.866905212402344,
+      "learning_rate": 5e-06,
+      "loss": 0.9902,
+      "num_input_tokens_seen": 16343184,
+      "step": 261
+    },
+    {
+      "epoch": 0.8685524126455907,
+      "loss": 0.8425556421279907,
+      "loss_ce": 0.005885718856006861,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 16343184,
+      "step": 261
+    },
+    {
+      "epoch": 0.8718801996672213,
+      "grad_norm": 26.4387264251709,
+      "learning_rate": 5e-06,
+      "loss": 1.1488,
+      "num_input_tokens_seen": 16405096,
+      "step": 262
+    },
+    {
+      "epoch": 0.8718801996672213,
+      "loss": 1.1173707246780396,
+      "loss_ce": 0.0014038896188139915,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.09033203125,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 16405096,
+      "step": 262
+    },
+    {
+      "epoch": 0.8752079866888519,
+      "grad_norm": 44.49690246582031,
+      "learning_rate": 5e-06,
+      "loss": 1.1406,
+      "num_input_tokens_seen": 16468060,
+      "step": 263
+    },
+    {
+      "epoch": 0.8752079866888519,
+      "loss": 1.2327797412872314,
+      "loss_ce": 0.004508184269070625,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0908203125,
+      "loss_xval": 1.2265625,
+      "num_input_tokens_seen": 16468060,
+      "step": 263
+    },
+    {
+      "epoch": 0.8785357737104825,
+      "grad_norm": 26.186443328857422,
+      "learning_rate": 5e-06,
+      "loss": 1.6099,
+      "num_input_tokens_seen": 16532308,
+      "step": 264
+    },
+    {
+      "epoch": 0.8785357737104825,
+      "loss": 1.6475975513458252,
+      "loss_ce": 0.0040428778156638145,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.1083984375,
+      "loss_xval": 1.640625,
+      "num_input_tokens_seen": 16532308,
+      "step": 264
+    },
+    {
+      "epoch": 0.8818635607321131,
+      "grad_norm": 16.094806671142578,
+      "learning_rate": 5e-06,
+      "loss": 1.2654,
+      "num_input_tokens_seen": 16595464,
+      "step": 265
+    },
+    {
+      "epoch": 0.8818635607321131,
+      "loss": 1.439080834388733,
+      "loss_ce": 0.015496812760829926,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.1123046875,
+      "loss_xval": 1.421875,
+      "num_input_tokens_seen": 16595464,
+      "step": 265
+    },
+    {
+      "epoch": 0.8851913477537438,
+      "grad_norm": 13.568002700805664,
+      "learning_rate": 5e-06,
+      "loss": 1.3184,
+      "num_input_tokens_seen": 16659512,
+      "step": 266
+    },
+    {
+      "epoch": 0.8851913477537438,
+      "loss": 1.2902538776397705,
+      "loss_ce": 0.0011914295610040426,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.10107421875,
+      "loss_xval": 1.2890625,
+      "num_input_tokens_seen": 16659512,
+      "step": 266
+    },
+    {
+      "epoch": 0.8885191347753744,
+      "grad_norm": 17.312501907348633,
+      "learning_rate": 5e-06,
+      "loss": 1.2592,
+      "num_input_tokens_seen": 16723264,
+      "step": 267
+    },
+    {
+      "epoch": 0.8885191347753744,
+      "loss": 1.0298858880996704,
+      "loss_ce": 0.006448333151638508,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0810546875,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 16723264,
+      "step": 267
+    },
+    {
+      "epoch": 0.891846921797005,
+      "grad_norm": 11.952759742736816,
+      "learning_rate": 5e-06,
+      "loss": 1.1869,
+      "num_input_tokens_seen": 16786628,
+      "step": 268
+    },
+    {
+      "epoch": 0.891846921797005,
+      "loss": 1.2707010507583618,
+      "loss_ce": 0.002634685719385743,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.08984375,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 16786628,
+      "step": 268
+    },
+    {
+      "epoch": 0.8951747088186356,
+      "grad_norm": 81.69204711914062,
+      "learning_rate": 5e-06,
+      "loss": 1.0619,
+      "num_input_tokens_seen": 16848308,
+      "step": 269
+    },
+    {
+      "epoch": 0.8951747088186356,
+      "loss": 1.1395946741104126,
+      "loss_ce": 0.0033642093185335398,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.1123046875,
+      "loss_xval": 1.1328125,
+      "num_input_tokens_seen": 16848308,
+      "step": 269
+    },
+    {
+      "epoch": 0.8985024958402662,
+      "grad_norm": 17.892982482910156,
+      "learning_rate": 5e-06,
+      "loss": 1.0715,
+      "num_input_tokens_seen": 16910492,
+      "step": 270
+    },
+    {
+      "epoch": 0.8985024958402662,
+      "loss": 0.9769110679626465,
+      "loss_ce": 0.0023017562925815582,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0888671875,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 16910492,
+      "step": 270
+    },
+    {
+      "epoch": 0.9018302828618968,
+      "grad_norm": 14.101139068603516,
+      "learning_rate": 5e-06,
+      "loss": 1.093,
+      "num_input_tokens_seen": 16973704,
+      "step": 271
+    },
+    {
+      "epoch": 0.9018302828618968,
+      "loss": 0.8828328847885132,
+      "loss_ce": 0.00368250603787601,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0791015625,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 16973704,
+      "step": 271
+    },
+    {
+      "epoch": 0.9051580698835274,
+      "grad_norm": 17.054306030273438,
+      "learning_rate": 5e-06,
+      "loss": 1.1851,
+      "num_input_tokens_seen": 17037116,
+      "step": 272
+    },
+    {
+      "epoch": 0.9051580698835274,
+      "loss": 1.2159790992736816,
+      "loss_ce": 0.0040650563314557076,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0947265625,
+      "loss_xval": 1.2109375,
+      "num_input_tokens_seen": 17037116,
+      "step": 272
+    },
+    {
+      "epoch": 0.908485856905158,
+      "grad_norm": 19.807126998901367,
+      "learning_rate": 5e-06,
+      "loss": 1.0618,
+      "num_input_tokens_seen": 17098136,
+      "step": 273
+    },
+    {
+      "epoch": 0.908485856905158,
+      "loss": 1.046678066253662,
+      "loss_ce": 0.020799197256565094,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.1220703125,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 17098136,
+      "step": 273
+    },
+    {
+      "epoch": 0.9118136439267887,
+      "grad_norm": 19.801219940185547,
+      "learning_rate": 5e-06,
+      "loss": 1.4729,
+      "num_input_tokens_seen": 17160376,
+      "step": 274
+    },
+    {
+      "epoch": 0.9118136439267887,
+      "loss": 1.3921507596969604,
+      "loss_ce": 0.038146864622831345,
+      "loss_iou": 0.4375,
+      "loss_num": 0.095703125,
+      "loss_xval": 1.3515625,
+      "num_input_tokens_seen": 17160376,
+      "step": 274
+    },
+    {
+      "epoch": 0.9151414309484193,
+      "grad_norm": 28.975116729736328,
+      "learning_rate": 5e-06,
+      "loss": 1.058,
+      "num_input_tokens_seen": 17223484,
+      "step": 275
+    },
+    {
+      "epoch": 0.9151414309484193,
+      "loss": 1.0364493131637573,
+      "loss_ce": 0.0027578980661928654,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0703125,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 17223484,
+      "step": 275
+    },
+    {
+      "epoch": 0.9184692179700499,
+      "grad_norm": 16.811128616333008,
+      "learning_rate": 5e-06,
+      "loss": 0.8941,
+      "num_input_tokens_seen": 17285864,
+      "step": 276
+    },
+    {
+      "epoch": 0.9184692179700499,
+      "loss": 0.9024735689163208,
+      "loss_ce": 0.01306929811835289,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.08349609375,
+      "loss_xval": 0.890625,
+      "num_input_tokens_seen": 17285864,
+      "step": 276
+    },
+    {
+      "epoch": 0.9217970049916805,
+      "grad_norm": 16.22093391418457,
+      "learning_rate": 5e-06,
+      "loss": 1.0589,
+      "num_input_tokens_seen": 17349464,
+      "step": 277
+    },
+    {
+      "epoch": 0.9217970049916805,
+      "loss": 0.9893227219581604,
+      "loss_ce": 0.0005531868082471192,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.099609375,
+      "loss_xval": 0.98828125,
+      "num_input_tokens_seen": 17349464,
+      "step": 277
+    },
+    {
+      "epoch": 0.9251247920133111,
+      "grad_norm": 12.684296607971191,
+      "learning_rate": 5e-06,
+      "loss": 1.1196,
+      "num_input_tokens_seen": 17412492,
+      "step": 278
+    },
+    {
+      "epoch": 0.9251247920133111,
+      "loss": 0.9944457411766052,
+      "loss_ce": 0.0027464870363473892,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.09033203125,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 17412492,
+      "step": 278
+    },
+    {
+      "epoch": 0.9284525790349417,
+      "grad_norm": 54.076019287109375,
+      "learning_rate": 5e-06,
+      "loss": 1.2471,
+      "num_input_tokens_seen": 17475792,
+      "step": 279
+    },
+    {
+      "epoch": 0.9284525790349417,
+      "loss": 1.5113483667373657,
+      "loss_ce": 0.0035358648747205734,
+      "loss_iou": 0.453125,
+      "loss_num": 0.12060546875,
+      "loss_xval": 1.5078125,
+      "num_input_tokens_seen": 17475792,
+      "step": 279
+    },
+    {
+      "epoch": 0.9317803660565723,
+      "grad_norm": 46.81892013549805,
+      "learning_rate": 5e-06,
+      "loss": 1.3139,
+      "num_input_tokens_seen": 17539360,
+      "step": 280
+    },
+    {
+      "epoch": 0.9317803660565723,
+      "loss": 0.9593971967697144,
+      "loss_ce": 0.013108117505908012,
+      "loss_iou": 0.3125,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 17539360,
+      "step": 280
+    },
+    {
+      "epoch": 0.9351081530782029,
+      "grad_norm": 26.500545501708984,
+      "learning_rate": 5e-06,
+      "loss": 1.0319,
+      "num_input_tokens_seen": 17602120,
+      "step": 281
+    },
+    {
+      "epoch": 0.9351081530782029,
+      "loss": 0.7355974316596985,
+      "loss_ce": 0.0031755524687469006,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 17602120,
+      "step": 281
+    },
+    {
+      "epoch": 0.9384359400998337,
+      "grad_norm": 26.911073684692383,
+      "learning_rate": 5e-06,
+      "loss": 1.0786,
+      "num_input_tokens_seen": 17665200,
+      "step": 282
+    },
+    {
+      "epoch": 0.9384359400998337,
+      "loss": 1.0884435176849365,
+      "loss_ce": 0.0012852513464167714,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.107421875,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 17665200,
+      "step": 282
+    },
+    {
+      "epoch": 0.9417637271214643,
+      "grad_norm": 10.468093872070312,
+      "learning_rate": 5e-06,
+      "loss": 0.7664,
+      "num_input_tokens_seen": 17727580,
+      "step": 283
+    },
+    {
+      "epoch": 0.9417637271214643,
+      "loss": 0.8793269991874695,
+      "loss_ce": 0.0031062541529536247,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.875,
+      "num_input_tokens_seen": 17727580,
+      "step": 283
+    },
+    {
+      "epoch": 0.9450915141430949,
+      "grad_norm": 9.961268424987793,
+      "learning_rate": 5e-06,
+      "loss": 0.9138,
+      "num_input_tokens_seen": 17789316,
+      "step": 284
+    },
+    {
+      "epoch": 0.9450915141430949,
+      "loss": 1.0281856060028076,
+      "loss_ce": 0.002794938860461116,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.1181640625,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 17789316,
+      "step": 284
+    },
+    {
+      "epoch": 0.9484193011647255,
+      "grad_norm": 12.167577743530273,
+      "learning_rate": 5e-06,
+      "loss": 1.1296,
+      "num_input_tokens_seen": 17852072,
+      "step": 285
+    },
+    {
+      "epoch": 0.9484193011647255,
+      "loss": 1.0344905853271484,
+      "loss_ce": 0.005193705670535564,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0771484375,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 17852072,
+      "step": 285
+    },
+    {
+      "epoch": 0.9517470881863561,
+      "grad_norm": 12.375018119812012,
+      "learning_rate": 5e-06,
+      "loss": 1.0284,
+      "num_input_tokens_seen": 17914960,
+      "step": 286
+    },
+    {
+      "epoch": 0.9517470881863561,
+      "loss": 0.9589194059371948,
+      "loss_ce": 0.009944751858711243,
+      "loss_iou": 0.3125,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.94921875,
+      "num_input_tokens_seen": 17914960,
+      "step": 286
+    },
+    {
+      "epoch": 0.9550748752079867,
+      "grad_norm": 27.08910369873047,
+      "learning_rate": 5e-06,
+      "loss": 1.1156,
+      "num_input_tokens_seen": 17976848,
+      "step": 287
+    },
+    {
+      "epoch": 0.9550748752079867,
+      "loss": 0.7211456298828125,
+      "loss_ce": 0.0006866899784654379,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.08056640625,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 17976848,
+      "step": 287
+    },
+    {
+      "epoch": 0.9584026622296173,
+      "grad_norm": 20.21120262145996,
+      "learning_rate": 5e-06,
+      "loss": 0.9142,
+      "num_input_tokens_seen": 18038416,
+      "step": 288
+    },
+    {
+      "epoch": 0.9584026622296173,
+      "loss": 1.0173907279968262,
+      "loss_ce": 0.010310713201761246,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0771484375,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 18038416,
+      "step": 288
+    },
+    {
+      "epoch": 0.961730449251248,
+      "grad_norm": 25.247774124145508,
+      "learning_rate": 5e-06,
+      "loss": 1.3575,
+      "num_input_tokens_seen": 18101284,
+      "step": 289
+    },
+    {
+      "epoch": 0.961730449251248,
+      "loss": 1.379800796508789,
+      "loss_ce": 0.005777302198112011,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.111328125,
+      "loss_xval": 1.375,
+      "num_input_tokens_seen": 18101284,
+      "step": 289
+    },
+    {
+      "epoch": 0.9650582362728786,
+      "grad_norm": 18.5815486907959,
+      "learning_rate": 5e-06,
+      "loss": 1.1872,
+      "num_input_tokens_seen": 18163408,
+      "step": 290
+    },
+    {
+      "epoch": 0.9650582362728786,
+      "loss": 1.0184900760650635,
+      "loss_ce": 0.000667768414132297,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.09619140625,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 18163408,
+      "step": 290
+    },
+    {
+      "epoch": 0.9683860232945092,
+      "grad_norm": 15.90703010559082,
+      "learning_rate": 5e-06,
+      "loss": 1.0381,
+      "num_input_tokens_seen": 18225688,
+      "step": 291
+    },
+    {
+      "epoch": 0.9683860232945092,
+      "loss": 0.9319148659706116,
+      "loss_ce": 0.00027425718144513667,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 18225688,
+      "step": 291
+    },
+    {
+      "epoch": 0.9717138103161398,
+      "grad_norm": 16.51078987121582,
+      "learning_rate": 5e-06,
+      "loss": 1.0699,
+      "num_input_tokens_seen": 18288152,
+      "step": 292
+    },
+    {
+      "epoch": 0.9717138103161398,
+      "loss": 1.2005267143249512,
+      "loss_ce": 0.0009417659603059292,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.07568359375,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 18288152,
+      "step": 292
+    },
+    {
+      "epoch": 0.9750415973377704,
+      "grad_norm": 21.68274688720703,
+      "learning_rate": 5e-06,
+      "loss": 1.1143,
+      "num_input_tokens_seen": 18351572,
+      "step": 293
+    },
+    {
+      "epoch": 0.9750415973377704,
+      "loss": 1.3064453601837158,
+      "loss_ce": 0.0044434284791350365,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0927734375,
+      "loss_xval": 1.3046875,
+      "num_input_tokens_seen": 18351572,
+      "step": 293
+    },
+    {
+      "epoch": 0.978369384359401,
+      "grad_norm": 17.310152053833008,
+      "learning_rate": 5e-06,
+      "loss": 1.0826,
+      "num_input_tokens_seen": 18415056,
+      "step": 294
+    },
+    {
+      "epoch": 0.978369384359401,
+      "loss": 0.8527212142944336,
+      "loss_ce": 0.004210504237562418,
+      "loss_iou": 0.265625,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.84765625,
+      "num_input_tokens_seen": 18415056,
+      "step": 294
+    },
+    {
+      "epoch": 0.9816971713810316,
+      "grad_norm": 48.573822021484375,
+      "learning_rate": 5e-06,
+      "loss": 1.1994,
+      "num_input_tokens_seen": 18477488,
+      "step": 295
+    },
+    {
+      "epoch": 0.9816971713810316,
+      "loss": 1.2597250938415527,
+      "loss_ce": 0.020955566316843033,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.08740234375,
+      "loss_xval": 1.2421875,
+      "num_input_tokens_seen": 18477488,
+      "step": 295
+    },
+    {
+      "epoch": 0.9850249584026622,
+      "grad_norm": 25.190378189086914,
+      "learning_rate": 5e-06,
+      "loss": 1.2747,
+      "num_input_tokens_seen": 18541468,
+      "step": 296
+    },
+    {
+      "epoch": 0.9850249584026622,
+      "loss": 1.3697137832641602,
+      "loss_ce": 0.002770456252619624,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.107421875,
+      "loss_xval": 1.3671875,
+      "num_input_tokens_seen": 18541468,
+      "step": 296
+    },
+    {
+      "epoch": 0.9883527454242929,
+      "grad_norm": 13.586545944213867,
+      "learning_rate": 5e-06,
+      "loss": 1.2876,
+      "num_input_tokens_seen": 18604076,
+      "step": 297
+    },
+    {
+      "epoch": 0.9883527454242929,
+      "loss": 1.3679777383804321,
+      "loss_ce": 0.004940649960190058,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0986328125,
+      "loss_xval": 1.359375,
+      "num_input_tokens_seen": 18604076,
+      "step": 297
+    },
+    {
+      "epoch": 0.9916805324459235,
+      "grad_norm": 20.70176887512207,
+      "learning_rate": 5e-06,
+      "loss": 0.8859,
+      "num_input_tokens_seen": 18667228,
+      "step": 298
+    },
+    {
+      "epoch": 0.9916805324459235,
+      "loss": 0.9582436084747314,
+      "loss_ce": 0.0007241345010697842,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.07958984375,
+      "loss_xval": 0.95703125,
+      "num_input_tokens_seen": 18667228,
+      "step": 298
+    },
+    {
+      "epoch": 0.9950083194675541,
+      "grad_norm": 11.641081809997559,
+      "learning_rate": 5e-06,
+      "loss": 1.1695,
+      "num_input_tokens_seen": 18730164,
+      "step": 299
+    },
+    {
+      "epoch": 0.9950083194675541,
+      "loss": 1.2555705308914185,
+      "loss_ce": 0.029008055105805397,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.09814453125,
+      "loss_xval": 1.2265625,
+      "num_input_tokens_seen": 18730164,
+      "step": 299
+    },
+    {
+      "epoch": 0.9983361064891847,
+      "grad_norm": 8.146730422973633,
+      "learning_rate": 5e-06,
+      "loss": 1.0957,
+      "num_input_tokens_seen": 18795388,
+      "step": 300
+    },
+    {
+      "epoch": 0.9983361064891847,
+      "loss": 1.0049164295196533,
+      "loss_ce": 0.002475087298080325,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 1.0,
+      "num_input_tokens_seen": 18795388,
+      "step": 300
+    },
+    {
+      "epoch": 0.9983361064891847,
+      "loss": 1.238393783569336,
+      "loss_ce": 0.0025539840571582317,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.09375,
+      "loss_xval": 1.234375,
+      "num_input_tokens_seen": 18827556,
+      "step": 300
+    },
+    {
+      "epoch": 1.0016638935108153,
+      "grad_norm": 13.772902488708496,
+      "learning_rate": 5e-06,
+      "loss": 1.0861,
+      "num_input_tokens_seen": 18858660,
+      "step": 301
+    },
+    {
+      "epoch": 1.0016638935108153,
+      "loss": 0.933795690536499,
+      "loss_ce": 0.000690196524374187,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.08154296875,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 18858660,
+      "step": 301
+    },
+    {
+      "epoch": 1.004991680532446,
+      "grad_norm": 92.42483520507812,
+      "learning_rate": 5e-06,
+      "loss": 1.1483,
+      "num_input_tokens_seen": 18921372,
+      "step": 302
+    },
+    {
+      "epoch": 1.004991680532446,
+      "loss": 1.1891447305679321,
+      "loss_ce": 0.002133029280230403,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0908203125,
+      "loss_xval": 1.1875,
+      "num_input_tokens_seen": 18921372,
+      "step": 302
+    },
+    {
+      "epoch": 1.0083194675540765,
+      "grad_norm": 13.548688888549805,
+      "learning_rate": 5e-06,
+      "loss": 0.8687,
+      "num_input_tokens_seen": 18982324,
+      "step": 303
+    },
+    {
+      "epoch": 1.0083194675540765,
+      "loss": 0.7147096395492554,
+      "loss_ce": 0.003283819416537881,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.07666015625,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 18982324,
+      "step": 303
+    },
+    {
+      "epoch": 1.0116472545757071,
+      "grad_norm": 16.49581527709961,
+      "learning_rate": 5e-06,
+      "loss": 1.1634,
+      "num_input_tokens_seen": 19044756,
+      "step": 304
+    },
+    {
+      "epoch": 1.0116472545757071,
+      "loss": 1.4042516946792603,
+      "loss_ce": 0.0009313594782724977,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.107421875,
+      "loss_xval": 1.40625,
+      "num_input_tokens_seen": 19044756,
+      "step": 304
+    },
+    {
+      "epoch": 1.0149750415973378,
+      "grad_norm": 26.810819625854492,
+      "learning_rate": 5e-06,
+      "loss": 0.9443,
+      "num_input_tokens_seen": 19106748,
+      "step": 305
+    },
+    {
+      "epoch": 1.0149750415973378,
+      "loss": 0.8466845750808716,
+      "loss_ce": 0.002446266822516918,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.076171875,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 19106748,
+      "step": 305
+    },
+    {
+      "epoch": 1.0183028286189684,
+      "grad_norm": 11.85273551940918,
+      "learning_rate": 5e-06,
+      "loss": 1.1287,
+      "num_input_tokens_seen": 19169128,
+      "step": 306
+    },
+    {
+      "epoch": 1.0183028286189684,
+      "loss": 1.0434765815734863,
+      "loss_ce": 0.00026368399267084897,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.053466796875,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 19169128,
+      "step": 306
+    },
+    {
+      "epoch": 1.021630615640599,
+      "grad_norm": 14.573290824890137,
+      "learning_rate": 5e-06,
+      "loss": 1.0402,
+      "num_input_tokens_seen": 19232736,
+      "step": 307
+    },
+    {
+      "epoch": 1.021630615640599,
+      "loss": 1.0792515277862549,
+      "loss_ce": 0.003812062554061413,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0703125,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 19232736,
+      "step": 307
+    },
+    {
+      "epoch": 1.0249584026622296,
+      "grad_norm": 16.600948333740234,
+      "learning_rate": 5e-06,
+      "loss": 1.3143,
+      "num_input_tokens_seen": 19294736,
+      "step": 308
+    },
+    {
+      "epoch": 1.0249584026622296,
+      "loss": 1.2582886219024658,
+      "loss_ce": 0.008776895701885223,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.091796875,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 19294736,
+      "step": 308
+    },
+    {
+      "epoch": 1.0282861896838602,
+      "grad_norm": 11.475916862487793,
+      "learning_rate": 5e-06,
+      "loss": 0.8611,
+      "num_input_tokens_seen": 19354768,
+      "step": 309
+    },
+    {
+      "epoch": 1.0282861896838602,
+      "loss": 0.5999712944030762,
+      "loss_ce": 0.002803354524075985,
+      "loss_iou": 0.01251220703125,
+      "loss_num": 0.11474609375,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 19354768,
+      "step": 309
+    },
+    {
+      "epoch": 1.0316139767054908,
+      "grad_norm": 16.11930274963379,
+      "learning_rate": 5e-06,
+      "loss": 1.2539,
+      "num_input_tokens_seen": 19418164,
+      "step": 310
+    },
+    {
+      "epoch": 1.0316139767054908,
+      "loss": 1.3940266370773315,
+      "loss_ce": 0.0014484911225736141,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.1123046875,
+      "loss_xval": 1.390625,
+      "num_input_tokens_seen": 19418164,
+      "step": 310
+    },
+    {
+      "epoch": 1.0349417637271214,
+      "grad_norm": 25.261287689208984,
+      "learning_rate": 5e-06,
+      "loss": 1.2662,
+      "num_input_tokens_seen": 19481684,
+      "step": 311
+    },
+    {
+      "epoch": 1.0349417637271214,
+      "loss": 1.263882040977478,
+      "loss_ce": 0.009487524628639221,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.07958984375,
+      "loss_xval": 1.2578125,
+      "num_input_tokens_seen": 19481684,
+      "step": 311
+    },
+    {
+      "epoch": 1.038269550748752,
+      "grad_norm": 11.63999080657959,
+      "learning_rate": 5e-06,
+      "loss": 1.2871,
+      "num_input_tokens_seen": 19544500,
+      "step": 312
+    },
+    {
+      "epoch": 1.038269550748752,
+      "loss": 1.305168628692627,
+      "loss_ce": 0.0012135988799855113,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0888671875,
+      "loss_xval": 1.3046875,
+      "num_input_tokens_seen": 19544500,
+      "step": 312
+    },
+    {
+      "epoch": 1.0415973377703827,
+      "grad_norm": 12.800460815429688,
+      "learning_rate": 5e-06,
+      "loss": 1.0853,
+      "num_input_tokens_seen": 19609040,
+      "step": 313
+    },
+    {
+      "epoch": 1.0415973377703827,
+      "loss": 1.041963815689087,
+      "loss_ce": 0.0007040311465971172,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0693359375,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 19609040,
+      "step": 313
+    },
+    {
+      "epoch": 1.0449251247920133,
+      "grad_norm": 54.40293502807617,
+      "learning_rate": 5e-06,
+      "loss": 0.9638,
+      "num_input_tokens_seen": 19671612,
+      "step": 314
+    },
+    {
+      "epoch": 1.0449251247920133,
+      "loss": 0.9310311079025269,
+      "loss_ce": 0.004761609248816967,
+      "loss_iou": 0.3125,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.92578125,
+      "num_input_tokens_seen": 19671612,
+      "step": 314
+    },
+    {
+      "epoch": 1.0482529118136439,
+      "grad_norm": 16.393136978149414,
+      "learning_rate": 5e-06,
+      "loss": 1.0914,
+      "num_input_tokens_seen": 19733824,
+      "step": 315
+    },
+    {
+      "epoch": 1.0482529118136439,
+      "loss": 1.1248711347579956,
+      "loss_ce": 0.002312490250915289,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.06494140625,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 19733824,
+      "step": 315
+    },
+    {
+      "epoch": 1.0515806988352745,
+      "grad_norm": 19.917030334472656,
+      "learning_rate": 5e-06,
+      "loss": 0.8806,
+      "num_input_tokens_seen": 19794408,
+      "step": 316
+    },
+    {
+      "epoch": 1.0515806988352745,
+      "loss": 0.5999534130096436,
+      "loss_ce": 0.0005881565157324076,
+      "loss_iou": 0.0,
+      "loss_num": 0.1201171875,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 19794408,
+      "step": 316
+    },
+    {
+      "epoch": 1.054908485856905,
+      "grad_norm": 23.361522674560547,
+      "learning_rate": 5e-06,
+      "loss": 1.2846,
+      "num_input_tokens_seen": 19857392,
+      "step": 317
+    },
+    {
+      "epoch": 1.054908485856905,
+      "loss": 1.1950931549072266,
+      "loss_ce": 0.0005131502402946353,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.09619140625,
+      "loss_xval": 1.1953125,
+      "num_input_tokens_seen": 19857392,
+      "step": 317
+    },
+    {
+      "epoch": 1.0582362728785357,
+      "grad_norm": 15.505134582519531,
+      "learning_rate": 5e-06,
+      "loss": 0.9283,
+      "num_input_tokens_seen": 19921056,
+      "step": 318
+    },
+    {
+      "epoch": 1.0582362728785357,
+      "loss": 1.0836975574493408,
+      "loss_ce": 0.0006896366830915213,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.07666015625,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 19921056,
+      "step": 318
+    },
+    {
+      "epoch": 1.0615640599001663,
+      "grad_norm": 15.914534568786621,
+      "learning_rate": 5e-06,
+      "loss": 0.8054,
+      "num_input_tokens_seen": 19981520,
+      "step": 319
+    },
+    {
+      "epoch": 1.0615640599001663,
+      "loss": 0.7295636534690857,
+      "loss_ce": 0.004862701054662466,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.09130859375,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 19981520,
+      "step": 319
+    },
+    {
+      "epoch": 1.064891846921797,
+      "grad_norm": 13.378053665161133,
+      "learning_rate": 5e-06,
+      "loss": 1.1706,
+      "num_input_tokens_seen": 20043756,
+      "step": 320
+    },
+    {
+      "epoch": 1.064891846921797,
+      "loss": 1.3541194200515747,
+      "loss_ce": 0.0010921121574938297,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0751953125,
+      "loss_xval": 1.3515625,
+      "num_input_tokens_seen": 20043756,
+      "step": 320
+    },
+    {
+      "epoch": 1.0682196339434276,
+      "grad_norm": 27.88616943359375,
+      "learning_rate": 5e-06,
+      "loss": 1.2254,
+      "num_input_tokens_seen": 20107020,
+      "step": 321
+    },
+    {
+      "epoch": 1.0682196339434276,
+      "loss": 1.242746114730835,
+      "loss_ce": 0.0015352306654676795,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.08740234375,
+      "loss_xval": 1.2421875,
+      "num_input_tokens_seen": 20107020,
+      "step": 321
+    },
+    {
+      "epoch": 1.0715474209650582,
+      "grad_norm": 16.343183517456055,
+      "learning_rate": 5e-06,
+      "loss": 1.1786,
+      "num_input_tokens_seen": 20169684,
+      "step": 322
+    },
+    {
+      "epoch": 1.0715474209650582,
+      "loss": 1.4232938289642334,
+      "loss_ce": 0.00923123024404049,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.12109375,
+      "loss_xval": 1.4140625,
+      "num_input_tokens_seen": 20169684,
+      "step": 322
+    },
+    {
+      "epoch": 1.0748752079866888,
+      "grad_norm": 19.044235229492188,
+      "learning_rate": 5e-06,
+      "loss": 1.1389,
+      "num_input_tokens_seen": 20231040,
+      "step": 323
+    },
+    {
+      "epoch": 1.0748752079866888,
+      "loss": 1.2635955810546875,
+      "loss_ce": 0.0021209523547440767,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.08447265625,
+      "loss_xval": 1.2578125,
+      "num_input_tokens_seen": 20231040,
+      "step": 323
+    },
+    {
+      "epoch": 1.0782029950083194,
+      "grad_norm": 22.810884475708008,
+      "learning_rate": 5e-06,
+      "loss": 1.1854,
+      "num_input_tokens_seen": 20294960,
+      "step": 324
+    },
+    {
+      "epoch": 1.0782029950083194,
+      "loss": 1.3016592264175415,
+      "loss_ce": 0.005028393119573593,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.087890625,
+      "loss_xval": 1.296875,
+      "num_input_tokens_seen": 20294960,
+      "step": 324
+    },
+    {
+      "epoch": 1.08153078202995,
+      "grad_norm": 26.073562622070312,
+      "learning_rate": 5e-06,
+      "loss": 1.0587,
+      "num_input_tokens_seen": 20358040,
+      "step": 325
+    },
+    {
+      "epoch": 1.08153078202995,
+      "loss": 1.235098123550415,
+      "loss_ce": 0.001699719694443047,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.08154296875,
+      "loss_xval": 1.234375,
+      "num_input_tokens_seen": 20358040,
+      "step": 325
+    },
+    {
+      "epoch": 1.0848585690515806,
+      "grad_norm": 13.904311180114746,
+      "learning_rate": 5e-06,
+      "loss": 1.253,
+      "num_input_tokens_seen": 20421232,
+      "step": 326
+    },
+    {
+      "epoch": 1.0848585690515806,
+      "loss": 1.2734218835830688,
+      "loss_ce": 0.001449216390028596,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.09130859375,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 20421232,
+      "step": 326
+    },
+    {
+      "epoch": 1.0881863560732112,
+      "grad_norm": 15.164936065673828,
+      "learning_rate": 5e-06,
+      "loss": 1.1673,
+      "num_input_tokens_seen": 20483820,
+      "step": 327
+    },
+    {
+      "epoch": 1.0881863560732112,
+      "loss": 1.1625075340270996,
+      "loss_ce": 0.0007643033168278635,
+      "loss_iou": 0.390625,
+      "loss_num": 0.076171875,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 20483820,
+      "step": 327
+    },
+    {
+      "epoch": 1.0915141430948418,
+      "grad_norm": 24.25714874267578,
+      "learning_rate": 5e-06,
+      "loss": 1.2235,
+      "num_input_tokens_seen": 20547208,
+      "step": 328
+    },
+    {
+      "epoch": 1.0915141430948418,
+      "loss": 1.3560025691986084,
+      "loss_ce": 0.0029753113631159067,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.09130859375,
+      "loss_xval": 1.3515625,
+      "num_input_tokens_seen": 20547208,
+      "step": 328
+    },
+    {
+      "epoch": 1.0948419301164725,
+      "grad_norm": 27.510494232177734,
+      "learning_rate": 5e-06,
+      "loss": 0.8721,
+      "num_input_tokens_seen": 20610188,
+      "step": 329
+    },
+    {
+      "epoch": 1.0948419301164725,
+      "loss": 0.7758044600486755,
+      "loss_ce": 0.00029176787938922644,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 20610188,
+      "step": 329
+    },
+    {
+      "epoch": 1.098169717138103,
+      "grad_norm": 162.73048400878906,
+      "learning_rate": 5e-06,
+      "loss": 1.1872,
+      "num_input_tokens_seen": 20674032,
+      "step": 330
+    },
+    {
+      "epoch": 1.098169717138103,
+      "loss": 0.9909277558326721,
+      "loss_ce": 0.0016699021216481924,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.98828125,
+      "num_input_tokens_seen": 20674032,
+      "step": 330
+    },
+    {
+      "epoch": 1.1014975041597337,
+      "grad_norm": 10.696892738342285,
+      "learning_rate": 5e-06,
+      "loss": 1.1888,
+      "num_input_tokens_seen": 20735628,
+      "step": 331
+    },
+    {
+      "epoch": 1.1014975041597337,
+      "loss": 1.2723336219787598,
+      "loss_ce": 0.0003609945997595787,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.095703125,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 20735628,
+      "step": 331
+    },
+    {
+      "epoch": 1.1048252911813643,
+      "grad_norm": 51.85942459106445,
+      "learning_rate": 5e-06,
+      "loss": 1.2714,
+      "num_input_tokens_seen": 20799440,
+      "step": 332
+    },
+    {
+      "epoch": 1.1048252911813643,
+      "loss": 1.278044581413269,
+      "loss_ce": 0.001189130125567317,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.10009765625,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 20799440,
+      "step": 332
+    },
+    {
+      "epoch": 1.108153078202995,
+      "grad_norm": 10.22545051574707,
+      "learning_rate": 5e-06,
+      "loss": 0.7951,
+      "num_input_tokens_seen": 20861468,
+      "step": 333
+    },
+    {
+      "epoch": 1.108153078202995,
+      "loss": 1.0232528448104858,
+      "loss_ce": 0.000547763193026185,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.08056640625,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 20861468,
+      "step": 333
+    },
+    {
+      "epoch": 1.1114808652246255,
+      "grad_norm": 8.053681373596191,
+      "learning_rate": 5e-06,
+      "loss": 1.1034,
+      "num_input_tokens_seen": 20923164,
+      "step": 334
+    },
+    {
+      "epoch": 1.1114808652246255,
+      "loss": 1.012281060218811,
+      "loss_ce": 0.0009285883279517293,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.08154296875,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 20923164,
+      "step": 334
+    },
+    {
+      "epoch": 1.1148086522462561,
+      "grad_norm": 15.463237762451172,
+      "learning_rate": 5e-06,
+      "loss": 1.1601,
+      "num_input_tokens_seen": 20987752,
+      "step": 335
+    },
+    {
+      "epoch": 1.1148086522462561,
+      "loss": 0.9524677395820618,
+      "loss_ce": 0.000807545380666852,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.062255859375,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 20987752,
+      "step": 335
+    },
+    {
+      "epoch": 1.1181364392678868,
+      "grad_norm": 10.856858253479004,
+      "learning_rate": 5e-06,
+      "loss": 1.1705,
+      "num_input_tokens_seen": 21050632,
+      "step": 336
+    },
+    {
+      "epoch": 1.1181364392678868,
+      "loss": 1.1689019203186035,
+      "loss_ce": 0.003130426863208413,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.07421875,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 21050632,
+      "step": 336
+    },
+    {
+      "epoch": 1.1214642262895174,
+      "grad_norm": 13.99282455444336,
+      "learning_rate": 5e-06,
+      "loss": 1.4048,
+      "num_input_tokens_seen": 21112928,
+      "step": 337
+    },
+    {
+      "epoch": 1.1214642262895174,
+      "loss": 1.4012787342071533,
+      "loss_ce": 0.0016206144355237484,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.11083984375,
+      "loss_xval": 1.3984375,
+      "num_input_tokens_seen": 21112928,
+      "step": 337
+    },
+    {
+      "epoch": 1.124792013311148,
+      "grad_norm": 16.16678237915039,
+      "learning_rate": 5e-06,
+      "loss": 1.147,
+      "num_input_tokens_seen": 21176408,
+      "step": 338
+    },
+    {
+      "epoch": 1.124792013311148,
+      "loss": 0.9964649081230164,
+      "loss_ce": 0.00281256134621799,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.09423828125,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 21176408,
+      "step": 338
+    },
+    {
+      "epoch": 1.1281198003327786,
+      "grad_norm": 16.422903060913086,
+      "learning_rate": 5e-06,
+      "loss": 1.1554,
+      "num_input_tokens_seen": 21238804,
+      "step": 339
+    },
+    {
+      "epoch": 1.1281198003327786,
+      "loss": 1.3646725416183472,
+      "loss_ce": 0.006762400269508362,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.1337890625,
+      "loss_xval": 1.359375,
+      "num_input_tokens_seen": 21238804,
+      "step": 339
+    },
+    {
+      "epoch": 1.1314475873544092,
+      "grad_norm": 10.378382682800293,
+      "learning_rate": 5e-06,
+      "loss": 1.0336,
+      "num_input_tokens_seen": 21300804,
+      "step": 340
+    },
+    {
+      "epoch": 1.1314475873544092,
+      "loss": 1.2877411842346191,
+      "loss_ce": 0.03554388880729675,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0859375,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 21300804,
+      "step": 340
+    },
+    {
+      "epoch": 1.1347753743760398,
+      "grad_norm": 18.60308074951172,
+      "learning_rate": 5e-06,
+      "loss": 1.0494,
+      "num_input_tokens_seen": 21364188,
+      "step": 341
+    },
+    {
+      "epoch": 1.1347753743760398,
+      "loss": 1.0969760417938232,
+      "loss_ce": 0.0027377367950975895,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.06103515625,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 21364188,
+      "step": 341
+    },
+    {
+      "epoch": 1.1381031613976704,
+      "grad_norm": 20.462738037109375,
+      "learning_rate": 5e-06,
+      "loss": 1.1857,
+      "num_input_tokens_seen": 21426420,
+      "step": 342
+    },
+    {
+      "epoch": 1.1381031613976704,
+      "loss": 1.0656057596206665,
+      "loss_ce": 0.0026175249367952347,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.08447265625,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 21426420,
+      "step": 342
+    },
+    {
+      "epoch": 1.1414309484193013,
+      "grad_norm": 21.535289764404297,
+      "learning_rate": 5e-06,
+      "loss": 1.1397,
+      "num_input_tokens_seen": 21489396,
+      "step": 343
+    },
+    {
+      "epoch": 1.1414309484193013,
+      "loss": 1.322596549987793,
+      "loss_ce": 0.0036267684772610664,
+      "loss_iou": 0.421875,
+      "loss_num": 0.09521484375,
+      "loss_xval": 1.3203125,
+      "num_input_tokens_seen": 21489396,
+      "step": 343
+    },
+    {
+      "epoch": 1.1447587354409319,
+      "grad_norm": 17.741491317749023,
+      "learning_rate": 5e-06,
+      "loss": 1.0668,
+      "num_input_tokens_seen": 21553004,
+      "step": 344
+    },
+    {
+      "epoch": 1.1447587354409319,
+      "loss": 1.1139004230499268,
+      "loss_ce": 0.0011073811911046505,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.07763671875,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 21553004,
+      "step": 344
+    },
+    {
+      "epoch": 1.1480865224625625,
+      "grad_norm": 16.684860229492188,
+      "learning_rate": 5e-06,
+      "loss": 1.2185,
+      "num_input_tokens_seen": 21616504,
+      "step": 345
+    },
+    {
+      "epoch": 1.1480865224625625,
+      "loss": 1.2288806438446045,
+      "loss_ce": 0.002318133134394884,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.087890625,
+      "loss_xval": 1.2265625,
+      "num_input_tokens_seen": 21616504,
+      "step": 345
+    },
+    {
+      "epoch": 1.151414309484193,
+      "grad_norm": 13.319676399230957,
+      "learning_rate": 5e-06,
+      "loss": 0.9034,
+      "num_input_tokens_seen": 21679532,
+      "step": 346
+    },
+    {
+      "epoch": 1.151414309484193,
+      "loss": 0.8316326141357422,
+      "loss_ce": 0.002897227182984352,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 21679532,
+      "step": 346
+    },
+    {
+      "epoch": 1.1547420965058237,
+      "grad_norm": 11.575480461120605,
+      "learning_rate": 5e-06,
+      "loss": 1.139,
+      "num_input_tokens_seen": 21742168,
+      "step": 347
+    },
+    {
+      "epoch": 1.1547420965058237,
+      "loss": 0.9771150946617126,
+      "loss_ce": 0.0007967862184159458,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 21742168,
+      "step": 347
+    },
+    {
+      "epoch": 1.1580698835274543,
+      "grad_norm": 10.588814735412598,
+      "learning_rate": 5e-06,
+      "loss": 0.9637,
+      "num_input_tokens_seen": 21803688,
+      "step": 348
+    },
+    {
+      "epoch": 1.1580698835274543,
+      "loss": 0.6981990933418274,
+      "loss_ce": 0.00020107255841139704,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.69921875,
+      "num_input_tokens_seen": 21803688,
+      "step": 348
+    },
+    {
+      "epoch": 1.161397670549085,
+      "grad_norm": 13.000744819641113,
+      "learning_rate": 5e-06,
+      "loss": 0.7941,
+      "num_input_tokens_seen": 21867184,
+      "step": 349
+    },
+    {
+      "epoch": 1.161397670549085,
+      "loss": 1.015153169631958,
+      "loss_ce": 0.0019696177914738655,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.09130859375,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 21867184,
+      "step": 349
+    },
+    {
+      "epoch": 1.1647254575707155,
+      "grad_norm": 8.500694274902344,
+      "learning_rate": 5e-06,
+      "loss": 0.9189,
+      "num_input_tokens_seen": 21930068,
+      "step": 350
+    },
+    {
+      "epoch": 1.1647254575707155,
+      "loss": 0.8839759826660156,
+      "loss_ce": 0.004825552459806204,
+      "loss_iou": 0.296875,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 21930068,
+      "step": 350
+    },
+    {
+      "epoch": 1.1680532445923462,
+      "grad_norm": 22.050308227539062,
+      "learning_rate": 5e-06,
+      "loss": 0.85,
+      "num_input_tokens_seen": 21992112,
+      "step": 351
+    },
+    {
+      "epoch": 1.1680532445923462,
+      "loss": 0.871414065361023,
+      "loss_ce": 0.005691413767635822,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.10302734375,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 21992112,
+      "step": 351
+    },
+    {
+      "epoch": 1.1713810316139768,
+      "grad_norm": 15.947305679321289,
+      "learning_rate": 5e-06,
+      "loss": 0.9005,
+      "num_input_tokens_seen": 22055200,
+      "step": 352
+    },
+    {
+      "epoch": 1.1713810316139768,
+      "loss": 0.8017796277999878,
+      "loss_ce": 0.00209699384868145,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 22055200,
+      "step": 352
+    },
+    {
+      "epoch": 1.1747088186356074,
+      "grad_norm": 15.728245735168457,
+      "learning_rate": 5e-06,
+      "loss": 0.9721,
+      "num_input_tokens_seen": 22117372,
+      "step": 353
+    },
+    {
+      "epoch": 1.1747088186356074,
+      "loss": 0.8560371398925781,
+      "loss_ce": 0.026935596019029617,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 22117372,
+      "step": 353
+    },
+    {
+      "epoch": 1.178036605657238,
+      "grad_norm": 9.21013355255127,
+      "learning_rate": 5e-06,
+      "loss": 1.0148,
+      "num_input_tokens_seen": 22180280,
+      "step": 354
+    },
+    {
+      "epoch": 1.178036605657238,
+      "loss": 1.0702283382415771,
+      "loss_ce": 0.0018690190045163035,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0791015625,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 22180280,
+      "step": 354
+    },
+    {
+      "epoch": 1.1813643926788686,
+      "grad_norm": 13.137001991271973,
+      "learning_rate": 5e-06,
+      "loss": 0.9983,
+      "num_input_tokens_seen": 22243800,
+      "step": 355
+    },
+    {
+      "epoch": 1.1813643926788686,
+      "loss": 0.9866027235984802,
+      "loss_ce": 0.002227720571681857,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 22243800,
+      "step": 355
+    },
+    {
+      "epoch": 1.1846921797004992,
+      "grad_norm": 13.438755989074707,
+      "learning_rate": 5e-06,
+      "loss": 0.9054,
+      "num_input_tokens_seen": 22306152,
+      "step": 356
+    },
+    {
+      "epoch": 1.1846921797004992,
+      "loss": 0.93604576587677,
+      "loss_ce": 0.0034285818692296743,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 22306152,
+      "step": 356
+    },
+    {
+      "epoch": 1.1880199667221298,
+      "grad_norm": 9.37387466430664,
+      "learning_rate": 5e-06,
+      "loss": 0.9098,
+      "num_input_tokens_seen": 22368308,
+      "step": 357
+    },
+    {
+      "epoch": 1.1880199667221298,
+      "loss": 0.8839547038078308,
+      "loss_ce": 0.0011421950766816735,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.08740234375,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 22368308,
+      "step": 357
+    },
+    {
+      "epoch": 1.1913477537437605,
+      "grad_norm": 15.862778663635254,
+      "learning_rate": 5e-06,
+      "loss": 0.9651,
+      "num_input_tokens_seen": 22430760,
+      "step": 358
+    },
+    {
+      "epoch": 1.1913477537437605,
+      "loss": 0.8542115688323975,
+      "loss_ce": 0.0004518293426372111,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 22430760,
+      "step": 358
+    },
+    {
+      "epoch": 1.194675540765391,
+      "grad_norm": 13.395066261291504,
+      "learning_rate": 5e-06,
+      "loss": 1.2366,
+      "num_input_tokens_seen": 22494188,
+      "step": 359
+    },
+    {
+      "epoch": 1.194675540765391,
+      "loss": 1.1658010482788086,
+      "loss_ce": 0.0014943606220185757,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.08154296875,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 22494188,
+      "step": 359
+    },
+    {
+      "epoch": 1.1980033277870217,
+      "grad_norm": 17.724477767944336,
+      "learning_rate": 5e-06,
+      "loss": 1.1109,
+      "num_input_tokens_seen": 22557396,
+      "step": 360
+    },
+    {
+      "epoch": 1.1980033277870217,
+      "loss": 0.9321660995483398,
+      "loss_ce": 0.0002813843311741948,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.08251953125,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 22557396,
+      "step": 360
+    },
+    {
+      "epoch": 1.2013311148086523,
+      "grad_norm": 15.220552444458008,
+      "learning_rate": 5e-06,
+      "loss": 1.3183,
+      "num_input_tokens_seen": 22619512,
+      "step": 361
+    },
+    {
+      "epoch": 1.2013311148086523,
+      "loss": 1.3823212385177612,
+      "loss_ce": 0.002438361756503582,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.08056640625,
+      "loss_xval": 1.3828125,
+      "num_input_tokens_seen": 22619512,
+      "step": 361
+    },
+    {
+      "epoch": 1.204658901830283,
+      "grad_norm": 13.84691047668457,
+      "learning_rate": 5e-06,
+      "loss": 0.9889,
+      "num_input_tokens_seen": 22683192,
+      "step": 362
+    },
+    {
+      "epoch": 1.204658901830283,
+      "loss": 1.0336413383483887,
+      "loss_ce": 0.005320955533534288,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.07568359375,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 22683192,
+      "step": 362
+    },
+    {
+      "epoch": 1.2079866888519135,
+      "grad_norm": 8.67003345489502,
+      "learning_rate": 5e-06,
+      "loss": 1.0009,
+      "num_input_tokens_seen": 22746804,
+      "step": 363
+    },
+    {
+      "epoch": 1.2079866888519135,
+      "loss": 0.8072459697723389,
+      "loss_ce": 0.001337698893621564,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 22746804,
+      "step": 363
+    },
+    {
+      "epoch": 1.2113144758735441,
+      "grad_norm": 19.87843894958496,
+      "learning_rate": 5e-06,
+      "loss": 1.1418,
+      "num_input_tokens_seen": 22809236,
+      "step": 364
+    },
+    {
+      "epoch": 1.2113144758735441,
+      "loss": 1.1470714807510376,
+      "loss_ce": 0.001075398176908493,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.083984375,
+      "loss_xval": 1.1484375,
+      "num_input_tokens_seen": 22809236,
+      "step": 364
+    },
+    {
+      "epoch": 1.2146422628951747,
+      "grad_norm": 19.612537384033203,
+      "learning_rate": 5e-06,
+      "loss": 1.0437,
+      "num_input_tokens_seen": 22871224,
+      "step": 365
+    },
+    {
+      "epoch": 1.2146422628951747,
+      "loss": 1.0723521709442139,
+      "loss_ce": 0.0015513792168349028,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.06591796875,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 22871224,
+      "step": 365
+    },
+    {
+      "epoch": 1.2179700499168054,
+      "grad_norm": 13.617435455322266,
+      "learning_rate": 5e-06,
+      "loss": 1.1042,
+      "num_input_tokens_seen": 22934344,
+      "step": 366
+    },
+    {
+      "epoch": 1.2179700499168054,
+      "loss": 1.1154537200927734,
+      "loss_ce": 0.0016841854667291045,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.07568359375,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 22934344,
+      "step": 366
+    },
+    {
+      "epoch": 1.221297836938436,
+      "grad_norm": 8.669763565063477,
+      "learning_rate": 5e-06,
+      "loss": 0.6066,
+      "num_input_tokens_seen": 22996164,
+      "step": 367
+    },
+    {
+      "epoch": 1.221297836938436,
+      "loss": 0.5188974142074585,
+      "loss_ce": 0.005957929417490959,
+      "loss_iou": 0.1201171875,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 22996164,
+      "step": 367
+    },
+    {
+      "epoch": 1.2246256239600666,
+      "grad_norm": 12.283041000366211,
+      "learning_rate": 5e-06,
+      "loss": 0.9514,
+      "num_input_tokens_seen": 23058328,
+      "step": 368
+    },
+    {
+      "epoch": 1.2246256239600666,
+      "loss": 1.0867336988449097,
+      "loss_ce": 0.0010403767228126526,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.09521484375,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 23058328,
+      "step": 368
+    },
+    {
+      "epoch": 1.2279534109816972,
+      "grad_norm": 25.49207305908203,
+      "learning_rate": 5e-06,
+      "loss": 1.1151,
+      "num_input_tokens_seen": 23120736,
+      "step": 369
+    },
+    {
+      "epoch": 1.2279534109816972,
+      "loss": 1.349541425704956,
+      "loss_ce": 0.0018851247150450945,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.08740234375,
+      "loss_xval": 1.34375,
+      "num_input_tokens_seen": 23120736,
+      "step": 369
+    },
+    {
+      "epoch": 1.2312811980033278,
+      "grad_norm": 20.782428741455078,
+      "learning_rate": 5e-06,
+      "loss": 1.2506,
+      "num_input_tokens_seen": 23182684,
+      "step": 370
+    },
+    {
+      "epoch": 1.2312811980033278,
+      "loss": 1.2768054008483887,
+      "loss_ce": 0.00043815511162392795,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.103515625,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 23182684,
+      "step": 370
+    },
+    {
+      "epoch": 1.2346089850249584,
+      "grad_norm": 12.254620552062988,
+      "learning_rate": 5e-06,
+      "loss": 1.2855,
+      "num_input_tokens_seen": 23244864,
+      "step": 371
+    },
+    {
+      "epoch": 1.2346089850249584,
+      "loss": 1.1182758808135986,
+      "loss_ce": 0.002309112809598446,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.08349609375,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 23244864,
+      "step": 371
+    },
+    {
+      "epoch": 1.237936772046589,
+      "grad_norm": 20.859880447387695,
+      "learning_rate": 5e-06,
+      "loss": 1.0792,
+      "num_input_tokens_seen": 23308204,
+      "step": 372
+    },
+    {
+      "epoch": 1.237936772046589,
+      "loss": 1.1883625984191895,
+      "loss_ce": 0.00037434539990499616,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.091796875,
+      "loss_xval": 1.1875,
+      "num_input_tokens_seen": 23308204,
+      "step": 372
+    },
+    {
+      "epoch": 1.2412645590682196,
+      "grad_norm": 10.013941764831543,
+      "learning_rate": 5e-06,
+      "loss": 0.9596,
+      "num_input_tokens_seen": 23371756,
+      "step": 373
+    },
+    {
+      "epoch": 1.2412645590682196,
+      "loss": 1.0599653720855713,
+      "loss_ce": 0.0008833592291921377,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.06982421875,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 23371756,
+      "step": 373
+    },
+    {
+      "epoch": 1.2445923460898503,
+      "grad_norm": 15.680611610412598,
+      "learning_rate": 5e-06,
+      "loss": 1.147,
+      "num_input_tokens_seen": 23434380,
+      "step": 374
+    },
+    {
+      "epoch": 1.2445923460898503,
+      "loss": 1.3000984191894531,
+      "loss_ce": 0.0020027763675898314,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.11962890625,
+      "loss_xval": 1.296875,
+      "num_input_tokens_seen": 23434380,
+      "step": 374
+    },
+    {
+      "epoch": 1.2479201331114809,
+      "grad_norm": 55.765045166015625,
+      "learning_rate": 5e-06,
+      "loss": 0.9942,
+      "num_input_tokens_seen": 23496936,
+      "step": 375
+    },
+    {
+      "epoch": 1.2479201331114809,
+      "loss": 0.8713576793670654,
+      "loss_ce": 0.0001418392639607191,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 23496936,
+      "step": 375
+    },
+    {
+      "epoch": 1.2512479201331115,
+      "grad_norm": 11.986567497253418,
+      "learning_rate": 5e-06,
+      "loss": 1.1553,
+      "num_input_tokens_seen": 23561228,
+      "step": 376
+    },
+    {
+      "epoch": 1.2512479201331115,
+      "loss": 1.2353582382202148,
+      "loss_ce": 0.008063295856118202,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.078125,
+      "loss_xval": 1.2265625,
+      "num_input_tokens_seen": 23561228,
+      "step": 376
+    },
+    {
+      "epoch": 1.254575707154742,
+      "grad_norm": 15.518547058105469,
+      "learning_rate": 5e-06,
+      "loss": 1.1645,
+      "num_input_tokens_seen": 23623508,
+      "step": 377
+    },
+    {
+      "epoch": 1.254575707154742,
+      "loss": 1.1245479583740234,
+      "loss_ce": 0.011266733519732952,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.056396484375,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 23623508,
+      "step": 377
+    },
+    {
+      "epoch": 1.2579034941763727,
+      "grad_norm": 12.487281799316406,
+      "learning_rate": 5e-06,
+      "loss": 1.1126,
+      "num_input_tokens_seen": 23686012,
+      "step": 378
+    },
+    {
+      "epoch": 1.2579034941763727,
+      "loss": 1.173827052116394,
+      "loss_ce": 0.0009754931088536978,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.1005859375,
+      "loss_xval": 1.171875,
+      "num_input_tokens_seen": 23686012,
+      "step": 378
+    },
+    {
+      "epoch": 1.2612312811980033,
+      "grad_norm": 14.35226821899414,
+      "learning_rate": 5e-06,
+      "loss": 1.0547,
+      "num_input_tokens_seen": 23749124,
+      "step": 379
+    },
+    {
+      "epoch": 1.2612312811980033,
+      "loss": 1.0610747337341309,
+      "loss_ce": 0.0007720579742453992,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0673828125,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 23749124,
+      "step": 379
+    },
+    {
+      "epoch": 1.264559068219634,
+      "grad_norm": 24.53921890258789,
+      "learning_rate": 5e-06,
+      "loss": 1.2042,
+      "num_input_tokens_seen": 23811144,
+      "step": 380
+    },
+    {
+      "epoch": 1.264559068219634,
+      "loss": 1.225796103477478,
+      "loss_ce": 0.0036280876956880093,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.08984375,
+      "loss_xval": 1.21875,
+      "num_input_tokens_seen": 23811144,
+      "step": 380
+    },
+    {
+      "epoch": 1.2678868552412645,
+      "grad_norm": 13.439510345458984,
+      "learning_rate": 5e-06,
+      "loss": 0.9866,
+      "num_input_tokens_seen": 23874480,
+      "step": 381
+    },
+    {
+      "epoch": 1.2678868552412645,
+      "loss": 0.8618694543838501,
+      "loss_ce": 0.0004192190826870501,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.86328125,
+      "num_input_tokens_seen": 23874480,
+      "step": 381
+    },
+    {
+      "epoch": 1.2712146422628952,
+      "grad_norm": 16.41143035888672,
+      "learning_rate": 5e-06,
+      "loss": 0.9888,
+      "num_input_tokens_seen": 23936680,
+      "step": 382
+    },
+    {
+      "epoch": 1.2712146422628952,
+      "loss": 0.9557693004608154,
+      "loss_ce": 0.00514674698933959,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.09033203125,
+      "loss_xval": 0.94921875,
+      "num_input_tokens_seen": 23936680,
+      "step": 382
+    },
+    {
+      "epoch": 1.2745424292845258,
+      "grad_norm": 20.497072219848633,
+      "learning_rate": 5e-06,
+      "loss": 0.873,
+      "num_input_tokens_seen": 23997596,
+      "step": 383
+    },
+    {
+      "epoch": 1.2745424292845258,
+      "loss": 0.9353244304656982,
+      "loss_ce": 0.0036837609950453043,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 23997596,
+      "step": 383
+    },
+    {
+      "epoch": 1.2778702163061564,
+      "grad_norm": 16.8009033203125,
+      "learning_rate": 5e-06,
+      "loss": 0.836,
+      "num_input_tokens_seen": 24059808,
+      "step": 384
+    },
+    {
+      "epoch": 1.2778702163061564,
+      "loss": 0.957974910736084,
+      "loss_ce": 0.0016761153237894177,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.95703125,
+      "num_input_tokens_seen": 24059808,
+      "step": 384
+    },
+    {
+      "epoch": 1.281198003327787,
+      "grad_norm": 16.275964736938477,
+      "learning_rate": 5e-06,
+      "loss": 0.9754,
+      "num_input_tokens_seen": 24123296,
+      "step": 385
+    },
+    {
+      "epoch": 1.281198003327787,
+      "loss": 1.200371503829956,
+      "loss_ce": 0.0004203752614557743,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.09130859375,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 24123296,
+      "step": 385
+    },
+    {
+      "epoch": 1.2845257903494176,
+      "grad_norm": 15.432588577270508,
+      "learning_rate": 5e-06,
+      "loss": 1.1704,
+      "num_input_tokens_seen": 24186340,
+      "step": 386
+    },
+    {
+      "epoch": 1.2845257903494176,
+      "loss": 1.2510305643081665,
+      "loss_ce": 0.006401653401553631,
+      "loss_iou": 0.453125,
+      "loss_num": 0.068359375,
+      "loss_xval": 1.2421875,
+      "num_input_tokens_seen": 24186340,
+      "step": 386
+    },
+    {
+      "epoch": 1.2878535773710482,
+      "grad_norm": 12.922508239746094,
+      "learning_rate": 5e-06,
+      "loss": 1.2301,
+      "num_input_tokens_seen": 24250244,
+      "step": 387
+    },
+    {
+      "epoch": 1.2878535773710482,
+      "loss": 1.315189242362976,
+      "loss_ce": 0.0007360831368714571,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.068359375,
+      "loss_xval": 1.3125,
+      "num_input_tokens_seen": 24250244,
+      "step": 387
+    },
+    {
+      "epoch": 1.2911813643926788,
+      "grad_norm": 16.12157440185547,
+      "learning_rate": 5e-06,
+      "loss": 1.0807,
+      "num_input_tokens_seen": 24312700,
+      "step": 388
+    },
+    {
+      "epoch": 1.2911813643926788,
+      "loss": 0.9249197840690613,
+      "loss_ce": 0.00011512526543810964,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.92578125,
+      "num_input_tokens_seen": 24312700,
+      "step": 388
+    },
+    {
+      "epoch": 1.2945091514143094,
+      "grad_norm": 15.61776351928711,
+      "learning_rate": 5e-06,
+      "loss": 1.206,
+      "num_input_tokens_seen": 24372944,
+      "step": 389
+    },
+    {
+      "epoch": 1.2945091514143094,
+      "loss": 1.2747079133987427,
+      "loss_ce": 0.00664151506498456,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.1376953125,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 24372944,
+      "step": 389
+    },
+    {
+      "epoch": 1.29783693843594,
+      "grad_norm": 10.26531982421875,
+      "learning_rate": 5e-06,
+      "loss": 1.1097,
+      "num_input_tokens_seen": 24436248,
+      "step": 390
+    },
+    {
+      "epoch": 1.29783693843594,
+      "loss": 1.227935791015625,
+      "loss_ce": 0.0003966076474171132,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0791015625,
+      "loss_xval": 1.2265625,
+      "num_input_tokens_seen": 24436248,
+      "step": 390
+    },
+    {
+      "epoch": 1.3011647254575707,
+      "grad_norm": 10.401841163635254,
+      "learning_rate": 5e-06,
+      "loss": 1.0097,
+      "num_input_tokens_seen": 24498940,
+      "step": 391
+    },
+    {
+      "epoch": 1.3011647254575707,
+      "loss": 1.0491610765457153,
+      "loss_ce": 0.004727460443973541,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.08349609375,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 24498940,
+      "step": 391
+    },
+    {
+      "epoch": 1.3044925124792013,
+      "grad_norm": 17.991519927978516,
+      "learning_rate": 5e-06,
+      "loss": 1.1516,
+      "num_input_tokens_seen": 24563372,
+      "step": 392
+    },
+    {
+      "epoch": 1.3044925124792013,
+      "loss": 1.063659906387329,
+      "loss_ce": 0.0001832506968639791,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 24563372,
+      "step": 392
+    },
+    {
+      "epoch": 1.307820299500832,
+      "grad_norm": 16.1484375,
+      "learning_rate": 5e-06,
+      "loss": 0.8481,
+      "num_input_tokens_seen": 24625780,
+      "step": 393
+    },
+    {
+      "epoch": 1.307820299500832,
+      "loss": 0.6541198492050171,
+      "loss_ce": 6.708937871735543e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 24625780,
+      "step": 393
+    },
+    {
+      "epoch": 1.3111480865224625,
+      "grad_norm": 15.9221773147583,
+      "learning_rate": 5e-06,
+      "loss": 0.9679,
+      "num_input_tokens_seen": 24687492,
+      "step": 394
+    },
+    {
+      "epoch": 1.3111480865224625,
+      "loss": 1.267278790473938,
+      "loss_ce": 0.0021421266719698906,
+      "loss_iou": 0.40625,
+      "loss_num": 0.09033203125,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 24687492,
+      "step": 394
+    },
+    {
+      "epoch": 1.3144758735440931,
+      "grad_norm": 23.325969696044922,
+      "learning_rate": 5e-06,
+      "loss": 1.177,
+      "num_input_tokens_seen": 24752340,
+      "step": 395
+    },
+    {
+      "epoch": 1.3144758735440931,
+      "loss": 1.2171975374221802,
+      "loss_ce": 0.001377263804897666,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 1.21875,
+      "num_input_tokens_seen": 24752340,
+      "step": 395
+    },
+    {
+      "epoch": 1.3178036605657237,
+      "grad_norm": 43.90242004394531,
+      "learning_rate": 5e-06,
+      "loss": 1.1792,
+      "num_input_tokens_seen": 24814652,
+      "step": 396
+    },
+    {
+      "epoch": 1.3178036605657237,
+      "loss": 1.2642714977264404,
+      "loss_ce": 0.0020644143223762512,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.08642578125,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 24814652,
+      "step": 396
+    },
+    {
+      "epoch": 1.3211314475873543,
+      "grad_norm": 45.1225700378418,
+      "learning_rate": 5e-06,
+      "loss": 1.1999,
+      "num_input_tokens_seen": 24878016,
+      "step": 397
+    },
+    {
+      "epoch": 1.3211314475873543,
+      "loss": 1.1546871662139893,
+      "loss_ce": 0.003564041806384921,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0751953125,
+      "loss_xval": 1.1484375,
+      "num_input_tokens_seen": 24878016,
+      "step": 397
+    },
+    {
+      "epoch": 1.324459234608985,
+      "grad_norm": 19.703289031982422,
+      "learning_rate": 5e-06,
+      "loss": 1.1245,
+      "num_input_tokens_seen": 24939340,
+      "step": 398
+    },
+    {
+      "epoch": 1.324459234608985,
+      "loss": 0.9508374333381653,
+      "loss_ce": 0.000153833330841735,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.94921875,
+      "num_input_tokens_seen": 24939340,
+      "step": 398
+    },
+    {
+      "epoch": 1.3277870216306156,
+      "grad_norm": 11.191904067993164,
+      "learning_rate": 5e-06,
+      "loss": 0.963,
+      "num_input_tokens_seen": 25002132,
+      "step": 399
+    },
+    {
+      "epoch": 1.3277870216306156,
+      "loss": 0.7988940477371216,
+      "loss_ce": 0.006901869084686041,
+      "loss_iou": 0.265625,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 25002132,
+      "step": 399
+    },
+    {
+      "epoch": 1.3311148086522462,
+      "grad_norm": 10.622575759887695,
+      "learning_rate": 5e-06,
+      "loss": 1.1994,
+      "num_input_tokens_seen": 25066044,
+      "step": 400
+    },
+    {
+      "epoch": 1.3311148086522462,
+      "loss": 1.0515964031219482,
+      "loss_ce": 0.0037447987124323845,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.068359375,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 25066044,
+      "step": 400
+    },
+    {
+      "epoch": 1.3344425956738768,
+      "grad_norm": 24.427770614624023,
+      "learning_rate": 5e-06,
+      "loss": 0.9622,
+      "num_input_tokens_seen": 25128340,
+      "step": 401
+    },
+    {
+      "epoch": 1.3344425956738768,
+      "loss": 0.9643534421920776,
+      "loss_ce": 0.002195254433900118,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.10595703125,
+      "loss_xval": 0.9609375,
+      "num_input_tokens_seen": 25128340,
+      "step": 401
+    },
+    {
+      "epoch": 1.3377703826955074,
+      "grad_norm": 10.355712890625,
+      "learning_rate": 5e-06,
+      "loss": 1.0325,
+      "num_input_tokens_seen": 25192192,
+      "step": 402
+    },
+    {
+      "epoch": 1.3377703826955074,
+      "loss": 1.0074443817138672,
+      "loss_ce": 0.001585077028721571,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 25192192,
+      "step": 402
+    },
+    {
+      "epoch": 1.341098169717138,
+      "grad_norm": 21.185884475708008,
+      "learning_rate": 5e-06,
+      "loss": 1.0896,
+      "num_input_tokens_seen": 25255636,
+      "step": 403
+    },
+    {
+      "epoch": 1.341098169717138,
+      "loss": 0.9336959719657898,
+      "loss_ce": 0.00144498934969306,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 25255636,
+      "step": 403
+    },
+    {
+      "epoch": 1.3444259567387689,
+      "grad_norm": 9.39505386352539,
+      "learning_rate": 5e-06,
+      "loss": 1.0355,
+      "num_input_tokens_seen": 25317992,
+      "step": 404
+    },
+    {
+      "epoch": 1.3444259567387689,
+      "loss": 1.0250355005264282,
+      "loss_ce": 0.00013315524847712368,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0634765625,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 25317992,
+      "step": 404
+    },
+    {
+      "epoch": 1.3477537437603995,
+      "grad_norm": 20.77029037475586,
+      "learning_rate": 5e-06,
+      "loss": 1.1133,
+      "num_input_tokens_seen": 25380772,
+      "step": 405
+    },
+    {
+      "epoch": 1.3477537437603995,
+      "loss": 1.0792059898376465,
+      "loss_ce": 0.011090747080743313,
+      "loss_iou": 0.296875,
+      "loss_num": 0.09521484375,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 25380772,
+      "step": 405
+    },
+    {
+      "epoch": 1.35108153078203,
+      "grad_norm": 31.16495132446289,
+      "learning_rate": 5e-06,
+      "loss": 1.1511,
+      "num_input_tokens_seen": 25445244,
+      "step": 406
+    },
+    {
+      "epoch": 1.35108153078203,
+      "loss": 1.2152575254440308,
+      "loss_ce": 0.0004138269869145006,
+      "loss_iou": 0.390625,
+      "loss_num": 0.08740234375,
+      "loss_xval": 1.21875,
+      "num_input_tokens_seen": 25445244,
+      "step": 406
+    },
+    {
+      "epoch": 1.3544093178036607,
+      "grad_norm": 20.450408935546875,
+      "learning_rate": 5e-06,
+      "loss": 0.8806,
+      "num_input_tokens_seen": 25507188,
+      "step": 407
+    },
+    {
+      "epoch": 1.3544093178036607,
+      "loss": 0.6218196153640747,
+      "loss_ce": 0.00023759223404340446,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 25507188,
+      "step": 407
+    },
+    {
+      "epoch": 1.3577371048252913,
+      "grad_norm": 16.568866729736328,
+      "learning_rate": 5e-06,
+      "loss": 0.9924,
+      "num_input_tokens_seen": 25570620,
+      "step": 408
+    },
+    {
+      "epoch": 1.3577371048252913,
+      "loss": 0.8238029479980469,
+      "loss_ce": 0.022045187652111053,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 25570620,
+      "step": 408
+    },
+    {
+      "epoch": 1.361064891846922,
+      "grad_norm": 17.278478622436523,
+      "learning_rate": 5e-06,
+      "loss": 1.0307,
+      "num_input_tokens_seen": 25633892,
+      "step": 409
+    },
+    {
+      "epoch": 1.361064891846922,
+      "loss": 1.0489603281021118,
+      "loss_ce": 0.0023294282145798206,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0791015625,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 25633892,
+      "step": 409
+    },
+    {
+      "epoch": 1.3643926788685525,
+      "grad_norm": 12.66992473602295,
+      "learning_rate": 5e-06,
+      "loss": 1.148,
+      "num_input_tokens_seen": 25697340,
+      "step": 410
+    },
+    {
+      "epoch": 1.3643926788685525,
+      "loss": 1.0088579654693604,
+      "loss_ce": 0.003456423059105873,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.08544921875,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 25697340,
+      "step": 410
+    },
+    {
+      "epoch": 1.3677204658901831,
+      "grad_norm": 6.634986400604248,
+      "learning_rate": 5e-06,
+      "loss": 0.8108,
+      "num_input_tokens_seen": 25758432,
+      "step": 411
+    },
+    {
+      "epoch": 1.3677204658901831,
+      "loss": 0.9283370971679688,
+      "loss_ce": 0.00426485575735569,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.07763671875,
+      "loss_xval": 0.92578125,
+      "num_input_tokens_seen": 25758432,
+      "step": 411
+    },
+    {
+      "epoch": 1.3710482529118138,
+      "grad_norm": 10.549354553222656,
+      "learning_rate": 5e-06,
+      "loss": 1.0254,
+      "num_input_tokens_seen": 25822248,
+      "step": 412
+    },
+    {
+      "epoch": 1.3710482529118138,
+      "loss": 0.9352825284004211,
+      "loss_ce": 0.005350937135517597,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 25822248,
+      "step": 412
+    },
+    {
+      "epoch": 1.3743760399334444,
+      "grad_norm": 14.273979187011719,
+      "learning_rate": 5e-06,
+      "loss": 1.1039,
+      "num_input_tokens_seen": 25884780,
+      "step": 413
+    },
+    {
+      "epoch": 1.3743760399334444,
+      "loss": 1.2133036851882935,
+      "loss_ce": 0.0009013470844365656,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.072265625,
+      "loss_xval": 1.2109375,
+      "num_input_tokens_seen": 25884780,
+      "step": 413
+    },
+    {
+      "epoch": 1.377703826955075,
+      "grad_norm": 14.600478172302246,
+      "learning_rate": 5e-06,
+      "loss": 0.9331,
+      "num_input_tokens_seen": 25946912,
+      "step": 414
+    },
+    {
+      "epoch": 1.377703826955075,
+      "loss": 0.6742814183235168,
+      "loss_ce": 0.00045326852705329657,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 25946912,
+      "step": 414
+    },
+    {
+      "epoch": 1.3810316139767056,
+      "grad_norm": 11.724282264709473,
+      "learning_rate": 5e-06,
+      "loss": 1.0044,
+      "num_input_tokens_seen": 26009668,
+      "step": 415
+    },
+    {
+      "epoch": 1.3810316139767056,
+      "loss": 1.0532835721969604,
+      "loss_ce": 6.0868831496918574e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0771484375,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 26009668,
+      "step": 415
+    },
+    {
+      "epoch": 1.3843594009983362,
+      "grad_norm": 10.00571346282959,
+      "learning_rate": 5e-06,
+      "loss": 0.9009,
+      "num_input_tokens_seen": 26071008,
+      "step": 416
+    },
+    {
+      "epoch": 1.3843594009983362,
+      "loss": 1.1453657150268555,
+      "loss_ce": 0.044779855757951736,
+      "loss_iou": 0.3125,
+      "loss_num": 0.09521484375,
+      "loss_xval": 1.1015625,
+      "num_input_tokens_seen": 26071008,
+      "step": 416
+    },
+    {
+      "epoch": 1.3876871880199668,
+      "grad_norm": 15.58684253692627,
+      "learning_rate": 5e-06,
+      "loss": 1.1956,
+      "num_input_tokens_seen": 26132352,
+      "step": 417
+    },
+    {
+      "epoch": 1.3876871880199668,
+      "loss": 0.8832725882530212,
+      "loss_ce": 0.00046005373587831855,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.08154296875,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 26132352,
+      "step": 417
+    },
+    {
+      "epoch": 1.3910149750415974,
+      "grad_norm": 19.847135543823242,
+      "learning_rate": 5e-06,
+      "loss": 0.9647,
+      "num_input_tokens_seen": 26195748,
+      "step": 418
+    },
+    {
+      "epoch": 1.3910149750415974,
+      "loss": 0.9432509541511536,
+      "loss_ce": 0.0030654240399599075,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 26195748,
+      "step": 418
+    },
+    {
+      "epoch": 1.394342762063228,
+      "grad_norm": 34.97438049316406,
+      "learning_rate": 5e-06,
+      "loss": 1.1494,
+      "num_input_tokens_seen": 26257164,
+      "step": 419
+    },
+    {
+      "epoch": 1.394342762063228,
+      "loss": 1.2391083240509033,
+      "loss_ce": 9.462250454816967e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.08349609375,
+      "loss_xval": 1.2421875,
+      "num_input_tokens_seen": 26257164,
+      "step": 419
+    },
+    {
+      "epoch": 1.3976705490848587,
+      "grad_norm": 14.589425086975098,
+      "learning_rate": 5e-06,
+      "loss": 1.1584,
+      "num_input_tokens_seen": 26320716,
+      "step": 420
+    },
+    {
+      "epoch": 1.3976705490848587,
+      "loss": 1.2718900442123413,
+      "loss_ce": 0.004800152964890003,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.10107421875,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 26320716,
+      "step": 420
+    },
+    {
+      "epoch": 1.4009983361064893,
+      "grad_norm": 13.36715316772461,
+      "learning_rate": 5e-06,
+      "loss": 1.1376,
+      "num_input_tokens_seen": 26384520,
+      "step": 421
+    },
+    {
+      "epoch": 1.4009983361064893,
+      "loss": 0.8977671265602112,
+      "loss_ce": 0.0003061619936488569,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 26384520,
+      "step": 421
+    },
+    {
+      "epoch": 1.4043261231281199,
+      "grad_norm": 23.56789779663086,
+      "learning_rate": 5e-06,
+      "loss": 1.0789,
+      "num_input_tokens_seen": 26447724,
+      "step": 422
+    },
+    {
+      "epoch": 1.4043261231281199,
+      "loss": 1.1214494705200195,
+      "loss_ce": 0.003285370534285903,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.07080078125,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 26447724,
+      "step": 422
+    },
+    {
+      "epoch": 1.4076539101497505,
+      "grad_norm": 17.013761520385742,
+      "learning_rate": 5e-06,
+      "loss": 1.0713,
+      "num_input_tokens_seen": 26510680,
+      "step": 423
+    },
+    {
+      "epoch": 1.4076539101497505,
+      "loss": 1.0345571041107178,
+      "loss_ce": 0.00037738040555268526,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.09130859375,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 26510680,
+      "step": 423
+    },
+    {
+      "epoch": 1.410981697171381,
+      "grad_norm": 41.10530471801758,
+      "learning_rate": 5e-06,
+      "loss": 1.1384,
+      "num_input_tokens_seen": 26574016,
+      "step": 424
+    },
+    {
+      "epoch": 1.410981697171381,
+      "loss": 1.054764747619629,
+      "loss_ce": 7.73034553276375e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.07666015625,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 26574016,
+      "step": 424
+    },
+    {
+      "epoch": 1.4143094841930117,
+      "grad_norm": 81.72950744628906,
+      "learning_rate": 5e-06,
+      "loss": 1.1902,
+      "num_input_tokens_seen": 26637704,
+      "step": 425
+    },
+    {
+      "epoch": 1.4143094841930117,
+      "loss": 0.9354544878005981,
+      "loss_ce": 0.0003958155866712332,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 26637704,
+      "step": 425
+    },
+    {
+      "epoch": 1.4176372712146423,
+      "grad_norm": 29.790283203125,
+      "learning_rate": 5e-06,
+      "loss": 1.1938,
+      "num_input_tokens_seen": 26701468,
+      "step": 426
+    },
+    {
+      "epoch": 1.4176372712146423,
+      "loss": 0.8883048892021179,
+      "loss_ce": 0.002562721725553274,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.08544921875,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 26701468,
+      "step": 426
+    },
+    {
+      "epoch": 1.420965058236273,
+      "grad_norm": 16.44548797607422,
+      "learning_rate": 5e-06,
+      "loss": 1.1486,
+      "num_input_tokens_seen": 26764044,
+      "step": 427
+    },
+    {
+      "epoch": 1.420965058236273,
+      "loss": 1.1505101919174194,
+      "loss_ce": 0.0006078255828469992,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0947265625,
+      "loss_xval": 1.1484375,
+      "num_input_tokens_seen": 26764044,
+      "step": 427
+    },
+    {
+      "epoch": 1.4242928452579036,
+      "grad_norm": 194.185791015625,
+      "learning_rate": 5e-06,
+      "loss": 0.9832,
+      "num_input_tokens_seen": 26826320,
+      "step": 428
+    },
+    {
+      "epoch": 1.4242928452579036,
+      "loss": 0.9561614990234375,
+      "loss_ce": 0.004989705514162779,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.10546875,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 26826320,
+      "step": 428
+    },
+    {
+      "epoch": 1.4276206322795342,
+      "grad_norm": 27.536348342895508,
+      "learning_rate": 5e-06,
+      "loss": 1.0146,
+      "num_input_tokens_seen": 26889240,
+      "step": 429
+    },
+    {
+      "epoch": 1.4276206322795342,
+      "loss": 1.2331770658493042,
+      "loss_ce": 0.004661452490836382,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0830078125,
+      "loss_xval": 1.2265625,
+      "num_input_tokens_seen": 26889240,
+      "step": 429
+    },
+    {
+      "epoch": 1.4309484193011648,
+      "grad_norm": 23.240528106689453,
+      "learning_rate": 5e-06,
+      "loss": 1.2288,
+      "num_input_tokens_seen": 26952020,
+      "step": 430
+    },
+    {
+      "epoch": 1.4309484193011648,
+      "loss": 1.34483003616333,
+      "loss_ce": 0.00010352435492677614,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.09912109375,
+      "loss_xval": 1.34375,
+      "num_input_tokens_seen": 26952020,
+      "step": 430
+    },
+    {
+      "epoch": 1.4342762063227954,
+      "grad_norm": 14.340412139892578,
+      "learning_rate": 5e-06,
+      "loss": 1.1576,
+      "num_input_tokens_seen": 27015052,
+      "step": 431
+    },
+    {
+      "epoch": 1.4342762063227954,
+      "loss": 1.1539814472198486,
+      "loss_ce": 0.0004169994790572673,
+      "loss_iou": 0.375,
+      "loss_num": 0.0810546875,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 27015052,
+      "step": 431
+    },
+    {
+      "epoch": 1.437603993344426,
+      "grad_norm": 13.505350112915039,
+      "learning_rate": 5e-06,
+      "loss": 0.9107,
+      "num_input_tokens_seen": 27077640,
+      "step": 432
+    },
+    {
+      "epoch": 1.437603993344426,
+      "loss": 0.8342275023460388,
+      "loss_ce": 0.0010976643534377217,
+      "loss_iou": 0.265625,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 27077640,
+      "step": 432
+    },
+    {
+      "epoch": 1.4409317803660566,
+      "grad_norm": 15.690736770629883,
+      "learning_rate": 5e-06,
+      "loss": 0.8302,
+      "num_input_tokens_seen": 27140080,
+      "step": 433
+    },
+    {
+      "epoch": 1.4409317803660566,
+      "loss": 0.9211246967315674,
+      "loss_ce": 0.0016911044949665666,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 27140080,
+      "step": 433
+    },
+    {
+      "epoch": 1.4442595673876872,
+      "grad_norm": 13.289794921875,
+      "learning_rate": 5e-06,
+      "loss": 0.8306,
+      "num_input_tokens_seen": 27203336,
+      "step": 434
+    },
+    {
+      "epoch": 1.4442595673876872,
+      "loss": 0.7346397042274475,
+      "loss_ce": 0.00050883594667539,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 27203336,
+      "step": 434
+    },
+    {
+      "epoch": 1.4475873544093179,
+      "grad_norm": 11.401854515075684,
+      "learning_rate": 5e-06,
+      "loss": 0.9116,
+      "num_input_tokens_seen": 27265880,
+      "step": 435
+    },
+    {
+      "epoch": 1.4475873544093179,
+      "loss": 0.9152499437332153,
+      "loss_ce": 0.0006992316339164972,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 27265880,
+      "step": 435
+    },
+    {
+      "epoch": 1.4509151414309485,
+      "grad_norm": 10.400731086730957,
+      "learning_rate": 5e-06,
+      "loss": 1.1403,
+      "num_input_tokens_seen": 27329740,
+      "step": 436
+    },
+    {
+      "epoch": 1.4509151414309485,
+      "loss": 1.0320229530334473,
+      "loss_ce": 0.0005287755047902465,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.068359375,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 27329740,
+      "step": 436
+    },
+    {
+      "epoch": 1.454242928452579,
+      "grad_norm": 18.541240692138672,
+      "learning_rate": 5e-06,
+      "loss": 0.7472,
+      "num_input_tokens_seen": 27392388,
+      "step": 437
+    },
+    {
+      "epoch": 1.454242928452579,
+      "loss": 0.6203223466873169,
+      "loss_ce": 0.0013037655735388398,
+      "loss_iou": 0.130859375,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 27392388,
+      "step": 437
+    },
+    {
+      "epoch": 1.4575707154742097,
+      "grad_norm": 22.43004035949707,
+      "learning_rate": 5e-06,
+      "loss": 0.9605,
+      "num_input_tokens_seen": 27453448,
+      "step": 438
+    },
+    {
+      "epoch": 1.4575707154742097,
+      "loss": 0.6307517290115356,
+      "loss_ce": 0.00013653574569616467,
+      "loss_iou": 0.103515625,
+      "loss_num": 0.08447265625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 27453448,
+      "step": 438
+    },
+    {
+      "epoch": 1.4608985024958403,
+      "grad_norm": 13.829190254211426,
+      "learning_rate": 5e-06,
+      "loss": 0.9866,
+      "num_input_tokens_seen": 27515228,
+      "step": 439
+    },
+    {
+      "epoch": 1.4608985024958403,
+      "loss": 0.9368977546691895,
+      "loss_ce": 0.004280570894479752,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 27515228,
+      "step": 439
+    },
+    {
+      "epoch": 1.464226289517471,
+      "grad_norm": 14.281774520874023,
+      "learning_rate": 5e-06,
+      "loss": 0.8164,
+      "num_input_tokens_seen": 27577956,
+      "step": 440
+    },
+    {
+      "epoch": 1.464226289517471,
+      "loss": 0.7943712472915649,
+      "loss_ce": 0.002379045821726322,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 27577956,
+      "step": 440
+    },
+    {
+      "epoch": 1.4675540765391015,
+      "grad_norm": 11.133589744567871,
+      "learning_rate": 5e-06,
+      "loss": 1.0627,
+      "num_input_tokens_seen": 27641140,
+      "step": 441
+    },
+    {
+      "epoch": 1.4675540765391015,
+      "loss": 1.2072147130966187,
+      "loss_ce": 0.0011600162833929062,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.09814453125,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 27641140,
+      "step": 441
+    },
+    {
+      "epoch": 1.4708818635607321,
+      "grad_norm": 15.000788688659668,
+      "learning_rate": 5e-06,
+      "loss": 0.919,
+      "num_input_tokens_seen": 27702864,
+      "step": 442
+    },
+    {
+      "epoch": 1.4708818635607321,
+      "loss": 1.0833325386047363,
+      "loss_ce": 0.0015454718377441168,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0751953125,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 27702864,
+      "step": 442
+    },
+    {
+      "epoch": 1.4742096505823628,
+      "grad_norm": 10.785703659057617,
+      "learning_rate": 5e-06,
+      "loss": 0.9502,
+      "num_input_tokens_seen": 27766516,
+      "step": 443
+    },
+    {
+      "epoch": 1.4742096505823628,
+      "loss": 0.9589688777923584,
+      "loss_ce": 0.0007169640157371759,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.95703125,
+      "num_input_tokens_seen": 27766516,
+      "step": 443
+    },
+    {
+      "epoch": 1.4775374376039934,
+      "grad_norm": 14.798559188842773,
+      "learning_rate": 5e-06,
+      "loss": 1.033,
+      "num_input_tokens_seen": 27829568,
+      "step": 444
+    },
+    {
+      "epoch": 1.4775374376039934,
+      "loss": 1.023010015487671,
+      "loss_ce": 0.00030495758983306587,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0634765625,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 27829568,
+      "step": 444
+    },
+    {
+      "epoch": 1.480865224625624,
+      "grad_norm": 11.827593803405762,
+      "learning_rate": 5e-06,
+      "loss": 0.6733,
+      "num_input_tokens_seen": 27892536,
+      "step": 445
+    },
+    {
+      "epoch": 1.480865224625624,
+      "loss": 0.5998960733413696,
+      "loss_ce": 0.0006834049127064645,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 27892536,
+      "step": 445
+    },
+    {
+      "epoch": 1.4841930116472546,
+      "grad_norm": 12.027179718017578,
+      "learning_rate": 5e-06,
+      "loss": 0.9374,
+      "num_input_tokens_seen": 27954480,
+      "step": 446
+    },
+    {
+      "epoch": 1.4841930116472546,
+      "loss": 0.8107286095619202,
+      "loss_ce": 0.005674911662936211,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 27954480,
+      "step": 446
+    },
+    {
+      "epoch": 1.4875207986688852,
+      "grad_norm": 12.46907901763916,
+      "learning_rate": 5e-06,
+      "loss": 0.9105,
+      "num_input_tokens_seen": 28015384,
+      "step": 447
+    },
+    {
+      "epoch": 1.4875207986688852,
+      "loss": 1.007506012916565,
+      "loss_ce": 0.00018174726574216038,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.10302734375,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 28015384,
+      "step": 447
+    },
+    {
+      "epoch": 1.4908485856905158,
+      "grad_norm": 19.959171295166016,
+      "learning_rate": 5e-06,
+      "loss": 0.984,
+      "num_input_tokens_seen": 28077524,
+      "step": 448
+    },
+    {
+      "epoch": 1.4908485856905158,
+      "loss": 1.0535316467285156,
+      "loss_ce": 0.0010414018761366606,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.08837890625,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 28077524,
+      "step": 448
+    },
+    {
+      "epoch": 1.4941763727121464,
+      "grad_norm": 20.844703674316406,
+      "learning_rate": 5e-06,
+      "loss": 0.9459,
+      "num_input_tokens_seen": 28139404,
+      "step": 449
+    },
+    {
+      "epoch": 1.4941763727121464,
+      "loss": 0.9492704272270203,
+      "loss_ce": 0.0007841180195100605,
+      "loss_iou": 0.21875,
+      "loss_num": 0.10205078125,
+      "loss_xval": 0.94921875,
+      "num_input_tokens_seen": 28139404,
+      "step": 449
+    },
+    {
+      "epoch": 1.497504159733777,
+      "grad_norm": 25.41834831237793,
+      "learning_rate": 5e-06,
+      "loss": 1.3058,
+      "num_input_tokens_seen": 28201620,
+      "step": 450
+    },
+    {
+      "epoch": 1.497504159733777,
+      "loss": 1.2704012393951416,
+      "loss_ce": 0.0028230701573193073,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0830078125,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 28201620,
+      "step": 450
+    },
+    {
+      "epoch": 1.5008319467554077,
+      "grad_norm": 35.88543701171875,
+      "learning_rate": 5e-06,
+      "loss": 1.1012,
+      "num_input_tokens_seen": 28265268,
+      "step": 451
+    },
+    {
+      "epoch": 1.5008319467554077,
+      "loss": 0.7575756311416626,
+      "loss_ce": 0.001472145551815629,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 28265268,
+      "step": 451
+    },
+    {
+      "epoch": 1.5041597337770383,
+      "grad_norm": 9.400871276855469,
+      "learning_rate": 5e-06,
+      "loss": 0.7297,
+      "num_input_tokens_seen": 28328344,
+      "step": 452
+    },
+    {
+      "epoch": 1.5041597337770383,
+      "loss": 0.6579368114471436,
+      "loss_ce": 0.0014427044661715627,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 28328344,
+      "step": 452
+    },
+    {
+      "epoch": 1.5074875207986689,
+      "grad_norm": 16.661739349365234,
+      "learning_rate": 5e-06,
+      "loss": 1.0671,
+      "num_input_tokens_seen": 28390980,
+      "step": 453
+    },
+    {
+      "epoch": 1.5074875207986689,
+      "loss": 1.226120114326477,
+      "loss_ce": 0.0010225145379081368,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.1064453125,
+      "loss_xval": 1.2265625,
+      "num_input_tokens_seen": 28390980,
+      "step": 453
+    },
+    {
+      "epoch": 1.5108153078202995,
+      "grad_norm": 16.2419490814209,
+      "learning_rate": 5e-06,
+      "loss": 0.9537,
+      "num_input_tokens_seen": 28453244,
+      "step": 454
+    },
+    {
+      "epoch": 1.5108153078202995,
+      "loss": 0.9552655816078186,
+      "loss_ce": 0.0018964293412864208,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.1123046875,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 28453244,
+      "step": 454
+    },
+    {
+      "epoch": 1.51414309484193,
+      "grad_norm": 22.404499053955078,
+      "learning_rate": 5e-06,
+      "loss": 0.8511,
+      "num_input_tokens_seen": 28515280,
+      "step": 455
+    },
+    {
+      "epoch": 1.51414309484193,
+      "loss": 0.9287418127059937,
+      "loss_ce": 0.006378548685461283,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 28515280,
+      "step": 455
+    },
+    {
+      "epoch": 1.5174708818635607,
+      "grad_norm": 23.939455032348633,
+      "learning_rate": 5e-06,
+      "loss": 1.3932,
+      "num_input_tokens_seen": 28578200,
+      "step": 456
+    },
+    {
+      "epoch": 1.5174708818635607,
+      "loss": 1.485808253288269,
+      "loss_ce": 0.004362954758107662,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.10498046875,
+      "loss_xval": 1.484375,
+      "num_input_tokens_seen": 28578200,
+      "step": 456
+    },
+    {
+      "epoch": 1.5207986688851913,
+      "grad_norm": 19.551912307739258,
+      "learning_rate": 5e-06,
+      "loss": 0.9394,
+      "num_input_tokens_seen": 28642336,
+      "step": 457
+    },
+    {
+      "epoch": 1.5207986688851913,
+      "loss": 0.8611233234405518,
+      "loss_ce": 3.935792119591497e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 28642336,
+      "step": 457
+    },
+    {
+      "epoch": 1.524126455906822,
+      "grad_norm": 33.463558197021484,
+      "learning_rate": 5e-06,
+      "loss": 1.3023,
+      "num_input_tokens_seen": 28705660,
+      "step": 458
+    },
+    {
+      "epoch": 1.524126455906822,
+      "loss": 1.4212629795074463,
+      "loss_ce": 0.0020735724829137325,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.1123046875,
+      "loss_xval": 1.421875,
+      "num_input_tokens_seen": 28705660,
+      "step": 458
+    },
+    {
+      "epoch": 1.5274542429284526,
+      "grad_norm": 32.06713104248047,
+      "learning_rate": 5e-06,
+      "loss": 0.8497,
+      "num_input_tokens_seen": 28768376,
+      "step": 459
+    },
+    {
+      "epoch": 1.5274542429284526,
+      "loss": 0.9550895690917969,
+      "loss_ce": 0.0004996892530471087,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 28768376,
+      "step": 459
+    },
+    {
+      "epoch": 1.5307820299500832,
+      "grad_norm": 21.96576690673828,
+      "learning_rate": 5e-06,
+      "loss": 0.8911,
+      "num_input_tokens_seen": 28831008,
+      "step": 460
+    },
+    {
+      "epoch": 1.5307820299500832,
+      "loss": 0.796554684638977,
+      "loss_ce": 0.0021211060229688883,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.0810546875,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 28831008,
+      "step": 460
+    },
+    {
+      "epoch": 1.5341098169717138,
+      "grad_norm": 12.529973030090332,
+      "learning_rate": 5e-06,
+      "loss": 1.0927,
+      "num_input_tokens_seen": 28894052,
+      "step": 461
+    },
+    {
+      "epoch": 1.5341098169717138,
+      "loss": 0.7951915264129639,
+      "loss_ce": 0.004420042969286442,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.0986328125,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 28894052,
+      "step": 461
+    },
+    {
+      "epoch": 1.5374376039933444,
+      "grad_norm": 21.059383392333984,
+      "learning_rate": 5e-06,
+      "loss": 0.8035,
+      "num_input_tokens_seen": 28956284,
+      "step": 462
+    },
+    {
+      "epoch": 1.5374376039933444,
+      "loss": 0.6268208026885986,
+      "loss_ce": 0.001698711421340704,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.05126953125,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 28956284,
+      "step": 462
+    },
+    {
+      "epoch": 1.540765391014975,
+      "grad_norm": 17.505203247070312,
+      "learning_rate": 5e-06,
+      "loss": 1.2985,
+      "num_input_tokens_seen": 29020736,
+      "step": 463
+    },
+    {
+      "epoch": 1.540765391014975,
+      "loss": 1.3152070045471191,
+      "loss_ce": 0.0017303972272202373,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.09228515625,
+      "loss_xval": 1.3125,
+      "num_input_tokens_seen": 29020736,
+      "step": 463
+    },
+    {
+      "epoch": 1.5440931780366056,
+      "grad_norm": 13.60715389251709,
+      "learning_rate": 5e-06,
+      "loss": 0.8746,
+      "num_input_tokens_seen": 29082664,
+      "step": 464
+    },
+    {
+      "epoch": 1.5440931780366056,
+      "loss": 1.051095962524414,
+      "loss_ce": 0.002511953003704548,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 29082664,
+      "step": 464
+    },
+    {
+      "epoch": 1.5474209650582362,
+      "grad_norm": 22.727663040161133,
+      "learning_rate": 5e-06,
+      "loss": 0.8494,
+      "num_input_tokens_seen": 29146680,
+      "step": 465
+    },
+    {
+      "epoch": 1.5474209650582362,
+      "loss": 1.0247268676757812,
+      "loss_ce": 0.0010452390415593982,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.06884765625,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 29146680,
+      "step": 465
+    },
+    {
+      "epoch": 1.5507487520798668,
+      "grad_norm": 19.767471313476562,
+      "learning_rate": 5e-06,
+      "loss": 0.9335,
+      "num_input_tokens_seen": 29210188,
+      "step": 466
+    },
+    {
+      "epoch": 1.5507487520798668,
+      "loss": 1.0263903141021729,
+      "loss_ce": 0.00368527346290648,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 29210188,
+      "step": 466
+    },
+    {
+      "epoch": 1.5540765391014975,
+      "grad_norm": 11.171165466308594,
+      "learning_rate": 5e-06,
+      "loss": 0.9697,
+      "num_input_tokens_seen": 29272836,
+      "step": 467
+    },
+    {
+      "epoch": 1.5540765391014975,
+      "loss": 0.898423969745636,
+      "loss_ce": 0.006089954171329737,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.07763671875,
+      "loss_xval": 0.890625,
+      "num_input_tokens_seen": 29272836,
+      "step": 467
+    },
+    {
+      "epoch": 1.557404326123128,
+      "grad_norm": 12.950803756713867,
+      "learning_rate": 5e-06,
+      "loss": 0.8517,
+      "num_input_tokens_seen": 29334744,
+      "step": 468
+    },
+    {
+      "epoch": 1.557404326123128,
+      "loss": 1.0840175151824951,
+      "loss_ce": 0.008089832961559296,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.08203125,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 29334744,
+      "step": 468
+    },
+    {
+      "epoch": 1.5607321131447587,
+      "grad_norm": 43.52952194213867,
+      "learning_rate": 5e-06,
+      "loss": 1.2985,
+      "num_input_tokens_seen": 29398080,
+      "step": 469
+    },
+    {
+      "epoch": 1.5607321131447587,
+      "loss": 1.0282366275787354,
+      "loss_ce": 0.003090148326009512,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0927734375,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 29398080,
+      "step": 469
+    },
+    {
+      "epoch": 1.5640599001663893,
+      "grad_norm": 49.79926681518555,
+      "learning_rate": 5e-06,
+      "loss": 0.9525,
+      "num_input_tokens_seen": 29461576,
+      "step": 470
+    },
+    {
+      "epoch": 1.5640599001663893,
+      "loss": 0.6735060214996338,
+      "loss_ce": 0.0019971991423517466,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 29461576,
+      "step": 470
+    },
+    {
+      "epoch": 1.56738768718802,
+      "grad_norm": 13.252382278442383,
+      "learning_rate": 5e-06,
+      "loss": 0.929,
+      "num_input_tokens_seen": 29524476,
+      "step": 471
+    },
+    {
+      "epoch": 1.56738768718802,
+      "loss": 0.9817473888397217,
+      "loss_ce": 0.002011068630963564,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.98046875,
+      "num_input_tokens_seen": 29524476,
+      "step": 471
+    },
+    {
+      "epoch": 1.5707154742096505,
+      "grad_norm": 14.923233985900879,
+      "learning_rate": 5e-06,
+      "loss": 1.1227,
+      "num_input_tokens_seen": 29587796,
+      "step": 472
+    },
+    {
+      "epoch": 1.5707154742096505,
+      "loss": 1.3451569080352783,
+      "loss_ce": 0.00043034314876422286,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0908203125,
+      "loss_xval": 1.34375,
+      "num_input_tokens_seen": 29587796,
+      "step": 472
+    },
+    {
+      "epoch": 1.5740432612312811,
+      "grad_norm": 15.69069766998291,
+      "learning_rate": 5e-06,
+      "loss": 0.9072,
+      "num_input_tokens_seen": 29650672,
+      "step": 473
+    },
+    {
+      "epoch": 1.5740432612312811,
+      "loss": 0.7996507883071899,
+      "loss_ce": 0.003264050930738449,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 29650672,
+      "step": 473
+    },
+    {
+      "epoch": 1.5773710482529117,
+      "grad_norm": 28.606983184814453,
+      "learning_rate": 5e-06,
+      "loss": 0.9883,
+      "num_input_tokens_seen": 29714604,
+      "step": 474
+    },
+    {
+      "epoch": 1.5773710482529117,
+      "loss": 0.6658027768135071,
+      "loss_ce": 0.0002754491288214922,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 29714604,
+      "step": 474
+    },
+    {
+      "epoch": 1.5806988352745424,
+      "grad_norm": 14.635608673095703,
+      "learning_rate": 5e-06,
+      "loss": 0.9022,
+      "num_input_tokens_seen": 29777760,
+      "step": 475
+    },
+    {
+      "epoch": 1.5806988352745424,
+      "loss": 0.7488625049591064,
+      "loss_ce": 0.012534376233816147,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 29777760,
+      "step": 475
+    },
+    {
+      "epoch": 1.584026622296173,
+      "grad_norm": 23.2712459564209,
+      "learning_rate": 5e-06,
+      "loss": 0.7148,
+      "num_input_tokens_seen": 29838992,
+      "step": 476
+    },
+    {
+      "epoch": 1.584026622296173,
+      "loss": 0.6552135944366455,
+      "loss_ce": 0.00030633312417194247,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 29838992,
+      "step": 476
+    },
+    {
+      "epoch": 1.5873544093178036,
+      "grad_norm": 15.995981216430664,
+      "learning_rate": 5e-06,
+      "loss": 0.807,
+      "num_input_tokens_seen": 29902632,
+      "step": 477
+    },
+    {
+      "epoch": 1.5873544093178036,
+      "loss": 0.7012355327606201,
+      "loss_ce": 0.0023830283898860216,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.69921875,
+      "num_input_tokens_seen": 29902632,
+      "step": 477
+    },
+    {
+      "epoch": 1.5906821963394342,
+      "grad_norm": 17.334726333618164,
+      "learning_rate": 5e-06,
+      "loss": 0.8927,
+      "num_input_tokens_seen": 29965592,
+      "step": 478
+    },
+    {
+      "epoch": 1.5906821963394342,
+      "loss": 1.0914649963378906,
+      "loss_ce": 0.0006445984472520649,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.080078125,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 29965592,
+      "step": 478
+    },
+    {
+      "epoch": 1.5940099833610648,
+      "grad_norm": 22.14893913269043,
+      "learning_rate": 5e-06,
+      "loss": 0.9355,
+      "num_input_tokens_seen": 30029944,
+      "step": 479
+    },
+    {
+      "epoch": 1.5940099833610648,
+      "loss": 0.8586336374282837,
+      "loss_ce": 0.0011202013120055199,
+      "loss_iou": 0.296875,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 30029944,
+      "step": 479
+    },
+    {
+      "epoch": 1.5973377703826954,
+      "grad_norm": 15.957133293151855,
+      "learning_rate": 5e-06,
+      "loss": 0.7887,
+      "num_input_tokens_seen": 30091948,
+      "step": 480
+    },
+    {
+      "epoch": 1.5973377703826954,
+      "loss": 0.7341344356536865,
+      "loss_ce": 0.0012242539087310433,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 30091948,
+      "step": 480
+    },
+    {
+      "epoch": 1.600665557404326,
+      "grad_norm": 12.82165241241455,
+      "learning_rate": 5e-06,
+      "loss": 0.9523,
+      "num_input_tokens_seen": 30153712,
+      "step": 481
+    },
+    {
+      "epoch": 1.600665557404326,
+      "loss": 0.8892770409584045,
+      "loss_ce": 0.0006051433738321066,
+      "loss_iou": 0.328125,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.890625,
+      "num_input_tokens_seen": 30153712,
+      "step": 481
+    },
+    {
+      "epoch": 1.6039933444259566,
+      "grad_norm": 16.93235206604004,
+      "learning_rate": 5e-06,
+      "loss": 1.0522,
+      "num_input_tokens_seen": 30216844,
+      "step": 482
+    },
+    {
+      "epoch": 1.6039933444259566,
+      "loss": 1.1036779880523682,
+      "loss_ce": 0.0011389621067792177,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.11376953125,
+      "loss_xval": 1.1015625,
+      "num_input_tokens_seen": 30216844,
+      "step": 482
+    },
+    {
+      "epoch": 1.6073211314475873,
+      "grad_norm": 12.133223533630371,
+      "learning_rate": 5e-06,
+      "loss": 0.8506,
+      "num_input_tokens_seen": 30278368,
+      "step": 483
+    },
+    {
+      "epoch": 1.6073211314475873,
+      "loss": 0.9850232601165771,
+      "loss_ce": 0.0001599617680767551,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 30278368,
+      "step": 483
+    },
+    {
+      "epoch": 1.6106489184692179,
+      "grad_norm": 44.32785415649414,
+      "learning_rate": 5e-06,
+      "loss": 0.9432,
+      "num_input_tokens_seen": 30339888,
+      "step": 484
+    },
+    {
+      "epoch": 1.6106489184692179,
+      "loss": 1.1375561952590942,
+      "loss_ce": 0.00022712742793373764,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.10498046875,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 30339888,
+      "step": 484
+    },
+    {
+      "epoch": 1.6139767054908485,
+      "grad_norm": 10.50809383392334,
+      "learning_rate": 5e-06,
+      "loss": 0.9693,
+      "num_input_tokens_seen": 30402144,
+      "step": 485
+    },
+    {
+      "epoch": 1.6139767054908485,
+      "loss": 0.9972724914550781,
+      "loss_ce": 0.000446375401224941,
+      "loss_iou": 0.328125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.99609375,
+      "num_input_tokens_seen": 30402144,
+      "step": 485
+    },
+    {
+      "epoch": 1.617304492512479,
+      "grad_norm": 15.496356010437012,
+      "learning_rate": 5e-06,
+      "loss": 1.2141,
+      "num_input_tokens_seen": 30465452,
+      "step": 486
+    },
+    {
+      "epoch": 1.617304492512479,
+      "loss": 1.3970156908035278,
+      "loss_ce": 0.0010195414070039988,
+      "loss_iou": 0.421875,
+      "loss_num": 0.1103515625,
+      "loss_xval": 1.3984375,
+      "num_input_tokens_seen": 30465452,
+      "step": 486
+    },
+    {
+      "epoch": 1.6206322795341097,
+      "grad_norm": 36.90744400024414,
+      "learning_rate": 5e-06,
+      "loss": 1.0141,
+      "num_input_tokens_seen": 30529608,
+      "step": 487
+    },
+    {
+      "epoch": 1.6206322795341097,
+      "loss": 1.0115238428115845,
+      "loss_ce": 0.0037113595753908157,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.05517578125,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 30529608,
+      "step": 487
+    },
+    {
+      "epoch": 1.6239600665557403,
+      "grad_norm": 20.59255027770996,
+      "learning_rate": 5e-06,
+      "loss": 1.0354,
+      "num_input_tokens_seen": 30591308,
+      "step": 488
+    },
+    {
+      "epoch": 1.6239600665557403,
+      "loss": 1.2096441984176636,
+      "loss_ce": 0.0016364282928407192,
+      "loss_iou": 0.375,
+      "loss_num": 0.091796875,
+      "loss_xval": 1.2109375,
+      "num_input_tokens_seen": 30591308,
+      "step": 488
+    },
+    {
+      "epoch": 1.627287853577371,
+      "grad_norm": 24.88413429260254,
+      "learning_rate": 5e-06,
+      "loss": 0.9571,
+      "num_input_tokens_seen": 30654808,
+      "step": 489
+    },
+    {
+      "epoch": 1.627287853577371,
+      "loss": 0.9968461394309998,
+      "loss_ce": 0.003926191478967667,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 30654808,
+      "step": 489
+    },
+    {
+      "epoch": 1.6306156405990015,
+      "grad_norm": 24.321149826049805,
+      "learning_rate": 5e-06,
+      "loss": 1.1118,
+      "num_input_tokens_seen": 30719144,
+      "step": 490
+    },
+    {
+      "epoch": 1.6306156405990015,
+      "loss": 0.9674986600875854,
+      "loss_ce": 0.00192247552331537,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.96484375,
+      "num_input_tokens_seen": 30719144,
+      "step": 490
+    },
+    {
+      "epoch": 1.6339434276206322,
+      "grad_norm": 18.081993103027344,
+      "learning_rate": 5e-06,
+      "loss": 0.905,
+      "num_input_tokens_seen": 30780976,
+      "step": 491
+    },
+    {
+      "epoch": 1.6339434276206322,
+      "loss": 0.8123141527175903,
+      "loss_ce": 0.0009127893717959523,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 30780976,
+      "step": 491
+    },
+    {
+      "epoch": 1.6372712146422628,
+      "grad_norm": 38.994476318359375,
+      "learning_rate": 5e-06,
+      "loss": 0.9762,
+      "num_input_tokens_seen": 30843968,
+      "step": 492
+    },
+    {
+      "epoch": 1.6372712146422628,
+      "loss": 0.8970677852630615,
+      "loss_ce": 9.50983667280525e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 30843968,
+      "step": 492
+    },
+    {
+      "epoch": 1.6405990016638934,
+      "grad_norm": 19.654754638671875,
+      "learning_rate": 5e-06,
+      "loss": 1.1088,
+      "num_input_tokens_seen": 30904740,
+      "step": 493
+    },
+    {
+      "epoch": 1.6405990016638934,
+      "loss": 1.1129732131958008,
+      "loss_ce": 0.0016451734118163586,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.09033203125,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 30904740,
+      "step": 493
+    },
+    {
+      "epoch": 1.643926788685524,
+      "grad_norm": 18.252052307128906,
+      "learning_rate": 5e-06,
+      "loss": 0.7697,
+      "num_input_tokens_seen": 30968000,
+      "step": 494
+    },
+    {
+      "epoch": 1.643926788685524,
+      "loss": 0.8988114595413208,
+      "loss_ce": 0.0008622044115327299,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 30968000,
+      "step": 494
+    },
+    {
+      "epoch": 1.6472545757071546,
+      "grad_norm": 11.877856254577637,
+      "learning_rate": 5e-06,
+      "loss": 0.7778,
+      "num_input_tokens_seen": 31031844,
+      "step": 495
+    },
+    {
+      "epoch": 1.6472545757071546,
+      "loss": 0.8768675327301025,
+      "loss_ce": 0.0011351365828886628,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.08251953125,
+      "loss_xval": 0.875,
+      "num_input_tokens_seen": 31031844,
+      "step": 495
+    },
+    {
+      "epoch": 1.6505823627287852,
+      "grad_norm": 16.35790252685547,
+      "learning_rate": 5e-06,
+      "loss": 1.1891,
+      "num_input_tokens_seen": 31095556,
+      "step": 496
+    },
+    {
+      "epoch": 1.6505823627287852,
+      "loss": 1.4308223724365234,
+      "loss_ce": 0.0001583620032761246,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.1083984375,
+      "loss_xval": 1.4296875,
+      "num_input_tokens_seen": 31095556,
+      "step": 496
+    },
+    {
+      "epoch": 1.6539101497504158,
+      "grad_norm": 12.555450439453125,
+      "learning_rate": 5e-06,
+      "loss": 0.8734,
+      "num_input_tokens_seen": 31158156,
+      "step": 497
+    },
+    {
+      "epoch": 1.6539101497504158,
+      "loss": 0.7806690335273743,
+      "loss_ce": 0.007719771936535835,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 31158156,
+      "step": 497
+    },
+    {
+      "epoch": 1.6572379367720464,
+      "grad_norm": 12.337371826171875,
+      "learning_rate": 5e-06,
+      "loss": 1.1544,
+      "num_input_tokens_seen": 31221496,
+      "step": 498
+    },
+    {
+      "epoch": 1.6572379367720464,
+      "loss": 1.3280014991760254,
+      "loss_ce": 0.0008530584746040404,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0908203125,
+      "loss_xval": 1.328125,
+      "num_input_tokens_seen": 31221496,
+      "step": 498
+    },
+    {
+      "epoch": 1.660565723793677,
+      "grad_norm": 16.143665313720703,
+      "learning_rate": 5e-06,
+      "loss": 1.252,
+      "num_input_tokens_seen": 31285016,
+      "step": 499
+    },
+    {
+      "epoch": 1.660565723793677,
+      "loss": 1.2564626932144165,
+      "loss_ce": 0.0003592088760342449,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.07666015625,
+      "loss_xval": 1.2578125,
+      "num_input_tokens_seen": 31285016,
+      "step": 499
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "grad_norm": 10.787676811218262,
+      "learning_rate": 5e-06,
+      "loss": 0.975,
+      "num_input_tokens_seen": 31348120,
+      "step": 500
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "eval_seeclick_CIoU": 0.13085231557488441,
+      "eval_seeclick_GIoU": 0.1558702141046524,
+      "eval_seeclick_IoU": 0.23901727050542831,
+      "eval_seeclick_MAE_all": 0.18714486062526703,
+      "eval_seeclick_MAE_h": 0.07685794495046139,
+      "eval_seeclick_MAE_w": 0.13620607554912567,
+      "eval_seeclick_MAE_x_boxes": 0.32132112979888916,
+      "eval_seeclick_MAE_y_boxes": 0.09800946339964867,
+      "eval_seeclick_NUM_probability": 0.9999272227287292,
+      "eval_seeclick_inside_bbox": 0.3541666716337204,
+      "eval_seeclick_loss": 2.6794745922088623,
+      "eval_seeclick_loss_ce": 0.06953983008861542,
+      "eval_seeclick_loss_iou": 0.837890625,
+      "eval_seeclick_loss_num": 0.179595947265625,
+      "eval_seeclick_loss_xval": 2.574462890625,
+      "eval_seeclick_runtime": 65.7063,
+      "eval_seeclick_samples_per_second": 0.715,
+      "eval_seeclick_steps_per_second": 0.03,
+      "num_input_tokens_seen": 31348120,
+      "step": 500
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "eval_icons_CIoU": 0.07393881864845753,
+      "eval_icons_GIoU": 0.19662170112133026,
+      "eval_icons_IoU": 0.21981175243854523,
+      "eval_icons_MAE_all": 0.13982237502932549,
+      "eval_icons_MAE_h": 0.07564222812652588,
+      "eval_icons_MAE_w": 0.15035200864076614,
+      "eval_icons_MAE_x_boxes": 0.12051251530647278,
+      "eval_icons_MAE_y_boxes": 0.049114322289824486,
+      "eval_icons_NUM_probability": 0.9999871253967285,
+      "eval_icons_inside_bbox": 0.3420138955116272,
+      "eval_icons_loss": 2.3178176879882812,
+      "eval_icons_loss_ce": 5.301104465615936e-06,
+      "eval_icons_loss_iou": 0.798095703125,
+      "eval_icons_loss_num": 0.14259719848632812,
+      "eval_icons_loss_xval": 2.310546875,
+      "eval_icons_runtime": 65.5349,
+      "eval_icons_samples_per_second": 0.763,
+      "eval_icons_steps_per_second": 0.031,
+      "num_input_tokens_seen": 31348120,
+      "step": 500
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "eval_screenspot_CIoU": 0.03867738569776217,
+      "eval_screenspot_GIoU": 0.1068417305747668,
+      "eval_screenspot_IoU": 0.18955722451210022,
+      "eval_screenspot_MAE_all": 0.20505708952744803,
+      "eval_screenspot_MAE_h": 0.07092766960461934,
+      "eval_screenspot_MAE_w": 0.17401040097077689,
+      "eval_screenspot_MAE_x_boxes": 0.2769670287768046,
+      "eval_screenspot_MAE_y_boxes": 0.1273154765367508,
+      "eval_screenspot_NUM_probability": 0.9997655948003134,
+      "eval_screenspot_inside_bbox": 0.31791667143503827,
+      "eval_screenspot_loss": 2.826397657394409,
+      "eval_screenspot_loss_ce": 0.0007081345732634267,
+      "eval_screenspot_loss_iou": 0.9080403645833334,
+      "eval_screenspot_loss_num": 0.20873006184895834,
+      "eval_screenspot_loss_xval": 2.8587239583333335,
+      "eval_screenspot_runtime": 123.0498,
+      "eval_screenspot_samples_per_second": 0.723,
+      "eval_screenspot_steps_per_second": 0.024,
+      "num_input_tokens_seen": 31348120,
+      "step": 500
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "eval_compot_CIoU": -0.07353978231549263,
+      "eval_compot_GIoU": 0.04984145052731037,
+      "eval_compot_IoU": 0.10792786628007889,
+      "eval_compot_MAE_all": 0.2465325966477394,
+      "eval_compot_MAE_h": 0.08729888498783112,
+      "eval_compot_MAE_w": 0.27648504078388214,
+      "eval_compot_MAE_x_boxes": 0.19644539058208466,
+      "eval_compot_MAE_y_boxes": 0.16193149238824844,
+      "eval_compot_NUM_probability": 0.9998366832733154,
+      "eval_compot_inside_bbox": 0.16840277798473835,
+      "eval_compot_loss": 3.0365827083587646,
+      "eval_compot_loss_ce": 0.005284860031679273,
+      "eval_compot_loss_iou": 0.947509765625,
+      "eval_compot_loss_num": 0.235687255859375,
+      "eval_compot_loss_xval": 3.072265625,
+      "eval_compot_runtime": 67.1863,
+      "eval_compot_samples_per_second": 0.744,
+      "eval_compot_steps_per_second": 0.03,
+      "num_input_tokens_seen": 31348120,
+      "step": 500
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "eval_custom_ui_MAE_all": 0.09588127583265305,
+      "eval_custom_ui_MAE_x": 0.10331301391124725,
+      "eval_custom_ui_MAE_y": 0.08844954147934914,
+      "eval_custom_ui_NUM_probability": 0.9999941885471344,
+      "eval_custom_ui_loss": 0.48714783787727356,
+      "eval_custom_ui_loss_ce": 0.00020886939091724344,
+      "eval_custom_ui_loss_num": 0.0944366455078125,
+      "eval_custom_ui_loss_xval": 0.4720458984375,
+      "eval_custom_ui_runtime": 51.6926,
+      "eval_custom_ui_samples_per_second": 0.967,
+      "eval_custom_ui_steps_per_second": 0.039,
+      "num_input_tokens_seen": 31348120,
+      "step": 500
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "loss": 0.44073396921157837,
+      "loss_ce": 0.0003042669268324971,
+      "loss_iou": 0.0,
+      "loss_num": 0.087890625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 31348120,
+      "step": 500
+    },
+    {
+      "epoch": 1.6672212978369383,
+      "grad_norm": 19.1532039642334,
+      "learning_rate": 5e-06,
+      "loss": 1.0931,
+      "num_input_tokens_seen": 31411648,
+      "step": 501
+    },
+    {
+      "epoch": 1.6672212978369383,
+      "loss": 1.0893843173980713,
+      "loss_ce": 0.0014936134684830904,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0791015625,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 31411648,
+      "step": 501
+    },
+    {
+      "epoch": 1.670549084858569,
+      "grad_norm": 41.87156295776367,
+      "learning_rate": 5e-06,
+      "loss": 1.1134,
+      "num_input_tokens_seen": 31474148,
+      "step": 502
+    },
+    {
+      "epoch": 1.670549084858569,
+      "loss": 0.7424819469451904,
+      "loss_ce": 0.0009047660860233009,
+      "loss_iou": 0.21875,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 31474148,
+      "step": 502
+    },
+    {
+      "epoch": 1.6738768718801995,
+      "grad_norm": 26.744535446166992,
+      "learning_rate": 5e-06,
+      "loss": 0.9912,
+      "num_input_tokens_seen": 31538180,
+      "step": 503
+    },
+    {
+      "epoch": 1.6738768718801995,
+      "loss": 1.1642227172851562,
+      "loss_ce": 0.0023575148079544306,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.07421875,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 31538180,
+      "step": 503
+    },
+    {
+      "epoch": 1.6772046589018301,
+      "grad_norm": 10.769326210021973,
+      "learning_rate": 5e-06,
+      "loss": 1.1929,
+      "num_input_tokens_seen": 31601972,
+      "step": 504
+    },
+    {
+      "epoch": 1.6772046589018301,
+      "loss": 1.1387523412704468,
+      "loss_ce": 0.0005688150995410979,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.08056640625,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 31601972,
+      "step": 504
+    },
+    {
+      "epoch": 1.6805324459234607,
+      "grad_norm": 24.930686950683594,
+      "learning_rate": 5e-06,
+      "loss": 1.0095,
+      "num_input_tokens_seen": 31665256,
+      "step": 505
+    },
+    {
+      "epoch": 1.6805324459234607,
+      "loss": 1.1174432039260864,
+      "loss_ce": 0.000499797286465764,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.07470703125,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 31665256,
+      "step": 505
+    },
+    {
+      "epoch": 1.6838602329450914,
+      "grad_norm": 11.738724708557129,
+      "learning_rate": 5e-06,
+      "loss": 0.7514,
+      "num_input_tokens_seen": 31726716,
+      "step": 506
+    },
+    {
+      "epoch": 1.6838602329450914,
+      "loss": 0.5557652711868286,
+      "loss_ce": 0.0007115780026651919,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.5546875,
+      "num_input_tokens_seen": 31726716,
+      "step": 506
+    },
+    {
+      "epoch": 1.687188019966722,
+      "grad_norm": 8.881688117980957,
+      "learning_rate": 5e-06,
+      "loss": 0.8519,
+      "num_input_tokens_seen": 31789836,
+      "step": 507
+    },
+    {
+      "epoch": 1.687188019966722,
+      "loss": 0.9876424670219421,
+      "loss_ce": 0.00033773150062188506,
+      "loss_iou": 0.25,
+      "loss_num": 0.09716796875,
+      "loss_xval": 0.98828125,
+      "num_input_tokens_seen": 31789836,
+      "step": 507
+    },
+    {
+      "epoch": 1.6905158069883528,
+      "grad_norm": 13.23851490020752,
+      "learning_rate": 5e-06,
+      "loss": 1.0906,
+      "num_input_tokens_seen": 31853316,
+      "step": 508
+    },
+    {
+      "epoch": 1.6905158069883528,
+      "loss": 1.2013055086135864,
+      "loss_ce": 0.0006219673086889088,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.11328125,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 31853316,
+      "step": 508
+    },
+    {
+      "epoch": 1.6938435940099834,
+      "grad_norm": 32.30921173095703,
+      "learning_rate": 5e-06,
+      "loss": 1.0942,
+      "num_input_tokens_seen": 31916364,
+      "step": 509
+    },
+    {
+      "epoch": 1.6938435940099834,
+      "loss": 1.0613774061203003,
+      "loss_ce": 0.0005864131962880492,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.09033203125,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 31916364,
+      "step": 509
+    },
+    {
+      "epoch": 1.697171381031614,
+      "grad_norm": 21.38906478881836,
+      "learning_rate": 5e-06,
+      "loss": 1.1213,
+      "num_input_tokens_seen": 31979748,
+      "step": 510
+    },
+    {
+      "epoch": 1.697171381031614,
+      "loss": 1.2083408832550049,
+      "loss_ce": 0.04061625152826309,
+      "loss_iou": 0.359375,
+      "loss_num": 0.08984375,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 31979748,
+      "step": 510
+    },
+    {
+      "epoch": 1.7004991680532446,
+      "grad_norm": 27.368276596069336,
+      "learning_rate": 5e-06,
+      "loss": 1.2503,
+      "num_input_tokens_seen": 32042528,
+      "step": 511
+    },
+    {
+      "epoch": 1.7004991680532446,
+      "loss": 1.3059027194976807,
+      "loss_ce": 0.0012151505798101425,
+      "loss_iou": 0.5,
+      "loss_num": 0.060791015625,
+      "loss_xval": 1.3046875,
+      "num_input_tokens_seen": 32042528,
+      "step": 511
+    },
+    {
+      "epoch": 1.7038269550748752,
+      "grad_norm": 11.777514457702637,
+      "learning_rate": 5e-06,
+      "loss": 1.0625,
+      "num_input_tokens_seen": 32105152,
+      "step": 512
+    },
+    {
+      "epoch": 1.7038269550748752,
+      "loss": 0.9886508584022522,
+      "loss_ce": 0.000857874343637377,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.98828125,
+      "num_input_tokens_seen": 32105152,
+      "step": 512
+    },
+    {
+      "epoch": 1.7071547420965059,
+      "grad_norm": 19.91946792602539,
+      "learning_rate": 5e-06,
+      "loss": 0.7893,
+      "num_input_tokens_seen": 32167704,
+      "step": 513
+    },
+    {
+      "epoch": 1.7071547420965059,
+      "loss": 0.8008160591125488,
+      "loss_ce": 0.0010113224852830172,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 32167704,
+      "step": 513
+    },
+    {
+      "epoch": 1.7104825291181365,
+      "grad_norm": 24.025108337402344,
+      "learning_rate": 5e-06,
+      "loss": 1.0882,
+      "num_input_tokens_seen": 32231292,
+      "step": 514
+    },
+    {
+      "epoch": 1.7104825291181365,
+      "loss": 1.2957994937896729,
+      "loss_ce": 0.0008775498135946691,
+      "loss_iou": 0.453125,
+      "loss_num": 0.07763671875,
+      "loss_xval": 1.296875,
+      "num_input_tokens_seen": 32231292,
+      "step": 514
+    },
+    {
+      "epoch": 1.713810316139767,
+      "grad_norm": 22.541730880737305,
+      "learning_rate": 5e-06,
+      "loss": 0.9815,
+      "num_input_tokens_seen": 32294084,
+      "step": 515
+    },
+    {
+      "epoch": 1.713810316139767,
+      "loss": 1.1235849857330322,
+      "loss_ce": 0.0005380419315770268,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.07275390625,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 32294084,
+      "step": 515
+    },
+    {
+      "epoch": 1.7171381031613977,
+      "grad_norm": 17.039142608642578,
+      "learning_rate": 5e-06,
+      "loss": 0.8633,
+      "num_input_tokens_seen": 32355804,
+      "step": 516
+    },
+    {
+      "epoch": 1.7171381031613977,
+      "loss": 0.8609045147895813,
+      "loss_ce": 0.0012853904627263546,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.080078125,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 32355804,
+      "step": 516
+    },
+    {
+      "epoch": 1.7204658901830283,
+      "grad_norm": 17.313488006591797,
+      "learning_rate": 5e-06,
+      "loss": 1.0722,
+      "num_input_tokens_seen": 32418416,
+      "step": 517
+    },
+    {
+      "epoch": 1.7204658901830283,
+      "loss": 0.8409963846206665,
+      "loss_ce": 0.0006643726374022663,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.83984375,
+      "num_input_tokens_seen": 32418416,
+      "step": 517
+    },
+    {
+      "epoch": 1.723793677204659,
+      "grad_norm": 21.147789001464844,
+      "learning_rate": 5e-06,
+      "loss": 0.9084,
+      "num_input_tokens_seen": 32480452,
+      "step": 518
+    },
+    {
+      "epoch": 1.723793677204659,
+      "loss": 0.6569137573242188,
+      "loss_ce": 0.00041967950528487563,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 32480452,
+      "step": 518
+    },
+    {
+      "epoch": 1.7271214642262895,
+      "grad_norm": 14.255158424377441,
+      "learning_rate": 5e-06,
+      "loss": 0.9042,
+      "num_input_tokens_seen": 32543352,
+      "step": 519
+    },
+    {
+      "epoch": 1.7271214642262895,
+      "loss": 0.9606277346611023,
+      "loss_ce": 0.0011550780618563294,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.9609375,
+      "num_input_tokens_seen": 32543352,
+      "step": 519
+    },
+    {
+      "epoch": 1.7304492512479202,
+      "grad_norm": 23.895132064819336,
+      "learning_rate": 5e-06,
+      "loss": 1.0736,
+      "num_input_tokens_seen": 32606656,
+      "step": 520
+    },
+    {
+      "epoch": 1.7304492512479202,
+      "loss": 0.9719319343566895,
+      "loss_ce": 0.004036390222609043,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.96875,
+      "num_input_tokens_seen": 32606656,
+      "step": 520
+    },
+    {
+      "epoch": 1.7337770382695508,
+      "grad_norm": 28.39651870727539,
+      "learning_rate": 5e-06,
+      "loss": 1.2689,
+      "num_input_tokens_seen": 32668332,
+      "step": 521
+    },
+    {
+      "epoch": 1.7337770382695508,
+      "loss": 1.0842338800430298,
+      "loss_ce": 0.001958527136594057,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0888671875,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 32668332,
+      "step": 521
+    },
+    {
+      "epoch": 1.7371048252911814,
+      "grad_norm": 14.18319320678711,
+      "learning_rate": 5e-06,
+      "loss": 1.1421,
+      "num_input_tokens_seen": 32730852,
+      "step": 522
+    },
+    {
+      "epoch": 1.7371048252911814,
+      "loss": 1.0944100618362427,
+      "loss_ce": 4.9661710363579914e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0859375,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 32730852,
+      "step": 522
+    },
+    {
+      "epoch": 1.740432612312812,
+      "grad_norm": 20.83157730102539,
+      "learning_rate": 5e-06,
+      "loss": 1.2773,
+      "num_input_tokens_seen": 32795496,
+      "step": 523
+    },
+    {
+      "epoch": 1.740432612312812,
+      "loss": 1.1278488636016846,
+      "loss_ce": 0.004802019335329533,
+      "loss_iou": 0.421875,
+      "loss_num": 0.055908203125,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 32795496,
+      "step": 523
+    },
+    {
+      "epoch": 1.7437603993344426,
+      "grad_norm": 16.72551918029785,
+      "learning_rate": 5e-06,
+      "loss": 0.9721,
+      "num_input_tokens_seen": 32860204,
+      "step": 524
+    },
+    {
+      "epoch": 1.7437603993344426,
+      "loss": 1.024550437927246,
+      "loss_ce": 0.0004415763833094388,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.05810546875,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 32860204,
+      "step": 524
+    },
+    {
+      "epoch": 1.7470881863560732,
+      "grad_norm": 11.438901901245117,
+      "learning_rate": 5e-06,
+      "loss": 0.8762,
+      "num_input_tokens_seen": 32923856,
+      "step": 525
+    },
+    {
+      "epoch": 1.7470881863560732,
+      "loss": 1.0130120515823364,
+      "loss_ce": 0.001415381790138781,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0556640625,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 32923856,
+      "step": 525
+    },
+    {
+      "epoch": 1.7504159733777038,
+      "grad_norm": 8.21237850189209,
+      "learning_rate": 5e-06,
+      "loss": 0.7951,
+      "num_input_tokens_seen": 32985220,
+      "step": 526
+    },
+    {
+      "epoch": 1.7504159733777038,
+      "loss": 0.9805971384048462,
+      "loss_ce": 0.00037258450174704194,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.98046875,
+      "num_input_tokens_seen": 32985220,
+      "step": 526
+    },
+    {
+      "epoch": 1.7537437603993344,
+      "grad_norm": 9.287155151367188,
+      "learning_rate": 5e-06,
+      "loss": 0.9109,
+      "num_input_tokens_seen": 33048680,
+      "step": 527
+    },
+    {
+      "epoch": 1.7537437603993344,
+      "loss": 0.9192065596580505,
+      "loss_ce": 0.0012378038372844458,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 33048680,
+      "step": 527
+    },
+    {
+      "epoch": 1.757071547420965,
+      "grad_norm": 46.055049896240234,
+      "learning_rate": 5e-06,
+      "loss": 1.0704,
+      "num_input_tokens_seen": 33112220,
+      "step": 528
+    },
+    {
+      "epoch": 1.757071547420965,
+      "loss": 1.060153841972351,
+      "loss_ce": 0.0010718015255406499,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.07275390625,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 33112220,
+      "step": 528
+    },
+    {
+      "epoch": 1.7603993344425957,
+      "grad_norm": 27.12957000732422,
+      "learning_rate": 5e-06,
+      "loss": 0.865,
+      "num_input_tokens_seen": 33172044,
+      "step": 529
+    },
+    {
+      "epoch": 1.7603993344425957,
+      "loss": 0.9046726226806641,
+      "loss_ce": 0.0001316153648076579,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.076171875,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 33172044,
+      "step": 529
+    },
+    {
+      "epoch": 1.7637271214642263,
+      "grad_norm": 14.900341987609863,
+      "learning_rate": 5e-06,
+      "loss": 0.811,
+      "num_input_tokens_seen": 33234976,
+      "step": 530
+    },
+    {
+      "epoch": 1.7637271214642263,
+      "loss": 0.8328965306282043,
+      "loss_ce": 0.00037697955849580467,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 33234976,
+      "step": 530
+    },
+    {
+      "epoch": 1.767054908485857,
+      "grad_norm": 19.717370986938477,
+      "learning_rate": 5e-06,
+      "loss": 1.0421,
+      "num_input_tokens_seen": 33298716,
+      "step": 531
+    },
+    {
+      "epoch": 1.767054908485857,
+      "loss": 1.0288456678390503,
+      "loss_ce": 0.0005254416028037667,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 33298716,
+      "step": 531
+    },
+    {
+      "epoch": 1.7703826955074875,
+      "grad_norm": 14.97089672088623,
+      "learning_rate": 5e-06,
+      "loss": 0.7139,
+      "num_input_tokens_seen": 33359296,
+      "step": 532
+    },
+    {
+      "epoch": 1.7703826955074875,
+      "loss": 0.6194471716880798,
+      "loss_ce": 0.00262587983161211,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 33359296,
+      "step": 532
+    },
+    {
+      "epoch": 1.7737104825291181,
+      "grad_norm": 10.585949897766113,
+      "learning_rate": 5e-06,
+      "loss": 0.893,
+      "num_input_tokens_seen": 33420788,
+      "step": 533
+    },
+    {
+      "epoch": 1.7737104825291181,
+      "loss": 0.8950399160385132,
+      "loss_ce": 0.0002645118802320212,
+      "loss_iou": 0.3125,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.89453125,
+      "num_input_tokens_seen": 33420788,
+      "step": 533
+    },
+    {
+      "epoch": 1.7770382695507487,
+      "grad_norm": 10.35478687286377,
+      "learning_rate": 5e-06,
+      "loss": 1.0367,
+      "num_input_tokens_seen": 33484684,
+      "step": 534
+    },
+    {
+      "epoch": 1.7770382695507487,
+      "loss": 1.3992263078689575,
+      "loss_ce": 5.641989628202282e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.1083984375,
+      "loss_xval": 1.3984375,
+      "num_input_tokens_seen": 33484684,
+      "step": 534
+    },
+    {
+      "epoch": 1.7803660565723793,
+      "grad_norm": 13.746817588806152,
+      "learning_rate": 5e-06,
+      "loss": 0.9671,
+      "num_input_tokens_seen": 33547376,
+      "step": 535
+    },
+    {
+      "epoch": 1.7803660565723793,
+      "loss": 1.0142176151275635,
+      "loss_ce": 5.745934322476387e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.06298828125,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 33547376,
+      "step": 535
+    },
+    {
+      "epoch": 1.78369384359401,
+      "grad_norm": 16.677959442138672,
+      "learning_rate": 5e-06,
+      "loss": 1.0859,
+      "num_input_tokens_seen": 33610036,
+      "step": 536
+    },
+    {
+      "epoch": 1.78369384359401,
+      "loss": 1.0913352966308594,
+      "loss_ce": 0.00027076914557255805,
+      "loss_iou": 0.328125,
+      "loss_num": 0.08642578125,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 33610036,
+      "step": 536
+    },
+    {
+      "epoch": 1.7870216306156406,
+      "grad_norm": 10.253283500671387,
+      "learning_rate": 5e-06,
+      "loss": 0.7858,
+      "num_input_tokens_seen": 33672984,
+      "step": 537
+    },
+    {
+      "epoch": 1.7870216306156406,
+      "loss": 0.7678014636039734,
+      "loss_ce": 0.0027867835015058517,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 33672984,
+      "step": 537
+    },
+    {
+      "epoch": 1.7903494176372712,
+      "grad_norm": 35.10622024536133,
+      "learning_rate": 5e-06,
+      "loss": 1.0201,
+      "num_input_tokens_seen": 33736132,
+      "step": 538
+    },
+    {
+      "epoch": 1.7903494176372712,
+      "loss": 1.0910940170288086,
+      "loss_ce": 0.0012502449098974466,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0927734375,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 33736132,
+      "step": 538
+    },
+    {
+      "epoch": 1.7936772046589018,
+      "grad_norm": 12.398226737976074,
+      "learning_rate": 5e-06,
+      "loss": 0.8778,
+      "num_input_tokens_seen": 33799380,
+      "step": 539
+    },
+    {
+      "epoch": 1.7936772046589018,
+      "loss": 0.7171542644500732,
+      "loss_ce": 0.0008456680225208402,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 33799380,
+      "step": 539
+    },
+    {
+      "epoch": 1.7970049916805324,
+      "grad_norm": 14.9580659866333,
+      "learning_rate": 5e-06,
+      "loss": 0.7666,
+      "num_input_tokens_seen": 33862656,
+      "step": 540
+    },
+    {
+      "epoch": 1.7970049916805324,
+      "loss": 0.9555954933166504,
+      "loss_ce": 0.0010056574828922749,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 33862656,
+      "step": 540
+    },
+    {
+      "epoch": 1.800332778702163,
+      "grad_norm": 20.136131286621094,
+      "learning_rate": 5e-06,
+      "loss": 0.7634,
+      "num_input_tokens_seen": 33925932,
+      "step": 541
+    },
+    {
+      "epoch": 1.800332778702163,
+      "loss": 0.685050368309021,
+      "loss_ce": 0.004142177756875753,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 33925932,
+      "step": 541
+    },
+    {
+      "epoch": 1.8036605657237936,
+      "grad_norm": 26.61849021911621,
+      "learning_rate": 5e-06,
+      "loss": 1.1369,
+      "num_input_tokens_seen": 33989324,
+      "step": 542
+    },
+    {
+      "epoch": 1.8036605657237936,
+      "loss": 1.1945197582244873,
+      "loss_ce": 0.0018927238415926695,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.10400390625,
+      "loss_xval": 1.1953125,
+      "num_input_tokens_seen": 33989324,
+      "step": 542
+    },
+    {
+      "epoch": 1.8069883527454242,
+      "grad_norm": 12.676218032836914,
+      "learning_rate": 5e-06,
+      "loss": 1.0277,
+      "num_input_tokens_seen": 34054000,
+      "step": 543
+    },
+    {
+      "epoch": 1.8069883527454242,
+      "loss": 0.9486079216003418,
+      "loss_ce": 0.0018306173151358962,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 34054000,
+      "step": 543
+    },
+    {
+      "epoch": 1.8103161397670549,
+      "grad_norm": 14.024618148803711,
+      "learning_rate": 5e-06,
+      "loss": 1.0651,
+      "num_input_tokens_seen": 34118096,
+      "step": 544
+    },
+    {
+      "epoch": 1.8103161397670549,
+      "loss": 1.2064943313598633,
+      "loss_ce": 0.0009278804645873606,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.08837890625,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 34118096,
+      "step": 544
+    },
+    {
+      "epoch": 1.8136439267886857,
+      "grad_norm": 15.398441314697266,
+      "learning_rate": 5e-06,
+      "loss": 0.8867,
+      "num_input_tokens_seen": 34181016,
+      "step": 545
+    },
+    {
+      "epoch": 1.8136439267886857,
+      "loss": 0.6198215484619141,
+      "loss_ce": 0.00019265212176833302,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 34181016,
+      "step": 545
+    },
+    {
+      "epoch": 1.8169717138103163,
+      "grad_norm": 18.64105987548828,
+      "learning_rate": 5e-06,
+      "loss": 1.0866,
+      "num_input_tokens_seen": 34242020,
+      "step": 546
+    },
+    {
+      "epoch": 1.8169717138103163,
+      "loss": 0.9077228903770447,
+      "loss_ce": 0.0011066340375691652,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 34242020,
+      "step": 546
+    },
+    {
+      "epoch": 1.820299500831947,
+      "grad_norm": 13.642433166503906,
+      "learning_rate": 5e-06,
+      "loss": 0.9115,
+      "num_input_tokens_seen": 34305132,
+      "step": 547
+    },
+    {
+      "epoch": 1.820299500831947,
+      "loss": 0.6709655523300171,
+      "loss_ce": 0.0017760555492714047,
+      "loss_iou": 0.203125,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 34305132,
+      "step": 547
+    },
+    {
+      "epoch": 1.8236272878535775,
+      "grad_norm": 37.51459503173828,
+      "learning_rate": 5e-06,
+      "loss": 1.0159,
+      "num_input_tokens_seen": 34367900,
+      "step": 548
+    },
+    {
+      "epoch": 1.8236272878535775,
+      "loss": 1.0078930854797363,
+      "loss_ce": 0.0005688223754987121,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.050048828125,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 34367900,
+      "step": 548
+    },
+    {
+      "epoch": 1.8269550748752081,
+      "grad_norm": 27.26789093017578,
+      "learning_rate": 5e-06,
+      "loss": 1.1389,
+      "num_input_tokens_seen": 34429256,
+      "step": 549
+    },
+    {
+      "epoch": 1.8269550748752081,
+      "loss": 1.143343448638916,
+      "loss_ce": 0.001985959243029356,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.09814453125,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 34429256,
+      "step": 549
+    },
+    {
+      "epoch": 1.8302828618968388,
+      "grad_norm": 14.803655624389648,
+      "learning_rate": 5e-06,
+      "loss": 1.3364,
+      "num_input_tokens_seen": 34492152,
+      "step": 550
+    },
+    {
+      "epoch": 1.8302828618968388,
+      "loss": 1.4599002599716187,
+      "loss_ce": 0.0004276781983207911,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.126953125,
+      "loss_xval": 1.4609375,
+      "num_input_tokens_seen": 34492152,
+      "step": 550
+    },
+    {
+      "epoch": 1.8336106489184694,
+      "grad_norm": 26.294727325439453,
+      "learning_rate": 5e-06,
+      "loss": 1.1966,
+      "num_input_tokens_seen": 34555988,
+      "step": 551
+    },
+    {
+      "epoch": 1.8336106489184694,
+      "loss": 1.2220561504364014,
+      "loss_ce": 0.002573746722191572,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.08203125,
+      "loss_xval": 1.21875,
+      "num_input_tokens_seen": 34555988,
+      "step": 551
+    },
+    {
+      "epoch": 1.8369384359401,
+      "grad_norm": 14.449104309082031,
+      "learning_rate": 5e-06,
+      "loss": 0.9287,
+      "num_input_tokens_seen": 34619232,
+      "step": 552
+    },
+    {
+      "epoch": 1.8369384359401,
+      "loss": 0.9966768026351929,
+      "loss_ce": 0.0013155062915757298,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.99609375,
+      "num_input_tokens_seen": 34619232,
+      "step": 552
+    },
+    {
+      "epoch": 1.8402662229617306,
+      "grad_norm": 23.511919021606445,
+      "learning_rate": 5e-06,
+      "loss": 1.0271,
+      "num_input_tokens_seen": 34682704,
+      "step": 553
+    },
+    {
+      "epoch": 1.8402662229617306,
+      "loss": 1.2029976844787598,
+      "loss_ce": 0.0035348027013242245,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 34682704,
+      "step": 553
+    },
+    {
+      "epoch": 1.8435940099833612,
+      "grad_norm": 12.143444061279297,
+      "learning_rate": 5e-06,
+      "loss": 0.8477,
+      "num_input_tokens_seen": 34745964,
+      "step": 554
+    },
+    {
+      "epoch": 1.8435940099833612,
+      "loss": 1.0100423097610474,
+      "loss_ce": 0.00012407411122694612,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0537109375,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 34745964,
+      "step": 554
+    },
+    {
+      "epoch": 1.8469217970049918,
+      "grad_norm": 26.108985900878906,
+      "learning_rate": 5e-06,
+      "loss": 0.9329,
+      "num_input_tokens_seen": 34808712,
+      "step": 555
+    },
+    {
+      "epoch": 1.8469217970049918,
+      "loss": 0.9639697074890137,
+      "loss_ce": 0.0013231942430138588,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.9609375,
+      "num_input_tokens_seen": 34808712,
+      "step": 555
+    },
+    {
+      "epoch": 1.8502495840266224,
+      "grad_norm": 18.90593147277832,
+      "learning_rate": 5e-06,
+      "loss": 1.0983,
+      "num_input_tokens_seen": 34871224,
+      "step": 556
+    },
+    {
+      "epoch": 1.8502495840266224,
+      "loss": 1.1608165502548218,
+      "loss_ce": 0.0004161183023825288,
+      "loss_iou": 0.4375,
+      "loss_num": 0.056884765625,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 34871224,
+      "step": 556
+    },
+    {
+      "epoch": 1.853577371048253,
+      "grad_norm": 19.406909942626953,
+      "learning_rate": 5e-06,
+      "loss": 1.0324,
+      "num_input_tokens_seen": 34931504,
+      "step": 557
+    },
+    {
+      "epoch": 1.853577371048253,
+      "loss": 1.2264602184295654,
+      "loss_ce": 0.0003859272401314229,
+      "loss_iou": 0.421875,
+      "loss_num": 0.07666015625,
+      "loss_xval": 1.2265625,
+      "num_input_tokens_seen": 34931504,
+      "step": 557
+    },
+    {
+      "epoch": 1.8569051580698837,
+      "grad_norm": 12.974056243896484,
+      "learning_rate": 5e-06,
+      "loss": 0.7937,
+      "num_input_tokens_seen": 34994100,
+      "step": 558
+    },
+    {
+      "epoch": 1.8569051580698837,
+      "loss": 0.8865928649902344,
+      "loss_ce": 0.0032920846715569496,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 34994100,
+      "step": 558
+    },
+    {
+      "epoch": 1.8602329450915143,
+      "grad_norm": 13.932268142700195,
+      "learning_rate": 5e-06,
+      "loss": 1.1015,
+      "num_input_tokens_seen": 35056884,
+      "step": 559
+    },
+    {
+      "epoch": 1.8602329450915143,
+      "loss": 1.1955981254577637,
+      "loss_ce": 0.0007738770800642669,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.064453125,
+      "loss_xval": 1.1953125,
+      "num_input_tokens_seen": 35056884,
+      "step": 559
+    },
+    {
+      "epoch": 1.8635607321131449,
+      "grad_norm": 14.666391372680664,
+      "learning_rate": 5e-06,
+      "loss": 0.8643,
+      "num_input_tokens_seen": 35120340,
+      "step": 560
+    },
+    {
+      "epoch": 1.8635607321131449,
+      "loss": 0.886032223701477,
+      "loss_ce": 0.0012666526017710567,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 35120340,
+      "step": 560
+    },
+    {
+      "epoch": 1.8668885191347755,
+      "grad_norm": 13.323609352111816,
+      "learning_rate": 5e-06,
+      "loss": 1.1881,
+      "num_input_tokens_seen": 35183468,
+      "step": 561
+    },
+    {
+      "epoch": 1.8668885191347755,
+      "loss": 1.152817964553833,
+      "loss_ce": 0.0019389993976801634,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.08544921875,
+      "loss_xval": 1.1484375,
+      "num_input_tokens_seen": 35183468,
+      "step": 561
+    },
+    {
+      "epoch": 1.870216306156406,
+      "grad_norm": 9.693658828735352,
+      "learning_rate": 5e-06,
+      "loss": 1.1869,
+      "num_input_tokens_seen": 35247208,
+      "step": 562
+    },
+    {
+      "epoch": 1.870216306156406,
+      "loss": 1.3116389513015747,
+      "loss_ce": 0.0006038228166289628,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.08251953125,
+      "loss_xval": 1.3125,
+      "num_input_tokens_seen": 35247208,
+      "step": 562
+    },
+    {
+      "epoch": 1.8735440931780367,
+      "grad_norm": 29.306072235107422,
+      "learning_rate": 5e-06,
+      "loss": 1.1472,
+      "num_input_tokens_seen": 35310828,
+      "step": 563
+    },
+    {
+      "epoch": 1.8735440931780367,
+      "loss": 1.2778418064117432,
+      "loss_ce": 0.0068457284942269325,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.078125,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 35310828,
+      "step": 563
+    },
+    {
+      "epoch": 1.8768718801996673,
+      "grad_norm": 27.0247745513916,
+      "learning_rate": 5e-06,
+      "loss": 0.9282,
+      "num_input_tokens_seen": 35373164,
+      "step": 564
+    },
+    {
+      "epoch": 1.8768718801996673,
+      "loss": 0.9917428493499756,
+      "loss_ce": 0.00028781441505998373,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 35373164,
+      "step": 564
+    },
+    {
+      "epoch": 1.880199667221298,
+      "grad_norm": 14.91286563873291,
+      "learning_rate": 5e-06,
+      "loss": 0.9329,
+      "num_input_tokens_seen": 35435044,
+      "step": 565
+    },
+    {
+      "epoch": 1.880199667221298,
+      "loss": 1.015300989151001,
+      "loss_ce": 0.00016425539797637612,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.07958984375,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 35435044,
+      "step": 565
+    },
+    {
+      "epoch": 1.8835274542429286,
+      "grad_norm": 10.378997802734375,
+      "learning_rate": 5e-06,
+      "loss": 0.9905,
+      "num_input_tokens_seen": 35497864,
+      "step": 566
+    },
+    {
+      "epoch": 1.8835274542429286,
+      "loss": 1.1247049570083618,
+      "loss_ce": 0.0009256677003577352,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0849609375,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 35497864,
+      "step": 566
+    },
+    {
+      "epoch": 1.8868552412645592,
+      "grad_norm": 17.83684539794922,
+      "learning_rate": 5e-06,
+      "loss": 0.929,
+      "num_input_tokens_seen": 35560536,
+      "step": 567
+    },
+    {
+      "epoch": 1.8868552412645592,
+      "loss": 0.6144291162490845,
+      "loss_ce": 0.00017130836204160005,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 35560536,
+      "step": 567
+    },
+    {
+      "epoch": 1.8901830282861898,
+      "grad_norm": 49.60697937011719,
+      "learning_rate": 5e-06,
+      "loss": 0.9559,
+      "num_input_tokens_seen": 35622356,
+      "step": 568
+    },
+    {
+      "epoch": 1.8901830282861898,
+      "loss": 1.1087956428527832,
+      "loss_ce": 0.0001530621957499534,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.1259765625,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 35622356,
+      "step": 568
+    },
+    {
+      "epoch": 1.8935108153078204,
+      "grad_norm": 14.60826301574707,
+      "learning_rate": 5e-06,
+      "loss": 1.121,
+      "num_input_tokens_seen": 35685972,
+      "step": 569
+    },
+    {
+      "epoch": 1.8935108153078204,
+      "loss": 1.326939582824707,
+      "loss_ce": 0.0022325818426907063,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.11279296875,
+      "loss_xval": 1.328125,
+      "num_input_tokens_seen": 35685972,
+      "step": 569
+    },
+    {
+      "epoch": 1.896838602329451,
+      "grad_norm": 25.41444969177246,
+      "learning_rate": 5e-06,
+      "loss": 0.8218,
+      "num_input_tokens_seen": 35747280,
+      "step": 570
+    },
+    {
+      "epoch": 1.896838602329451,
+      "loss": 0.8077418804168701,
+      "loss_ce": 0.0006130048423074186,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.80859375,
+      "num_input_tokens_seen": 35747280,
+      "step": 570
+    },
+    {
+      "epoch": 1.9001663893510816,
+      "grad_norm": 24.91752052307129,
+      "learning_rate": 5e-06,
+      "loss": 1.1637,
+      "num_input_tokens_seen": 35811528,
+      "step": 571
+    },
+    {
+      "epoch": 1.9001663893510816,
+      "loss": 1.3153760433197021,
+      "loss_ce": 0.0033644186332821846,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 1.3125,
+      "num_input_tokens_seen": 35811528,
+      "step": 571
+    },
+    {
+      "epoch": 1.9034941763727122,
+      "grad_norm": 25.198698043823242,
+      "learning_rate": 5e-06,
+      "loss": 1.0152,
+      "num_input_tokens_seen": 35874204,
+      "step": 572
+    },
+    {
+      "epoch": 1.9034941763727122,
+      "loss": 1.3468836545944214,
+      "loss_ce": 0.0002039041864918545,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.09130859375,
+      "loss_xval": 1.34375,
+      "num_input_tokens_seen": 35874204,
+      "step": 572
+    },
+    {
+      "epoch": 1.9068219633943428,
+      "grad_norm": 17.966543197631836,
+      "learning_rate": 5e-06,
+      "loss": 0.8496,
+      "num_input_tokens_seen": 35936360,
+      "step": 573
+    },
+    {
+      "epoch": 1.9068219633943428,
+      "loss": 1.1128754615783691,
+      "loss_ce": 0.00032671194639988244,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 35936360,
+      "step": 573
+    },
+    {
+      "epoch": 1.9101497504159735,
+      "grad_norm": 22.120344161987305,
+      "learning_rate": 5e-06,
+      "loss": 1.0125,
+      "num_input_tokens_seen": 35998784,
+      "step": 574
+    },
+    {
+      "epoch": 1.9101497504159735,
+      "loss": 1.294684648513794,
+      "loss_ce": 0.0002511721686460078,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.09228515625,
+      "loss_xval": 1.296875,
+      "num_input_tokens_seen": 35998784,
+      "step": 574
+    },
+    {
+      "epoch": 1.913477537437604,
+      "grad_norm": 14.170149803161621,
+      "learning_rate": 5e-06,
+      "loss": 0.9052,
+      "num_input_tokens_seen": 36061612,
+      "step": 575
+    },
+    {
+      "epoch": 1.913477537437604,
+      "loss": 0.7781501412391663,
+      "loss_ce": 0.0025154012255370617,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 36061612,
+      "step": 575
+    },
+    {
+      "epoch": 1.9168053244592347,
+      "grad_norm": 38.99104309082031,
+      "learning_rate": 5e-06,
+      "loss": 1.1207,
+      "num_input_tokens_seen": 36124752,
+      "step": 576
+    },
+    {
+      "epoch": 1.9168053244592347,
+      "loss": 1.3191872835159302,
+      "loss_ce": 0.0008279454777948558,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0859375,
+      "loss_xval": 1.3203125,
+      "num_input_tokens_seen": 36124752,
+      "step": 576
+    },
+    {
+      "epoch": 1.9201331114808653,
+      "grad_norm": 18.79828453063965,
+      "learning_rate": 5e-06,
+      "loss": 1.0237,
+      "num_input_tokens_seen": 36185564,
+      "step": 577
+    },
+    {
+      "epoch": 1.9201331114808653,
+      "loss": 1.0448555946350098,
+      "loss_ce": 0.0010322753805667162,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.08447265625,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 36185564,
+      "step": 577
+    },
+    {
+      "epoch": 1.923460898502496,
+      "grad_norm": 8.662841796875,
+      "learning_rate": 5e-06,
+      "loss": 0.8287,
+      "num_input_tokens_seen": 36248000,
+      "step": 578
+    },
+    {
+      "epoch": 1.923460898502496,
+      "loss": 0.7125368118286133,
+      "loss_ce": 0.00013448702520690858,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 36248000,
+      "step": 578
+    },
+    {
+      "epoch": 1.9267886855241265,
+      "grad_norm": 24.12902069091797,
+      "learning_rate": 5e-06,
+      "loss": 0.9049,
+      "num_input_tokens_seen": 36310656,
+      "step": 579
+    },
+    {
+      "epoch": 1.9267886855241265,
+      "loss": 0.8042501211166382,
+      "loss_ce": 0.0005391854792833328,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 36310656,
+      "step": 579
+    },
+    {
+      "epoch": 1.9301164725457571,
+      "grad_norm": 14.56750202178955,
+      "learning_rate": 5e-06,
+      "loss": 0.8137,
+      "num_input_tokens_seen": 36373116,
+      "step": 580
+    },
+    {
+      "epoch": 1.9301164725457571,
+      "loss": 0.6355822086334229,
+      "loss_ce": 0.003990426659584045,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 36373116,
+      "step": 580
+    },
+    {
+      "epoch": 1.9334442595673877,
+      "grad_norm": 17.040950775146484,
+      "learning_rate": 5e-06,
+      "loss": 0.9652,
+      "num_input_tokens_seen": 36434072,
+      "step": 581
+    },
+    {
+      "epoch": 1.9334442595673877,
+      "loss": 1.1782926321029663,
+      "loss_ce": 0.004586563445627689,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0986328125,
+      "loss_xval": 1.171875,
+      "num_input_tokens_seen": 36434072,
+      "step": 581
+    },
+    {
+      "epoch": 1.9367720465890184,
+      "grad_norm": 10.302577018737793,
+      "learning_rate": 5e-06,
+      "loss": 0.7532,
+      "num_input_tokens_seen": 36496096,
+      "step": 582
+    },
+    {
+      "epoch": 1.9367720465890184,
+      "loss": 0.8137335777282715,
+      "loss_ce": 0.0005012074834667146,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 36496096,
+      "step": 582
+    },
+    {
+      "epoch": 1.940099833610649,
+      "grad_norm": 15.283492088317871,
+      "learning_rate": 5e-06,
+      "loss": 0.9993,
+      "num_input_tokens_seen": 36558260,
+      "step": 583
+    },
+    {
+      "epoch": 1.940099833610649,
+      "loss": 0.8162379860877991,
+      "loss_ce": 0.00044212467037141323,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 36558260,
+      "step": 583
+    },
+    {
+      "epoch": 1.9434276206322796,
+      "grad_norm": 10.120014190673828,
+      "learning_rate": 5e-06,
+      "loss": 0.6606,
+      "num_input_tokens_seen": 36620356,
+      "step": 584
+    },
+    {
+      "epoch": 1.9434276206322796,
+      "loss": 0.6671352386474609,
+      "loss_ce": 0.00014302069030236453,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 36620356,
+      "step": 584
+    },
+    {
+      "epoch": 1.9467554076539102,
+      "grad_norm": 10.893766403198242,
+      "learning_rate": 5e-06,
+      "loss": 0.9976,
+      "num_input_tokens_seen": 36683052,
+      "step": 585
+    },
+    {
+      "epoch": 1.9467554076539102,
+      "loss": 0.9768415689468384,
+      "loss_ce": 0.0027204775251448154,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.97265625,
+      "num_input_tokens_seen": 36683052,
+      "step": 585
+    },
+    {
+      "epoch": 1.9500831946755408,
+      "grad_norm": 22.243104934692383,
+      "learning_rate": 5e-06,
+      "loss": 1.0094,
+      "num_input_tokens_seen": 36745780,
+      "step": 586
+    },
+    {
+      "epoch": 1.9500831946755408,
+      "loss": 0.975989043712616,
+      "loss_ce": 0.0006472546374425292,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 36745780,
+      "step": 586
+    },
+    {
+      "epoch": 1.9534109816971714,
+      "grad_norm": 17.508132934570312,
+      "learning_rate": 5e-06,
+      "loss": 1.1263,
+      "num_input_tokens_seen": 36809268,
+      "step": 587
+    },
+    {
+      "epoch": 1.9534109816971714,
+      "loss": 0.9999065399169922,
+      "loss_ce": 0.0018596657318994403,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.08837890625,
+      "loss_xval": 1.0,
+      "num_input_tokens_seen": 36809268,
+      "step": 587
+    },
+    {
+      "epoch": 1.956738768718802,
+      "grad_norm": 13.245006561279297,
+      "learning_rate": 5e-06,
+      "loss": 0.8024,
+      "num_input_tokens_seen": 36873044,
+      "step": 588
+    },
+    {
+      "epoch": 1.956738768718802,
+      "loss": 0.6673635244369507,
+      "loss_ce": 0.00024929430219344795,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 36873044,
+      "step": 588
+    },
+    {
+      "epoch": 1.9600665557404326,
+      "grad_norm": 9.61874008178711,
+      "learning_rate": 5e-06,
+      "loss": 1.037,
+      "num_input_tokens_seen": 36935692,
+      "step": 589
+    },
+    {
+      "epoch": 1.9600665557404326,
+      "loss": 0.9211355447769165,
+      "loss_ce": 0.0014577455585822463,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 36935692,
+      "step": 589
+    },
+    {
+      "epoch": 1.9633943427620633,
+      "grad_norm": 10.554235458374023,
+      "learning_rate": 5e-06,
+      "loss": 0.9399,
+      "num_input_tokens_seen": 36997520,
+      "step": 590
+    },
+    {
+      "epoch": 1.9633943427620633,
+      "loss": 1.0528907775878906,
+      "loss_ce": 0.0004004701040685177,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0703125,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 36997520,
+      "step": 590
+    },
+    {
+      "epoch": 1.9667221297836939,
+      "grad_norm": 17.416133880615234,
+      "learning_rate": 5e-06,
+      "loss": 0.7433,
+      "num_input_tokens_seen": 37060460,
+      "step": 591
+    },
+    {
+      "epoch": 1.9667221297836939,
+      "loss": 0.858521044254303,
+      "loss_ce": 0.0008550334605388343,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 37060460,
+      "step": 591
+    },
+    {
+      "epoch": 1.9700499168053245,
+      "grad_norm": 14.890344619750977,
+      "learning_rate": 5e-06,
+      "loss": 0.9326,
+      "num_input_tokens_seen": 37123204,
+      "step": 592
+    },
+    {
+      "epoch": 1.9700499168053245,
+      "loss": 0.8313470482826233,
+      "loss_ce": 0.013231809251010418,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 37123204,
+      "step": 592
+    },
+    {
+      "epoch": 1.973377703826955,
+      "grad_norm": 14.116703987121582,
+      "learning_rate": 5e-06,
+      "loss": 1.1915,
+      "num_input_tokens_seen": 37185468,
+      "step": 593
+    },
+    {
+      "epoch": 1.973377703826955,
+      "loss": 1.041740894317627,
+      "loss_ce": 0.000511517224367708,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.053466796875,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 37185468,
+      "step": 593
+    },
+    {
+      "epoch": 1.9767054908485857,
+      "grad_norm": 10.910758018493652,
+      "learning_rate": 5e-06,
+      "loss": 0.7212,
+      "num_input_tokens_seen": 37245908,
+      "step": 594
+    },
+    {
+      "epoch": 1.9767054908485857,
+      "loss": 0.6745211482048035,
+      "loss_ce": 0.001547462772578001,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 37245908,
+      "step": 594
+    },
+    {
+      "epoch": 1.9800332778702163,
+      "grad_norm": 11.645059585571289,
+      "learning_rate": 5e-06,
+      "loss": 0.9004,
+      "num_input_tokens_seen": 37310236,
+      "step": 595
+    },
+    {
+      "epoch": 1.9800332778702163,
+      "loss": 0.8556256294250488,
+      "loss_ce": 0.006992822512984276,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.84765625,
+      "num_input_tokens_seen": 37310236,
+      "step": 595
+    },
+    {
+      "epoch": 1.983361064891847,
+      "grad_norm": 12.184340476989746,
+      "learning_rate": 5e-06,
+      "loss": 1.0324,
+      "num_input_tokens_seen": 37371568,
+      "step": 596
+    },
+    {
+      "epoch": 1.983361064891847,
+      "loss": 1.0166329145431519,
+      "loss_ce": 0.0027778963558375835,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0703125,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 37371568,
+      "step": 596
+    },
+    {
+      "epoch": 1.9866888519134775,
+      "grad_norm": 10.037638664245605,
+      "learning_rate": 5e-06,
+      "loss": 0.6938,
+      "num_input_tokens_seen": 37434188,
+      "step": 597
+    },
+    {
+      "epoch": 1.9866888519134775,
+      "loss": 0.8472949266433716,
+      "loss_ce": 0.0015917927958071232,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.08154296875,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 37434188,
+      "step": 597
+    },
+    {
+      "epoch": 1.9900166389351082,
+      "grad_norm": 28.653364181518555,
+      "learning_rate": 5e-06,
+      "loss": 1.3632,
+      "num_input_tokens_seen": 37497236,
+      "step": 598
+    },
+    {
+      "epoch": 1.9900166389351082,
+      "loss": 1.1080653667449951,
+      "loss_ce": 0.00015516526764258742,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.07421875,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 37497236,
+      "step": 598
+    },
+    {
+      "epoch": 1.9933444259567388,
+      "grad_norm": 63.16008377075195,
+      "learning_rate": 5e-06,
+      "loss": 0.984,
+      "num_input_tokens_seen": 37559724,
+      "step": 599
+    },
+    {
+      "epoch": 1.9933444259567388,
+      "loss": 1.0816211700439453,
+      "loss_ce": 0.00044438272016122937,
+      "loss_iou": 0.359375,
+      "loss_num": 0.072265625,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 37559724,
+      "step": 599
+    },
+    {
+      "epoch": 1.9966722129783694,
+      "grad_norm": 12.790926933288574,
+      "learning_rate": 5e-06,
+      "loss": 1.2626,
+      "num_input_tokens_seen": 37623232,
+      "step": 600
+    },
+    {
+      "epoch": 1.9966722129783694,
+      "loss": 1.700850009918213,
+      "loss_ce": 0.0013871309347450733,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.13671875,
+      "loss_xval": 1.703125,
+      "num_input_tokens_seen": 37623232,
+      "step": 600
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 17.36669158935547,
+      "learning_rate": 5e-06,
+      "loss": 0.7352,
+      "num_input_tokens_seen": 37686136,
+      "step": 601
+    },
+    {
+      "epoch": 2.0,
+      "loss": 0.7832919359207153,
+      "loss_ce": 0.0008212359971366823,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 37686136,
+      "step": 601
+    },
+    {
+      "epoch": 2.0033277870216306,
+      "grad_norm": 18.40623664855957,
+      "learning_rate": 5e-06,
+      "loss": 1.1938,
+      "num_input_tokens_seen": 37749472,
+      "step": 602
+    },
+    {
+      "epoch": 2.0033277870216306,
+      "loss": 1.2827110290527344,
+      "loss_ce": 0.00048443087143823504,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.07763671875,
+      "loss_xval": 1.28125,
+      "num_input_tokens_seen": 37749472,
+      "step": 602
+    },
+    {
+      "epoch": 2.0066555740432612,
+      "grad_norm": 12.205756187438965,
+      "learning_rate": 5e-06,
+      "loss": 0.5865,
+      "num_input_tokens_seen": 37811236,
+      "step": 603
+    },
+    {
+      "epoch": 2.0066555740432612,
+      "loss": 0.42705678939819336,
+      "loss_ce": 0.00029895632178522646,
+      "loss_iou": 0.0,
+      "loss_num": 0.08544921875,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 37811236,
+      "step": 603
+    },
+    {
+      "epoch": 2.009983361064892,
+      "grad_norm": 10.160212516784668,
+      "learning_rate": 5e-06,
+      "loss": 0.5894,
+      "num_input_tokens_seen": 37872200,
+      "step": 604
+    },
+    {
+      "epoch": 2.009983361064892,
+      "loss": 0.7009243965148926,
+      "loss_ce": 0.00024079754075501114,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.69921875,
+      "num_input_tokens_seen": 37872200,
+      "step": 604
+    },
+    {
+      "epoch": 2.0133111480865225,
+      "grad_norm": 13.080687522888184,
+      "learning_rate": 5e-06,
+      "loss": 0.7835,
+      "num_input_tokens_seen": 37933716,
+      "step": 605
+    },
+    {
+      "epoch": 2.0133111480865225,
+      "loss": 1.0986953973770142,
+      "loss_ce": 6.257939094211906e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.06640625,
+      "loss_xval": 1.1015625,
+      "num_input_tokens_seen": 37933716,
+      "step": 605
+    },
+    {
+      "epoch": 2.016638935108153,
+      "grad_norm": 10.649163246154785,
+      "learning_rate": 5e-06,
+      "loss": 0.7663,
+      "num_input_tokens_seen": 37994536,
+      "step": 606
+    },
+    {
+      "epoch": 2.016638935108153,
+      "loss": 0.9843522310256958,
+      "loss_ce": 0.0007096048793755472,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0869140625,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 37994536,
+      "step": 606
+    },
+    {
+      "epoch": 2.0199667221297837,
+      "grad_norm": 10.702330589294434,
+      "learning_rate": 5e-06,
+      "loss": 0.6246,
+      "num_input_tokens_seen": 38057508,
+      "step": 607
+    },
+    {
+      "epoch": 2.0199667221297837,
+      "loss": 0.6482642889022827,
+      "loss_ce": 0.0010474994778633118,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 38057508,
+      "step": 607
+    },
+    {
+      "epoch": 2.0232945091514143,
+      "grad_norm": 12.19349479675293,
+      "learning_rate": 5e-06,
+      "loss": 0.7891,
+      "num_input_tokens_seen": 38119664,
+      "step": 608
+    },
+    {
+      "epoch": 2.0232945091514143,
+      "loss": 0.8228746652603149,
+      "loss_ce": 0.0001207729583256878,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 38119664,
+      "step": 608
+    },
+    {
+      "epoch": 2.026622296173045,
+      "grad_norm": 30.340932846069336,
+      "learning_rate": 5e-06,
+      "loss": 1.0127,
+      "num_input_tokens_seen": 38184528,
+      "step": 609
+    },
+    {
+      "epoch": 2.026622296173045,
+      "loss": 1.135079026222229,
+      "loss_ce": 0.0008016878855414689,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 1.1328125,
+      "num_input_tokens_seen": 38184528,
+      "step": 609
+    },
+    {
+      "epoch": 2.0299500831946755,
+      "grad_norm": 19.663864135742188,
+      "learning_rate": 5e-06,
+      "loss": 1.0414,
+      "num_input_tokens_seen": 38247428,
+      "step": 610
+    },
+    {
+      "epoch": 2.0299500831946755,
+      "loss": 1.2719979286193848,
+      "loss_ce": 0.00026946913567371666,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0634765625,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 38247428,
+      "step": 610
+    },
+    {
+      "epoch": 2.033277870216306,
+      "grad_norm": 38.78046798706055,
+      "learning_rate": 5e-06,
+      "loss": 0.6704,
+      "num_input_tokens_seen": 38309384,
+      "step": 611
+    },
+    {
+      "epoch": 2.033277870216306,
+      "loss": 0.9232205152511597,
+      "loss_ce": 0.0001247778272954747,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 38309384,
+      "step": 611
+    },
+    {
+      "epoch": 2.0366056572379367,
+      "grad_norm": 21.593708038330078,
+      "learning_rate": 5e-06,
+      "loss": 0.8872,
+      "num_input_tokens_seen": 38373688,
+      "step": 612
+    },
+    {
+      "epoch": 2.0366056572379367,
+      "loss": 0.8260167837142944,
+      "loss_ce": 8.90689916559495e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 38373688,
+      "step": 612
+    },
+    {
+      "epoch": 2.0399334442595674,
+      "grad_norm": 28.829166412353516,
+      "learning_rate": 5e-06,
+      "loss": 1.195,
+      "num_input_tokens_seen": 38435916,
+      "step": 613
+    },
+    {
+      "epoch": 2.0399334442595674,
+      "loss": 1.4014410972595215,
+      "loss_ce": 0.000562186527531594,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.091796875,
+      "loss_xval": 1.3984375,
+      "num_input_tokens_seen": 38435916,
+      "step": 613
+    },
+    {
+      "epoch": 2.043261231281198,
+      "grad_norm": 30.802501678466797,
+      "learning_rate": 5e-06,
+      "loss": 0.928,
+      "num_input_tokens_seen": 38499540,
+      "step": 614
+    },
+    {
+      "epoch": 2.043261231281198,
+      "loss": 1.0729480981826782,
+      "loss_ce": 0.017283970490098,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0751953125,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 38499540,
+      "step": 614
+    },
+    {
+      "epoch": 2.0465890183028286,
+      "grad_norm": 14.069050788879395,
+      "learning_rate": 5e-06,
+      "loss": 0.9339,
+      "num_input_tokens_seen": 38563296,
+      "step": 615
+    },
+    {
+      "epoch": 2.0465890183028286,
+      "loss": 1.0433433055877686,
+      "loss_ce": 0.01477888599038124,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 38563296,
+      "step": 615
+    },
+    {
+      "epoch": 2.049916805324459,
+      "grad_norm": 8.753564834594727,
+      "learning_rate": 5e-06,
+      "loss": 0.71,
+      "num_input_tokens_seen": 38624788,
+      "step": 616
+    },
+    {
+      "epoch": 2.049916805324459,
+      "loss": 0.8744981288909912,
+      "loss_ce": 0.006578153930604458,
+      "loss_iou": 0.28125,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 38624788,
+      "step": 616
+    },
+    {
+      "epoch": 2.05324459234609,
+      "grad_norm": 21.490896224975586,
+      "learning_rate": 5e-06,
+      "loss": 1.1,
+      "num_input_tokens_seen": 38689684,
+      "step": 617
+    },
+    {
+      "epoch": 2.05324459234609,
+      "loss": 1.2144798040390015,
+      "loss_ce": 0.0028098488692194223,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.064453125,
+      "loss_xval": 1.2109375,
+      "num_input_tokens_seen": 38689684,
+      "step": 617
+    },
+    {
+      "epoch": 2.0565723793677204,
+      "grad_norm": 14.222006797790527,
+      "learning_rate": 5e-06,
+      "loss": 0.6892,
+      "num_input_tokens_seen": 38752120,
+      "step": 618
+    },
+    {
+      "epoch": 2.0565723793677204,
+      "loss": 0.7122447490692139,
+      "loss_ce": 0.0017955549992620945,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 38752120,
+      "step": 618
+    },
+    {
+      "epoch": 2.059900166389351,
+      "grad_norm": 13.638813972473145,
+      "learning_rate": 5e-06,
+      "loss": 0.8149,
+      "num_input_tokens_seen": 38812068,
+      "step": 619
+    },
+    {
+      "epoch": 2.059900166389351,
+      "loss": 1.018181562423706,
+      "loss_ce": 0.00011516415543155745,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.087890625,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 38812068,
+      "step": 619
+    },
+    {
+      "epoch": 2.0632279534109816,
+      "grad_norm": 16.652631759643555,
+      "learning_rate": 5e-06,
+      "loss": 0.7474,
+      "num_input_tokens_seen": 38874664,
+      "step": 620
+    },
+    {
+      "epoch": 2.0632279534109816,
+      "loss": 0.42710453271865845,
+      "loss_ce": 0.00028570188442245126,
+      "loss_iou": 0.09228515625,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 38874664,
+      "step": 620
+    },
+    {
+      "epoch": 2.0665557404326123,
+      "grad_norm": 22.291889190673828,
+      "learning_rate": 5e-06,
+      "loss": 0.9719,
+      "num_input_tokens_seen": 38938248,
+      "step": 621
+    },
+    {
+      "epoch": 2.0665557404326123,
+      "loss": 0.9289200305938721,
+      "loss_ce": 0.00020914892957080156,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 38938248,
+      "step": 621
+    },
+    {
+      "epoch": 2.069883527454243,
+      "grad_norm": 21.66976547241211,
+      "learning_rate": 5e-06,
+      "loss": 0.6807,
+      "num_input_tokens_seen": 39000808,
+      "step": 622
+    },
+    {
+      "epoch": 2.069883527454243,
+      "loss": 0.7833365201950073,
+      "loss_ce": 0.0003775203076656908,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 39000808,
+      "step": 622
+    },
+    {
+      "epoch": 2.0732113144758735,
+      "grad_norm": 13.521284103393555,
+      "learning_rate": 5e-06,
+      "loss": 0.9122,
+      "num_input_tokens_seen": 39064556,
+      "step": 623
+    },
+    {
+      "epoch": 2.0732113144758735,
+      "loss": 0.8654112815856934,
+      "loss_ce": 0.0004210532642900944,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.86328125,
+      "num_input_tokens_seen": 39064556,
+      "step": 623
+    },
+    {
+      "epoch": 2.076539101497504,
+      "grad_norm": 9.097267150878906,
+      "learning_rate": 5e-06,
+      "loss": 0.7518,
+      "num_input_tokens_seen": 39127724,
+      "step": 624
+    },
+    {
+      "epoch": 2.076539101497504,
+      "loss": 0.8574149012565613,
+      "loss_ce": 0.0007254519150592387,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 39127724,
+      "step": 624
+    },
+    {
+      "epoch": 2.0798668885191347,
+      "grad_norm": 62.03600311279297,
+      "learning_rate": 5e-06,
+      "loss": 0.779,
+      "num_input_tokens_seen": 39191480,
+      "step": 625
+    },
+    {
+      "epoch": 2.0798668885191347,
+      "loss": 0.8742235898971558,
+      "loss_ce": 7.806658686604351e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.875,
+      "num_input_tokens_seen": 39191480,
+      "step": 625
+    },
+    {
+      "epoch": 2.0831946755407653,
+      "grad_norm": 37.185428619384766,
+      "learning_rate": 5e-06,
+      "loss": 0.7275,
+      "num_input_tokens_seen": 39252676,
+      "step": 626
+    },
+    {
+      "epoch": 2.0831946755407653,
+      "loss": 0.7918392419815063,
+      "loss_ce": 0.0015560202300548553,
+      "loss_iou": 0.234375,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 39252676,
+      "step": 626
+    },
+    {
+      "epoch": 2.086522462562396,
+      "grad_norm": 13.905028343200684,
+      "learning_rate": 5e-06,
+      "loss": 0.5983,
+      "num_input_tokens_seen": 39314092,
+      "step": 627
+    },
+    {
+      "epoch": 2.086522462562396,
+      "loss": 0.29983216524124146,
+      "loss_ce": 0.00045470561599358916,
+      "loss_iou": 0.0,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 39314092,
+      "step": 627
+    },
+    {
+      "epoch": 2.0898502495840265,
+      "grad_norm": 21.244510650634766,
+      "learning_rate": 5e-06,
+      "loss": 0.8911,
+      "num_input_tokens_seen": 39377644,
+      "step": 628
+    },
+    {
+      "epoch": 2.0898502495840265,
+      "loss": 0.9056867361068726,
+      "loss_ce": 0.00016913384024519473,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 39377644,
+      "step": 628
+    },
+    {
+      "epoch": 2.093178036605657,
+      "grad_norm": 20.721181869506836,
+      "learning_rate": 5e-06,
+      "loss": 1.0455,
+      "num_input_tokens_seen": 39441688,
+      "step": 629
+    },
+    {
+      "epoch": 2.093178036605657,
+      "loss": 1.0174716711044312,
+      "loss_ce": 0.004288055002689362,
+      "loss_iou": 0.328125,
+      "loss_num": 0.07177734375,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 39441688,
+      "step": 629
+    },
+    {
+      "epoch": 2.0965058236272878,
+      "grad_norm": 12.86941909790039,
+      "learning_rate": 5e-06,
+      "loss": 0.7168,
+      "num_input_tokens_seen": 39503164,
+      "step": 630
+    },
+    {
+      "epoch": 2.0965058236272878,
+      "loss": 0.9291834831237793,
+      "loss_ce": 0.0034022473264485598,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.92578125,
+      "num_input_tokens_seen": 39503164,
+      "step": 630
+    },
+    {
+      "epoch": 2.0998336106489184,
+      "grad_norm": 10.400750160217285,
+      "learning_rate": 5e-06,
+      "loss": 0.8992,
+      "num_input_tokens_seen": 39566940,
+      "step": 631
+    },
+    {
+      "epoch": 2.0998336106489184,
+      "loss": 0.94353187084198,
+      "loss_ce": 0.001393245765939355,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 39566940,
+      "step": 631
+    },
+    {
+      "epoch": 2.103161397670549,
+      "grad_norm": 9.461636543273926,
+      "learning_rate": 5e-06,
+      "loss": 0.6036,
+      "num_input_tokens_seen": 39629024,
+      "step": 632
+    },
+    {
+      "epoch": 2.103161397670549,
+      "loss": 0.33993566036224365,
+      "loss_ce": 0.006317485123872757,
+      "loss_iou": 0.06982421875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 39629024,
+      "step": 632
+    },
+    {
+      "epoch": 2.1064891846921796,
+      "grad_norm": 12.25512981414795,
+      "learning_rate": 5e-06,
+      "loss": 0.6903,
+      "num_input_tokens_seen": 39691612,
+      "step": 633
+    },
+    {
+      "epoch": 2.1064891846921796,
+      "loss": 0.667500376701355,
+      "loss_ce": 1.9936600438086316e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 39691612,
+      "step": 633
+    },
+    {
+      "epoch": 2.10981697171381,
+      "grad_norm": 12.845260620117188,
+      "learning_rate": 5e-06,
+      "loss": 1.0146,
+      "num_input_tokens_seen": 39753308,
+      "step": 634
+    },
+    {
+      "epoch": 2.10981697171381,
+      "loss": 0.9783458113670349,
+      "loss_ce": 0.0008067445596680045,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 39753308,
+      "step": 634
+    },
+    {
+      "epoch": 2.113144758735441,
+      "grad_norm": 12.062701225280762,
+      "learning_rate": 5e-06,
+      "loss": 0.9782,
+      "num_input_tokens_seen": 39815888,
+      "step": 635
+    },
+    {
+      "epoch": 2.113144758735441,
+      "loss": 1.166348934173584,
+      "loss_ce": 0.0003332831838633865,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.11669921875,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 39815888,
+      "step": 635
+    },
+    {
+      "epoch": 2.1164725457570714,
+      "grad_norm": 13.48031234741211,
+      "learning_rate": 5e-06,
+      "loss": 0.9052,
+      "num_input_tokens_seen": 39877392,
+      "step": 636
+    },
+    {
+      "epoch": 2.1164725457570714,
+      "loss": 0.8594918251037598,
+      "loss_ce": 0.0008491812041029334,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0859375,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 39877392,
+      "step": 636
+    },
+    {
+      "epoch": 2.119800332778702,
+      "grad_norm": 11.958724975585938,
+      "learning_rate": 5e-06,
+      "loss": 0.8913,
+      "num_input_tokens_seen": 39940516,
+      "step": 637
+    },
+    {
+      "epoch": 2.119800332778702,
+      "loss": 0.792966902256012,
+      "loss_ce": 0.009763746522367,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 39940516,
+      "step": 637
+    },
+    {
+      "epoch": 2.1231281198003327,
+      "grad_norm": 15.40645694732666,
+      "learning_rate": 5e-06,
+      "loss": 0.718,
+      "num_input_tokens_seen": 40003388,
+      "step": 638
+    },
+    {
+      "epoch": 2.1231281198003327,
+      "loss": 0.7531487941741943,
+      "loss_ce": 0.00021912308875471354,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 40003388,
+      "step": 638
+    },
+    {
+      "epoch": 2.1264559068219633,
+      "grad_norm": 32.53299331665039,
+      "learning_rate": 5e-06,
+      "loss": 1.1154,
+      "num_input_tokens_seen": 40066328,
+      "step": 639
+    },
+    {
+      "epoch": 2.1264559068219633,
+      "loss": 1.099750280380249,
+      "loss_ce": 0.0001409617834724486,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.061279296875,
+      "loss_xval": 1.1015625,
+      "num_input_tokens_seen": 40066328,
+      "step": 639
+    },
+    {
+      "epoch": 2.129783693843594,
+      "grad_norm": 33.41072082519531,
+      "learning_rate": 5e-06,
+      "loss": 0.8616,
+      "num_input_tokens_seen": 40127316,
+      "step": 640
+    },
+    {
+      "epoch": 2.129783693843594,
+      "loss": 0.9770931601524353,
+      "loss_ce": 0.0015072214882820845,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 40127316,
+      "step": 640
+    },
+    {
+      "epoch": 2.1331114808652245,
+      "grad_norm": 13.928618431091309,
+      "learning_rate": 5e-06,
+      "loss": 1.0466,
+      "num_input_tokens_seen": 40190528,
+      "step": 641
+    },
+    {
+      "epoch": 2.1331114808652245,
+      "loss": 1.031181812286377,
+      "loss_ce": 0.0015187868848443031,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 40190528,
+      "step": 641
+    },
+    {
+      "epoch": 2.136439267886855,
+      "grad_norm": 19.280794143676758,
+      "learning_rate": 5e-06,
+      "loss": 0.9631,
+      "num_input_tokens_seen": 40252724,
+      "step": 642
+    },
+    {
+      "epoch": 2.136439267886855,
+      "loss": 0.7468453645706177,
+      "loss_ce": 0.0005074806977063417,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 40252724,
+      "step": 642
+    },
+    {
+      "epoch": 2.1397670549084857,
+      "grad_norm": 9.668257713317871,
+      "learning_rate": 5e-06,
+      "loss": 0.8991,
+      "num_input_tokens_seen": 40315860,
+      "step": 643
+    },
+    {
+      "epoch": 2.1397670549084857,
+      "loss": 0.768240213394165,
+      "loss_ce": 0.0001738157297950238,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 40315860,
+      "step": 643
+    },
+    {
+      "epoch": 2.1430948419301163,
+      "grad_norm": 16.013214111328125,
+      "learning_rate": 5e-06,
+      "loss": 0.97,
+      "num_input_tokens_seen": 40379176,
+      "step": 644
+    },
+    {
+      "epoch": 2.1430948419301163,
+      "loss": 0.7990410327911377,
+      "loss_ce": 0.001677798223681748,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 40379176,
+      "step": 644
+    },
+    {
+      "epoch": 2.146422628951747,
+      "grad_norm": 8.998737335205078,
+      "learning_rate": 5e-06,
+      "loss": 0.8889,
+      "num_input_tokens_seen": 40440688,
+      "step": 645
+    },
+    {
+      "epoch": 2.146422628951747,
+      "loss": 1.1174098253250122,
+      "loss_ce": 0.00022229723981581628,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.1240234375,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 40440688,
+      "step": 645
+    },
+    {
+      "epoch": 2.1497504159733776,
+      "grad_norm": 12.142661094665527,
+      "learning_rate": 5e-06,
+      "loss": 0.7721,
+      "num_input_tokens_seen": 40502032,
+      "step": 646
+    },
+    {
+      "epoch": 2.1497504159733776,
+      "loss": 0.9419033527374268,
+      "loss_ce": 8.781585165706929e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.08544921875,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 40502032,
+      "step": 646
+    },
+    {
+      "epoch": 2.153078202995008,
+      "grad_norm": 12.413509368896484,
+      "learning_rate": 5e-06,
+      "loss": 0.6306,
+      "num_input_tokens_seen": 40563216,
+      "step": 647
+    },
+    {
+      "epoch": 2.153078202995008,
+      "loss": 0.6236965656280518,
+      "loss_ce": 0.0004055765166413039,
+      "loss_iou": 0.15625,
+      "loss_num": 0.062255859375,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 40563216,
+      "step": 647
+    },
+    {
+      "epoch": 2.156405990016639,
+      "grad_norm": 12.308266639709473,
+      "learning_rate": 5e-06,
+      "loss": 1.0149,
+      "num_input_tokens_seen": 40626760,
+      "step": 648
+    },
+    {
+      "epoch": 2.156405990016639,
+      "loss": 1.0296001434326172,
+      "loss_ce": 0.0006693830946460366,
+      "loss_iou": 0.328125,
+      "loss_num": 0.07421875,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 40626760,
+      "step": 648
+    },
+    {
+      "epoch": 2.1597337770382694,
+      "grad_norm": 8.219746589660645,
+      "learning_rate": 5e-06,
+      "loss": 1.0916,
+      "num_input_tokens_seen": 40689820,
+      "step": 649
+    },
+    {
+      "epoch": 2.1597337770382694,
+      "loss": 1.043705940246582,
+      "loss_ce": 0.0009814061922952533,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.07470703125,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 40689820,
+      "step": 649
+    },
+    {
+      "epoch": 2.1630615640599,
+      "grad_norm": 16.483530044555664,
+      "learning_rate": 5e-06,
+      "loss": 1.3246,
+      "num_input_tokens_seen": 40753476,
+      "step": 650
+    },
+    {
+      "epoch": 2.1630615640599,
+      "loss": 1.3722883462905884,
+      "loss_ce": 0.0011946188751608133,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.08740234375,
+      "loss_xval": 1.375,
+      "num_input_tokens_seen": 40753476,
+      "step": 650
+    },
+    {
+      "epoch": 2.1663893510815306,
+      "grad_norm": 15.433051109313965,
+      "learning_rate": 5e-06,
+      "loss": 0.8102,
+      "num_input_tokens_seen": 40816652,
+      "step": 651
+    },
+    {
+      "epoch": 2.1663893510815306,
+      "loss": 0.710527241230011,
+      "loss_ce": 0.0022752864751964808,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.70703125,
+      "num_input_tokens_seen": 40816652,
+      "step": 651
+    },
+    {
+      "epoch": 2.1697171381031612,
+      "grad_norm": 6.271609306335449,
+      "learning_rate": 5e-06,
+      "loss": 0.4691,
+      "num_input_tokens_seen": 40877216,
+      "step": 652
+    },
+    {
+      "epoch": 2.1697171381031612,
+      "loss": 0.44094425439834595,
+      "loss_ce": 2.6259316655341536e-05,
+      "loss_iou": 0.07275390625,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 40877216,
+      "step": 652
+    },
+    {
+      "epoch": 2.173044925124792,
+      "grad_norm": 13.27111530303955,
+      "learning_rate": 5e-06,
+      "loss": 0.815,
+      "num_input_tokens_seen": 40938780,
+      "step": 653
+    },
+    {
+      "epoch": 2.173044925124792,
+      "loss": 0.9975978136062622,
+      "loss_ce": 0.000771641731262207,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0859375,
+      "loss_xval": 0.99609375,
+      "num_input_tokens_seen": 40938780,
+      "step": 653
+    },
+    {
+      "epoch": 2.1763727121464225,
+      "grad_norm": 13.250811576843262,
+      "learning_rate": 5e-06,
+      "loss": 0.8086,
+      "num_input_tokens_seen": 41002692,
+      "step": 654
+    },
+    {
+      "epoch": 2.1763727121464225,
+      "loss": 0.5700994729995728,
+      "loss_ce": 0.00027528181090019643,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 41002692,
+      "step": 654
+    },
+    {
+      "epoch": 2.179700499168053,
+      "grad_norm": 43.69986343383789,
+      "learning_rate": 5e-06,
+      "loss": 1.2292,
+      "num_input_tokens_seen": 41065588,
+      "step": 655
+    },
+    {
+      "epoch": 2.179700499168053,
+      "loss": 1.126779317855835,
+      "loss_ce": 0.0010468140244483948,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.07080078125,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 41065588,
+      "step": 655
+    },
+    {
+      "epoch": 2.1830282861896837,
+      "grad_norm": 23.3741455078125,
+      "learning_rate": 5e-06,
+      "loss": 0.7644,
+      "num_input_tokens_seen": 41127908,
+      "step": 656
+    },
+    {
+      "epoch": 2.1830282861896837,
+      "loss": 0.8155651092529297,
+      "loss_ce": 0.0005016563227400184,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 41127908,
+      "step": 656
+    },
+    {
+      "epoch": 2.1863560732113143,
+      "grad_norm": 16.306617736816406,
+      "learning_rate": 5e-06,
+      "loss": 1.0647,
+      "num_input_tokens_seen": 41190336,
+      "step": 657
+    },
+    {
+      "epoch": 2.1863560732113143,
+      "loss": 1.1568994522094727,
+      "loss_ce": 0.00016120978398248553,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.07763671875,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 41190336,
+      "step": 657
+    },
+    {
+      "epoch": 2.189683860232945,
+      "grad_norm": 11.455361366271973,
+      "learning_rate": 5e-06,
+      "loss": 0.7928,
+      "num_input_tokens_seen": 41252080,
+      "step": 658
+    },
+    {
+      "epoch": 2.189683860232945,
+      "loss": 0.7935343980789185,
+      "loss_ce": 0.0005656481371261179,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 41252080,
+      "step": 658
+    },
+    {
+      "epoch": 2.1930116472545755,
+      "grad_norm": 14.715596199035645,
+      "learning_rate": 5e-06,
+      "loss": 0.9227,
+      "num_input_tokens_seen": 41314652,
+      "step": 659
+    },
+    {
+      "epoch": 2.1930116472545755,
+      "loss": 0.932908833026886,
+      "loss_ce": 4.7503406676696613e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 41314652,
+      "step": 659
+    },
+    {
+      "epoch": 2.196339434276206,
+      "grad_norm": 31.348485946655273,
+      "learning_rate": 5e-06,
+      "loss": 0.9761,
+      "num_input_tokens_seen": 41377348,
+      "step": 660
+    },
+    {
+      "epoch": 2.196339434276206,
+      "loss": 0.9302453398704529,
+      "loss_ce": 6.955982826184481e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 41377348,
+      "step": 660
+    },
+    {
+      "epoch": 2.1996672212978368,
+      "grad_norm": 35.344242095947266,
+      "learning_rate": 5e-06,
+      "loss": 1.0837,
+      "num_input_tokens_seen": 41440992,
+      "step": 661
+    },
+    {
+      "epoch": 2.1996672212978368,
+      "loss": 0.8363819122314453,
+      "loss_ce": 0.00020026677520945668,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 41440992,
+      "step": 661
+    },
+    {
+      "epoch": 2.2029950083194674,
+      "grad_norm": 26.1990909576416,
+      "learning_rate": 5e-06,
+      "loss": 0.8925,
+      "num_input_tokens_seen": 41504132,
+      "step": 662
+    },
+    {
+      "epoch": 2.2029950083194674,
+      "loss": 0.7424072027206421,
+      "loss_ce": 0.00046388505143113434,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 41504132,
+      "step": 662
+    },
+    {
+      "epoch": 2.206322795341098,
+      "grad_norm": 9.237807273864746,
+      "learning_rate": 5e-06,
+      "loss": 0.6939,
+      "num_input_tokens_seen": 41565620,
+      "step": 663
+    },
+    {
+      "epoch": 2.206322795341098,
+      "loss": 0.7877798080444336,
+      "loss_ce": 6.009737990098074e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.07568359375,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 41565620,
+      "step": 663
+    },
+    {
+      "epoch": 2.2096505823627286,
+      "grad_norm": 15.479573249816895,
+      "learning_rate": 5e-06,
+      "loss": 1.0272,
+      "num_input_tokens_seen": 41629236,
+      "step": 664
+    },
+    {
+      "epoch": 2.2096505823627286,
+      "loss": 1.0712847709655762,
+      "loss_ce": 0.00048389771836809814,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0810546875,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 41629236,
+      "step": 664
+    },
+    {
+      "epoch": 2.212978369384359,
+      "grad_norm": 32.5822868347168,
+      "learning_rate": 5e-06,
+      "loss": 0.9038,
+      "num_input_tokens_seen": 41692904,
+      "step": 665
+    },
+    {
+      "epoch": 2.212978369384359,
+      "loss": 0.9757400751113892,
+      "loss_ce": 0.0011306863743811846,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.07958984375,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 41692904,
+      "step": 665
+    },
+    {
+      "epoch": 2.21630615640599,
+      "grad_norm": 15.892915725708008,
+      "learning_rate": 5e-06,
+      "loss": 0.9478,
+      "num_input_tokens_seen": 41755028,
+      "step": 666
+    },
+    {
+      "epoch": 2.21630615640599,
+      "loss": 1.2628767490386963,
+      "loss_ce": 0.0004256761458236724,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.08251953125,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 41755028,
+      "step": 666
+    },
+    {
+      "epoch": 2.2196339434276204,
+      "grad_norm": 32.91965103149414,
+      "learning_rate": 5e-06,
+      "loss": 1.0484,
+      "num_input_tokens_seen": 41819808,
+      "step": 667
+    },
+    {
+      "epoch": 2.2196339434276204,
+      "loss": 1.0486059188842773,
+      "loss_ce": 0.001242599100805819,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 41819808,
+      "step": 667
+    },
+    {
+      "epoch": 2.222961730449251,
+      "grad_norm": 6.644551753997803,
+      "learning_rate": 5e-06,
+      "loss": 0.8515,
+      "num_input_tokens_seen": 41883088,
+      "step": 668
+    },
+    {
+      "epoch": 2.222961730449251,
+      "loss": 0.8012328147888184,
+      "loss_ce": 0.00032947398722171783,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.078125,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 41883088,
+      "step": 668
+    },
+    {
+      "epoch": 2.2262895174708817,
+      "grad_norm": 19.913354873657227,
+      "learning_rate": 5e-06,
+      "loss": 0.9162,
+      "num_input_tokens_seen": 41946620,
+      "step": 669
+    },
+    {
+      "epoch": 2.2262895174708817,
+      "loss": 0.6975362300872803,
+      "loss_ce": 0.0007588434964418411,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 41946620,
+      "step": 669
+    },
+    {
+      "epoch": 2.2296173044925123,
+      "grad_norm": 25.868154525756836,
+      "learning_rate": 5e-06,
+      "loss": 0.8179,
+      "num_input_tokens_seen": 42009140,
+      "step": 670
+    },
+    {
+      "epoch": 2.2296173044925123,
+      "loss": 0.7486177086830139,
+      "loss_ce": 8.254170825239271e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 42009140,
+      "step": 670
+    },
+    {
+      "epoch": 2.232945091514143,
+      "grad_norm": 16.261491775512695,
+      "learning_rate": 5e-06,
+      "loss": 0.8723,
+      "num_input_tokens_seen": 42069640,
+      "step": 671
+    },
+    {
+      "epoch": 2.232945091514143,
+      "loss": 1.1612319946289062,
+      "loss_ce": 0.0005875469068996608,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.087890625,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 42069640,
+      "step": 671
+    },
+    {
+      "epoch": 2.2362728785357735,
+      "grad_norm": 9.516688346862793,
+      "learning_rate": 5e-06,
+      "loss": 0.8249,
+      "num_input_tokens_seen": 42132272,
+      "step": 672
+    },
+    {
+      "epoch": 2.2362728785357735,
+      "loss": 0.815399706363678,
+      "loss_ce": 0.0001836877636378631,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 42132272,
+      "step": 672
+    },
+    {
+      "epoch": 2.239600665557404,
+      "grad_norm": 12.779966354370117,
+      "learning_rate": 5e-06,
+      "loss": 0.7785,
+      "num_input_tokens_seen": 42194892,
+      "step": 673
+    },
+    {
+      "epoch": 2.239600665557404,
+      "loss": 0.6607878804206848,
+      "loss_ce": 0.00014333476428873837,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 42194892,
+      "step": 673
+    },
+    {
+      "epoch": 2.2429284525790347,
+      "grad_norm": 10.946259498596191,
+      "learning_rate": 5e-06,
+      "loss": 0.7948,
+      "num_input_tokens_seen": 42256156,
+      "step": 674
+    },
+    {
+      "epoch": 2.2429284525790347,
+      "loss": 0.764301061630249,
+      "loss_ce": 0.0003850538341794163,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 42256156,
+      "step": 674
+    },
+    {
+      "epoch": 2.2462562396006653,
+      "grad_norm": 9.564322471618652,
+      "learning_rate": 5e-06,
+      "loss": 0.6524,
+      "num_input_tokens_seen": 42318292,
+      "step": 675
+    },
+    {
+      "epoch": 2.2462562396006653,
+      "loss": 0.6682754158973694,
+      "loss_ce": 6.24852254986763e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 42318292,
+      "step": 675
+    },
+    {
+      "epoch": 2.249584026622296,
+      "grad_norm": 13.924878120422363,
+      "learning_rate": 5e-06,
+      "loss": 0.8124,
+      "num_input_tokens_seen": 42377848,
+      "step": 676
+    },
+    {
+      "epoch": 2.249584026622296,
+      "loss": 0.6967849731445312,
+      "loss_ce": 0.0017471231985837221,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 42377848,
+      "step": 676
+    },
+    {
+      "epoch": 2.2529118136439266,
+      "grad_norm": 16.81814956665039,
+      "learning_rate": 5e-06,
+      "loss": 0.902,
+      "num_input_tokens_seen": 42439512,
+      "step": 677
+    },
+    {
+      "epoch": 2.2529118136439266,
+      "loss": 0.8860781788825989,
+      "loss_ce": 0.00033597912988625467,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 42439512,
+      "step": 677
+    },
+    {
+      "epoch": 2.256239600665557,
+      "grad_norm": 11.292144775390625,
+      "learning_rate": 5e-06,
+      "loss": 1.0798,
+      "num_input_tokens_seen": 42502844,
+      "step": 678
+    },
+    {
+      "epoch": 2.256239600665557,
+      "loss": 1.1715800762176514,
+      "loss_ce": 0.00019335387332830578,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0771484375,
+      "loss_xval": 1.171875,
+      "num_input_tokens_seen": 42502844,
+      "step": 678
+    },
+    {
+      "epoch": 2.259567387687188,
+      "grad_norm": 7.578566551208496,
+      "learning_rate": 5e-06,
+      "loss": 1.047,
+      "num_input_tokens_seen": 42565372,
+      "step": 679
+    },
+    {
+      "epoch": 2.259567387687188,
+      "loss": 1.0372211933135986,
+      "loss_ce": 0.0010883715003728867,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0615234375,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 42565372,
+      "step": 679
+    },
+    {
+      "epoch": 2.2628951747088184,
+      "grad_norm": 17.18895721435547,
+      "learning_rate": 5e-06,
+      "loss": 0.9563,
+      "num_input_tokens_seen": 42627684,
+      "step": 680
+    },
+    {
+      "epoch": 2.2628951747088184,
+      "loss": 1.0924670696258545,
+      "loss_ce": 0.0006701992242597044,
+      "loss_iou": 0.34375,
+      "loss_num": 0.08056640625,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 42627684,
+      "step": 680
+    },
+    {
+      "epoch": 2.266222961730449,
+      "grad_norm": 9.680091857910156,
+      "learning_rate": 5e-06,
+      "loss": 1.0404,
+      "num_input_tokens_seen": 42691788,
+      "step": 681
+    },
+    {
+      "epoch": 2.266222961730449,
+      "loss": 0.8576643466949463,
+      "loss_ce": 0.00048660385073162615,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 42691788,
+      "step": 681
+    },
+    {
+      "epoch": 2.2695507487520796,
+      "grad_norm": 23.31703758239746,
+      "learning_rate": 5e-06,
+      "loss": 0.8207,
+      "num_input_tokens_seen": 42754224,
+      "step": 682
+    },
+    {
+      "epoch": 2.2695507487520796,
+      "loss": 0.6044555902481079,
+      "loss_ce": 8.549916674382985e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 42754224,
+      "step": 682
+    },
+    {
+      "epoch": 2.2728785357737102,
+      "grad_norm": 10.370177268981934,
+      "learning_rate": 5e-06,
+      "loss": 0.9969,
+      "num_input_tokens_seen": 42817536,
+      "step": 683
+    },
+    {
+      "epoch": 2.2728785357737102,
+      "loss": 1.0521934032440186,
+      "loss_ce": 0.0008017598884180188,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.08544921875,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 42817536,
+      "step": 683
+    },
+    {
+      "epoch": 2.276206322795341,
+      "grad_norm": 11.924843788146973,
+      "learning_rate": 5e-06,
+      "loss": 1.1041,
+      "num_input_tokens_seen": 42880584,
+      "step": 684
+    },
+    {
+      "epoch": 2.276206322795341,
+      "loss": 0.924105703830719,
+      "loss_ce": 0.0005217483267188072,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 42880584,
+      "step": 684
+    },
+    {
+      "epoch": 2.2795341098169715,
+      "grad_norm": 12.28469467163086,
+      "learning_rate": 5e-06,
+      "loss": 0.7665,
+      "num_input_tokens_seen": 42943528,
+      "step": 685
+    },
+    {
+      "epoch": 2.2795341098169715,
+      "loss": 0.8234906196594238,
+      "loss_ce": 0.00012639828491955996,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 42943528,
+      "step": 685
+    },
+    {
+      "epoch": 2.2828618968386025,
+      "grad_norm": 38.641693115234375,
+      "learning_rate": 5e-06,
+      "loss": 0.9702,
+      "num_input_tokens_seen": 43007544,
+      "step": 686
+    },
+    {
+      "epoch": 2.2828618968386025,
+      "loss": 0.8214001655578613,
+      "loss_ce": 0.00011106484453193843,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 43007544,
+      "step": 686
+    },
+    {
+      "epoch": 2.286189683860233,
+      "grad_norm": 22.426542282104492,
+      "learning_rate": 5e-06,
+      "loss": 0.6703,
+      "num_input_tokens_seen": 43069176,
+      "step": 687
+    },
+    {
+      "epoch": 2.286189683860233,
+      "loss": 0.6367377042770386,
+      "loss_ce": 0.0007513945456594229,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 43069176,
+      "step": 687
+    },
+    {
+      "epoch": 2.2895174708818637,
+      "grad_norm": 27.776212692260742,
+      "learning_rate": 5e-06,
+      "loss": 0.798,
+      "num_input_tokens_seen": 43133544,
+      "step": 688
+    },
+    {
+      "epoch": 2.2895174708818637,
+      "loss": 0.882362961769104,
+      "loss_ce": 0.00016075739404186606,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 43133544,
+      "step": 688
+    },
+    {
+      "epoch": 2.2928452579034944,
+      "grad_norm": 9.201436996459961,
+      "learning_rate": 5e-06,
+      "loss": 1.0724,
+      "num_input_tokens_seen": 43197028,
+      "step": 689
+    },
+    {
+      "epoch": 2.2928452579034944,
+      "loss": 1.2933428287506104,
+      "loss_ce": 0.001350653124973178,
+      "loss_iou": 0.453125,
+      "loss_num": 0.07666015625,
+      "loss_xval": 1.2890625,
+      "num_input_tokens_seen": 43197028,
+      "step": 689
+    },
+    {
+      "epoch": 2.296173044925125,
+      "grad_norm": 11.568137168884277,
+      "learning_rate": 5e-06,
+      "loss": 0.728,
+      "num_input_tokens_seen": 43260444,
+      "step": 690
+    },
+    {
+      "epoch": 2.296173044925125,
+      "loss": 0.6836668252944946,
+      "loss_ce": 0.001293812645599246,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.68359375,
+      "num_input_tokens_seen": 43260444,
+      "step": 690
+    },
+    {
+      "epoch": 2.2995008319467556,
+      "grad_norm": 14.878931999206543,
+      "learning_rate": 5e-06,
+      "loss": 1.0874,
+      "num_input_tokens_seen": 43324336,
+      "step": 691
+    },
+    {
+      "epoch": 2.2995008319467556,
+      "loss": 1.3032047748565674,
+      "loss_ce": 0.0009586900705471635,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.1005859375,
+      "loss_xval": 1.3046875,
+      "num_input_tokens_seen": 43324336,
+      "step": 691
+    },
+    {
+      "epoch": 2.302828618968386,
+      "grad_norm": 15.7121000289917,
+      "learning_rate": 5e-06,
+      "loss": 0.6999,
+      "num_input_tokens_seen": 43386868,
+      "step": 692
+    },
+    {
+      "epoch": 2.302828618968386,
+      "loss": 0.6661006212234497,
+      "loss_ce": 0.00032910649315454066,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 43386868,
+      "step": 692
+    },
+    {
+      "epoch": 2.306156405990017,
+      "grad_norm": 10.086871147155762,
+      "learning_rate": 5e-06,
+      "loss": 0.7448,
+      "num_input_tokens_seen": 43449464,
+      "step": 693
+    },
+    {
+      "epoch": 2.306156405990017,
+      "loss": 0.49396997690200806,
+      "loss_ce": 0.0011721242917701602,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 43449464,
+      "step": 693
+    },
+    {
+      "epoch": 2.3094841930116474,
+      "grad_norm": 29.50054931640625,
+      "learning_rate": 5e-06,
+      "loss": 0.9566,
+      "num_input_tokens_seen": 43513148,
+      "step": 694
+    },
+    {
+      "epoch": 2.3094841930116474,
+      "loss": 1.0682692527770996,
+      "loss_ce": 0.0008863506955094635,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.06640625,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 43513148,
+      "step": 694
+    },
+    {
+      "epoch": 2.312811980033278,
+      "grad_norm": 11.244392395019531,
+      "learning_rate": 5e-06,
+      "loss": 0.7949,
+      "num_input_tokens_seen": 43572220,
+      "step": 695
+    },
+    {
+      "epoch": 2.312811980033278,
+      "loss": 0.8442493677139282,
+      "loss_ce": 0.00025524571537971497,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 43572220,
+      "step": 695
+    },
+    {
+      "epoch": 2.3161397670549086,
+      "grad_norm": 34.46859359741211,
+      "learning_rate": 5e-06,
+      "loss": 0.8627,
+      "num_input_tokens_seen": 43635076,
+      "step": 696
+    },
+    {
+      "epoch": 2.3161397670549086,
+      "loss": 0.9726455211639404,
+      "loss_ce": 0.00023341998166870326,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.97265625,
+      "num_input_tokens_seen": 43635076,
+      "step": 696
+    },
+    {
+      "epoch": 2.3194675540765393,
+      "grad_norm": 32.053955078125,
+      "learning_rate": 5e-06,
+      "loss": 1.0028,
+      "num_input_tokens_seen": 43698032,
+      "step": 697
+    },
+    {
+      "epoch": 2.3194675540765393,
+      "loss": 0.7314814925193787,
+      "loss_ce": 3.620800271164626e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 43698032,
+      "step": 697
+    },
+    {
+      "epoch": 2.32279534109817,
+      "grad_norm": 14.377496719360352,
+      "learning_rate": 5e-06,
+      "loss": 0.7126,
+      "num_input_tokens_seen": 43759968,
+      "step": 698
+    },
+    {
+      "epoch": 2.32279534109817,
+      "loss": 0.6350549459457397,
+      "loss_ce": 0.0002893089840654284,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 43759968,
+      "step": 698
+    },
+    {
+      "epoch": 2.3261231281198005,
+      "grad_norm": 10.184778213500977,
+      "learning_rate": 5e-06,
+      "loss": 0.8103,
+      "num_input_tokens_seen": 43823276,
+      "step": 699
+    },
+    {
+      "epoch": 2.3261231281198005,
+      "loss": 0.973152220249176,
+      "loss_ce": 0.0004960413789376616,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.97265625,
+      "num_input_tokens_seen": 43823276,
+      "step": 699
+    },
+    {
+      "epoch": 2.329450915141431,
+      "grad_norm": 10.185461044311523,
+      "learning_rate": 5e-06,
+      "loss": 0.9135,
+      "num_input_tokens_seen": 43886252,
+      "step": 700
+    },
+    {
+      "epoch": 2.329450915141431,
+      "loss": 0.7423095107078552,
+      "loss_ce": 0.0008544151787646115,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 43886252,
+      "step": 700
+    },
+    {
+      "epoch": 2.3327787021630617,
+      "grad_norm": 14.283385276794434,
+      "learning_rate": 5e-06,
+      "loss": 0.8628,
+      "num_input_tokens_seen": 43950104,
+      "step": 701
+    },
+    {
+      "epoch": 2.3327787021630617,
+      "loss": 0.9078635573387146,
+      "loss_ce": 2.6576359232421964e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 43950104,
+      "step": 701
+    },
+    {
+      "epoch": 2.3361064891846923,
+      "grad_norm": 17.961149215698242,
+      "learning_rate": 5e-06,
+      "loss": 0.8288,
+      "num_input_tokens_seen": 44013476,
+      "step": 702
+    },
+    {
+      "epoch": 2.3361064891846923,
+      "loss": 0.6979770660400391,
+      "loss_ce": 0.0004673382209148258,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.69921875,
+      "num_input_tokens_seen": 44013476,
+      "step": 702
+    },
+    {
+      "epoch": 2.339434276206323,
+      "grad_norm": 16.723167419433594,
+      "learning_rate": 5e-06,
+      "loss": 0.8697,
+      "num_input_tokens_seen": 44078112,
+      "step": 703
+    },
+    {
+      "epoch": 2.339434276206323,
+      "loss": 0.748289167881012,
+      "loss_ce": 0.000974722090177238,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 44078112,
+      "step": 703
+    },
+    {
+      "epoch": 2.3427620632279536,
+      "grad_norm": 22.591569900512695,
+      "learning_rate": 5e-06,
+      "loss": 0.8053,
+      "num_input_tokens_seen": 44140928,
+      "step": 704
+    },
+    {
+      "epoch": 2.3427620632279536,
+      "loss": 0.524970293045044,
+      "loss_ce": 0.000800370064098388,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 44140928,
+      "step": 704
+    },
+    {
+      "epoch": 2.346089850249584,
+      "grad_norm": 49.59932327270508,
+      "learning_rate": 5e-06,
+      "loss": 1.0129,
+      "num_input_tokens_seen": 44203212,
+      "step": 705
+    },
+    {
+      "epoch": 2.346089850249584,
+      "loss": 0.6107718348503113,
+      "loss_ce": 5.408008291851729e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 44203212,
+      "step": 705
+    },
+    {
+      "epoch": 2.3494176372712148,
+      "grad_norm": 12.25649642944336,
+      "learning_rate": 5e-06,
+      "loss": 1.0834,
+      "num_input_tokens_seen": 44267888,
+      "step": 706
+    },
+    {
+      "epoch": 2.3494176372712148,
+      "loss": 1.069253921508789,
+      "loss_ce": 0.0007725384784862399,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.068359375,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 44267888,
+      "step": 706
+    },
+    {
+      "epoch": 2.3527454242928454,
+      "grad_norm": 8.59157943725586,
+      "learning_rate": 5e-06,
+      "loss": 0.9064,
+      "num_input_tokens_seen": 44331788,
+      "step": 707
+    },
+    {
+      "epoch": 2.3527454242928454,
+      "loss": 0.8050553798675537,
+      "loss_ce": 0.0013445127988234162,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 44331788,
+      "step": 707
+    },
+    {
+      "epoch": 2.356073211314476,
+      "grad_norm": 13.44419002532959,
+      "learning_rate": 5e-06,
+      "loss": 0.8174,
+      "num_input_tokens_seen": 44394832,
+      "step": 708
+    },
+    {
+      "epoch": 2.356073211314476,
+      "loss": 0.922235906124115,
+      "loss_ce": 0.0008491869666613638,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 44394832,
+      "step": 708
+    },
+    {
+      "epoch": 2.3594009983361066,
+      "grad_norm": 9.479907989501953,
+      "learning_rate": 5e-06,
+      "loss": 1.0606,
+      "num_input_tokens_seen": 44459416,
+      "step": 709
+    },
+    {
+      "epoch": 2.3594009983361066,
+      "loss": 0.9726129770278931,
+      "loss_ce": 0.0014215761329978704,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.97265625,
+      "num_input_tokens_seen": 44459416,
+      "step": 709
+    },
+    {
+      "epoch": 2.3627287853577372,
+      "grad_norm": 7.282090187072754,
+      "learning_rate": 5e-06,
+      "loss": 0.5709,
+      "num_input_tokens_seen": 44519432,
+      "step": 710
+    },
+    {
+      "epoch": 2.3627287853577372,
+      "loss": 0.7231759428977966,
+      "loss_ce": 0.0001534855255158618,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 44519432,
+      "step": 710
+    },
+    {
+      "epoch": 2.366056572379368,
+      "grad_norm": 10.938211441040039,
+      "learning_rate": 5e-06,
+      "loss": 0.9808,
+      "num_input_tokens_seen": 44583072,
+      "step": 711
+    },
+    {
+      "epoch": 2.366056572379368,
+      "loss": 0.8792444467544556,
+      "loss_ce": 0.0010706019820645452,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 44583072,
+      "step": 711
+    },
+    {
+      "epoch": 2.3693843594009985,
+      "grad_norm": 15.567341804504395,
+      "learning_rate": 5e-06,
+      "loss": 0.9657,
+      "num_input_tokens_seen": 44646108,
+      "step": 712
+    },
+    {
+      "epoch": 2.3693843594009985,
+      "loss": 1.0545854568481445,
+      "loss_ce": 0.0018511018715798855,
+      "loss_iou": 0.390625,
+      "loss_num": 0.05419921875,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 44646108,
+      "step": 712
+    },
+    {
+      "epoch": 2.372712146422629,
+      "grad_norm": 10.228503227233887,
+      "learning_rate": 5e-06,
+      "loss": 0.9219,
+      "num_input_tokens_seen": 44708204,
+      "step": 713
+    },
+    {
+      "epoch": 2.372712146422629,
+      "loss": 1.1827502250671387,
+      "loss_ce": 0.0006212661974132061,
+      "loss_iou": 0.40625,
+      "loss_num": 0.07421875,
+      "loss_xval": 1.1796875,
+      "num_input_tokens_seen": 44708204,
+      "step": 713
+    },
+    {
+      "epoch": 2.3760399334442597,
+      "grad_norm": 12.394549369812012,
+      "learning_rate": 5e-06,
+      "loss": 0.6104,
+      "num_input_tokens_seen": 44769100,
+      "step": 714
+    },
+    {
+      "epoch": 2.3760399334442597,
+      "loss": 0.5975984334945679,
+      "loss_ce": 0.0037263473495841026,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 44769100,
+      "step": 714
+    },
+    {
+      "epoch": 2.3793677204658903,
+      "grad_norm": 13.236189842224121,
+      "learning_rate": 5e-06,
+      "loss": 0.9356,
+      "num_input_tokens_seen": 44832944,
+      "step": 715
+    },
+    {
+      "epoch": 2.3793677204658903,
+      "loss": 0.889470100402832,
+      "loss_ce": 0.003239629790186882,
+      "loss_iou": 0.3125,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 44832944,
+      "step": 715
+    },
+    {
+      "epoch": 2.382695507487521,
+      "grad_norm": 15.01264762878418,
+      "learning_rate": 5e-06,
+      "loss": 0.937,
+      "num_input_tokens_seen": 44894316,
+      "step": 716
+    },
+    {
+      "epoch": 2.382695507487521,
+      "loss": 0.9371458292007446,
+      "loss_ce": 0.0013547912240028381,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0966796875,
+      "loss_xval": 0.9375,
+      "num_input_tokens_seen": 44894316,
+      "step": 716
+    },
+    {
+      "epoch": 2.3860232945091515,
+      "grad_norm": 40.445865631103516,
+      "learning_rate": 5e-06,
+      "loss": 1.1198,
+      "num_input_tokens_seen": 44958448,
+      "step": 717
+    },
+    {
+      "epoch": 2.3860232945091515,
+      "loss": 1.0740020275115967,
+      "loss_ce": 0.00027151533868163824,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0595703125,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 44958448,
+      "step": 717
+    },
+    {
+      "epoch": 2.389351081530782,
+      "grad_norm": 21.32187271118164,
+      "learning_rate": 5e-06,
+      "loss": 0.8088,
+      "num_input_tokens_seen": 45020540,
+      "step": 718
+    },
+    {
+      "epoch": 2.389351081530782,
+      "loss": 0.8202319145202637,
+      "loss_ce": 0.00028564067906700075,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 45020540,
+      "step": 718
+    },
+    {
+      "epoch": 2.3926788685524127,
+      "grad_norm": 11.086956024169922,
+      "learning_rate": 5e-06,
+      "loss": 0.7241,
+      "num_input_tokens_seen": 45081188,
+      "step": 719
+    },
+    {
+      "epoch": 2.3926788685524127,
+      "loss": 0.6227037906646729,
+      "loss_ce": 0.00014521228149533272,
+      "loss_iou": 0.171875,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 45081188,
+      "step": 719
+    },
+    {
+      "epoch": 2.3960066555740434,
+      "grad_norm": 15.942041397094727,
+      "learning_rate": 5e-06,
+      "loss": 0.86,
+      "num_input_tokens_seen": 45142776,
+      "step": 720
+    },
+    {
+      "epoch": 2.3960066555740434,
+      "loss": 0.6935033798217773,
+      "loss_ce": 0.0011205670889467,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.69140625,
+      "num_input_tokens_seen": 45142776,
+      "step": 720
+    },
+    {
+      "epoch": 2.399334442595674,
+      "grad_norm": 13.693964004516602,
+      "learning_rate": 5e-06,
+      "loss": 1.0913,
+      "num_input_tokens_seen": 45205824,
+      "step": 721
+    },
+    {
+      "epoch": 2.399334442595674,
+      "loss": 1.0079128742218018,
+      "loss_ce": 0.0008326807874254882,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.08349609375,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 45205824,
+      "step": 721
+    },
+    {
+      "epoch": 2.4026622296173046,
+      "grad_norm": 15.026814460754395,
+      "learning_rate": 5e-06,
+      "loss": 0.7762,
+      "num_input_tokens_seen": 45268764,
+      "step": 722
+    },
+    {
+      "epoch": 2.4026622296173046,
+      "loss": 0.6626920700073242,
+      "loss_ce": 9.44572821026668e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 45268764,
+      "step": 722
+    },
+    {
+      "epoch": 2.405990016638935,
+      "grad_norm": 11.194849014282227,
+      "learning_rate": 5e-06,
+      "loss": 0.6871,
+      "num_input_tokens_seen": 45331492,
+      "step": 723
+    },
+    {
+      "epoch": 2.405990016638935,
+      "loss": 0.48289889097213745,
+      "loss_ce": 0.000721134536433965,
+      "loss_iou": 0.125,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.482421875,
+      "num_input_tokens_seen": 45331492,
+      "step": 723
+    },
+    {
+      "epoch": 2.409317803660566,
+      "grad_norm": 15.214365005493164,
+      "learning_rate": 5e-06,
+      "loss": 0.6964,
+      "num_input_tokens_seen": 45394532,
+      "step": 724
+    },
+    {
+      "epoch": 2.409317803660566,
+      "loss": 0.8721684217453003,
+      "loss_ce": 0.000769482518080622,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 45394532,
+      "step": 724
+    },
+    {
+      "epoch": 2.4126455906821964,
+      "grad_norm": 18.638185501098633,
+      "learning_rate": 5e-06,
+      "loss": 0.757,
+      "num_input_tokens_seen": 45457232,
+      "step": 725
+    },
+    {
+      "epoch": 2.4126455906821964,
+      "loss": 0.6739429235458374,
+      "loss_ce": 0.000358955207047984,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 45457232,
+      "step": 725
+    },
+    {
+      "epoch": 2.415973377703827,
+      "grad_norm": 18.782562255859375,
+      "learning_rate": 5e-06,
+      "loss": 0.8527,
+      "num_input_tokens_seen": 45519960,
+      "step": 726
+    },
+    {
+      "epoch": 2.415973377703827,
+      "loss": 0.7281821370124817,
+      "loss_ce": 0.00113136216532439,
+      "loss_iou": 0.28125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 45519960,
+      "step": 726
+    },
+    {
+      "epoch": 2.4193011647254576,
+      "grad_norm": 35.47547149658203,
+      "learning_rate": 5e-06,
+      "loss": 1.1733,
+      "num_input_tokens_seen": 45583772,
+      "step": 727
+    },
+    {
+      "epoch": 2.4193011647254576,
+      "loss": 1.2746448516845703,
+      "loss_ce": 0.00023083810810931027,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.06591796875,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 45583772,
+      "step": 727
+    },
+    {
+      "epoch": 2.4226289517470883,
+      "grad_norm": 30.0634708404541,
+      "learning_rate": 5e-06,
+      "loss": 0.9175,
+      "num_input_tokens_seen": 45646088,
+      "step": 728
+    },
+    {
+      "epoch": 2.4226289517470883,
+      "loss": 0.9010690450668335,
+      "loss_ce": 0.016364434733986855,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 45646088,
+      "step": 728
+    },
+    {
+      "epoch": 2.425956738768719,
+      "grad_norm": 15.3126859664917,
+      "learning_rate": 5e-06,
+      "loss": 0.6904,
+      "num_input_tokens_seen": 45708808,
+      "step": 729
+    },
+    {
+      "epoch": 2.425956738768719,
+      "loss": 0.5067150592803955,
+      "loss_ce": 0.0014660221058875322,
+      "loss_iou": 0.109375,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 45708808,
+      "step": 729
+    },
+    {
+      "epoch": 2.4292845257903495,
+      "grad_norm": 19.397184371948242,
+      "learning_rate": 5e-06,
+      "loss": 0.8657,
+      "num_input_tokens_seen": 45770068,
+      "step": 730
+    },
+    {
+      "epoch": 2.4292845257903495,
+      "loss": 0.9862344264984131,
+      "loss_ce": 0.001096484367735684,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 45770068,
+      "step": 730
+    },
+    {
+      "epoch": 2.43261231281198,
+      "grad_norm": 7.857912063598633,
+      "learning_rate": 5e-06,
+      "loss": 0.6581,
+      "num_input_tokens_seen": 45832556,
+      "step": 731
+    },
+    {
+      "epoch": 2.43261231281198,
+      "loss": 0.5868287682533264,
+      "loss_ce": 0.002844445873051882,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 45832556,
+      "step": 731
+    },
+    {
+      "epoch": 2.4359400998336107,
+      "grad_norm": 28.30218505859375,
+      "learning_rate": 5e-06,
+      "loss": 0.9152,
+      "num_input_tokens_seen": 45894836,
+      "step": 732
+    },
+    {
+      "epoch": 2.4359400998336107,
+      "loss": 1.060274362564087,
+      "loss_ce": 0.0003377725079189986,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.06494140625,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 45894836,
+      "step": 732
+    },
+    {
+      "epoch": 2.4392678868552413,
+      "grad_norm": 17.082120895385742,
+      "learning_rate": 5e-06,
+      "loss": 0.698,
+      "num_input_tokens_seen": 45957656,
+      "step": 733
+    },
+    {
+      "epoch": 2.4392678868552413,
+      "loss": 0.6111936569213867,
+      "loss_ce": 0.0005979241104796529,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 45957656,
+      "step": 733
+    },
+    {
+      "epoch": 2.442595673876872,
+      "grad_norm": 23.90694236755371,
+      "learning_rate": 5e-06,
+      "loss": 0.9885,
+      "num_input_tokens_seen": 46021108,
+      "step": 734
+    },
+    {
+      "epoch": 2.442595673876872,
+      "loss": 1.1393766403198242,
+      "loss_ce": 0.0013150431914255023,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0986328125,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 46021108,
+      "step": 734
+    },
+    {
+      "epoch": 2.4459234608985025,
+      "grad_norm": 11.997917175292969,
+      "learning_rate": 5e-06,
+      "loss": 0.8011,
+      "num_input_tokens_seen": 46083808,
+      "step": 735
+    },
+    {
+      "epoch": 2.4459234608985025,
+      "loss": 0.8153502345085144,
+      "loss_ce": 0.003277485491707921,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 46083808,
+      "step": 735
+    },
+    {
+      "epoch": 2.449251247920133,
+      "grad_norm": 27.563457489013672,
+      "learning_rate": 5e-06,
+      "loss": 0.8633,
+      "num_input_tokens_seen": 46146540,
+      "step": 736
+    },
+    {
+      "epoch": 2.449251247920133,
+      "loss": 0.8773728609085083,
+      "loss_ce": 0.0001755441480781883,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 46146540,
+      "step": 736
+    },
+    {
+      "epoch": 2.4525790349417638,
+      "grad_norm": 14.77198600769043,
+      "learning_rate": 5e-06,
+      "loss": 0.9703,
+      "num_input_tokens_seen": 46209148,
+      "step": 737
+    },
+    {
+      "epoch": 2.4525790349417638,
+      "loss": 0.757889986038208,
+      "loss_ce": 0.0027630457188934088,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 46209148,
+      "step": 737
+    },
+    {
+      "epoch": 2.4559068219633944,
+      "grad_norm": 11.89362907409668,
+      "learning_rate": 5e-06,
+      "loss": 0.7569,
+      "num_input_tokens_seen": 46272732,
+      "step": 738
+    },
+    {
+      "epoch": 2.4559068219633944,
+      "loss": 0.7210886478424072,
+      "loss_ce": 0.0008737589814700186,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 46272732,
+      "step": 738
+    },
+    {
+      "epoch": 2.459234608985025,
+      "grad_norm": 4.648507118225098,
+      "learning_rate": 5e-06,
+      "loss": 0.619,
+      "num_input_tokens_seen": 46336108,
+      "step": 739
+    },
+    {
+      "epoch": 2.459234608985025,
+      "loss": 0.7744088768959045,
+      "loss_ce": 0.00014738636673428118,
+      "loss_iou": 0.25,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 46336108,
+      "step": 739
+    },
+    {
+      "epoch": 2.4625623960066556,
+      "grad_norm": 21.0081844329834,
+      "learning_rate": 5e-06,
+      "loss": 0.7802,
+      "num_input_tokens_seen": 46397180,
+      "step": 740
+    },
+    {
+      "epoch": 2.4625623960066556,
+      "loss": 0.857671856880188,
+      "loss_ce": 0.0025693103671073914,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 46397180,
+      "step": 740
+    },
+    {
+      "epoch": 2.465890183028286,
+      "grad_norm": 13.0715913772583,
+      "learning_rate": 5e-06,
+      "loss": 0.7849,
+      "num_input_tokens_seen": 46459372,
+      "step": 741
+    },
+    {
+      "epoch": 2.465890183028286,
+      "loss": 0.6565080881118774,
+      "loss_ce": 0.006361585110425949,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.0625,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 46459372,
+      "step": 741
+    },
+    {
+      "epoch": 2.469217970049917,
+      "grad_norm": 9.451695442199707,
+      "learning_rate": 5e-06,
+      "loss": 0.991,
+      "num_input_tokens_seen": 46523256,
+      "step": 742
+    },
+    {
+      "epoch": 2.469217970049917,
+      "loss": 1.2594883441925049,
+      "loss_ce": 0.00021096415002830327,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.072265625,
+      "loss_xval": 1.2578125,
+      "num_input_tokens_seen": 46523256,
+      "step": 742
+    },
+    {
+      "epoch": 2.4725457570715474,
+      "grad_norm": 13.714978218078613,
+      "learning_rate": 5e-06,
+      "loss": 0.9132,
+      "num_input_tokens_seen": 46587516,
+      "step": 743
+    },
+    {
+      "epoch": 2.4725457570715474,
+      "loss": 1.0107247829437256,
+      "loss_ce": 0.0005928677855990827,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0673828125,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 46587516,
+      "step": 743
+    },
+    {
+      "epoch": 2.475873544093178,
+      "grad_norm": 12.65008544921875,
+      "learning_rate": 5e-06,
+      "loss": 1.0877,
+      "num_input_tokens_seen": 46650720,
+      "step": 744
+    },
+    {
+      "epoch": 2.475873544093178,
+      "loss": 1.1367709636688232,
+      "loss_ce": 0.0029819714836776257,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0830078125,
+      "loss_xval": 1.1328125,
+      "num_input_tokens_seen": 46650720,
+      "step": 744
+    },
+    {
+      "epoch": 2.4792013311148087,
+      "grad_norm": 36.57686996459961,
+      "learning_rate": 5e-06,
+      "loss": 0.8593,
+      "num_input_tokens_seen": 46715288,
+      "step": 745
+    },
+    {
+      "epoch": 2.4792013311148087,
+      "loss": 0.6860495209693909,
+      "loss_ce": 1.4360236491484102e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 46715288,
+      "step": 745
+    },
+    {
+      "epoch": 2.4825291181364393,
+      "grad_norm": 12.971878051757812,
+      "learning_rate": 5e-06,
+      "loss": 0.7946,
+      "num_input_tokens_seen": 46777768,
+      "step": 746
+    },
+    {
+      "epoch": 2.4825291181364393,
+      "loss": 0.78072589635849,
+      "loss_ce": 0.0009407766629010439,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 46777768,
+      "step": 746
+    },
+    {
+      "epoch": 2.48585690515807,
+      "grad_norm": 11.606049537658691,
+      "learning_rate": 5e-06,
+      "loss": 0.8739,
+      "num_input_tokens_seen": 46839216,
+      "step": 747
+    },
+    {
+      "epoch": 2.48585690515807,
+      "loss": 0.9141720533370972,
+      "loss_ce": 0.0010860441252589226,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 46839216,
+      "step": 747
+    },
+    {
+      "epoch": 2.4891846921797005,
+      "grad_norm": 17.032249450683594,
+      "learning_rate": 5e-06,
+      "loss": 0.5929,
+      "num_input_tokens_seen": 46899036,
+      "step": 748
+    },
+    {
+      "epoch": 2.4891846921797005,
+      "loss": 0.5191299319267273,
+      "loss_ce": 8.698241435922682e-05,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 46899036,
+      "step": 748
+    },
+    {
+      "epoch": 2.492512479201331,
+      "grad_norm": 11.303759574890137,
+      "learning_rate": 5e-06,
+      "loss": 0.805,
+      "num_input_tokens_seen": 46961588,
+      "step": 749
+    },
+    {
+      "epoch": 2.492512479201331,
+      "loss": 0.8356009721755981,
+      "loss_ce": 0.000151735614053905,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.076171875,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 46961588,
+      "step": 749
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "grad_norm": 23.632003784179688,
+      "learning_rate": 5e-06,
+      "loss": 0.833,
+      "num_input_tokens_seen": 47024696,
+      "step": 750
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "eval_seeclick_CIoU": 0.13491493463516235,
+      "eval_seeclick_GIoU": 0.15883130580186844,
+      "eval_seeclick_IoU": 0.24382416903972626,
+      "eval_seeclick_MAE_all": 0.19065534323453903,
+      "eval_seeclick_MAE_h": 0.05082565359771252,
+      "eval_seeclick_MAE_w": 0.13937117159366608,
+      "eval_seeclick_MAE_x_boxes": 0.30929259955883026,
+      "eval_seeclick_MAE_y_boxes": 0.13166731595993042,
+      "eval_seeclick_NUM_probability": 0.9999373257160187,
+      "eval_seeclick_inside_bbox": 0.35208334028720856,
+      "eval_seeclick_loss": 2.6928210258483887,
+      "eval_seeclick_loss_ce": 0.08299023285508156,
+      "eval_seeclick_loss_iou": 0.830810546875,
+      "eval_seeclick_loss_num": 0.1858978271484375,
+      "eval_seeclick_loss_xval": 2.59033203125,
+      "eval_seeclick_runtime": 62.8164,
+      "eval_seeclick_samples_per_second": 0.748,
+      "eval_seeclick_steps_per_second": 0.032,
+      "num_input_tokens_seen": 47024696,
+      "step": 750
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "eval_icons_CIoU": 0.05114769656211138,
+      "eval_icons_GIoU": 0.15702077746391296,
+      "eval_icons_IoU": 0.2036093920469284,
+      "eval_icons_MAE_all": 0.12747248262166977,
+      "eval_icons_MAE_h": 0.052048404701054096,
+      "eval_icons_MAE_w": 0.1445429064333439,
+      "eval_icons_MAE_x_boxes": 0.12461870163679123,
+      "eval_icons_MAE_y_boxes": 0.046514895744621754,
+      "eval_icons_NUM_probability": 0.9999918937683105,
+      "eval_icons_inside_bbox": 0.3576388955116272,
+      "eval_icons_loss": 2.3343775272369385,
+      "eval_icons_loss_ce": 2.2014152136762277e-06,
+      "eval_icons_loss_iou": 0.830078125,
+      "eval_icons_loss_num": 0.1331043243408203,
+      "eval_icons_loss_xval": 2.32666015625,
+      "eval_icons_runtime": 65.6403,
+      "eval_icons_samples_per_second": 0.762,
+      "eval_icons_steps_per_second": 0.03,
+      "num_input_tokens_seen": 47024696,
+      "step": 750
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "eval_screenspot_CIoU": 0.039610182866454124,
+      "eval_screenspot_GIoU": 0.10459545751412709,
+      "eval_screenspot_IoU": 0.1884987105925878,
+      "eval_screenspot_MAE_all": 0.20243261754512787,
+      "eval_screenspot_MAE_h": 0.0592109685142835,
+      "eval_screenspot_MAE_w": 0.18089259415864944,
+      "eval_screenspot_MAE_x_boxes": 0.250293031334877,
+      "eval_screenspot_MAE_y_boxes": 0.1454415942231814,
+      "eval_screenspot_NUM_probability": 0.9999621907869974,
+      "eval_screenspot_inside_bbox": 0.3387500047683716,
+      "eval_screenspot_loss": 2.80856990814209,
+      "eval_screenspot_loss_ce": 0.00027161008135105175,
+      "eval_screenspot_loss_iou": 0.9033203125,
+      "eval_screenspot_loss_num": 0.20716349283854166,
+      "eval_screenspot_loss_xval": 2.8421223958333335,
+      "eval_screenspot_runtime": 126.77,
+      "eval_screenspot_samples_per_second": 0.702,
+      "eval_screenspot_steps_per_second": 0.024,
+      "num_input_tokens_seen": 47024696,
+      "step": 750
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "eval_compot_CIoU": -0.054212167859077454,
+      "eval_compot_GIoU": 0.057059221900999546,
+      "eval_compot_IoU": 0.11786043643951416,
+      "eval_compot_MAE_all": 0.2351270318031311,
+      "eval_compot_MAE_h": 0.06904683262109756,
+      "eval_compot_MAE_w": 0.2527950033545494,
+      "eval_compot_MAE_x_boxes": 0.19278910011053085,
+      "eval_compot_MAE_y_boxes": 0.18391906470060349,
+      "eval_compot_NUM_probability": 0.9999668002128601,
+      "eval_compot_inside_bbox": 0.2239583358168602,
+      "eval_compot_loss": 2.9689693450927734,
+      "eval_compot_loss_ce": 0.006361398845911026,
+      "eval_compot_loss_iou": 0.924072265625,
+      "eval_compot_loss_num": 0.2135009765625,
+      "eval_compot_loss_xval": 2.916015625,
+      "eval_compot_runtime": 72.4042,
+      "eval_compot_samples_per_second": 0.691,
+      "eval_compot_steps_per_second": 0.028,
+      "num_input_tokens_seen": 47024696,
+      "step": 750
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "eval_custom_ui_MAE_all": 0.09127533435821533,
+      "eval_custom_ui_MAE_x": 0.09527997672557831,
+      "eval_custom_ui_MAE_y": 0.08727069199085236,
+      "eval_custom_ui_NUM_probability": 0.999992311000824,
+      "eval_custom_ui_loss": 0.4499455690383911,
+      "eval_custom_ui_loss_ce": 8.726405303605134e-05,
+      "eval_custom_ui_loss_num": 0.0854949951171875,
+      "eval_custom_ui_loss_xval": 0.4273681640625,
+      "eval_custom_ui_runtime": 56.9564,
+      "eval_custom_ui_samples_per_second": 0.878,
+      "eval_custom_ui_steps_per_second": 0.035,
+      "num_input_tokens_seen": 47024696,
+      "step": 750
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "loss": 0.40762990713119507,
+      "loss_ce": 0.00015917516429908574,
+      "loss_iou": 0.0,
+      "loss_num": 0.08154296875,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 47024696,
+      "step": 750
+    },
+    {
+      "epoch": 2.4991680532445923,
+      "grad_norm": 18.54668426513672,
+      "learning_rate": 5e-06,
+      "loss": 0.8896,
+      "num_input_tokens_seen": 47088064,
+      "step": 751
+    },
+    {
+      "epoch": 2.4991680532445923,
+      "loss": 0.5935692191123962,
+      "loss_ce": 0.0016502785729244351,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.05126953125,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 47088064,
+      "step": 751
+    },
+    {
+      "epoch": 2.502495840266223,
+      "grad_norm": 11.817997932434082,
+      "learning_rate": 5e-06,
+      "loss": 0.8837,
+      "num_input_tokens_seen": 47150452,
+      "step": 752
+    },
+    {
+      "epoch": 2.502495840266223,
+      "loss": 0.8448632955551147,
+      "loss_ce": 0.00044191486085765064,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 47150452,
+      "step": 752
+    },
+    {
+      "epoch": 2.5058236272878536,
+      "grad_norm": 23.670198440551758,
+      "learning_rate": 5e-06,
+      "loss": 0.7463,
+      "num_input_tokens_seen": 47212136,
+      "step": 753
+    },
+    {
+      "epoch": 2.5058236272878536,
+      "loss": 0.8123268485069275,
+      "loss_ce": 7.097484922269359e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 47212136,
+      "step": 753
+    },
+    {
+      "epoch": 2.509151414309484,
+      "grad_norm": 19.003517150878906,
+      "learning_rate": 5e-06,
+      "loss": 0.7155,
+      "num_input_tokens_seen": 47275172,
+      "step": 754
+    },
+    {
+      "epoch": 2.509151414309484,
+      "loss": 0.7601994276046753,
+      "loss_ce": 0.0021427858155220747,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 47275172,
+      "step": 754
+    },
+    {
+      "epoch": 2.512479201331115,
+      "grad_norm": 42.69208908081055,
+      "learning_rate": 5e-06,
+      "loss": 0.8689,
+      "num_input_tokens_seen": 47336960,
+      "step": 755
+    },
+    {
+      "epoch": 2.512479201331115,
+      "loss": 1.1230835914611816,
+      "loss_ce": 0.0015015102690085769,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 47336960,
+      "step": 755
+    },
+    {
+      "epoch": 2.5158069883527454,
+      "grad_norm": 18.97686004638672,
+      "learning_rate": 5e-06,
+      "loss": 0.9074,
+      "num_input_tokens_seen": 47400968,
+      "step": 756
+    },
+    {
+      "epoch": 2.5158069883527454,
+      "loss": 0.8372678756713867,
+      "loss_ce": 0.000353793177055195,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 47400968,
+      "step": 756
+    },
+    {
+      "epoch": 2.519134775374376,
+      "grad_norm": 19.407001495361328,
+      "learning_rate": 5e-06,
+      "loss": 0.7905,
+      "num_input_tokens_seen": 47462844,
+      "step": 757
+    },
+    {
+      "epoch": 2.519134775374376,
+      "loss": 0.7666225433349609,
+      "loss_ce": 0.0009975419379770756,
+      "loss_iou": 0.171875,
+      "loss_num": 0.08447265625,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 47462844,
+      "step": 757
+    },
+    {
+      "epoch": 2.5224625623960066,
+      "grad_norm": 21.340974807739258,
+      "learning_rate": 5e-06,
+      "loss": 0.8336,
+      "num_input_tokens_seen": 47524364,
+      "step": 758
+    },
+    {
+      "epoch": 2.5224625623960066,
+      "loss": 1.158284068107605,
+      "loss_ce": 0.0013016541488468647,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.08740234375,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 47524364,
+      "step": 758
+    },
+    {
+      "epoch": 2.5257903494176372,
+      "grad_norm": 12.445579528808594,
+      "learning_rate": 5e-06,
+      "loss": 0.7022,
+      "num_input_tokens_seen": 47587348,
+      "step": 759
+    },
+    {
+      "epoch": 2.5257903494176372,
+      "loss": 0.6259897947311401,
+      "loss_ce": 0.0009897334966808558,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 47587348,
+      "step": 759
+    },
+    {
+      "epoch": 2.529118136439268,
+      "grad_norm": 11.766027450561523,
+      "learning_rate": 5e-06,
+      "loss": 0.9098,
+      "num_input_tokens_seen": 47649600,
+      "step": 760
+    },
+    {
+      "epoch": 2.529118136439268,
+      "loss": 0.6773974895477295,
+      "loss_ce": 0.0006396822864189744,
+      "loss_iou": 0.234375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 47649600,
+      "step": 760
+    },
+    {
+      "epoch": 2.5324459234608985,
+      "grad_norm": 12.828213691711426,
+      "learning_rate": 5e-06,
+      "loss": 0.7839,
+      "num_input_tokens_seen": 47712632,
+      "step": 761
+    },
+    {
+      "epoch": 2.5324459234608985,
+      "loss": 0.7082505226135254,
+      "loss_ce": 0.00012061389134032652,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.70703125,
+      "num_input_tokens_seen": 47712632,
+      "step": 761
+    },
+    {
+      "epoch": 2.535773710482529,
+      "grad_norm": 21.702556610107422,
+      "learning_rate": 5e-06,
+      "loss": 0.9561,
+      "num_input_tokens_seen": 47776100,
+      "step": 762
+    },
+    {
+      "epoch": 2.535773710482529,
+      "loss": 0.7422810792922974,
+      "loss_ce": 9.361249976791441e-05,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 47776100,
+      "step": 762
+    },
+    {
+      "epoch": 2.5391014975041597,
+      "grad_norm": 15.470916748046875,
+      "learning_rate": 5e-06,
+      "loss": 0.9802,
+      "num_input_tokens_seen": 47837816,
+      "step": 763
+    },
+    {
+      "epoch": 2.5391014975041597,
+      "loss": 1.192439317703247,
+      "loss_ce": 0.000544731505215168,
+      "loss_iou": 0.390625,
+      "loss_num": 0.08251953125,
+      "loss_xval": 1.1953125,
+      "num_input_tokens_seen": 47837816,
+      "step": 763
+    },
+    {
+      "epoch": 2.5424292845257903,
+      "grad_norm": 21.050203323364258,
+      "learning_rate": 5e-06,
+      "loss": 0.906,
+      "num_input_tokens_seen": 47899704,
+      "step": 764
+    },
+    {
+      "epoch": 2.5424292845257903,
+      "loss": 0.768609344959259,
+      "loss_ce": 0.000665017869323492,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 47899704,
+      "step": 764
+    },
+    {
+      "epoch": 2.545757071547421,
+      "grad_norm": 10.072230339050293,
+      "learning_rate": 5e-06,
+      "loss": 0.5435,
+      "num_input_tokens_seen": 47961524,
+      "step": 765
+    },
+    {
+      "epoch": 2.545757071547421,
+      "loss": 0.4695216119289398,
+      "loss_ce": 0.0011378447525203228,
+      "loss_iou": 0.08349609375,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.46875,
+      "num_input_tokens_seen": 47961524,
+      "step": 765
+    },
+    {
+      "epoch": 2.5490848585690515,
+      "grad_norm": 7.9047136306762695,
+      "learning_rate": 5e-06,
+      "loss": 0.6468,
+      "num_input_tokens_seen": 48020096,
+      "step": 766
+    },
+    {
+      "epoch": 2.5490848585690515,
+      "loss": 0.7101080417633057,
+      "loss_ce": 0.0008795225294306874,
+      "loss_iou": 0.10986328125,
+      "loss_num": 0.09814453125,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 48020096,
+      "step": 766
+    },
+    {
+      "epoch": 2.552412645590682,
+      "grad_norm": 9.703407287597656,
+      "learning_rate": 5e-06,
+      "loss": 0.5771,
+      "num_input_tokens_seen": 48082036,
+      "step": 767
+    },
+    {
+      "epoch": 2.552412645590682,
+      "loss": 0.5914462208747864,
+      "loss_ce": 0.000625885440967977,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 48082036,
+      "step": 767
+    },
+    {
+      "epoch": 2.5557404326123128,
+      "grad_norm": 22.55636215209961,
+      "learning_rate": 5e-06,
+      "loss": 1.1754,
+      "num_input_tokens_seen": 48142276,
+      "step": 768
+    },
+    {
+      "epoch": 2.5557404326123128,
+      "loss": 0.9645742774009705,
+      "loss_ce": 0.00021879022824577987,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.96484375,
+      "num_input_tokens_seen": 48142276,
+      "step": 768
+    },
+    {
+      "epoch": 2.5590682196339434,
+      "grad_norm": 23.570049285888672,
+      "learning_rate": 5e-06,
+      "loss": 0.7371,
+      "num_input_tokens_seen": 48204608,
+      "step": 769
+    },
+    {
+      "epoch": 2.5590682196339434,
+      "loss": 0.6285191774368286,
+      "loss_ce": 0.0007115625194273889,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 48204608,
+      "step": 769
+    },
+    {
+      "epoch": 2.562396006655574,
+      "grad_norm": 30.504138946533203,
+      "learning_rate": 5e-06,
+      "loss": 0.7577,
+      "num_input_tokens_seen": 48268100,
+      "step": 770
+    },
+    {
+      "epoch": 2.562396006655574,
+      "loss": 0.6577082872390747,
+      "loss_ce": 0.00011549589544301853,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 48268100,
+      "step": 770
+    },
+    {
+      "epoch": 2.5657237936772046,
+      "grad_norm": 25.44117546081543,
+      "learning_rate": 5e-06,
+      "loss": 0.9569,
+      "num_input_tokens_seen": 48330448,
+      "step": 771
+    },
+    {
+      "epoch": 2.5657237936772046,
+      "loss": 0.9606022834777832,
+      "loss_ce": 0.0006413036026060581,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.9609375,
+      "num_input_tokens_seen": 48330448,
+      "step": 771
+    },
+    {
+      "epoch": 2.569051580698835,
+      "grad_norm": 26.2667293548584,
+      "learning_rate": 5e-06,
+      "loss": 0.8804,
+      "num_input_tokens_seen": 48393420,
+      "step": 772
+    },
+    {
+      "epoch": 2.569051580698835,
+      "loss": 0.798115074634552,
+      "loss_ce": 0.004169770982116461,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 48393420,
+      "step": 772
+    },
+    {
+      "epoch": 2.572379367720466,
+      "grad_norm": 11.151239395141602,
+      "learning_rate": 5e-06,
+      "loss": 0.6429,
+      "num_input_tokens_seen": 48455548,
+      "step": 773
+    },
+    {
+      "epoch": 2.572379367720466,
+      "loss": 0.5176081657409668,
+      "loss_ce": 0.0030817792285233736,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 48455548,
+      "step": 773
+    },
+    {
+      "epoch": 2.5757071547420964,
+      "grad_norm": 17.321378707885742,
+      "learning_rate": 5e-06,
+      "loss": 0.9045,
+      "num_input_tokens_seen": 48519460,
+      "step": 774
+    },
+    {
+      "epoch": 2.5757071547420964,
+      "loss": 1.2099051475524902,
+      "loss_ce": 0.0021415799856185913,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 1.2109375,
+      "num_input_tokens_seen": 48519460,
+      "step": 774
+    },
+    {
+      "epoch": 2.579034941763727,
+      "grad_norm": 13.804886817932129,
+      "learning_rate": 5e-06,
+      "loss": 0.8553,
+      "num_input_tokens_seen": 48581736,
+      "step": 775
+    },
+    {
+      "epoch": 2.579034941763727,
+      "loss": 0.8312492966651917,
+      "loss_ce": 0.00019460837938822806,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 48581736,
+      "step": 775
+    },
+    {
+      "epoch": 2.5823627287853577,
+      "grad_norm": 28.333999633789062,
+      "learning_rate": 5e-06,
+      "loss": 0.7878,
+      "num_input_tokens_seen": 48645052,
+      "step": 776
+    },
+    {
+      "epoch": 2.5823627287853577,
+      "loss": 0.7494910359382629,
+      "loss_ce": 0.0012000187998637557,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 48645052,
+      "step": 776
+    },
+    {
+      "epoch": 2.5856905158069883,
+      "grad_norm": 13.613855361938477,
+      "learning_rate": 5e-06,
+      "loss": 0.8614,
+      "num_input_tokens_seen": 48708828,
+      "step": 777
+    },
+    {
+      "epoch": 2.5856905158069883,
+      "loss": 0.8165592551231384,
+      "loss_ce": 0.0006412834627553821,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 48708828,
+      "step": 777
+    },
+    {
+      "epoch": 2.589018302828619,
+      "grad_norm": 19.142955780029297,
+      "learning_rate": 5e-06,
+      "loss": 0.829,
+      "num_input_tokens_seen": 48771620,
+      "step": 778
+    },
+    {
+      "epoch": 2.589018302828619,
+      "loss": 1.0290508270263672,
+      "loss_ce": 0.0012187063694000244,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.06396484375,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 48771620,
+      "step": 778
+    },
+    {
+      "epoch": 2.5923460898502495,
+      "grad_norm": 18.934194564819336,
+      "learning_rate": 5e-06,
+      "loss": 0.8152,
+      "num_input_tokens_seen": 48833540,
+      "step": 779
+    },
+    {
+      "epoch": 2.5923460898502495,
+      "loss": 0.8391866087913513,
+      "loss_ce": 0.001295918715186417,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 48833540,
+      "step": 779
+    },
+    {
+      "epoch": 2.59567387687188,
+      "grad_norm": 18.425006866455078,
+      "learning_rate": 5e-06,
+      "loss": 0.8615,
+      "num_input_tokens_seen": 48896456,
+      "step": 780
+    },
+    {
+      "epoch": 2.59567387687188,
+      "loss": 0.8006330132484436,
+      "loss_ce": 0.00046212831512093544,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 48896456,
+      "step": 780
+    },
+    {
+      "epoch": 2.5990016638935107,
+      "grad_norm": 18.229389190673828,
+      "learning_rate": 5e-06,
+      "loss": 0.7638,
+      "num_input_tokens_seen": 48959876,
+      "step": 781
+    },
+    {
+      "epoch": 2.5990016638935107,
+      "loss": 0.8060437440872192,
+      "loss_ce": 0.00013553237658925354,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 48959876,
+      "step": 781
+    },
+    {
+      "epoch": 2.6023294509151413,
+      "grad_norm": 10.035086631774902,
+      "learning_rate": 5e-06,
+      "loss": 0.5497,
+      "num_input_tokens_seen": 49020532,
+      "step": 782
+    },
+    {
+      "epoch": 2.6023294509151413,
+      "loss": 0.5629553198814392,
+      "loss_ce": 8.909497410058975e-05,
+      "loss_iou": 0.12451171875,
+      "loss_num": 0.0625,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 49020532,
+      "step": 782
+    },
+    {
+      "epoch": 2.605657237936772,
+      "grad_norm": 26.17829704284668,
+      "learning_rate": 5e-06,
+      "loss": 0.7714,
+      "num_input_tokens_seen": 49082484,
+      "step": 783
+    },
+    {
+      "epoch": 2.605657237936772,
+      "loss": 0.8501694202423096,
+      "loss_ce": 0.0005600237054750323,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.8515625,
+      "num_input_tokens_seen": 49082484,
+      "step": 783
+    },
+    {
+      "epoch": 2.6089850249584026,
+      "grad_norm": 22.57184410095215,
+      "learning_rate": 5e-06,
+      "loss": 0.9777,
+      "num_input_tokens_seen": 49145952,
+      "step": 784
+    },
+    {
+      "epoch": 2.6089850249584026,
+      "loss": 0.7366130352020264,
+      "loss_ce": 4.0785289456835017e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.062255859375,
+      "loss_xval": 0.73828125,
+      "num_input_tokens_seen": 49145952,
+      "step": 784
+    },
+    {
+      "epoch": 2.612312811980033,
+      "grad_norm": 12.318331718444824,
+      "learning_rate": 5e-06,
+      "loss": 0.8321,
+      "num_input_tokens_seen": 49206352,
+      "step": 785
+    },
+    {
+      "epoch": 2.612312811980033,
+      "loss": 0.8576995134353638,
+      "loss_ce": 0.004183888901025057,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.8515625,
+      "num_input_tokens_seen": 49206352,
+      "step": 785
+    },
+    {
+      "epoch": 2.615640599001664,
+      "grad_norm": 11.607172966003418,
+      "learning_rate": 5e-06,
+      "loss": 0.7514,
+      "num_input_tokens_seen": 49268404,
+      "step": 786
+    },
+    {
+      "epoch": 2.615640599001664,
+      "loss": 0.8183543086051941,
+      "loss_ce": 0.0019480856135487556,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 49268404,
+      "step": 786
+    },
+    {
+      "epoch": 2.6189683860232944,
+      "grad_norm": 14.248104095458984,
+      "learning_rate": 5e-06,
+      "loss": 0.875,
+      "num_input_tokens_seen": 49332036,
+      "step": 787
+    },
+    {
+      "epoch": 2.6189683860232944,
+      "loss": 0.7565048336982727,
+      "loss_ce": 0.0011337447213009,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 49332036,
+      "step": 787
+    },
+    {
+      "epoch": 2.622296173044925,
+      "grad_norm": 17.36334800720215,
+      "learning_rate": 5e-06,
+      "loss": 0.7014,
+      "num_input_tokens_seen": 49394740,
+      "step": 788
+    },
+    {
+      "epoch": 2.622296173044925,
+      "loss": 0.5740655064582825,
+      "loss_ce": 0.0005791678559035063,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.57421875,
+      "num_input_tokens_seen": 49394740,
+      "step": 788
+    },
+    {
+      "epoch": 2.6256239600665556,
+      "grad_norm": 20.616777420043945,
+      "learning_rate": 5e-06,
+      "loss": 0.8368,
+      "num_input_tokens_seen": 49456944,
+      "step": 789
+    },
+    {
+      "epoch": 2.6256239600665556,
+      "loss": 0.8422057628631592,
+      "loss_ce": 0.00028675797511823475,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 49456944,
+      "step": 789
+    },
+    {
+      "epoch": 2.6289517470881862,
+      "grad_norm": 19.112476348876953,
+      "learning_rate": 5e-06,
+      "loss": 0.9446,
+      "num_input_tokens_seen": 49520124,
+      "step": 790
+    },
+    {
+      "epoch": 2.6289517470881862,
+      "loss": 0.8822909593582153,
+      "loss_ce": 8.882155816536397e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 49520124,
+      "step": 790
+    },
+    {
+      "epoch": 2.632279534109817,
+      "grad_norm": 11.602209091186523,
+      "learning_rate": 5e-06,
+      "loss": 0.9648,
+      "num_input_tokens_seen": 49583060,
+      "step": 791
+    },
+    {
+      "epoch": 2.632279534109817,
+      "loss": 0.8904417753219604,
+      "loss_ce": 0.001037474605254829,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.890625,
+      "num_input_tokens_seen": 49583060,
+      "step": 791
+    },
+    {
+      "epoch": 2.6356073211314475,
+      "grad_norm": 19.73798179626465,
+      "learning_rate": 5e-06,
+      "loss": 0.9234,
+      "num_input_tokens_seen": 49643248,
+      "step": 792
+    },
+    {
+      "epoch": 2.6356073211314475,
+      "loss": 0.8780515193939209,
+      "loss_ce": 0.00012182131467852741,
+      "loss_iou": 0.296875,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 49643248,
+      "step": 792
+    },
+    {
+      "epoch": 2.638935108153078,
+      "grad_norm": 31.99167251586914,
+      "learning_rate": 5e-06,
+      "loss": 1.2363,
+      "num_input_tokens_seen": 49706308,
+      "step": 793
+    },
+    {
+      "epoch": 2.638935108153078,
+      "loss": 1.3618634939193726,
+      "loss_ce": 4.7138204536167905e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.08837890625,
+      "loss_xval": 1.359375,
+      "num_input_tokens_seen": 49706308,
+      "step": 793
+    },
+    {
+      "epoch": 2.6422628951747087,
+      "grad_norm": 27.228073120117188,
+      "learning_rate": 5e-06,
+      "loss": 0.9539,
+      "num_input_tokens_seen": 49769864,
+      "step": 794
+    },
+    {
+      "epoch": 2.6422628951747087,
+      "loss": 1.108590006828308,
+      "loss_ce": 0.00019164662808179855,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 49769864,
+      "step": 794
+    },
+    {
+      "epoch": 2.6455906821963393,
+      "grad_norm": 21.955646514892578,
+      "learning_rate": 5e-06,
+      "loss": 1.0233,
+      "num_input_tokens_seen": 49834096,
+      "step": 795
+    },
+    {
+      "epoch": 2.6455906821963393,
+      "loss": 0.9163686037063599,
+      "loss_ce": 0.00010885349911404774,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 49834096,
+      "step": 795
+    },
+    {
+      "epoch": 2.64891846921797,
+      "grad_norm": 17.316287994384766,
+      "learning_rate": 5e-06,
+      "loss": 0.6946,
+      "num_input_tokens_seen": 49897404,
+      "step": 796
+    },
+    {
+      "epoch": 2.64891846921797,
+      "loss": 0.49226897954940796,
+      "loss_ce": 0.0015463390154764056,
+      "loss_iou": 0.12060546875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.490234375,
+      "num_input_tokens_seen": 49897404,
+      "step": 796
+    },
+    {
+      "epoch": 2.6522462562396005,
+      "grad_norm": 9.91407299041748,
+      "learning_rate": 5e-06,
+      "loss": 0.9557,
+      "num_input_tokens_seen": 49960132,
+      "step": 797
+    },
+    {
+      "epoch": 2.6522462562396005,
+      "loss": 1.2538728713989258,
+      "loss_ce": 0.0002108162734657526,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 49960132,
+      "step": 797
+    },
+    {
+      "epoch": 2.655574043261231,
+      "grad_norm": 13.048734664916992,
+      "learning_rate": 5e-06,
+      "loss": 0.9346,
+      "num_input_tokens_seen": 50021972,
+      "step": 798
+    },
+    {
+      "epoch": 2.655574043261231,
+      "loss": 0.8867144584655762,
+      "loss_ce": 0.00023978884564712644,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 50021972,
+      "step": 798
+    },
+    {
+      "epoch": 2.6589018302828618,
+      "grad_norm": 20.037155151367188,
+      "learning_rate": 5e-06,
+      "loss": 1.0502,
+      "num_input_tokens_seen": 50084912,
+      "step": 799
+    },
+    {
+      "epoch": 2.6589018302828618,
+      "loss": 1.0424444675445557,
+      "loss_ce": 0.00020819506607949734,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.06201171875,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 50084912,
+      "step": 799
+    },
+    {
+      "epoch": 2.6622296173044924,
+      "grad_norm": 23.562681198120117,
+      "learning_rate": 5e-06,
+      "loss": 1.143,
+      "num_input_tokens_seen": 50148712,
+      "step": 800
+    },
+    {
+      "epoch": 2.6622296173044924,
+      "loss": 1.1583603620529175,
+      "loss_ce": 0.001622087205760181,
+      "loss_iou": 0.375,
+      "loss_num": 0.0810546875,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 50148712,
+      "step": 800
+    },
+    {
+      "epoch": 2.665557404326123,
+      "grad_norm": 12.925877571105957,
+      "learning_rate": 5e-06,
+      "loss": 0.807,
+      "num_input_tokens_seen": 50210512,
+      "step": 801
+    },
+    {
+      "epoch": 2.665557404326123,
+      "loss": 1.0540846586227417,
+      "loss_ce": 0.0003736392827704549,
+      "loss_iou": 0.390625,
+      "loss_num": 0.054443359375,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 50210512,
+      "step": 801
+    },
+    {
+      "epoch": 2.6688851913477536,
+      "grad_norm": 11.798089027404785,
+      "learning_rate": 5e-06,
+      "loss": 0.9187,
+      "num_input_tokens_seen": 50274604,
+      "step": 802
+    },
+    {
+      "epoch": 2.6688851913477536,
+      "loss": 0.7706761956214905,
+      "loss_ce": 0.00041252042865380645,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 50274604,
+      "step": 802
+    },
+    {
+      "epoch": 2.672212978369384,
+      "grad_norm": 9.151251792907715,
+      "learning_rate": 5e-06,
+      "loss": 0.5951,
+      "num_input_tokens_seen": 50336456,
+      "step": 803
+    },
+    {
+      "epoch": 2.672212978369384,
+      "loss": 0.6084632873535156,
+      "loss_ce": 6.486372876679525e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 50336456,
+      "step": 803
+    },
+    {
+      "epoch": 2.675540765391015,
+      "grad_norm": 66.70578002929688,
+      "learning_rate": 5e-06,
+      "loss": 0.7686,
+      "num_input_tokens_seen": 50400404,
+      "step": 804
+    },
+    {
+      "epoch": 2.675540765391015,
+      "loss": 0.7586301565170288,
+      "loss_ce": 8.528407488483936e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 50400404,
+      "step": 804
+    },
+    {
+      "epoch": 2.6788685524126454,
+      "grad_norm": 14.901700019836426,
+      "learning_rate": 5e-06,
+      "loss": 0.8289,
+      "num_input_tokens_seen": 50464096,
+      "step": 805
+    },
+    {
+      "epoch": 2.6788685524126454,
+      "loss": 0.9449521899223328,
+      "loss_ce": 0.0013486790703609586,
+      "loss_iou": 0.359375,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 50464096,
+      "step": 805
+    },
+    {
+      "epoch": 2.682196339434276,
+      "grad_norm": 16.723764419555664,
+      "learning_rate": 5e-06,
+      "loss": 0.6458,
+      "num_input_tokens_seen": 50526736,
+      "step": 806
+    },
+    {
+      "epoch": 2.682196339434276,
+      "loss": 0.7537314891815186,
+      "loss_ce": 0.0008018329390324652,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 50526736,
+      "step": 806
+    },
+    {
+      "epoch": 2.6855241264559067,
+      "grad_norm": 15.801563262939453,
+      "learning_rate": 5e-06,
+      "loss": 0.6904,
+      "num_input_tokens_seen": 50589548,
+      "step": 807
+    },
+    {
+      "epoch": 2.6855241264559067,
+      "loss": 0.5793936252593994,
+      "loss_ce": 4.7907709813443944e-05,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 50589548,
+      "step": 807
+    },
+    {
+      "epoch": 2.6888519134775377,
+      "grad_norm": 12.457715034484863,
+      "learning_rate": 5e-06,
+      "loss": 1.0248,
+      "num_input_tokens_seen": 50650376,
+      "step": 808
+    },
+    {
+      "epoch": 2.6888519134775377,
+      "loss": 1.0397467613220215,
+      "loss_ce": 0.00019596036872826517,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.05615234375,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 50650376,
+      "step": 808
+    },
+    {
+      "epoch": 2.6921797004991683,
+      "grad_norm": 23.163949966430664,
+      "learning_rate": 5e-06,
+      "loss": 0.6915,
+      "num_input_tokens_seen": 50713052,
+      "step": 809
+    },
+    {
+      "epoch": 2.6921797004991683,
+      "loss": 0.7694122791290283,
+      "loss_ce": 0.00012515315029304475,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 50713052,
+      "step": 809
+    },
+    {
+      "epoch": 2.695507487520799,
+      "grad_norm": 18.702228546142578,
+      "learning_rate": 5e-06,
+      "loss": 0.9216,
+      "num_input_tokens_seen": 50776796,
+      "step": 810
+    },
+    {
+      "epoch": 2.695507487520799,
+      "loss": 0.9874943494796753,
+      "loss_ce": 0.00018971107783727348,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.98828125,
+      "num_input_tokens_seen": 50776796,
+      "step": 810
+    },
+    {
+      "epoch": 2.6988352745424296,
+      "grad_norm": 9.986546516418457,
+      "learning_rate": 5e-06,
+      "loss": 0.9254,
+      "num_input_tokens_seen": 50839964,
+      "step": 811
+    },
+    {
+      "epoch": 2.6988352745424296,
+      "loss": 0.8390403985977173,
+      "loss_ce": 5.112246435601264e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.83984375,
+      "num_input_tokens_seen": 50839964,
+      "step": 811
+    },
+    {
+      "epoch": 2.70216306156406,
+      "grad_norm": 15.212403297424316,
+      "learning_rate": 5e-06,
+      "loss": 0.8605,
+      "num_input_tokens_seen": 50901932,
+      "step": 812
+    },
+    {
+      "epoch": 2.70216306156406,
+      "loss": 0.7976114749908447,
+      "loss_ce": 0.0004923460073769093,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 50901932,
+      "step": 812
+    },
+    {
+      "epoch": 2.7054908485856908,
+      "grad_norm": 25.891193389892578,
+      "learning_rate": 5e-06,
+      "loss": 0.8721,
+      "num_input_tokens_seen": 50964524,
+      "step": 813
+    },
+    {
+      "epoch": 2.7054908485856908,
+      "loss": 0.8218099474906921,
+      "loss_ce": 3.2617710530757904e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 50964524,
+      "step": 813
+    },
+    {
+      "epoch": 2.7088186356073214,
+      "grad_norm": 15.289278984069824,
+      "learning_rate": 5e-06,
+      "loss": 0.7168,
+      "num_input_tokens_seen": 51027040,
+      "step": 814
+    },
+    {
+      "epoch": 2.7088186356073214,
+      "loss": 0.5973207950592041,
+      "loss_ce": 0.0008241605246439576,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 51027040,
+      "step": 814
+    },
+    {
+      "epoch": 2.712146422628952,
+      "grad_norm": 12.39792537689209,
+      "learning_rate": 5e-06,
+      "loss": 0.803,
+      "num_input_tokens_seen": 51089884,
+      "step": 815
+    },
+    {
+      "epoch": 2.712146422628952,
+      "loss": 0.7842392325401306,
+      "loss_ce": 0.0014480899553745985,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0625,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 51089884,
+      "step": 815
+    },
+    {
+      "epoch": 2.7154742096505826,
+      "grad_norm": 74.54549407958984,
+      "learning_rate": 5e-06,
+      "loss": 0.8014,
+      "num_input_tokens_seen": 51152684,
+      "step": 816
+    },
+    {
+      "epoch": 2.7154742096505826,
+      "loss": 0.8306292295455933,
+      "loss_ce": 0.000184907199582085,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 51152684,
+      "step": 816
+    },
+    {
+      "epoch": 2.7188019966722132,
+      "grad_norm": 11.50522518157959,
+      "learning_rate": 5e-06,
+      "loss": 0.7829,
+      "num_input_tokens_seen": 51214500,
+      "step": 817
+    },
+    {
+      "epoch": 2.7188019966722132,
+      "loss": 0.881805419921875,
+      "loss_ce": 0.00045781210064888,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 51214500,
+      "step": 817
+    },
+    {
+      "epoch": 2.722129783693844,
+      "grad_norm": 13.687999725341797,
+      "learning_rate": 5e-06,
+      "loss": 0.8651,
+      "num_input_tokens_seen": 51276604,
+      "step": 818
+    },
+    {
+      "epoch": 2.722129783693844,
+      "loss": 0.8142944574356079,
+      "loss_ce": 0.0005737675237469375,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 51276604,
+      "step": 818
+    },
+    {
+      "epoch": 2.7254575707154745,
+      "grad_norm": 5.8822712898254395,
+      "learning_rate": 5e-06,
+      "loss": 0.5399,
+      "num_input_tokens_seen": 51337268,
+      "step": 819
+    },
+    {
+      "epoch": 2.7254575707154745,
+      "loss": 0.4156995415687561,
+      "loss_ce": 0.00017218466382473707,
+      "loss_iou": 0.1279296875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 51337268,
+      "step": 819
+    },
+    {
+      "epoch": 2.728785357737105,
+      "grad_norm": 10.577842712402344,
+      "learning_rate": 5e-06,
+      "loss": 0.5331,
+      "num_input_tokens_seen": 51398040,
+      "step": 820
+    },
+    {
+      "epoch": 2.728785357737105,
+      "loss": 0.44522571563720703,
+      "loss_ce": 0.0008897931547835469,
+      "loss_iou": 0.10986328125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.4453125,
+      "num_input_tokens_seen": 51398040,
+      "step": 820
+    },
+    {
+      "epoch": 2.7321131447587357,
+      "grad_norm": 14.673174858093262,
+      "learning_rate": 5e-06,
+      "loss": 0.9165,
+      "num_input_tokens_seen": 51459948,
+      "step": 821
+    },
+    {
+      "epoch": 2.7321131447587357,
+      "loss": 1.0592856407165527,
+      "loss_ce": 0.00020362591021694243,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.08203125,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 51459948,
+      "step": 821
+    },
+    {
+      "epoch": 2.7354409317803663,
+      "grad_norm": 10.287886619567871,
+      "learning_rate": 5e-06,
+      "loss": 0.8561,
+      "num_input_tokens_seen": 51524140,
+      "step": 822
+    },
+    {
+      "epoch": 2.7354409317803663,
+      "loss": 0.8826354742050171,
+      "loss_ce": 0.0007995132473297417,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 51524140,
+      "step": 822
+    },
+    {
+      "epoch": 2.738768718801997,
+      "grad_norm": 9.983254432678223,
+      "learning_rate": 5e-06,
+      "loss": 0.823,
+      "num_input_tokens_seen": 51587172,
+      "step": 823
+    },
+    {
+      "epoch": 2.738768718801997,
+      "loss": 1.0295748710632324,
+      "loss_ce": 3.3884258300531656e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.08056640625,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 51587172,
+      "step": 823
+    },
+    {
+      "epoch": 2.7420965058236275,
+      "grad_norm": 10.106569290161133,
+      "learning_rate": 5e-06,
+      "loss": 0.8517,
+      "num_input_tokens_seen": 51649148,
+      "step": 824
+    },
+    {
+      "epoch": 2.7420965058236275,
+      "loss": 0.7791653275489807,
+      "loss_ce": 0.0006008880445733666,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 51649148,
+      "step": 824
+    },
+    {
+      "epoch": 2.745424292845258,
+      "grad_norm": 20.371871948242188,
+      "learning_rate": 5e-06,
+      "loss": 1.0496,
+      "num_input_tokens_seen": 51712836,
+      "step": 825
+    },
+    {
+      "epoch": 2.745424292845258,
+      "loss": 0.855181097984314,
+      "loss_ce": 0.01924358680844307,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 51712836,
+      "step": 825
+    },
+    {
+      "epoch": 2.7487520798668887,
+      "grad_norm": 12.943856239318848,
+      "learning_rate": 5e-06,
+      "loss": 0.8533,
+      "num_input_tokens_seen": 51775596,
+      "step": 826
+    },
+    {
+      "epoch": 2.7487520798668887,
+      "loss": 0.914795994758606,
+      "loss_ce": 0.0004892984288744628,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 51775596,
+      "step": 826
+    },
+    {
+      "epoch": 2.7520798668885194,
+      "grad_norm": 8.180317878723145,
+      "learning_rate": 5e-06,
+      "loss": 0.6992,
+      "num_input_tokens_seen": 51838420,
+      "step": 827
+    },
+    {
+      "epoch": 2.7520798668885194,
+      "loss": 0.7848867177963257,
+      "loss_ce": 0.0020497734658420086,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 51838420,
+      "step": 827
+    },
+    {
+      "epoch": 2.75540765391015,
+      "grad_norm": 15.182528495788574,
+      "learning_rate": 5e-06,
+      "loss": 0.8481,
+      "num_input_tokens_seen": 51901276,
+      "step": 828
+    },
+    {
+      "epoch": 2.75540765391015,
+      "loss": 0.7813282012939453,
+      "loss_ce": 0.002519632689654827,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 51901276,
+      "step": 828
+    },
+    {
+      "epoch": 2.7587354409317806,
+      "grad_norm": 7.6279377937316895,
+      "learning_rate": 5e-06,
+      "loss": 0.6484,
+      "num_input_tokens_seen": 51962060,
+      "step": 829
+    },
+    {
+      "epoch": 2.7587354409317806,
+      "loss": 0.7650944590568542,
+      "loss_ce": 0.00203291280195117,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.76171875,
+      "num_input_tokens_seen": 51962060,
+      "step": 829
+    },
+    {
+      "epoch": 2.762063227953411,
+      "grad_norm": 10.674771308898926,
+      "learning_rate": 5e-06,
+      "loss": 0.817,
+      "num_input_tokens_seen": 52025592,
+      "step": 830
+    },
+    {
+      "epoch": 2.762063227953411,
+      "loss": 0.7599319815635681,
+      "loss_ce": 0.00016639150271657854,
+      "loss_iou": 0.296875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 52025592,
+      "step": 830
+    },
+    {
+      "epoch": 2.765391014975042,
+      "grad_norm": 19.160572052001953,
+      "learning_rate": 5e-06,
+      "loss": 0.9251,
+      "num_input_tokens_seen": 52088076,
+      "step": 831
+    },
+    {
+      "epoch": 2.765391014975042,
+      "loss": 0.805517315864563,
+      "loss_ce": 0.0005856447969563305,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 52088076,
+      "step": 831
+    },
+    {
+      "epoch": 2.7687188019966724,
+      "grad_norm": 6.1606125831604,
+      "learning_rate": 5e-06,
+      "loss": 0.5771,
+      "num_input_tokens_seen": 52150536,
+      "step": 832
+    },
+    {
+      "epoch": 2.7687188019966724,
+      "loss": 0.6053818464279175,
+      "loss_ce": 0.0006455372204072773,
+      "loss_iou": 0.1259765625,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 52150536,
+      "step": 832
+    },
+    {
+      "epoch": 2.772046589018303,
+      "grad_norm": 22.167266845703125,
+      "learning_rate": 5e-06,
+      "loss": 0.9371,
+      "num_input_tokens_seen": 52213100,
+      "step": 833
+    },
+    {
+      "epoch": 2.772046589018303,
+      "loss": 0.9295365810394287,
+      "loss_ce": 0.00021526089403778315,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 52213100,
+      "step": 833
+    },
+    {
+      "epoch": 2.7753743760399336,
+      "grad_norm": 10.496980667114258,
+      "learning_rate": 5e-06,
+      "loss": 0.6609,
+      "num_input_tokens_seen": 52276464,
+      "step": 834
+    },
+    {
+      "epoch": 2.7753743760399336,
+      "loss": 0.5460642576217651,
+      "loss_ce": 0.00016583751130383462,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 52276464,
+      "step": 834
+    },
+    {
+      "epoch": 2.7787021630615643,
+      "grad_norm": 10.732155799865723,
+      "learning_rate": 5e-06,
+      "loss": 0.8073,
+      "num_input_tokens_seen": 52340528,
+      "step": 835
+    },
+    {
+      "epoch": 2.7787021630615643,
+      "loss": 0.6755124926567078,
+      "loss_ce": 0.0002195164270233363,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 52340528,
+      "step": 835
+    },
+    {
+      "epoch": 2.782029950083195,
+      "grad_norm": 18.337326049804688,
+      "learning_rate": 5e-06,
+      "loss": 0.7644,
+      "num_input_tokens_seen": 52403600,
+      "step": 836
+    },
+    {
+      "epoch": 2.782029950083195,
+      "loss": 0.5846283435821533,
+      "loss_ce": 0.0003998374450020492,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 52403600,
+      "step": 836
+    },
+    {
+      "epoch": 2.7853577371048255,
+      "grad_norm": 68.70188903808594,
+      "learning_rate": 5e-06,
+      "loss": 0.884,
+      "num_input_tokens_seen": 52466592,
+      "step": 837
+    },
+    {
+      "epoch": 2.7853577371048255,
+      "loss": 1.1231420040130615,
+      "loss_ce": 0.0011938156094402075,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0703125,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 52466592,
+      "step": 837
+    },
+    {
+      "epoch": 2.788685524126456,
+      "grad_norm": 35.47993850708008,
+      "learning_rate": 5e-06,
+      "loss": 0.9647,
+      "num_input_tokens_seen": 52529452,
+      "step": 838
+    },
+    {
+      "epoch": 2.788685524126456,
+      "loss": 0.9949101209640503,
+      "loss_ce": 0.0007694980013184249,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 52529452,
+      "step": 838
+    },
+    {
+      "epoch": 2.7920133111480867,
+      "grad_norm": 21.51439094543457,
+      "learning_rate": 5e-06,
+      "loss": 0.8653,
+      "num_input_tokens_seen": 52592540,
+      "step": 839
+    },
+    {
+      "epoch": 2.7920133111480867,
+      "loss": 0.947022020816803,
+      "loss_ce": 0.0014653451507911086,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 52592540,
+      "step": 839
+    },
+    {
+      "epoch": 2.7953410981697173,
+      "grad_norm": 16.271461486816406,
+      "learning_rate": 5e-06,
+      "loss": 0.7983,
+      "num_input_tokens_seen": 52654704,
+      "step": 840
+    },
+    {
+      "epoch": 2.7953410981697173,
+      "loss": 0.980492353439331,
+      "loss_ce": 0.0014884801348671317,
+      "loss_iou": 0.28125,
+      "loss_num": 0.08349609375,
+      "loss_xval": 0.98046875,
+      "num_input_tokens_seen": 52654704,
+      "step": 840
+    },
+    {
+      "epoch": 2.798668885191348,
+      "grad_norm": 10.07279109954834,
+      "learning_rate": 5e-06,
+      "loss": 0.5959,
+      "num_input_tokens_seen": 52716292,
+      "step": 841
+    },
+    {
+      "epoch": 2.798668885191348,
+      "loss": 0.4731648564338684,
+      "loss_ce": 0.000630666094366461,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.47265625,
+      "num_input_tokens_seen": 52716292,
+      "step": 841
+    },
+    {
+      "epoch": 2.8019966722129785,
+      "grad_norm": 13.416818618774414,
+      "learning_rate": 5e-06,
+      "loss": 0.9959,
+      "num_input_tokens_seen": 52780908,
+      "step": 842
+    },
+    {
+      "epoch": 2.8019966722129785,
+      "loss": 1.2821091413497925,
+      "loss_ce": 0.0006150341359898448,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0859375,
+      "loss_xval": 1.28125,
+      "num_input_tokens_seen": 52780908,
+      "step": 842
+    },
+    {
+      "epoch": 2.805324459234609,
+      "grad_norm": 23.089929580688477,
+      "learning_rate": 5e-06,
+      "loss": 0.6385,
+      "num_input_tokens_seen": 52842208,
+      "step": 843
+    },
+    {
+      "epoch": 2.805324459234609,
+      "loss": 0.59052574634552,
+      "loss_ce": 0.0006820209673605859,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 52842208,
+      "step": 843
+    },
+    {
+      "epoch": 2.8086522462562398,
+      "grad_norm": 18.833816528320312,
+      "learning_rate": 5e-06,
+      "loss": 0.9746,
+      "num_input_tokens_seen": 52904996,
+      "step": 844
+    },
+    {
+      "epoch": 2.8086522462562398,
+      "loss": 1.0083526372909546,
+      "loss_ce": 5.187302303966135e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.046142578125,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 52904996,
+      "step": 844
+    },
+    {
+      "epoch": 2.8119800332778704,
+      "grad_norm": 14.17634105682373,
+      "learning_rate": 5e-06,
+      "loss": 0.884,
+      "num_input_tokens_seen": 52969668,
+      "step": 845
+    },
+    {
+      "epoch": 2.8119800332778704,
+      "loss": 1.1289470195770264,
+      "loss_ce": 0.0019939455669373274,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.050048828125,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 52969668,
+      "step": 845
+    },
+    {
+      "epoch": 2.815307820299501,
+      "grad_norm": 23.1187801361084,
+      "learning_rate": 5e-06,
+      "loss": 0.7,
+      "num_input_tokens_seen": 53031812,
+      "step": 846
+    },
+    {
+      "epoch": 2.815307820299501,
+      "loss": 0.7373853921890259,
+      "loss_ce": 8.065036672633141e-05,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.73828125,
+      "num_input_tokens_seen": 53031812,
+      "step": 846
+    },
+    {
+      "epoch": 2.8186356073211316,
+      "grad_norm": 9.363740921020508,
+      "learning_rate": 5e-06,
+      "loss": 0.9656,
+      "num_input_tokens_seen": 53095744,
+      "step": 847
+    },
+    {
+      "epoch": 2.8186356073211316,
+      "loss": 1.1773841381072998,
+      "loss_ce": 0.0018469768110662699,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.08154296875,
+      "loss_xval": 1.171875,
+      "num_input_tokens_seen": 53095744,
+      "step": 847
+    },
+    {
+      "epoch": 2.821963394342762,
+      "grad_norm": 11.104792594909668,
+      "learning_rate": 5e-06,
+      "loss": 0.8244,
+      "num_input_tokens_seen": 53158744,
+      "step": 848
+    },
+    {
+      "epoch": 2.821963394342762,
+      "loss": 0.9474537372589111,
+      "loss_ce": 0.0006763727869838476,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 53158744,
+      "step": 848
+    },
+    {
+      "epoch": 2.825291181364393,
+      "grad_norm": 12.321479797363281,
+      "learning_rate": 5e-06,
+      "loss": 0.9694,
+      "num_input_tokens_seen": 53221352,
+      "step": 849
+    },
+    {
+      "epoch": 2.825291181364393,
+      "loss": 0.9421533346176147,
+      "loss_ce": 0.0009911877568811178,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0810546875,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 53221352,
+      "step": 849
+    },
+    {
+      "epoch": 2.8286189683860234,
+      "grad_norm": 12.530994415283203,
+      "learning_rate": 5e-06,
+      "loss": 0.9866,
+      "num_input_tokens_seen": 53285452,
+      "step": 850
+    },
+    {
+      "epoch": 2.8286189683860234,
+      "loss": 1.0217738151550293,
+      "loss_ce": 0.001021851203404367,
+      "loss_iou": 0.375,
+      "loss_num": 0.05419921875,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 53285452,
+      "step": 850
+    },
+    {
+      "epoch": 2.831946755407654,
+      "grad_norm": 15.440099716186523,
+      "learning_rate": 5e-06,
+      "loss": 0.8441,
+      "num_input_tokens_seen": 53346748,
+      "step": 851
+    },
+    {
+      "epoch": 2.831946755407654,
+      "loss": 0.7872898578643799,
+      "loss_ce": 0.00018045374599751085,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 53346748,
+      "step": 851
+    },
+    {
+      "epoch": 2.8352745424292847,
+      "grad_norm": 28.69334602355957,
+      "learning_rate": 5e-06,
+      "loss": 0.8573,
+      "num_input_tokens_seen": 53409872,
+      "step": 852
+    },
+    {
+      "epoch": 2.8352745424292847,
+      "loss": 0.8631025552749634,
+      "loss_ce": 0.002262737834826112,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 53409872,
+      "step": 852
+    },
+    {
+      "epoch": 2.8386023294509153,
+      "grad_norm": 12.272283554077148,
+      "learning_rate": 5e-06,
+      "loss": 0.8724,
+      "num_input_tokens_seen": 53471764,
+      "step": 853
+    },
+    {
+      "epoch": 2.8386023294509153,
+      "loss": 1.037917971611023,
+      "loss_ce": 0.0020293924026191235,
+      "loss_iou": 0.3125,
+      "loss_num": 0.08203125,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 53471764,
+      "step": 853
+    },
+    {
+      "epoch": 2.841930116472546,
+      "grad_norm": 19.907730102539062,
+      "learning_rate": 5e-06,
+      "loss": 0.8915,
+      "num_input_tokens_seen": 53535256,
+      "step": 854
+    },
+    {
+      "epoch": 2.841930116472546,
+      "loss": 1.0508040189743042,
+      "loss_ce": 0.0014875233173370361,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.049560546875,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 53535256,
+      "step": 854
+    },
+    {
+      "epoch": 2.8452579034941765,
+      "grad_norm": 16.519927978515625,
+      "learning_rate": 5e-06,
+      "loss": 0.8323,
+      "num_input_tokens_seen": 53598600,
+      "step": 855
+    },
+    {
+      "epoch": 2.8452579034941765,
+      "loss": 0.9220311045646667,
+      "loss_ce": 0.0006444402388297021,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 53598600,
+      "step": 855
+    },
+    {
+      "epoch": 2.848585690515807,
+      "grad_norm": 21.156627655029297,
+      "learning_rate": 5e-06,
+      "loss": 0.8297,
+      "num_input_tokens_seen": 53660192,
+      "step": 856
+    },
+    {
+      "epoch": 2.848585690515807,
+      "loss": 0.6860744953155518,
+      "loss_ce": 3.937574365409091e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 53660192,
+      "step": 856
+    },
+    {
+      "epoch": 2.8519134775374377,
+      "grad_norm": 15.835363388061523,
+      "learning_rate": 5e-06,
+      "loss": 1.1557,
+      "num_input_tokens_seen": 53723952,
+      "step": 857
+    },
+    {
+      "epoch": 2.8519134775374377,
+      "loss": 1.1079440116882324,
+      "loss_ce": 0.0005221446044743061,
+      "loss_iou": 0.375,
+      "loss_num": 0.0712890625,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 53723952,
+      "step": 857
+    },
+    {
+      "epoch": 2.8552412645590683,
+      "grad_norm": 9.046875,
+      "learning_rate": 5e-06,
+      "loss": 0.8038,
+      "num_input_tokens_seen": 53785616,
+      "step": 858
+    },
+    {
+      "epoch": 2.8552412645590683,
+      "loss": 0.8684693574905396,
+      "loss_ce": 0.0004273601807653904,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 53785616,
+      "step": 858
+    },
+    {
+      "epoch": 2.858569051580699,
+      "grad_norm": 46.66500473022461,
+      "learning_rate": 5e-06,
+      "loss": 1.2536,
+      "num_input_tokens_seen": 53849740,
+      "step": 859
+    },
+    {
+      "epoch": 2.858569051580699,
+      "loss": 1.047816276550293,
+      "loss_ce": 0.0068005844950675964,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.05517578125,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 53849740,
+      "step": 859
+    },
+    {
+      "epoch": 2.8618968386023296,
+      "grad_norm": 25.487953186035156,
+      "learning_rate": 5e-06,
+      "loss": 0.8454,
+      "num_input_tokens_seen": 53911752,
+      "step": 860
+    },
+    {
+      "epoch": 2.8618968386023296,
+      "loss": 0.7410845160484314,
+      "loss_ce": 0.00017875817138701677,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 53911752,
+      "step": 860
+    },
+    {
+      "epoch": 2.86522462562396,
+      "grad_norm": 8.039831161499023,
+      "learning_rate": 5e-06,
+      "loss": 0.4574,
+      "num_input_tokens_seen": 53973596,
+      "step": 861
+    },
+    {
+      "epoch": 2.86522462562396,
+      "loss": 0.4054723381996155,
+      "loss_ce": 0.00019890097610186785,
+      "loss_iou": 0.107421875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 53973596,
+      "step": 861
+    },
+    {
+      "epoch": 2.868552412645591,
+      "grad_norm": 12.18427562713623,
+      "learning_rate": 5e-06,
+      "loss": 0.777,
+      "num_input_tokens_seen": 54037804,
+      "step": 862
+    },
+    {
+      "epoch": 2.868552412645591,
+      "loss": 0.7093939781188965,
+      "loss_ce": 0.004559976048767567,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 54037804,
+      "step": 862
+    },
+    {
+      "epoch": 2.8718801996672214,
+      "grad_norm": 11.70964527130127,
+      "learning_rate": 5e-06,
+      "loss": 0.8396,
+      "num_input_tokens_seen": 54099912,
+      "step": 863
+    },
+    {
+      "epoch": 2.8718801996672214,
+      "loss": 0.9924643635749817,
+      "loss_ce": 0.0005210142116993666,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 54099912,
+      "step": 863
+    },
+    {
+      "epoch": 2.875207986688852,
+      "grad_norm": 12.803565979003906,
+      "learning_rate": 5e-06,
+      "loss": 0.7378,
+      "num_input_tokens_seen": 54162264,
+      "step": 864
+    },
+    {
+      "epoch": 2.875207986688852,
+      "loss": 0.7761240005493164,
+      "loss_ce": 0.002198257017880678,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 54162264,
+      "step": 864
+    },
+    {
+      "epoch": 2.8785357737104826,
+      "grad_norm": 11.305477142333984,
+      "learning_rate": 5e-06,
+      "loss": 0.7493,
+      "num_input_tokens_seen": 54225144,
+      "step": 865
+    },
+    {
+      "epoch": 2.8785357737104826,
+      "loss": 0.8462469577789307,
+      "loss_ce": 0.0005437912186607718,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 54225144,
+      "step": 865
+    },
+    {
+      "epoch": 2.8818635607321132,
+      "grad_norm": 12.183488845825195,
+      "learning_rate": 5e-06,
+      "loss": 1.0525,
+      "num_input_tokens_seen": 54286592,
+      "step": 866
+    },
+    {
+      "epoch": 2.8818635607321132,
+      "loss": 1.0839440822601318,
+      "loss_ce": 0.0009362882701680064,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.08203125,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 54286592,
+      "step": 866
+    },
+    {
+      "epoch": 2.885191347753744,
+      "grad_norm": 10.725017547607422,
+      "learning_rate": 5e-06,
+      "loss": 1.0308,
+      "num_input_tokens_seen": 54348900,
+      "step": 867
+    },
+    {
+      "epoch": 2.885191347753744,
+      "loss": 0.9614525437355042,
+      "loss_ce": 2.6763778805616312e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.9609375,
+      "num_input_tokens_seen": 54348900,
+      "step": 867
+    },
+    {
+      "epoch": 2.8885191347753745,
+      "grad_norm": 12.298053741455078,
+      "learning_rate": 5e-06,
+      "loss": 0.7657,
+      "num_input_tokens_seen": 54411716,
+      "step": 868
+    },
+    {
+      "epoch": 2.8885191347753745,
+      "loss": 0.7437622547149658,
+      "loss_ce": 0.0010864399373531342,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 54411716,
+      "step": 868
+    },
+    {
+      "epoch": 2.891846921797005,
+      "grad_norm": 29.771516799926758,
+      "learning_rate": 5e-06,
+      "loss": 0.8699,
+      "num_input_tokens_seen": 54474996,
+      "step": 869
+    },
+    {
+      "epoch": 2.891846921797005,
+      "loss": 0.875381350517273,
+      "loss_ce": 0.0006255026091821492,
+      "loss_iou": 0.296875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.875,
+      "num_input_tokens_seen": 54474996,
+      "step": 869
+    },
+    {
+      "epoch": 2.8951747088186357,
+      "grad_norm": 22.231077194213867,
+      "learning_rate": 5e-06,
+      "loss": 0.6807,
+      "num_input_tokens_seen": 54538292,
+      "step": 870
+    },
+    {
+      "epoch": 2.8951747088186357,
+      "loss": 0.6149954795837402,
+      "loss_ce": 0.0007987120770849288,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 54538292,
+      "step": 870
+    },
+    {
+      "epoch": 2.8985024958402663,
+      "grad_norm": 13.072921752929688,
+      "learning_rate": 5e-06,
+      "loss": 0.9311,
+      "num_input_tokens_seen": 54600360,
+      "step": 871
+    },
+    {
+      "epoch": 2.8985024958402663,
+      "loss": 1.268268346786499,
+      "loss_ce": 0.0002019708917941898,
+      "loss_iou": 0.390625,
+      "loss_num": 0.09716796875,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 54600360,
+      "step": 871
+    },
+    {
+      "epoch": 2.901830282861897,
+      "grad_norm": 12.388557434082031,
+      "learning_rate": 5e-06,
+      "loss": 0.8199,
+      "num_input_tokens_seen": 54662656,
+      "step": 872
+    },
+    {
+      "epoch": 2.901830282861897,
+      "loss": 0.6191840767860413,
+      "loss_ce": 4.344484841567464e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 54662656,
+      "step": 872
+    },
+    {
+      "epoch": 2.9051580698835275,
+      "grad_norm": 7.256121635437012,
+      "learning_rate": 5e-06,
+      "loss": 0.7057,
+      "num_input_tokens_seen": 54723256,
+      "step": 873
+    },
+    {
+      "epoch": 2.9051580698835275,
+      "loss": 0.5848498344421387,
+      "loss_ce": 0.001109591918066144,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 54723256,
+      "step": 873
+    },
+    {
+      "epoch": 2.908485856905158,
+      "grad_norm": 30.038572311401367,
+      "learning_rate": 5e-06,
+      "loss": 1.0476,
+      "num_input_tokens_seen": 54786872,
+      "step": 874
+    },
+    {
+      "epoch": 2.908485856905158,
+      "loss": 0.7975156903266907,
+      "loss_ce": 0.000274513236945495,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 54786872,
+      "step": 874
+    },
+    {
+      "epoch": 2.9118136439267888,
+      "grad_norm": 27.706817626953125,
+      "learning_rate": 5e-06,
+      "loss": 0.8176,
+      "num_input_tokens_seen": 54848424,
+      "step": 875
+    },
+    {
+      "epoch": 2.9118136439267888,
+      "loss": 0.6453001499176025,
+      "loss_ce": 0.0006468780338764191,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 54848424,
+      "step": 875
+    },
+    {
+      "epoch": 2.9151414309484194,
+      "grad_norm": 16.334976196289062,
+      "learning_rate": 5e-06,
+      "loss": 0.828,
+      "num_input_tokens_seen": 54910524,
+      "step": 876
+    },
+    {
+      "epoch": 2.9151414309484194,
+      "loss": 0.9851783514022827,
+      "loss_ce": 7.095631735865027e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 54910524,
+      "step": 876
+    },
+    {
+      "epoch": 2.91846921797005,
+      "grad_norm": 26.634248733520508,
+      "learning_rate": 5e-06,
+      "loss": 0.8181,
+      "num_input_tokens_seen": 54973516,
+      "step": 877
+    },
+    {
+      "epoch": 2.91846921797005,
+      "loss": 0.83323734998703,
+      "loss_ce": 0.00035160701372660697,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 54973516,
+      "step": 877
+    },
+    {
+      "epoch": 2.9217970049916806,
+      "grad_norm": 22.770780563354492,
+      "learning_rate": 5e-06,
+      "loss": 0.897,
+      "num_input_tokens_seen": 55034692,
+      "step": 878
+    },
+    {
+      "epoch": 2.9217970049916806,
+      "loss": 0.9181663990020752,
+      "loss_ce": 0.00019761671137530357,
+      "loss_iou": 0.328125,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 55034692,
+      "step": 878
+    },
+    {
+      "epoch": 2.925124792013311,
+      "grad_norm": 14.436359405517578,
+      "learning_rate": 5e-06,
+      "loss": 0.7651,
+      "num_input_tokens_seen": 55096088,
+      "step": 879
+    },
+    {
+      "epoch": 2.925124792013311,
+      "loss": 0.8270174860954285,
+      "loss_ce": 0.00011315852316329256,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 55096088,
+      "step": 879
+    },
+    {
+      "epoch": 2.928452579034942,
+      "grad_norm": 21.74774742126465,
+      "learning_rate": 5e-06,
+      "loss": 0.9191,
+      "num_input_tokens_seen": 55159628,
+      "step": 880
+    },
+    {
+      "epoch": 2.928452579034942,
+      "loss": 0.9564664363861084,
+      "loss_ce": 0.0008694813586771488,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.95703125,
+      "num_input_tokens_seen": 55159628,
+      "step": 880
+    },
+    {
+      "epoch": 2.9317803660565724,
+      "grad_norm": 13.85302448272705,
+      "learning_rate": 5e-06,
+      "loss": 0.6639,
+      "num_input_tokens_seen": 55222304,
+      "step": 881
+    },
+    {
+      "epoch": 2.9317803660565724,
+      "loss": 0.6171444654464722,
+      "loss_ce": 0.0015438641421496868,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 55222304,
+      "step": 881
+    },
+    {
+      "epoch": 2.935108153078203,
+      "grad_norm": 15.573785781860352,
+      "learning_rate": 5e-06,
+      "loss": 0.9698,
+      "num_input_tokens_seen": 55285600,
+      "step": 882
+    },
+    {
+      "epoch": 2.935108153078203,
+      "loss": 1.1866347789764404,
+      "loss_ce": 0.0001112837198888883,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.060791015625,
+      "loss_xval": 1.1875,
+      "num_input_tokens_seen": 55285600,
+      "step": 882
+    },
+    {
+      "epoch": 2.9384359400998337,
+      "grad_norm": 9.72053337097168,
+      "learning_rate": 5e-06,
+      "loss": 1.011,
+      "num_input_tokens_seen": 55349256,
+      "step": 883
+    },
+    {
+      "epoch": 2.9384359400998337,
+      "loss": 0.9649059176445007,
+      "loss_ce": 0.0007946036057546735,
+      "loss_iou": 0.3125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.96484375,
+      "num_input_tokens_seen": 55349256,
+      "step": 883
+    },
+    {
+      "epoch": 2.9417637271214643,
+      "grad_norm": 40.047515869140625,
+      "learning_rate": 5e-06,
+      "loss": 0.9117,
+      "num_input_tokens_seen": 55413708,
+      "step": 884
+    },
+    {
+      "epoch": 2.9417637271214643,
+      "loss": 0.8685333132743835,
+      "loss_ce": 0.0026885762345045805,
+      "loss_iou": 0.28125,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 55413708,
+      "step": 884
+    },
+    {
+      "epoch": 2.945091514143095,
+      "grad_norm": 29.746788024902344,
+      "learning_rate": 5e-06,
+      "loss": 1.1525,
+      "num_input_tokens_seen": 55478204,
+      "step": 885
+    },
+    {
+      "epoch": 2.945091514143095,
+      "loss": 1.2547807693481445,
+      "loss_ce": 0.00624566525220871,
+      "loss_iou": 0.453125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 55478204,
+      "step": 885
+    },
+    {
+      "epoch": 2.9484193011647255,
+      "grad_norm": 14.427910804748535,
+      "learning_rate": 5e-06,
+      "loss": 0.8744,
+      "num_input_tokens_seen": 55540140,
+      "step": 886
+    },
+    {
+      "epoch": 2.9484193011647255,
+      "loss": 0.9295307993888855,
+      "loss_ce": 0.0006977797020226717,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 55540140,
+      "step": 886
+    },
+    {
+      "epoch": 2.951747088186356,
+      "grad_norm": 13.86884593963623,
+      "learning_rate": 5e-06,
+      "loss": 0.7577,
+      "num_input_tokens_seen": 55602724,
+      "step": 887
+    },
+    {
+      "epoch": 2.951747088186356,
+      "loss": 0.9324434399604797,
+      "loss_ce": 0.0006807436002418399,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 55602724,
+      "step": 887
+    },
+    {
+      "epoch": 2.9550748752079867,
+      "grad_norm": 14.230158805847168,
+      "learning_rate": 5e-06,
+      "loss": 1.1293,
+      "num_input_tokens_seen": 55667092,
+      "step": 888
+    },
+    {
+      "epoch": 2.9550748752079867,
+      "loss": 1.015046238899231,
+      "loss_ce": 0.0013743427116423845,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.06298828125,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 55667092,
+      "step": 888
+    },
+    {
+      "epoch": 2.9584026622296173,
+      "grad_norm": 13.532299041748047,
+      "learning_rate": 5e-06,
+      "loss": 0.7976,
+      "num_input_tokens_seen": 55729756,
+      "step": 889
+    },
+    {
+      "epoch": 2.9584026622296173,
+      "loss": 0.7759445905685425,
+      "loss_ce": 0.0006149307591840625,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 55729756,
+      "step": 889
+    },
+    {
+      "epoch": 2.961730449251248,
+      "grad_norm": 15.530501365661621,
+      "learning_rate": 5e-06,
+      "loss": 0.6693,
+      "num_input_tokens_seen": 55792828,
+      "step": 890
+    },
+    {
+      "epoch": 2.961730449251248,
+      "loss": 0.7313098907470703,
+      "loss_ce": 0.0003528632805682719,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 55792828,
+      "step": 890
+    },
+    {
+      "epoch": 2.9650582362728786,
+      "grad_norm": 15.952990531921387,
+      "learning_rate": 5e-06,
+      "loss": 0.8108,
+      "num_input_tokens_seen": 55854952,
+      "step": 891
+    },
+    {
+      "epoch": 2.9650582362728786,
+      "loss": 0.6857528686523438,
+      "loss_ce": 0.0009384113945998251,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.68359375,
+      "num_input_tokens_seen": 55854952,
+      "step": 891
+    },
+    {
+      "epoch": 2.968386023294509,
+      "grad_norm": 7.85128116607666,
+      "learning_rate": 5e-06,
+      "loss": 0.8289,
+      "num_input_tokens_seen": 55917956,
+      "step": 892
+    },
+    {
+      "epoch": 2.968386023294509,
+      "loss": 0.668018102645874,
+      "loss_ce": 4.93402112624608e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 55917956,
+      "step": 892
+    },
+    {
+      "epoch": 2.97171381031614,
+      "grad_norm": 17.0181884765625,
+      "learning_rate": 5e-06,
+      "loss": 0.6734,
+      "num_input_tokens_seen": 55979892,
+      "step": 893
+    },
+    {
+      "epoch": 2.97171381031614,
+      "loss": 0.6943376064300537,
+      "loss_ce": 0.0009782593697309494,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 55979892,
+      "step": 893
+    },
+    {
+      "epoch": 2.9750415973377704,
+      "grad_norm": 24.690635681152344,
+      "learning_rate": 5e-06,
+      "loss": 0.7435,
+      "num_input_tokens_seen": 56042384,
+      "step": 894
+    },
+    {
+      "epoch": 2.9750415973377704,
+      "loss": 0.47268909215927124,
+      "loss_ce": 0.00015488307690247893,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.47265625,
+      "num_input_tokens_seen": 56042384,
+      "step": 894
+    },
+    {
+      "epoch": 2.978369384359401,
+      "grad_norm": 18.475412368774414,
+      "learning_rate": 5e-06,
+      "loss": 0.8394,
+      "num_input_tokens_seen": 56104388,
+      "step": 895
+    },
+    {
+      "epoch": 2.978369384359401,
+      "loss": 0.7747071981430054,
+      "loss_ce": 4.900360363535583e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.0849609375,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 56104388,
+      "step": 895
+    },
+    {
+      "epoch": 2.9816971713810316,
+      "grad_norm": 10.097528457641602,
+      "learning_rate": 5e-06,
+      "loss": 0.5897,
+      "num_input_tokens_seen": 56167044,
+      "step": 896
+    },
+    {
+      "epoch": 2.9816971713810316,
+      "loss": 0.6344481706619263,
+      "loss_ce": 0.0010939212515950203,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 56167044,
+      "step": 896
+    },
+    {
+      "epoch": 2.9850249584026622,
+      "grad_norm": 15.803515434265137,
+      "learning_rate": 5e-06,
+      "loss": 0.623,
+      "num_input_tokens_seen": 56229968,
+      "step": 897
+    },
+    {
+      "epoch": 2.9850249584026622,
+      "loss": 0.4044834077358246,
+      "loss_ce": 0.00033913765219040215,
+      "loss_iou": 0.1103515625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.404296875,
+      "num_input_tokens_seen": 56229968,
+      "step": 897
+    },
+    {
+      "epoch": 2.988352745424293,
+      "grad_norm": 10.865888595581055,
+      "learning_rate": 5e-06,
+      "loss": 0.706,
+      "num_input_tokens_seen": 56293428,
+      "step": 898
+    },
+    {
+      "epoch": 2.988352745424293,
+      "loss": 0.6543700695037842,
+      "loss_ce": 1.2155430340499151e-05,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 56293428,
+      "step": 898
+    },
+    {
+      "epoch": 2.9916805324459235,
+      "grad_norm": 10.439217567443848,
+      "learning_rate": 5e-06,
+      "loss": 0.9566,
+      "num_input_tokens_seen": 56353876,
+      "step": 899
+    },
+    {
+      "epoch": 2.9916805324459235,
+      "loss": 0.7114546298980713,
+      "loss_ce": 2.8855994969489984e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 56353876,
+      "step": 899
+    },
+    {
+      "epoch": 2.995008319467554,
+      "grad_norm": 31.885540008544922,
+      "learning_rate": 5e-06,
+      "loss": 1.0362,
+      "num_input_tokens_seen": 56418016,
+      "step": 900
+    },
+    {
+      "epoch": 2.995008319467554,
+      "loss": 1.0725525617599487,
+      "loss_ce": 0.00028691417537629604,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0712890625,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 56418016,
+      "step": 900
+    },
+    {
+      "epoch": 2.9983361064891847,
+      "grad_norm": 31.542757034301758,
+      "learning_rate": 5e-06,
+      "loss": 0.915,
+      "num_input_tokens_seen": 56479904,
+      "step": 901
+    },
+    {
+      "epoch": 2.9983361064891847,
+      "loss": 1.0275923013687134,
+      "loss_ce": 0.0026899336371570826,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.06298828125,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 56479904,
+      "step": 901
+    },
+    {
+      "epoch": 2.9983361064891847,
+      "loss": 0.8745320439338684,
+      "loss_ce": 0.00026448973221704364,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.875,
+      "num_input_tokens_seen": 56512420,
+      "step": 901
+    },
+    {
+      "epoch": 3.0016638935108153,
+      "grad_norm": 20.686283111572266,
+      "learning_rate": 5e-06,
+      "loss": 0.9452,
+      "num_input_tokens_seen": 56544136,
+      "step": 902
+    },
+    {
+      "epoch": 3.0016638935108153,
+      "loss": 1.015965461730957,
+      "loss_ce": 0.0008286432130262256,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.060791015625,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 56544136,
+      "step": 902
+    },
+    {
+      "epoch": 3.004991680532446,
+      "grad_norm": 30.531023025512695,
+      "learning_rate": 5e-06,
+      "loss": 0.9768,
+      "num_input_tokens_seen": 56606324,
+      "step": 903
+    },
+    {
+      "epoch": 3.004991680532446,
+      "loss": 0.9323800802230835,
+      "loss_ce": 0.00012908241478726268,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 56606324,
+      "step": 903
+    },
+    {
+      "epoch": 3.0083194675540765,
+      "grad_norm": 13.145706176757812,
+      "learning_rate": 5e-06,
+      "loss": 0.7993,
+      "num_input_tokens_seen": 56668452,
+      "step": 904
+    },
+    {
+      "epoch": 3.0083194675540765,
+      "loss": 0.9373518228530884,
+      "loss_ce": 9.595306619303301e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.9375,
+      "num_input_tokens_seen": 56668452,
+      "step": 904
+    },
+    {
+      "epoch": 3.011647254575707,
+      "grad_norm": 157.36032104492188,
+      "learning_rate": 5e-06,
+      "loss": 0.6418,
+      "num_input_tokens_seen": 56730152,
+      "step": 905
+    },
+    {
+      "epoch": 3.011647254575707,
+      "loss": 0.5183830261230469,
+      "loss_ce": 0.002696982817724347,
+      "loss_iou": 0.0,
+      "loss_num": 0.10302734375,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 56730152,
+      "step": 905
+    },
+    {
+      "epoch": 3.0149750415973378,
+      "grad_norm": 36.10614013671875,
+      "learning_rate": 5e-06,
+      "loss": 0.7523,
+      "num_input_tokens_seen": 56792164,
+      "step": 906
+    },
+    {
+      "epoch": 3.0149750415973378,
+      "loss": 1.0320461988449097,
+      "loss_ce": 6.377464160323143e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 56792164,
+      "step": 906
+    },
+    {
+      "epoch": 3.0183028286189684,
+      "grad_norm": 6.202398777008057,
+      "learning_rate": 5e-06,
+      "loss": 0.6772,
+      "num_input_tokens_seen": 56852804,
+      "step": 907
+    },
+    {
+      "epoch": 3.0183028286189684,
+      "loss": 0.6682400107383728,
+      "loss_ce": 2.7143531042383984e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 56852804,
+      "step": 907
+    },
+    {
+      "epoch": 3.021630615640599,
+      "grad_norm": 15.27850341796875,
+      "learning_rate": 5e-06,
+      "loss": 0.8055,
+      "num_input_tokens_seen": 56915900,
+      "step": 908
+    },
+    {
+      "epoch": 3.021630615640599,
+      "loss": 0.9380273222923279,
+      "loss_ce": 3.903326069121249e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.9375,
+      "num_input_tokens_seen": 56915900,
+      "step": 908
+    },
+    {
+      "epoch": 3.0249584026622296,
+      "grad_norm": 24.6085147857666,
+      "learning_rate": 5e-06,
+      "loss": 1.1187,
+      "num_input_tokens_seen": 56978488,
+      "step": 909
+    },
+    {
+      "epoch": 3.0249584026622296,
+      "loss": 1.3314964771270752,
+      "loss_ce": 0.002883219625800848,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0771484375,
+      "loss_xval": 1.328125,
+      "num_input_tokens_seen": 56978488,
+      "step": 909
+    },
+    {
+      "epoch": 3.02828618968386,
+      "grad_norm": 21.691267013549805,
+      "learning_rate": 5e-06,
+      "loss": 0.7511,
+      "num_input_tokens_seen": 57042312,
+      "step": 910
+    },
+    {
+      "epoch": 3.02828618968386,
+      "loss": 0.7569243907928467,
+      "loss_ce": 0.0004546991258393973,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 57042312,
+      "step": 910
+    },
+    {
+      "epoch": 3.031613976705491,
+      "grad_norm": 18.6723690032959,
+      "learning_rate": 5e-06,
+      "loss": 0.6642,
+      "num_input_tokens_seen": 57103640,
+      "step": 911
+    },
+    {
+      "epoch": 3.031613976705491,
+      "loss": 0.7280317544937134,
+      "loss_ce": 0.00024860326084308326,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 57103640,
+      "step": 911
+    },
+    {
+      "epoch": 3.0349417637271214,
+      "grad_norm": 16.631738662719727,
+      "learning_rate": 5e-06,
+      "loss": 0.6276,
+      "num_input_tokens_seen": 57165720,
+      "step": 912
+    },
+    {
+      "epoch": 3.0349417637271214,
+      "loss": 0.6402130126953125,
+      "loss_ce": 0.0010528563288971782,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 57165720,
+      "step": 912
+    },
+    {
+      "epoch": 3.038269550748752,
+      "grad_norm": 11.869771957397461,
+      "learning_rate": 5e-06,
+      "loss": 0.8904,
+      "num_input_tokens_seen": 57228768,
+      "step": 913
+    },
+    {
+      "epoch": 3.038269550748752,
+      "loss": 0.7842072248458862,
+      "loss_ce": 2.7603602575254627e-05,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 57228768,
+      "step": 913
+    },
+    {
+      "epoch": 3.0415973377703827,
+      "grad_norm": 16.256175994873047,
+      "learning_rate": 5e-06,
+      "loss": 0.7901,
+      "num_input_tokens_seen": 57290928,
+      "step": 914
+    },
+    {
+      "epoch": 3.0415973377703827,
+      "loss": 0.9927595257759094,
+      "loss_ce": 8.371622243430465e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 57290928,
+      "step": 914
+    },
+    {
+      "epoch": 3.0449251247920133,
+      "grad_norm": 24.751188278198242,
+      "learning_rate": 5e-06,
+      "loss": 0.6958,
+      "num_input_tokens_seen": 57353872,
+      "step": 915
+    },
+    {
+      "epoch": 3.0449251247920133,
+      "loss": 0.7315420508384705,
+      "loss_ce": 0.00424710288643837,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 57353872,
+      "step": 915
+    },
+    {
+      "epoch": 3.048252911813644,
+      "grad_norm": 16.81283187866211,
+      "learning_rate": 5e-06,
+      "loss": 1.0033,
+      "num_input_tokens_seen": 57418248,
+      "step": 916
+    },
+    {
+      "epoch": 3.048252911813644,
+      "loss": 0.8678412437438965,
+      "loss_ce": 0.0008979164995253086,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 57418248,
+      "step": 916
+    },
+    {
+      "epoch": 3.0515806988352745,
+      "grad_norm": 8.283062934875488,
+      "learning_rate": 5e-06,
+      "loss": 0.8209,
+      "num_input_tokens_seen": 57481256,
+      "step": 917
+    },
+    {
+      "epoch": 3.0515806988352745,
+      "loss": 0.8476381897926331,
+      "loss_ce": 0.00047019918565638363,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.84765625,
+      "num_input_tokens_seen": 57481256,
+      "step": 917
+    },
+    {
+      "epoch": 3.054908485856905,
+      "grad_norm": 19.258445739746094,
+      "learning_rate": 5e-06,
+      "loss": 0.7102,
+      "num_input_tokens_seen": 57543436,
+      "step": 918
+    },
+    {
+      "epoch": 3.054908485856905,
+      "loss": 0.8542848825454712,
+      "loss_ce": 0.0001589024905115366,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 57543436,
+      "step": 918
+    },
+    {
+      "epoch": 3.0582362728785357,
+      "grad_norm": 24.21323013305664,
+      "learning_rate": 5e-06,
+      "loss": 0.7274,
+      "num_input_tokens_seen": 57606988,
+      "step": 919
+    },
+    {
+      "epoch": 3.0582362728785357,
+      "loss": 0.861693263053894,
+      "loss_ce": 0.0008534241351298988,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 57606988,
+      "step": 919
+    },
+    {
+      "epoch": 3.0615640599001663,
+      "grad_norm": 28.660486221313477,
+      "learning_rate": 5e-06,
+      "loss": 0.5461,
+      "num_input_tokens_seen": 57668872,
+      "step": 920
+    },
+    {
+      "epoch": 3.0615640599001663,
+      "loss": 0.6151924133300781,
+      "loss_ce": 0.0004463810473680496,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 57668872,
+      "step": 920
+    },
+    {
+      "epoch": 3.064891846921797,
+      "grad_norm": 10.633936882019043,
+      "learning_rate": 5e-06,
+      "loss": 0.9587,
+      "num_input_tokens_seen": 57732472,
+      "step": 921
+    },
+    {
+      "epoch": 3.064891846921797,
+      "loss": 1.0549280643463135,
+      "loss_ce": 0.00048464565770700574,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.06298828125,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 57732472,
+      "step": 921
+    },
+    {
+      "epoch": 3.0682196339434276,
+      "grad_norm": 26.227346420288086,
+      "learning_rate": 5e-06,
+      "loss": 0.966,
+      "num_input_tokens_seen": 57796336,
+      "step": 922
+    },
+    {
+      "epoch": 3.0682196339434276,
+      "loss": 0.8780733346939087,
+      "loss_ce": 0.0028292066417634487,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.875,
+      "num_input_tokens_seen": 57796336,
+      "step": 922
+    },
+    {
+      "epoch": 3.071547420965058,
+      "grad_norm": 20.741628646850586,
+      "learning_rate": 5e-06,
+      "loss": 0.8608,
+      "num_input_tokens_seen": 57861404,
+      "step": 923
+    },
+    {
+      "epoch": 3.071547420965058,
+      "loss": 0.9321870803833008,
+      "loss_ce": 0.00030230951961129904,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 57861404,
+      "step": 923
+    },
+    {
+      "epoch": 3.074875207986689,
+      "grad_norm": 10.478264808654785,
+      "learning_rate": 5e-06,
+      "loss": 0.598,
+      "num_input_tokens_seen": 57923204,
+      "step": 924
+    },
+    {
+      "epoch": 3.074875207986689,
+      "loss": 0.46035149693489075,
+      "loss_ce": 8.537084795534611e-05,
+      "loss_iou": 0.09716796875,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.4609375,
+      "num_input_tokens_seen": 57923204,
+      "step": 924
+    },
+    {
+      "epoch": 3.0782029950083194,
+      "grad_norm": 15.388328552246094,
+      "learning_rate": 5e-06,
+      "loss": 1.0425,
+      "num_input_tokens_seen": 57985892,
+      "step": 925
+    },
+    {
+      "epoch": 3.0782029950083194,
+      "loss": 1.1015738248825073,
+      "loss_ce": 0.001964439172297716,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0869140625,
+      "loss_xval": 1.1015625,
+      "num_input_tokens_seen": 57985892,
+      "step": 925
+    },
+    {
+      "epoch": 3.08153078202995,
+      "grad_norm": 10.75679874420166,
+      "learning_rate": 5e-06,
+      "loss": 0.796,
+      "num_input_tokens_seen": 58049620,
+      "step": 926
+    },
+    {
+      "epoch": 3.08153078202995,
+      "loss": 0.7979081273078918,
+      "loss_ce": 5.657908695866354e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 58049620,
+      "step": 926
+    },
+    {
+      "epoch": 3.0848585690515806,
+      "grad_norm": 11.969529151916504,
+      "learning_rate": 5e-06,
+      "loss": 0.577,
+      "num_input_tokens_seen": 58111152,
+      "step": 927
+    },
+    {
+      "epoch": 3.0848585690515806,
+      "loss": 0.7319214344024658,
+      "loss_ce": 0.00010991955059580505,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 58111152,
+      "step": 927
+    },
+    {
+      "epoch": 3.0881863560732112,
+      "grad_norm": 9.87148666381836,
+      "learning_rate": 5e-06,
+      "loss": 0.7442,
+      "num_input_tokens_seen": 58175032,
+      "step": 928
+    },
+    {
+      "epoch": 3.0881863560732112,
+      "loss": 0.6432963013648987,
+      "loss_ce": 0.0014505886938422918,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 58175032,
+      "step": 928
+    },
+    {
+      "epoch": 3.091514143094842,
+      "grad_norm": 52.29267501831055,
+      "learning_rate": 5e-06,
+      "loss": 0.9138,
+      "num_input_tokens_seen": 58237164,
+      "step": 929
+    },
+    {
+      "epoch": 3.091514143094842,
+      "loss": 0.9514411091804504,
+      "loss_ce": 0.0002692709385883063,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 58237164,
+      "step": 929
+    },
+    {
+      "epoch": 3.0948419301164725,
+      "grad_norm": 12.875478744506836,
+      "learning_rate": 5e-06,
+      "loss": 0.9315,
+      "num_input_tokens_seen": 58300844,
+      "step": 930
+    },
+    {
+      "epoch": 3.0948419301164725,
+      "loss": 0.9565946459770203,
+      "loss_ce": 0.0007841241895221174,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.95703125,
+      "num_input_tokens_seen": 58300844,
+      "step": 930
+    },
+    {
+      "epoch": 3.098169717138103,
+      "grad_norm": 22.145496368408203,
+      "learning_rate": 5e-06,
+      "loss": 0.7771,
+      "num_input_tokens_seen": 58365028,
+      "step": 931
+    },
+    {
+      "epoch": 3.098169717138103,
+      "loss": 0.6949498653411865,
+      "loss_ce": 0.00036975587136112154,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 58365028,
+      "step": 931
+    },
+    {
+      "epoch": 3.1014975041597337,
+      "grad_norm": 16.330707550048828,
+      "learning_rate": 5e-06,
+      "loss": 0.7964,
+      "num_input_tokens_seen": 58426292,
+      "step": 932
+    },
+    {
+      "epoch": 3.1014975041597337,
+      "loss": 0.9439092874526978,
+      "loss_ce": 0.0003057535504922271,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0869140625,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 58426292,
+      "step": 932
+    },
+    {
+      "epoch": 3.1048252911813643,
+      "grad_norm": 10.452964782714844,
+      "learning_rate": 5e-06,
+      "loss": 0.7588,
+      "num_input_tokens_seen": 58488240,
+      "step": 933
+    },
+    {
+      "epoch": 3.1048252911813643,
+      "loss": 0.8706070780754089,
+      "loss_ce": 0.001954704290255904,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 58488240,
+      "step": 933
+    },
+    {
+      "epoch": 3.108153078202995,
+      "grad_norm": 18.522306442260742,
+      "learning_rate": 5e-06,
+      "loss": 0.9628,
+      "num_input_tokens_seen": 58551624,
+      "step": 934
+    },
+    {
+      "epoch": 3.108153078202995,
+      "loss": 0.8793235421180725,
+      "loss_ce": 0.0013938324991613626,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 58551624,
+      "step": 934
+    },
+    {
+      "epoch": 3.1114808652246255,
+      "grad_norm": 7.724632740020752,
+      "learning_rate": 5e-06,
+      "loss": 0.778,
+      "num_input_tokens_seen": 58613036,
+      "step": 935
+    },
+    {
+      "epoch": 3.1114808652246255,
+      "loss": 1.0509037971496582,
+      "loss_ce": 0.007446764037013054,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.10400390625,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 58613036,
+      "step": 935
+    },
+    {
+      "epoch": 3.114808652246256,
+      "grad_norm": 12.31716537475586,
+      "learning_rate": 5e-06,
+      "loss": 0.9438,
+      "num_input_tokens_seen": 58675872,
+      "step": 936
+    },
+    {
+      "epoch": 3.114808652246256,
+      "loss": 0.9533208608627319,
+      "loss_ce": 0.0005620683077722788,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 58675872,
+      "step": 936
+    },
+    {
+      "epoch": 3.1181364392678868,
+      "grad_norm": 14.161133766174316,
+      "learning_rate": 5e-06,
+      "loss": 0.8589,
+      "num_input_tokens_seen": 58738884,
+      "step": 937
+    },
+    {
+      "epoch": 3.1181364392678868,
+      "loss": 0.8286876678466797,
+      "loss_ce": 0.0001964616822078824,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 58738884,
+      "step": 937
+    },
+    {
+      "epoch": 3.1214642262895174,
+      "grad_norm": 17.004718780517578,
+      "learning_rate": 5e-06,
+      "loss": 0.7602,
+      "num_input_tokens_seen": 58802072,
+      "step": 938
+    },
+    {
+      "epoch": 3.1214642262895174,
+      "loss": 0.821278989315033,
+      "loss_ce": 0.0004782435134984553,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 58802072,
+      "step": 938
+    },
+    {
+      "epoch": 3.124792013311148,
+      "grad_norm": 12.556039810180664,
+      "learning_rate": 5e-06,
+      "loss": 0.8359,
+      "num_input_tokens_seen": 58864708,
+      "step": 939
+    },
+    {
+      "epoch": 3.124792013311148,
+      "loss": 0.8781052827835083,
+      "loss_ce": 0.0005417764186859131,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 58864708,
+      "step": 939
+    },
+    {
+      "epoch": 3.1281198003327786,
+      "grad_norm": 14.815278053283691,
+      "learning_rate": 5e-06,
+      "loss": 0.8864,
+      "num_input_tokens_seen": 58926916,
+      "step": 940
+    },
+    {
+      "epoch": 3.1281198003327786,
+      "loss": 0.9597880840301514,
+      "loss_ce": 0.0008037795196287334,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.9609375,
+      "num_input_tokens_seen": 58926916,
+      "step": 940
+    },
+    {
+      "epoch": 3.131447587354409,
+      "grad_norm": 17.76370620727539,
+      "learning_rate": 5e-06,
+      "loss": 0.84,
+      "num_input_tokens_seen": 58988668,
+      "step": 941
+    },
+    {
+      "epoch": 3.131447587354409,
+      "loss": 0.88735431432724,
+      "loss_ce": 0.00042194739216938615,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 58988668,
+      "step": 941
+    },
+    {
+      "epoch": 3.13477537437604,
+      "grad_norm": 36.662315368652344,
+      "learning_rate": 5e-06,
+      "loss": 0.8738,
+      "num_input_tokens_seen": 59052520,
+      "step": 942
+    },
+    {
+      "epoch": 3.13477537437604,
+      "loss": 0.8464229106903076,
+      "loss_ce": 0.0009028796339407563,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 59052520,
+      "step": 942
+    },
+    {
+      "epoch": 3.1381031613976704,
+      "grad_norm": 8.20991039276123,
+      "learning_rate": 5e-06,
+      "loss": 0.5881,
+      "num_input_tokens_seen": 59114480,
+      "step": 943
+    },
+    {
+      "epoch": 3.1381031613976704,
+      "loss": 0.5021640658378601,
+      "loss_ce": 8.887949661584571e-05,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 59114480,
+      "step": 943
+    },
+    {
+      "epoch": 3.141430948419301,
+      "grad_norm": 18.209897994995117,
+      "learning_rate": 5e-06,
+      "loss": 1.0336,
+      "num_input_tokens_seen": 59175212,
+      "step": 944
+    },
+    {
+      "epoch": 3.141430948419301,
+      "loss": 1.2060582637786865,
+      "loss_ce": 0.002444992307573557,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.080078125,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 59175212,
+      "step": 944
+    },
+    {
+      "epoch": 3.1447587354409317,
+      "grad_norm": 19.541431427001953,
+      "learning_rate": 5e-06,
+      "loss": 0.9663,
+      "num_input_tokens_seen": 59239460,
+      "step": 945
+    },
+    {
+      "epoch": 3.1447587354409317,
+      "loss": 0.7803661823272705,
+      "loss_ce": 9.27629807847552e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 59239460,
+      "step": 945
+    },
+    {
+      "epoch": 3.1480865224625623,
+      "grad_norm": 10.669411659240723,
+      "learning_rate": 5e-06,
+      "loss": 1.0049,
+      "num_input_tokens_seen": 59301360,
+      "step": 946
+    },
+    {
+      "epoch": 3.1480865224625623,
+      "loss": 1.1545720100402832,
+      "loss_ce": 3.09227907564491e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0888671875,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 59301360,
+      "step": 946
+    },
+    {
+      "epoch": 3.151414309484193,
+      "grad_norm": 12.653473854064941,
+      "learning_rate": 5e-06,
+      "loss": 0.7067,
+      "num_input_tokens_seen": 59365016,
+      "step": 947
+    },
+    {
+      "epoch": 3.151414309484193,
+      "loss": 0.7654366493225098,
+      "loss_ce": 0.0005440429667942226,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 59365016,
+      "step": 947
+    },
+    {
+      "epoch": 3.1547420965058235,
+      "grad_norm": 11.480525016784668,
+      "learning_rate": 5e-06,
+      "loss": 0.6465,
+      "num_input_tokens_seen": 59427212,
+      "step": 948
+    },
+    {
+      "epoch": 3.1547420965058235,
+      "loss": 0.8721858263015747,
+      "loss_ce": 0.0003596378955990076,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 59427212,
+      "step": 948
+    },
+    {
+      "epoch": 3.158069883527454,
+      "grad_norm": 34.165103912353516,
+      "learning_rate": 5e-06,
+      "loss": 0.9143,
+      "num_input_tokens_seen": 59491056,
+      "step": 949
+    },
+    {
+      "epoch": 3.158069883527454,
+      "loss": 0.9038797616958618,
+      "loss_ce": 7.110174919944257e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.90234375,
+      "num_input_tokens_seen": 59491056,
+      "step": 949
+    },
+    {
+      "epoch": 3.1613976705490847,
+      "grad_norm": 24.460308074951172,
+      "learning_rate": 5e-06,
+      "loss": 0.827,
+      "num_input_tokens_seen": 59552212,
+      "step": 950
+    },
+    {
+      "epoch": 3.1613976705490847,
+      "loss": 0.9148125052452087,
+      "loss_ce": 1.7638567442190833e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 59552212,
+      "step": 950
+    },
+    {
+      "epoch": 3.1647254575707153,
+      "grad_norm": 16.602678298950195,
+      "learning_rate": 5e-06,
+      "loss": 0.7661,
+      "num_input_tokens_seen": 59615180,
+      "step": 951
+    },
+    {
+      "epoch": 3.1647254575707153,
+      "loss": 0.9075579047203064,
+      "loss_ce": 0.0015520007582381368,
+      "loss_iou": 0.25,
+      "loss_num": 0.0810546875,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 59615180,
+      "step": 951
+    },
+    {
+      "epoch": 3.168053244592346,
+      "grad_norm": 14.823358535766602,
+      "learning_rate": 5e-06,
+      "loss": 0.69,
+      "num_input_tokens_seen": 59678104,
+      "step": 952
+    },
+    {
+      "epoch": 3.168053244592346,
+      "loss": 0.7148585319519043,
+      "loss_ce": 1.4785388884774875e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 59678104,
+      "step": 952
+    },
+    {
+      "epoch": 3.1713810316139766,
+      "grad_norm": 63.57862091064453,
+      "learning_rate": 5e-06,
+      "loss": 0.7865,
+      "num_input_tokens_seen": 59740976,
+      "step": 953
+    },
+    {
+      "epoch": 3.1713810316139766,
+      "loss": 0.5665551424026489,
+      "loss_ce": 0.001247492036782205,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 59740976,
+      "step": 953
+    },
+    {
+      "epoch": 3.174708818635607,
+      "grad_norm": 34.066715240478516,
+      "learning_rate": 5e-06,
+      "loss": 0.812,
+      "num_input_tokens_seen": 59803580,
+      "step": 954
+    },
+    {
+      "epoch": 3.174708818635607,
+      "loss": 0.7728263139724731,
+      "loss_ce": 0.0002432619803585112,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 59803580,
+      "step": 954
+    },
+    {
+      "epoch": 3.178036605657238,
+      "grad_norm": 31.254560470581055,
+      "learning_rate": 5e-06,
+      "loss": 1.0825,
+      "num_input_tokens_seen": 59865328,
+      "step": 955
+    },
+    {
+      "epoch": 3.178036605657238,
+      "loss": 0.8610126972198486,
+      "loss_ce": 0.00017285677313338965,
+      "loss_iou": 0.25,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 59865328,
+      "step": 955
+    },
+    {
+      "epoch": 3.1813643926788684,
+      "grad_norm": 10.921689987182617,
+      "learning_rate": 5e-06,
+      "loss": 1.0658,
+      "num_input_tokens_seen": 59927552,
+      "step": 956
+    },
+    {
+      "epoch": 3.1813643926788684,
+      "loss": 1.1395654678344727,
+      "loss_ce": 0.0006494110566563904,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.06689453125,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 59927552,
+      "step": 956
+    },
+    {
+      "epoch": 3.184692179700499,
+      "grad_norm": 9.880757331848145,
+      "learning_rate": 5e-06,
+      "loss": 0.8318,
+      "num_input_tokens_seen": 59989664,
+      "step": 957
+    },
+    {
+      "epoch": 3.184692179700499,
+      "loss": 0.9110987186431885,
+      "loss_ce": 0.0009424776071682572,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.91015625,
+      "num_input_tokens_seen": 59989664,
+      "step": 957
+    },
+    {
+      "epoch": 3.1880199667221296,
+      "grad_norm": 9.38766860961914,
+      "learning_rate": 5e-06,
+      "loss": 0.6643,
+      "num_input_tokens_seen": 60052276,
+      "step": 958
+    },
+    {
+      "epoch": 3.1880199667221296,
+      "loss": 0.7620002031326294,
+      "loss_ce": 3.7309764593373984e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.76171875,
+      "num_input_tokens_seen": 60052276,
+      "step": 958
+    },
+    {
+      "epoch": 3.1913477537437602,
+      "grad_norm": 16.43718719482422,
+      "learning_rate": 5e-06,
+      "loss": 0.9924,
+      "num_input_tokens_seen": 60116120,
+      "step": 959
+    },
+    {
+      "epoch": 3.1913477537437602,
+      "loss": 1.1171211004257202,
+      "loss_ce": 0.0013984288088977337,
+      "loss_iou": 0.375,
+      "loss_num": 0.0732421875,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 60116120,
+      "step": 959
+    },
+    {
+      "epoch": 3.194675540765391,
+      "grad_norm": 12.237974166870117,
+      "learning_rate": 5e-06,
+      "loss": 0.8707,
+      "num_input_tokens_seen": 60179548,
+      "step": 960
+    },
+    {
+      "epoch": 3.194675540765391,
+      "loss": 0.8710145950317383,
+      "loss_ce": 0.0006532109691761434,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 60179548,
+      "step": 960
+    },
+    {
+      "epoch": 3.1980033277870215,
+      "grad_norm": 24.791534423828125,
+      "learning_rate": 5e-06,
+      "loss": 0.8278,
+      "num_input_tokens_seen": 60241708,
+      "step": 961
+    },
+    {
+      "epoch": 3.1980033277870215,
+      "loss": 0.7581548690795898,
+      "loss_ce": 0.0022954712621867657,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 60241708,
+      "step": 961
+    },
+    {
+      "epoch": 3.201331114808652,
+      "grad_norm": 28.264705657958984,
+      "learning_rate": 5e-06,
+      "loss": 0.473,
+      "num_input_tokens_seen": 60304508,
+      "step": 962
+    },
+    {
+      "epoch": 3.201331114808652,
+      "loss": 0.5443636178970337,
+      "loss_ce": 0.0008455467177554965,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 60304508,
+      "step": 962
+    },
+    {
+      "epoch": 3.2046589018302827,
+      "grad_norm": 10.311064720153809,
+      "learning_rate": 5e-06,
+      "loss": 0.6768,
+      "num_input_tokens_seen": 60367092,
+      "step": 963
+    },
+    {
+      "epoch": 3.2046589018302827,
+      "loss": 0.8569867610931396,
+      "loss_ce": 5.31335172127001e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 60367092,
+      "step": 963
+    },
+    {
+      "epoch": 3.2079866888519133,
+      "grad_norm": 10.463589668273926,
+      "learning_rate": 5e-06,
+      "loss": 0.838,
+      "num_input_tokens_seen": 60429284,
+      "step": 964
+    },
+    {
+      "epoch": 3.2079866888519133,
+      "loss": 0.6484445333480835,
+      "loss_ce": 0.0007394892745651305,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 60429284,
+      "step": 964
+    },
+    {
+      "epoch": 3.211314475873544,
+      "grad_norm": 7.585339546203613,
+      "learning_rate": 5e-06,
+      "loss": 0.5347,
+      "num_input_tokens_seen": 60491188,
+      "step": 965
+    },
+    {
+      "epoch": 3.211314475873544,
+      "loss": 0.5105918049812317,
+      "loss_ce": 0.0004599463427439332,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 60491188,
+      "step": 965
+    },
+    {
+      "epoch": 3.2146422628951745,
+      "grad_norm": 9.942543029785156,
+      "learning_rate": 5e-06,
+      "loss": 0.62,
+      "num_input_tokens_seen": 60553992,
+      "step": 966
+    },
+    {
+      "epoch": 3.2146422628951745,
+      "loss": 0.46438688039779663,
+      "loss_ce": 0.00039763684617355466,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.46484375,
+      "num_input_tokens_seen": 60553992,
+      "step": 966
+    },
+    {
+      "epoch": 3.217970049916805,
+      "grad_norm": 11.350386619567871,
+      "learning_rate": 5e-06,
+      "loss": 0.887,
+      "num_input_tokens_seen": 60615900,
+      "step": 967
+    },
+    {
+      "epoch": 3.217970049916805,
+      "loss": 0.9106756448745728,
+      "loss_ce": 3.116061998298392e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.91015625,
+      "num_input_tokens_seen": 60615900,
+      "step": 967
+    },
+    {
+      "epoch": 3.2212978369384357,
+      "grad_norm": 11.828593254089355,
+      "learning_rate": 5e-06,
+      "loss": 0.9105,
+      "num_input_tokens_seen": 60678052,
+      "step": 968
+    },
+    {
+      "epoch": 3.2212978369384357,
+      "loss": 0.7735366225242615,
+      "loss_ce": 9.917082934407517e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 60678052,
+      "step": 968
+    },
+    {
+      "epoch": 3.2246256239600664,
+      "grad_norm": 6.765110969543457,
+      "learning_rate": 5e-06,
+      "loss": 0.7719,
+      "num_input_tokens_seen": 60740316,
+      "step": 969
+    },
+    {
+      "epoch": 3.2246256239600664,
+      "loss": 1.1397228240966797,
+      "loss_ce": 0.0015393083449453115,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.08349609375,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 60740316,
+      "step": 969
+    },
+    {
+      "epoch": 3.227953410981697,
+      "grad_norm": 14.274063110351562,
+      "learning_rate": 5e-06,
+      "loss": 0.7065,
+      "num_input_tokens_seen": 60801452,
+      "step": 970
+    },
+    {
+      "epoch": 3.227953410981697,
+      "loss": 0.7572785019874573,
+      "loss_ce": 0.00019845366477966309,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 60801452,
+      "step": 970
+    },
+    {
+      "epoch": 3.2312811980033276,
+      "grad_norm": 7.189883232116699,
+      "learning_rate": 5e-06,
+      "loss": 0.8128,
+      "num_input_tokens_seen": 60863756,
+      "step": 971
+    },
+    {
+      "epoch": 3.2312811980033276,
+      "loss": 0.9057375192642212,
+      "loss_ce": 0.00021990106324665248,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 60863756,
+      "step": 971
+    },
+    {
+      "epoch": 3.234608985024958,
+      "grad_norm": 20.947919845581055,
+      "learning_rate": 5e-06,
+      "loss": 1.029,
+      "num_input_tokens_seen": 60926592,
+      "step": 972
+    },
+    {
+      "epoch": 3.234608985024958,
+      "loss": 0.9959653615951538,
+      "loss_ce": 0.0006040430162101984,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.99609375,
+      "num_input_tokens_seen": 60926592,
+      "step": 972
+    },
+    {
+      "epoch": 3.237936772046589,
+      "grad_norm": 14.039640426635742,
+      "learning_rate": 5e-06,
+      "loss": 0.8004,
+      "num_input_tokens_seen": 60990536,
+      "step": 973
+    },
+    {
+      "epoch": 3.237936772046589,
+      "loss": 0.8818901777267456,
+      "loss_ce": 5.422691538115032e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 60990536,
+      "step": 973
+    },
+    {
+      "epoch": 3.2412645590682194,
+      "grad_norm": 12.025979042053223,
+      "learning_rate": 5e-06,
+      "loss": 0.6277,
+      "num_input_tokens_seen": 61053216,
+      "step": 974
+    },
+    {
+      "epoch": 3.2412645590682194,
+      "loss": 0.6128069758415222,
+      "loss_ce": 0.0011126084718853235,
+      "loss_iou": 0.21875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 61053216,
+      "step": 974
+    },
+    {
+      "epoch": 3.24459234608985,
+      "grad_norm": 13.156522750854492,
+      "learning_rate": 5e-06,
+      "loss": 0.7027,
+      "num_input_tokens_seen": 61116740,
+      "step": 975
+    },
+    {
+      "epoch": 3.24459234608985,
+      "loss": 0.7115057706832886,
+      "loss_ce": 0.0006903592147864401,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 61116740,
+      "step": 975
+    },
+    {
+      "epoch": 3.2479201331114806,
+      "grad_norm": 7.883967876434326,
+      "learning_rate": 5e-06,
+      "loss": 0.5578,
+      "num_input_tokens_seen": 61178912,
+      "step": 976
+    },
+    {
+      "epoch": 3.2479201331114806,
+      "loss": 0.7233986258506775,
+      "loss_ce": 0.0003761877305805683,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 61178912,
+      "step": 976
+    },
+    {
+      "epoch": 3.2512479201331113,
+      "grad_norm": 18.929187774658203,
+      "learning_rate": 5e-06,
+      "loss": 0.6357,
+      "num_input_tokens_seen": 61240704,
+      "step": 977
+    },
+    {
+      "epoch": 3.2512479201331113,
+      "loss": 0.6345537900924683,
+      "loss_ce": 3.23405911331065e-05,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 61240704,
+      "step": 977
+    },
+    {
+      "epoch": 3.254575707154742,
+      "grad_norm": 11.569648742675781,
+      "learning_rate": 5e-06,
+      "loss": 0.9763,
+      "num_input_tokens_seen": 61303340,
+      "step": 978
+    },
+    {
+      "epoch": 3.254575707154742,
+      "loss": 0.850032389163971,
+      "loss_ce": 0.00042301719076931477,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.8515625,
+      "num_input_tokens_seen": 61303340,
+      "step": 978
+    },
+    {
+      "epoch": 3.2579034941763725,
+      "grad_norm": 18.81865692138672,
+      "learning_rate": 5e-06,
+      "loss": 0.7637,
+      "num_input_tokens_seen": 61366660,
+      "step": 979
+    },
+    {
+      "epoch": 3.2579034941763725,
+      "loss": 0.6620635986328125,
+      "loss_ce": 0.0005646056379191577,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 61366660,
+      "step": 979
+    },
+    {
+      "epoch": 3.261231281198003,
+      "grad_norm": 21.63203239440918,
+      "learning_rate": 5e-06,
+      "loss": 0.6904,
+      "num_input_tokens_seen": 61429692,
+      "step": 980
+    },
+    {
+      "epoch": 3.261231281198003,
+      "loss": 0.768730103969574,
+      "loss_ce": 0.000541669491212815,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 61429692,
+      "step": 980
+    },
+    {
+      "epoch": 3.2645590682196337,
+      "grad_norm": 26.75554656982422,
+      "learning_rate": 5e-06,
+      "loss": 0.9439,
+      "num_input_tokens_seen": 61493484,
+      "step": 981
+    },
+    {
+      "epoch": 3.2645590682196337,
+      "loss": 0.9156728982925415,
+      "loss_ce": 2.3398324628942646e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 61493484,
+      "step": 981
+    },
+    {
+      "epoch": 3.2678868552412643,
+      "grad_norm": 34.384639739990234,
+      "learning_rate": 5e-06,
+      "loss": 1.0029,
+      "num_input_tokens_seen": 61556876,
+      "step": 982
+    },
+    {
+      "epoch": 3.2678868552412643,
+      "loss": 1.1158475875854492,
+      "loss_ce": 0.00110146077349782,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.08349609375,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 61556876,
+      "step": 982
+    },
+    {
+      "epoch": 3.271214642262895,
+      "grad_norm": 19.943857192993164,
+      "learning_rate": 5e-06,
+      "loss": 0.9756,
+      "num_input_tokens_seen": 61620480,
+      "step": 983
+    },
+    {
+      "epoch": 3.271214642262895,
+      "loss": 1.0187878608703613,
+      "loss_ce": 0.0007214570650830865,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.04638671875,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 61620480,
+      "step": 983
+    },
+    {
+      "epoch": 3.2745424292845255,
+      "grad_norm": 17.966175079345703,
+      "learning_rate": 5e-06,
+      "loss": 0.8658,
+      "num_input_tokens_seen": 61684128,
+      "step": 984
+    },
+    {
+      "epoch": 3.2745424292845255,
+      "loss": 0.9176982641220093,
+      "loss_ce": 0.0013164564734324813,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 61684128,
+      "step": 984
+    },
+    {
+      "epoch": 3.277870216306156,
+      "grad_norm": 20.2591552734375,
+      "learning_rate": 5e-06,
+      "loss": 0.9037,
+      "num_input_tokens_seen": 61747004,
+      "step": 985
+    },
+    {
+      "epoch": 3.277870216306156,
+      "loss": 1.0413687229156494,
+      "loss_ce": 0.0015739183872938156,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 61747004,
+      "step": 985
+    },
+    {
+      "epoch": 3.2811980033277868,
+      "grad_norm": 17.5063419342041,
+      "learning_rate": 5e-06,
+      "loss": 0.5624,
+      "num_input_tokens_seen": 61807736,
+      "step": 986
+    },
+    {
+      "epoch": 3.2811980033277868,
+      "loss": 0.4131319522857666,
+      "loss_ce": 4.6039051085244864e-05,
+      "loss_iou": 0.10400390625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.4140625,
+      "num_input_tokens_seen": 61807736,
+      "step": 986
+    },
+    {
+      "epoch": 3.284525790349418,
+      "grad_norm": 21.377288818359375,
+      "learning_rate": 5e-06,
+      "loss": 0.8481,
+      "num_input_tokens_seen": 61870936,
+      "step": 987
+    },
+    {
+      "epoch": 3.284525790349418,
+      "loss": 0.8774704337120056,
+      "loss_ce": 2.90050265903119e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 61870936,
+      "step": 987
+    },
+    {
+      "epoch": 3.2878535773710484,
+      "grad_norm": 21.52035903930664,
+      "learning_rate": 5e-06,
+      "loss": 0.5701,
+      "num_input_tokens_seen": 61932556,
+      "step": 988
+    },
+    {
+      "epoch": 3.2878535773710484,
+      "loss": 0.8120724558830261,
+      "loss_ce": 0.00030482906731776893,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 61932556,
+      "step": 988
+    },
+    {
+      "epoch": 3.291181364392679,
+      "grad_norm": 18.4642276763916,
+      "learning_rate": 5e-06,
+      "loss": 0.9416,
+      "num_input_tokens_seen": 61996240,
+      "step": 989
+    },
+    {
+      "epoch": 3.291181364392679,
+      "loss": 0.9976834058761597,
+      "loss_ce": 0.00036900522536598146,
+      "loss_iou": 0.375,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.99609375,
+      "num_input_tokens_seen": 61996240,
+      "step": 989
+    },
+    {
+      "epoch": 3.2945091514143097,
+      "grad_norm": 12.169829368591309,
+      "learning_rate": 5e-06,
+      "loss": 0.9705,
+      "num_input_tokens_seen": 62058608,
+      "step": 990
+    },
+    {
+      "epoch": 3.2945091514143097,
+      "loss": 1.1562477350234985,
+      "loss_ce": 0.0007300969446077943,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.07275390625,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 62058608,
+      "step": 990
+    },
+    {
+      "epoch": 3.2978369384359403,
+      "grad_norm": 18.226551055908203,
+      "learning_rate": 5e-06,
+      "loss": 0.8279,
+      "num_input_tokens_seen": 62121952,
+      "step": 991
+    },
+    {
+      "epoch": 3.2978369384359403,
+      "loss": 0.7579995393753052,
+      "loss_ce": 6.495913839899004e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 62121952,
+      "step": 991
+    },
+    {
+      "epoch": 3.301164725457571,
+      "grad_norm": 5.881676197052002,
+      "learning_rate": 5e-06,
+      "loss": 0.4544,
+      "num_input_tokens_seen": 62182960,
+      "step": 992
+    },
+    {
+      "epoch": 3.301164725457571,
+      "loss": 0.5944963693618774,
+      "loss_ce": 1.391188743582461e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 62182960,
+      "step": 992
+    },
+    {
+      "epoch": 3.3044925124792015,
+      "grad_norm": 23.40874671936035,
+      "learning_rate": 5e-06,
+      "loss": 1.0811,
+      "num_input_tokens_seen": 62245376,
+      "step": 993
+    },
+    {
+      "epoch": 3.3044925124792015,
+      "loss": 1.324781894683838,
+      "loss_ce": 0.0012955316342413425,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.08740234375,
+      "loss_xval": 1.3203125,
+      "num_input_tokens_seen": 62245376,
+      "step": 993
+    },
+    {
+      "epoch": 3.307820299500832,
+      "grad_norm": 16.891769409179688,
+      "learning_rate": 5e-06,
+      "loss": 0.8996,
+      "num_input_tokens_seen": 62308924,
+      "step": 994
+    },
+    {
+      "epoch": 3.307820299500832,
+      "loss": 1.0149037837982178,
+      "loss_ce": 0.001720264321193099,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.050048828125,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 62308924,
+      "step": 994
+    },
+    {
+      "epoch": 3.3111480865224627,
+      "grad_norm": 8.293322563171387,
+      "learning_rate": 5e-06,
+      "loss": 0.8275,
+      "num_input_tokens_seen": 62368672,
+      "step": 995
+    },
+    {
+      "epoch": 3.3111480865224627,
+      "loss": 0.8809036016464233,
+      "loss_ce": 4.425490624271333e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.08056640625,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 62368672,
+      "step": 995
+    },
+    {
+      "epoch": 3.3144758735440933,
+      "grad_norm": 16.752803802490234,
+      "learning_rate": 5e-06,
+      "loss": 0.7262,
+      "num_input_tokens_seen": 62431712,
+      "step": 996
+    },
+    {
+      "epoch": 3.3144758735440933,
+      "loss": 0.7768771052360535,
+      "loss_ce": 2.1639707483700477e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 62431712,
+      "step": 996
+    },
+    {
+      "epoch": 3.317803660565724,
+      "grad_norm": 15.595096588134766,
+      "learning_rate": 5e-06,
+      "loss": 0.7665,
+      "num_input_tokens_seen": 62493536,
+      "step": 997
+    },
+    {
+      "epoch": 3.317803660565724,
+      "loss": 0.8226642608642578,
+      "loss_ce": 0.00015452262596227229,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 62493536,
+      "step": 997
+    },
+    {
+      "epoch": 3.3211314475873546,
+      "grad_norm": 24.1756649017334,
+      "learning_rate": 5e-06,
+      "loss": 0.785,
+      "num_input_tokens_seen": 62557440,
+      "step": 998
+    },
+    {
+      "epoch": 3.3211314475873546,
+      "loss": 0.7211722731590271,
+      "loss_ce": 0.0009574384312145412,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 62557440,
+      "step": 998
+    },
+    {
+      "epoch": 3.324459234608985,
+      "grad_norm": 8.496849060058594,
+      "learning_rate": 5e-06,
+      "loss": 0.8508,
+      "num_input_tokens_seen": 62620908,
+      "step": 999
+    },
+    {
+      "epoch": 3.324459234608985,
+      "loss": 0.9880321025848389,
+      "loss_ce": 0.0017039903905242682,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 62620908,
+      "step": 999
+    },
+    {
+      "epoch": 3.327787021630616,
+      "grad_norm": 13.95791244506836,
+      "learning_rate": 5e-06,
+      "loss": 0.8183,
+      "num_input_tokens_seen": 62684036,
+      "step": 1000
+    },
+    {
+      "epoch": 3.327787021630616,
+      "eval_seeclick_CIoU": 0.11919866502285004,
+      "eval_seeclick_GIoU": 0.13886269554495811,
+      "eval_seeclick_IoU": 0.22386732697486877,
+      "eval_seeclick_MAE_all": 0.18997914344072342,
+      "eval_seeclick_MAE_h": 0.04969533532857895,
+      "eval_seeclick_MAE_w": 0.14520900696516037,
+      "eval_seeclick_MAE_x_boxes": 0.2827809154987335,
+      "eval_seeclick_MAE_y_boxes": 0.14949627220630646,
+      "eval_seeclick_NUM_probability": 0.9996586740016937,
+      "eval_seeclick_inside_bbox": 0.32083334028720856,
+      "eval_seeclick_loss": 2.743774175643921,
+      "eval_seeclick_loss_ce": 0.07989468798041344,
+      "eval_seeclick_loss_iou": 0.8560791015625,
+      "eval_seeclick_loss_num": 0.1868896484375,
+      "eval_seeclick_loss_xval": 2.6455078125,
+      "eval_seeclick_runtime": 71.1388,
+      "eval_seeclick_samples_per_second": 0.661,
+      "eval_seeclick_steps_per_second": 0.028,
+      "num_input_tokens_seen": 62684036,
+      "step": 1000
+    },
+    {
+      "epoch": 3.327787021630616,
+      "eval_icons_CIoU": 0.05206053704023361,
+      "eval_icons_GIoU": 0.16786546260118484,
+      "eval_icons_IoU": 0.20200002193450928,
+      "eval_icons_MAE_all": 0.14681052416563034,
+      "eval_icons_MAE_h": 0.08216147124767303,
+      "eval_icons_MAE_w": 0.1477516144514084,
+      "eval_icons_MAE_x_boxes": 0.1450214460492134,
+      "eval_icons_MAE_y_boxes": 0.04793294984847307,
+      "eval_icons_NUM_probability": 0.999993085861206,
+      "eval_icons_inside_bbox": 0.3888888955116272,
+      "eval_icons_loss": 2.3937511444091797,
+      "eval_icons_loss_ce": 1.3247173171748727e-06,
+      "eval_icons_loss_iou": 0.817138671875,
+      "eval_icons_loss_num": 0.1392059326171875,
+      "eval_icons_loss_xval": 2.33203125,
+      "eval_icons_runtime": 73.4327,
+      "eval_icons_samples_per_second": 0.681,
+      "eval_icons_steps_per_second": 0.027,
+      "num_input_tokens_seen": 62684036,
+      "step": 1000
+    },
+    {
+      "epoch": 3.327787021630616,
+      "eval_screenspot_CIoU": 0.05590761390825113,
+      "eval_screenspot_GIoU": 0.1182707001765569,
+      "eval_screenspot_IoU": 0.20136764148871103,
+      "eval_screenspot_MAE_all": 0.19364242255687714,
+      "eval_screenspot_MAE_h": 0.06537577758232753,
+      "eval_screenspot_MAE_w": 0.1785052220026652,
+      "eval_screenspot_MAE_x_boxes": 0.2493977944056193,
+      "eval_screenspot_MAE_y_boxes": 0.13203182319800058,
+      "eval_screenspot_NUM_probability": 0.9998934666315714,
+      "eval_screenspot_inside_bbox": 0.3787499964237213,
+      "eval_screenspot_loss": 2.7400693893432617,
+      "eval_screenspot_loss_ce": 0.00023837910460618636,
+      "eval_screenspot_loss_iou": 0.8929036458333334,
+      "eval_screenspot_loss_num": 0.197662353515625,
+      "eval_screenspot_loss_xval": 2.7724609375,
+      "eval_screenspot_runtime": 118.7252,
+      "eval_screenspot_samples_per_second": 0.75,
+      "eval_screenspot_steps_per_second": 0.025,
+      "num_input_tokens_seen": 62684036,
+      "step": 1000
+    },
+    {
+      "epoch": 3.327787021630616,
+      "eval_compot_CIoU": -0.032527330331504345,
+      "eval_compot_GIoU": 0.07551112771034241,
+      "eval_compot_IoU": 0.1327061653137207,
+      "eval_compot_MAE_all": 0.22042576223611832,
+      "eval_compot_MAE_h": 0.06776861473917961,
+      "eval_compot_MAE_w": 0.24602457880973816,
+      "eval_compot_MAE_x_boxes": 0.19308777898550034,
+      "eval_compot_MAE_y_boxes": 0.1478528529405594,
+      "eval_compot_NUM_probability": 0.9999693036079407,
+      "eval_compot_inside_bbox": 0.2395833358168602,
+      "eval_compot_loss": 2.906449556350708,
+      "eval_compot_loss_ce": 0.005752389319241047,
+      "eval_compot_loss_iou": 0.91650390625,
+      "eval_compot_loss_num": 0.211029052734375,
+      "eval_compot_loss_xval": 2.8876953125,
+      "eval_compot_runtime": 67.807,
+      "eval_compot_samples_per_second": 0.737,
+      "eval_compot_steps_per_second": 0.029,
+      "num_input_tokens_seen": 62684036,
+      "step": 1000
+    },
+    {
+      "epoch": 3.327787021630616,
+      "eval_custom_ui_MAE_all": 0.08277507498860359,
+      "eval_custom_ui_MAE_x": 0.07837143167853355,
+      "eval_custom_ui_MAE_y": 0.08717871829867363,
+      "eval_custom_ui_NUM_probability": 0.9999922215938568,
+      "eval_custom_ui_loss": 0.4004429578781128,
+      "eval_custom_ui_loss_ce": 4.6469912376778666e-05,
+      "eval_custom_ui_loss_num": 0.0777587890625,
+      "eval_custom_ui_loss_xval": 0.3887939453125,
+      "eval_custom_ui_runtime": 50.8348,
+      "eval_custom_ui_samples_per_second": 0.984,
+      "eval_custom_ui_steps_per_second": 0.039,
+      "num_input_tokens_seen": 62684036,
+      "step": 1000
+    },
+    {
+      "epoch": 3.327787021630616,
+      "loss": 0.3899743854999542,
+      "loss_ce": 8.180466102203354e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.078125,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 62684036,
+      "step": 1000
+    },
+    {
+      "epoch": 3.3311148086522464,
+      "grad_norm": 11.472808837890625,
+      "learning_rate": 5e-06,
+      "loss": 0.683,
+      "num_input_tokens_seen": 62745984,
+      "step": 1001
+    },
+    {
+      "epoch": 3.3311148086522464,
+      "loss": 0.5718774199485779,
+      "loss_ce": 8.515124136465602e-06,
+      "loss_iou": 0.12451171875,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 62745984,
+      "step": 1001
+    },
+    {
+      "epoch": 3.334442595673877,
+      "grad_norm": 24.495885848999023,
+      "learning_rate": 5e-06,
+      "loss": 0.8243,
+      "num_input_tokens_seen": 62810288,
+      "step": 1002
+    },
+    {
+      "epoch": 3.334442595673877,
+      "loss": 0.7644026279449463,
+      "loss_ce": 0.00012040699948556721,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 62810288,
+      "step": 1002
+    },
+    {
+      "epoch": 3.3377703826955076,
+      "grad_norm": 31.466510772705078,
+      "learning_rate": 5e-06,
+      "loss": 0.9726,
+      "num_input_tokens_seen": 62873968,
+      "step": 1003
+    },
+    {
+      "epoch": 3.3377703826955076,
+      "loss": 0.8061179518699646,
+      "loss_ce": 0.00045388404396362603,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 62873968,
+      "step": 1003
+    },
+    {
+      "epoch": 3.3410981697171382,
+      "grad_norm": 17.76407241821289,
+      "learning_rate": 5e-06,
+      "loss": 0.83,
+      "num_input_tokens_seen": 62935872,
+      "step": 1004
+    },
+    {
+      "epoch": 3.3410981697171382,
+      "loss": 0.5958296656608582,
+      "loss_ce": 0.001591385342180729,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 62935872,
+      "step": 1004
+    },
+    {
+      "epoch": 3.344425956738769,
+      "grad_norm": 19.567127227783203,
+      "learning_rate": 5e-06,
+      "loss": 0.8515,
+      "num_input_tokens_seen": 62996516,
+      "step": 1005
+    },
+    {
+      "epoch": 3.344425956738769,
+      "loss": 0.9327386617660522,
+      "loss_ce": 0.00036568206269294024,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 62996516,
+      "step": 1005
+    },
+    {
+      "epoch": 3.3477537437603995,
+      "grad_norm": 28.632431030273438,
+      "learning_rate": 5e-06,
+      "loss": 0.9322,
+      "num_input_tokens_seen": 63059556,
+      "step": 1006
+    },
+    {
+      "epoch": 3.3477537437603995,
+      "loss": 0.8702507615089417,
+      "loss_ce": 0.000133581503177993,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 63059556,
+      "step": 1006
+    },
+    {
+      "epoch": 3.35108153078203,
+      "grad_norm": 13.562311172485352,
+      "learning_rate": 5e-06,
+      "loss": 0.6644,
+      "num_input_tokens_seen": 63122292,
+      "step": 1007
+    },
+    {
+      "epoch": 3.35108153078203,
+      "loss": 0.6380650401115417,
+      "loss_ce": 0.00018663291120901704,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 63122292,
+      "step": 1007
+    },
+    {
+      "epoch": 3.3544093178036607,
+      "grad_norm": 12.887714385986328,
+      "learning_rate": 5e-06,
+      "loss": 0.5642,
+      "num_input_tokens_seen": 63185440,
+      "step": 1008
+    },
+    {
+      "epoch": 3.3544093178036607,
+      "loss": 0.5903370380401611,
+      "loss_ce": 0.0017139973351731896,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 63185440,
+      "step": 1008
+    },
+    {
+      "epoch": 3.3577371048252913,
+      "grad_norm": 23.336519241333008,
+      "learning_rate": 5e-06,
+      "loss": 0.8652,
+      "num_input_tokens_seen": 63248932,
+      "step": 1009
+    },
+    {
+      "epoch": 3.3577371048252913,
+      "loss": 0.8400235176086426,
+      "loss_ce": 0.00042392255272716284,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.83984375,
+      "num_input_tokens_seen": 63248932,
+      "step": 1009
+    },
+    {
+      "epoch": 3.361064891846922,
+      "grad_norm": 9.43217945098877,
+      "learning_rate": 5e-06,
+      "loss": 0.7828,
+      "num_input_tokens_seen": 63311068,
+      "step": 1010
+    },
+    {
+      "epoch": 3.361064891846922,
+      "loss": 0.7335468530654907,
+      "loss_ce": 0.00039258040487766266,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 63311068,
+      "step": 1010
+    },
+    {
+      "epoch": 3.3643926788685525,
+      "grad_norm": 14.894099235534668,
+      "learning_rate": 5e-06,
+      "loss": 0.8479,
+      "num_input_tokens_seen": 63373624,
+      "step": 1011
+    },
+    {
+      "epoch": 3.3643926788685525,
+      "loss": 0.9643365144729614,
+      "loss_ce": 0.0007134462357498705,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.96484375,
+      "num_input_tokens_seen": 63373624,
+      "step": 1011
+    },
+    {
+      "epoch": 3.367720465890183,
+      "grad_norm": 21.609052658081055,
+      "learning_rate": 5e-06,
+      "loss": 0.6569,
+      "num_input_tokens_seen": 63435304,
+      "step": 1012
+    },
+    {
+      "epoch": 3.367720465890183,
+      "loss": 0.5942494869232178,
+      "loss_ce": 1.1208974683540873e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 63435304,
+      "step": 1012
+    },
+    {
+      "epoch": 3.3710482529118138,
+      "grad_norm": 17.91049575805664,
+      "learning_rate": 5e-06,
+      "loss": 0.8687,
+      "num_input_tokens_seen": 63496096,
+      "step": 1013
+    },
+    {
+      "epoch": 3.3710482529118138,
+      "loss": 0.6522290706634521,
+      "loss_ce": 0.00025154344621114433,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 63496096,
+      "step": 1013
+    },
+    {
+      "epoch": 3.3743760399334444,
+      "grad_norm": 4.573009490966797,
+      "learning_rate": 5e-06,
+      "loss": 0.5975,
+      "num_input_tokens_seen": 63557060,
+      "step": 1014
+    },
+    {
+      "epoch": 3.3743760399334444,
+      "loss": 0.42771950364112854,
+      "loss_ce": 0.0002292850404046476,
+      "loss_iou": 0.01611328125,
+      "loss_num": 0.0791015625,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 63557060,
+      "step": 1014
+    },
+    {
+      "epoch": 3.377703826955075,
+      "grad_norm": 13.058093070983887,
+      "learning_rate": 5e-06,
+      "loss": 0.6873,
+      "num_input_tokens_seen": 63619988,
+      "step": 1015
+    },
+    {
+      "epoch": 3.377703826955075,
+      "loss": 0.540930986404419,
+      "loss_ce": 3.743675188161433e-05,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 63619988,
+      "step": 1015
+    },
+    {
+      "epoch": 3.3810316139767056,
+      "grad_norm": 10.773247718811035,
+      "learning_rate": 5e-06,
+      "loss": 0.9057,
+      "num_input_tokens_seen": 63683036,
+      "step": 1016
+    },
+    {
+      "epoch": 3.3810316139767056,
+      "loss": 0.6826450228691101,
+      "loss_ce": 2.7821590265375562e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.68359375,
+      "num_input_tokens_seen": 63683036,
+      "step": 1016
+    },
+    {
+      "epoch": 3.384359400998336,
+      "grad_norm": 11.654807090759277,
+      "learning_rate": 5e-06,
+      "loss": 0.6626,
+      "num_input_tokens_seen": 63745488,
+      "step": 1017
+    },
+    {
+      "epoch": 3.384359400998336,
+      "loss": 0.5884538888931274,
+      "loss_ce": 7.501640357077122e-05,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 63745488,
+      "step": 1017
+    },
+    {
+      "epoch": 3.387687188019967,
+      "grad_norm": 32.220428466796875,
+      "learning_rate": 5e-06,
+      "loss": 0.7153,
+      "num_input_tokens_seen": 63808984,
+      "step": 1018
+    },
+    {
+      "epoch": 3.387687188019967,
+      "loss": 0.7895395755767822,
+      "loss_ce": 0.0002329161943634972,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 63808984,
+      "step": 1018
+    },
+    {
+      "epoch": 3.3910149750415974,
+      "grad_norm": 9.35084056854248,
+      "learning_rate": 5e-06,
+      "loss": 0.7474,
+      "num_input_tokens_seen": 63869352,
+      "step": 1019
+    },
+    {
+      "epoch": 3.3910149750415974,
+      "loss": 0.9051174521446228,
+      "loss_ce": 8.812104351818562e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 63869352,
+      "step": 1019
+    },
+    {
+      "epoch": 3.394342762063228,
+      "grad_norm": 21.432973861694336,
+      "learning_rate": 5e-06,
+      "loss": 0.7779,
+      "num_input_tokens_seen": 63931168,
+      "step": 1020
+    },
+    {
+      "epoch": 3.394342762063228,
+      "loss": 0.8875124454498291,
+      "loss_ce": 0.00018336158245801926,
+      "loss_iou": 0.328125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 63931168,
+      "step": 1020
+    },
+    {
+      "epoch": 3.3976705490848587,
+      "grad_norm": 16.27943992614746,
+      "learning_rate": 5e-06,
+      "loss": 0.8919,
+      "num_input_tokens_seen": 63994276,
+      "step": 1021
+    },
+    {
+      "epoch": 3.3976705490848587,
+      "loss": 0.7104352116584778,
+      "loss_ce": 0.0004742466553580016,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 63994276,
+      "step": 1021
+    },
+    {
+      "epoch": 3.4009983361064893,
+      "grad_norm": 18.430397033691406,
+      "learning_rate": 5e-06,
+      "loss": 0.8877,
+      "num_input_tokens_seen": 64056064,
+      "step": 1022
+    },
+    {
+      "epoch": 3.4009983361064893,
+      "loss": 0.7179223299026489,
+      "loss_ce": 0.0005151316290721297,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 64056064,
+      "step": 1022
+    },
+    {
+      "epoch": 3.40432612312812,
+      "grad_norm": 10.066780090332031,
+      "learning_rate": 5e-06,
+      "loss": 1.0067,
+      "num_input_tokens_seen": 64119668,
+      "step": 1023
+    },
+    {
+      "epoch": 3.40432612312812,
+      "loss": 0.8341488838195801,
+      "loss_ce": 4.246793105266988e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 64119668,
+      "step": 1023
+    },
+    {
+      "epoch": 3.4076539101497505,
+      "grad_norm": 7.158484935760498,
+      "learning_rate": 5e-06,
+      "loss": 0.9016,
+      "num_input_tokens_seen": 64183132,
+      "step": 1024
+    },
+    {
+      "epoch": 3.4076539101497505,
+      "loss": 0.7723921537399292,
+      "loss_ce": 0.0006635930621996522,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 64183132,
+      "step": 1024
+    },
+    {
+      "epoch": 3.410981697171381,
+      "grad_norm": 23.170434951782227,
+      "learning_rate": 5e-06,
+      "loss": 0.985,
+      "num_input_tokens_seen": 64244404,
+      "step": 1025
+    },
+    {
+      "epoch": 3.410981697171381,
+      "loss": 0.9742788076400757,
+      "loss_ce": 3.5607161407824606e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.97265625,
+      "num_input_tokens_seen": 64244404,
+      "step": 1025
+    },
+    {
+      "epoch": 3.4143094841930117,
+      "grad_norm": 15.608352661132812,
+      "learning_rate": 5e-06,
+      "loss": 0.764,
+      "num_input_tokens_seen": 64307508,
+      "step": 1026
+    },
+    {
+      "epoch": 3.4143094841930117,
+      "loss": 0.7705492973327637,
+      "loss_ce": 0.00016356556443497539,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 64307508,
+      "step": 1026
+    },
+    {
+      "epoch": 3.4176372712146423,
+      "grad_norm": 22.033626556396484,
+      "learning_rate": 5e-06,
+      "loss": 0.724,
+      "num_input_tokens_seen": 64369472,
+      "step": 1027
+    },
+    {
+      "epoch": 3.4176372712146423,
+      "loss": 0.7757673859596252,
+      "loss_ce": 0.0007429937249980867,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 64369472,
+      "step": 1027
+    },
+    {
+      "epoch": 3.420965058236273,
+      "grad_norm": 33.458152770996094,
+      "learning_rate": 5e-06,
+      "loss": 0.9622,
+      "num_input_tokens_seen": 64431116,
+      "step": 1028
+    },
+    {
+      "epoch": 3.420965058236273,
+      "loss": 1.2742621898651123,
+      "loss_ce": 0.0007026524399407208,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.09033203125,
+      "loss_xval": 1.2734375,
+      "num_input_tokens_seen": 64431116,
+      "step": 1028
+    },
+    {
+      "epoch": 3.4242928452579036,
+      "grad_norm": 10.938355445861816,
+      "learning_rate": 5e-06,
+      "loss": 0.9922,
+      "num_input_tokens_seen": 64491820,
+      "step": 1029
+    },
+    {
+      "epoch": 3.4242928452579036,
+      "loss": 1.0337855815887451,
+      "loss_ce": 0.00033826506114564836,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.08984375,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 64491820,
+      "step": 1029
+    },
+    {
+      "epoch": 3.427620632279534,
+      "grad_norm": 14.789898872375488,
+      "learning_rate": 5e-06,
+      "loss": 0.8743,
+      "num_input_tokens_seen": 64555604,
+      "step": 1030
+    },
+    {
+      "epoch": 3.427620632279534,
+      "loss": 0.8684331178665161,
+      "loss_ce": 2.4945697077782825e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 64555604,
+      "step": 1030
+    },
+    {
+      "epoch": 3.430948419301165,
+      "grad_norm": 13.57603645324707,
+      "learning_rate": 5e-06,
+      "loss": 0.7378,
+      "num_input_tokens_seen": 64617848,
+      "step": 1031
+    },
+    {
+      "epoch": 3.430948419301165,
+      "loss": 0.9734039902687073,
+      "loss_ce": 0.0007477752515114844,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.97265625,
+      "num_input_tokens_seen": 64617848,
+      "step": 1031
+    },
+    {
+      "epoch": 3.4342762063227954,
+      "grad_norm": 7.267496109008789,
+      "learning_rate": 5e-06,
+      "loss": 0.8297,
+      "num_input_tokens_seen": 64678864,
+      "step": 1032
+    },
+    {
+      "epoch": 3.4342762063227954,
+      "loss": 0.744103193283081,
+      "loss_ce": 0.0006950664101168513,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.0791015625,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 64678864,
+      "step": 1032
+    },
+    {
+      "epoch": 3.437603993344426,
+      "grad_norm": 9.682428359985352,
+      "learning_rate": 5e-06,
+      "loss": 0.9113,
+      "num_input_tokens_seen": 64741056,
+      "step": 1033
+    },
+    {
+      "epoch": 3.437603993344426,
+      "loss": 0.7183736562728882,
+      "loss_ce": 0.0008443902479484677,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 64741056,
+      "step": 1033
+    },
+    {
+      "epoch": 3.4409317803660566,
+      "grad_norm": 11.420270919799805,
+      "learning_rate": 5e-06,
+      "loss": 0.8273,
+      "num_input_tokens_seen": 64803852,
+      "step": 1034
+    },
+    {
+      "epoch": 3.4409317803660566,
+      "loss": 0.8885650038719177,
+      "loss_ce": 0.002090380061417818,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 64803852,
+      "step": 1034
+    },
+    {
+      "epoch": 3.4442595673876872,
+      "grad_norm": 10.965718269348145,
+      "learning_rate": 5e-06,
+      "loss": 0.9267,
+      "num_input_tokens_seen": 64865460,
+      "step": 1035
+    },
+    {
+      "epoch": 3.4442595673876872,
+      "loss": 0.7865220308303833,
+      "loss_ce": 2.2979209461482242e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 64865460,
+      "step": 1035
+    },
+    {
+      "epoch": 3.447587354409318,
+      "grad_norm": 12.703821182250977,
+      "learning_rate": 5e-06,
+      "loss": 0.8564,
+      "num_input_tokens_seen": 64929860,
+      "step": 1036
+    },
+    {
+      "epoch": 3.447587354409318,
+      "loss": 0.8818983435630798,
+      "loss_ce": 0.0012830996420234442,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 64929860,
+      "step": 1036
+    },
+    {
+      "epoch": 3.4509151414309485,
+      "grad_norm": 12.575376510620117,
+      "learning_rate": 5e-06,
+      "loss": 0.8592,
+      "num_input_tokens_seen": 64991228,
+      "step": 1037
+    },
+    {
+      "epoch": 3.4509151414309485,
+      "loss": 0.5446336269378662,
+      "loss_ce": 0.00010847233352251351,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 64991228,
+      "step": 1037
+    },
+    {
+      "epoch": 3.454242928452579,
+      "grad_norm": 18.06930160522461,
+      "learning_rate": 5e-06,
+      "loss": 0.8297,
+      "num_input_tokens_seen": 65052856,
+      "step": 1038
+    },
+    {
+      "epoch": 3.454242928452579,
+      "loss": 0.7931275963783264,
+      "loss_ce": 0.0008912733173929155,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 65052856,
+      "step": 1038
+    },
+    {
+      "epoch": 3.4575707154742097,
+      "grad_norm": 38.98135757446289,
+      "learning_rate": 5e-06,
+      "loss": 0.8898,
+      "num_input_tokens_seen": 65115964,
+      "step": 1039
+    },
+    {
+      "epoch": 3.4575707154742097,
+      "loss": 0.8223487138748169,
+      "loss_ce": 8.309633994940668e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 65115964,
+      "step": 1039
+    },
+    {
+      "epoch": 3.4608985024958403,
+      "grad_norm": 24.175413131713867,
+      "learning_rate": 5e-06,
+      "loss": 0.8058,
+      "num_input_tokens_seen": 65178592,
+      "step": 1040
+    },
+    {
+      "epoch": 3.4608985024958403,
+      "loss": 0.6951857209205627,
+      "loss_ce": 0.005732577759772539,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 65178592,
+      "step": 1040
+    },
+    {
+      "epoch": 3.464226289517471,
+      "grad_norm": 12.473821640014648,
+      "learning_rate": 5e-06,
+      "loss": 0.4723,
+      "num_input_tokens_seen": 65241096,
+      "step": 1041
+    },
+    {
+      "epoch": 3.464226289517471,
+      "loss": 0.5358234643936157,
+      "loss_ce": 0.00042302411748096347,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 65241096,
+      "step": 1041
+    },
+    {
+      "epoch": 3.4675540765391015,
+      "grad_norm": 7.259097576141357,
+      "learning_rate": 5e-06,
+      "loss": 0.6688,
+      "num_input_tokens_seen": 65304704,
+      "step": 1042
+    },
+    {
+      "epoch": 3.4675540765391015,
+      "loss": 0.8926037549972534,
+      "loss_ce": 0.0016125383554026484,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.890625,
+      "num_input_tokens_seen": 65304704,
+      "step": 1042
+    },
+    {
+      "epoch": 3.470881863560732,
+      "grad_norm": 9.943862915039062,
+      "learning_rate": 5e-06,
+      "loss": 0.7162,
+      "num_input_tokens_seen": 65366024,
+      "step": 1043
+    },
+    {
+      "epoch": 3.470881863560732,
+      "loss": 0.4994853138923645,
+      "loss_ce": 0.00033979519503191113,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 65366024,
+      "step": 1043
+    },
+    {
+      "epoch": 3.4742096505823628,
+      "grad_norm": 11.191426277160645,
+      "learning_rate": 5e-06,
+      "loss": 0.5539,
+      "num_input_tokens_seen": 65426728,
+      "step": 1044
+    },
+    {
+      "epoch": 3.4742096505823628,
+      "loss": 0.4565603733062744,
+      "loss_ce": 0.00013946183025836945,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.45703125,
+      "num_input_tokens_seen": 65426728,
+      "step": 1044
+    },
+    {
+      "epoch": 3.4775374376039934,
+      "grad_norm": 13.661975860595703,
+      "learning_rate": 5e-06,
+      "loss": 0.7925,
+      "num_input_tokens_seen": 65489952,
+      "step": 1045
+    },
+    {
+      "epoch": 3.4775374376039934,
+      "loss": 0.9170857071876526,
+      "loss_ce": 9.353942004963756e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 65489952,
+      "step": 1045
+    },
+    {
+      "epoch": 3.480865224625624,
+      "grad_norm": 12.216934204101562,
+      "learning_rate": 5e-06,
+      "loss": 0.8358,
+      "num_input_tokens_seen": 65553428,
+      "step": 1046
+    },
+    {
+      "epoch": 3.480865224625624,
+      "loss": 0.9110158085823059,
+      "loss_ce": 0.00037124729715287685,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.91015625,
+      "num_input_tokens_seen": 65553428,
+      "step": 1046
+    },
+    {
+      "epoch": 3.4841930116472546,
+      "grad_norm": 12.635567665100098,
+      "learning_rate": 5e-06,
+      "loss": 0.7208,
+      "num_input_tokens_seen": 65617964,
+      "step": 1047
+    },
+    {
+      "epoch": 3.4841930116472546,
+      "loss": 0.6810963153839111,
+      "loss_ce": 0.0006764110876247287,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 65617964,
+      "step": 1047
+    },
+    {
+      "epoch": 3.487520798668885,
+      "grad_norm": 15.042474746704102,
+      "learning_rate": 5e-06,
+      "loss": 0.9429,
+      "num_input_tokens_seen": 65681980,
+      "step": 1048
+    },
+    {
+      "epoch": 3.487520798668885,
+      "loss": 1.086167573928833,
+      "loss_ce": 0.001694836188107729,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.078125,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 65681980,
+      "step": 1048
+    },
+    {
+      "epoch": 3.490848585690516,
+      "grad_norm": 19.812339782714844,
+      "learning_rate": 5e-06,
+      "loss": 0.6651,
+      "num_input_tokens_seen": 65743944,
+      "step": 1049
+    },
+    {
+      "epoch": 3.490848585690516,
+      "loss": 0.5148214101791382,
+      "loss_ce": 5.08911361976061e-05,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 65743944,
+      "step": 1049
+    },
+    {
+      "epoch": 3.4941763727121464,
+      "grad_norm": 15.855693817138672,
+      "learning_rate": 5e-06,
+      "loss": 0.5802,
+      "num_input_tokens_seen": 65804424,
+      "step": 1050
+    },
+    {
+      "epoch": 3.4941763727121464,
+      "loss": 0.6694477796554565,
+      "loss_ce": 0.0002583569148555398,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 65804424,
+      "step": 1050
+    },
+    {
+      "epoch": 3.497504159733777,
+      "grad_norm": 12.423200607299805,
+      "learning_rate": 5e-06,
+      "loss": 0.8533,
+      "num_input_tokens_seen": 65866992,
+      "step": 1051
+    },
+    {
+      "epoch": 3.497504159733777,
+      "loss": 0.8513665199279785,
+      "loss_ce": 4.8082820285344496e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.8515625,
+      "num_input_tokens_seen": 65866992,
+      "step": 1051
+    },
+    {
+      "epoch": 3.5008319467554077,
+      "grad_norm": 14.53134536743164,
+      "learning_rate": 5e-06,
+      "loss": 0.7817,
+      "num_input_tokens_seen": 65930468,
+      "step": 1052
+    },
+    {
+      "epoch": 3.5008319467554077,
+      "loss": 0.7512681484222412,
+      "loss_ce": 0.0007798465667292476,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 65930468,
+      "step": 1052
+    },
+    {
+      "epoch": 3.5041597337770383,
+      "grad_norm": 19.564502716064453,
+      "learning_rate": 5e-06,
+      "loss": 0.8821,
+      "num_input_tokens_seen": 65994552,
+      "step": 1053
+    },
+    {
+      "epoch": 3.5041597337770383,
+      "loss": 1.1724635362625122,
+      "loss_ce": 0.0010767867788672447,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 1.171875,
+      "num_input_tokens_seen": 65994552,
+      "step": 1053
+    },
+    {
+      "epoch": 3.507487520798669,
+      "grad_norm": 13.443198204040527,
+      "learning_rate": 5e-06,
+      "loss": 0.7527,
+      "num_input_tokens_seen": 66057744,
+      "step": 1054
+    },
+    {
+      "epoch": 3.507487520798669,
+      "loss": 0.7170261740684509,
+      "loss_ce": 0.00047345724306069314,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 66057744,
+      "step": 1054
+    },
+    {
+      "epoch": 3.5108153078202995,
+      "grad_norm": 14.912182807922363,
+      "learning_rate": 5e-06,
+      "loss": 0.7259,
+      "num_input_tokens_seen": 66121440,
+      "step": 1055
+    },
+    {
+      "epoch": 3.5108153078202995,
+      "loss": 0.6320114135742188,
+      "loss_ce": 0.00017547918832860887,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 66121440,
+      "step": 1055
+    },
+    {
+      "epoch": 3.51414309484193,
+      "grad_norm": 24.48227310180664,
+      "learning_rate": 5e-06,
+      "loss": 0.8422,
+      "num_input_tokens_seen": 66184688,
+      "step": 1056
+    },
+    {
+      "epoch": 3.51414309484193,
+      "loss": 0.7655551433563232,
+      "loss_ce": 0.0006625619134865701,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 66184688,
+      "step": 1056
+    },
+    {
+      "epoch": 3.5174708818635607,
+      "grad_norm": 12.773367881774902,
+      "learning_rate": 5e-06,
+      "loss": 0.7156,
+      "num_input_tokens_seen": 66247292,
+      "step": 1057
+    },
+    {
+      "epoch": 3.5174708818635607,
+      "loss": 0.5347628593444824,
+      "loss_ce": 0.00046112615382298827,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 66247292,
+      "step": 1057
+    },
+    {
+      "epoch": 3.5207986688851913,
+      "grad_norm": 11.14990234375,
+      "learning_rate": 5e-06,
+      "loss": 0.5024,
+      "num_input_tokens_seen": 66310532,
+      "step": 1058
+    },
+    {
+      "epoch": 3.5207986688851913,
+      "loss": 0.6813678741455078,
+      "loss_ce": 0.0007648678729310632,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 66310532,
+      "step": 1058
+    },
+    {
+      "epoch": 3.524126455906822,
+      "grad_norm": 16.360013961791992,
+      "learning_rate": 5e-06,
+      "loss": 0.9234,
+      "num_input_tokens_seen": 66373648,
+      "step": 1059
+    },
+    {
+      "epoch": 3.524126455906822,
+      "loss": 0.9731760025024414,
+      "loss_ce": 3.146879680571146e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.97265625,
+      "num_input_tokens_seen": 66373648,
+      "step": 1059
+    },
+    {
+      "epoch": 3.5274542429284526,
+      "grad_norm": 9.46167278289795,
+      "learning_rate": 5e-06,
+      "loss": 0.4595,
+      "num_input_tokens_seen": 66436316,
+      "step": 1060
+    },
+    {
+      "epoch": 3.5274542429284526,
+      "loss": 0.5476481914520264,
+      "loss_ce": 4.0785591409076005e-05,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 66436316,
+      "step": 1060
+    },
+    {
+      "epoch": 3.530782029950083,
+      "grad_norm": 14.10741901397705,
+      "learning_rate": 5e-06,
+      "loss": 0.9308,
+      "num_input_tokens_seen": 66498344,
+      "step": 1061
+    },
+    {
+      "epoch": 3.530782029950083,
+      "loss": 0.9000440835952759,
+      "loss_ce": 0.002461038064211607,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 66498344,
+      "step": 1061
+    },
+    {
+      "epoch": 3.534109816971714,
+      "grad_norm": 23.609783172607422,
+      "learning_rate": 5e-06,
+      "loss": 0.9126,
+      "num_input_tokens_seen": 66562436,
+      "step": 1062
+    },
+    {
+      "epoch": 3.534109816971714,
+      "loss": 0.7102041840553284,
+      "loss_ce": 0.00036530819488689303,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 66562436,
+      "step": 1062
+    },
+    {
+      "epoch": 3.5374376039933444,
+      "grad_norm": 20.392234802246094,
+      "learning_rate": 5e-06,
+      "loss": 0.5586,
+      "num_input_tokens_seen": 66624320,
+      "step": 1063
+    },
+    {
+      "epoch": 3.5374376039933444,
+      "loss": 0.573874831199646,
+      "loss_ce": 0.0008768028346821666,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.57421875,
+      "num_input_tokens_seen": 66624320,
+      "step": 1063
+    },
+    {
+      "epoch": 3.540765391014975,
+      "grad_norm": 14.3955078125,
+      "learning_rate": 5e-06,
+      "loss": 0.772,
+      "num_input_tokens_seen": 66687376,
+      "step": 1064
+    },
+    {
+      "epoch": 3.540765391014975,
+      "loss": 0.6318628787994385,
+      "loss_ce": 2.6940453608403914e-05,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 66687376,
+      "step": 1064
+    },
+    {
+      "epoch": 3.5440931780366056,
+      "grad_norm": 16.033170700073242,
+      "learning_rate": 5e-06,
+      "loss": 0.6708,
+      "num_input_tokens_seen": 66751300,
+      "step": 1065
+    },
+    {
+      "epoch": 3.5440931780366056,
+      "loss": 0.6771686673164368,
+      "loss_ce": 0.0011432725004851818,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 66751300,
+      "step": 1065
+    },
+    {
+      "epoch": 3.5474209650582362,
+      "grad_norm": 19.82112693786621,
+      "learning_rate": 5e-06,
+      "loss": 1.0317,
+      "num_input_tokens_seen": 66814000,
+      "step": 1066
+    },
+    {
+      "epoch": 3.5474209650582362,
+      "loss": 1.0494093894958496,
+      "loss_ce": 9.300906822318211e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.07568359375,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 66814000,
+      "step": 1066
+    },
+    {
+      "epoch": 3.550748752079867,
+      "grad_norm": 25.609535217285156,
+      "learning_rate": 5e-06,
+      "loss": 0.7597,
+      "num_input_tokens_seen": 66877452,
+      "step": 1067
+    },
+    {
+      "epoch": 3.550748752079867,
+      "loss": 0.6766536831855774,
+      "loss_ce": 0.0016658806707710028,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 66877452,
+      "step": 1067
+    },
+    {
+      "epoch": 3.5540765391014975,
+      "grad_norm": 17.325336456298828,
+      "learning_rate": 5e-06,
+      "loss": 0.6623,
+      "num_input_tokens_seen": 66937972,
+      "step": 1068
+    },
+    {
+      "epoch": 3.5540765391014975,
+      "loss": 0.830532431602478,
+      "loss_ce": 0.0003322733100503683,
+      "loss_iou": 0.21875,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 66937972,
+      "step": 1068
+    },
+    {
+      "epoch": 3.557404326123128,
+      "grad_norm": 22.54002571105957,
+      "learning_rate": 5e-06,
+      "loss": 0.8791,
+      "num_input_tokens_seen": 67000040,
+      "step": 1069
+    },
+    {
+      "epoch": 3.557404326123128,
+      "loss": 1.1053575277328491,
+      "loss_ce": 0.00037706823786720634,
+      "loss_iou": 0.34375,
+      "loss_num": 0.08349609375,
+      "loss_xval": 1.1015625,
+      "num_input_tokens_seen": 67000040,
+      "step": 1069
+    },
+    {
+      "epoch": 3.5607321131447587,
+      "grad_norm": 9.65036678314209,
+      "learning_rate": 5e-06,
+      "loss": 0.699,
+      "num_input_tokens_seen": 67062404,
+      "step": 1070
+    },
+    {
+      "epoch": 3.5607321131447587,
+      "loss": 0.52567458152771,
+      "loss_ce": 3.979210669058375e-05,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 67062404,
+      "step": 1070
+    },
+    {
+      "epoch": 3.5640599001663893,
+      "grad_norm": 9.910517692565918,
+      "learning_rate": 5e-06,
+      "loss": 0.7048,
+      "num_input_tokens_seen": 67124276,
+      "step": 1071
+    },
+    {
+      "epoch": 3.5640599001663893,
+      "loss": 0.6014664173126221,
+      "loss_ce": 2.5981264116126113e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 67124276,
+      "step": 1071
+    },
+    {
+      "epoch": 3.56738768718802,
+      "grad_norm": 15.538228034973145,
+      "learning_rate": 5e-06,
+      "loss": 0.9398,
+      "num_input_tokens_seen": 67187400,
+      "step": 1072
+    },
+    {
+      "epoch": 3.56738768718802,
+      "loss": 1.0962104797363281,
+      "loss_ce": 1.9056995370192453e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.068359375,
+      "loss_xval": 1.09375,
+      "num_input_tokens_seen": 67187400,
+      "step": 1072
+    },
+    {
+      "epoch": 3.5707154742096505,
+      "grad_norm": 9.4950532913208,
+      "learning_rate": 5e-06,
+      "loss": 0.879,
+      "num_input_tokens_seen": 67249424,
+      "step": 1073
+    },
+    {
+      "epoch": 3.5707154742096505,
+      "loss": 0.9734940528869629,
+      "loss_ce": 0.00010540573566686362,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.97265625,
+      "num_input_tokens_seen": 67249424,
+      "step": 1073
+    },
+    {
+      "epoch": 3.574043261231281,
+      "grad_norm": 18.527990341186523,
+      "learning_rate": 5e-06,
+      "loss": 0.8317,
+      "num_input_tokens_seen": 67313036,
+      "step": 1074
+    },
+    {
+      "epoch": 3.574043261231281,
+      "loss": 0.6008124947547913,
+      "loss_ce": 0.0003486633358988911,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 67313036,
+      "step": 1074
+    },
+    {
+      "epoch": 3.5773710482529117,
+      "grad_norm": 26.33303451538086,
+      "learning_rate": 5e-06,
+      "loss": 0.8054,
+      "num_input_tokens_seen": 67375220,
+      "step": 1075
+    },
+    {
+      "epoch": 3.5773710482529117,
+      "loss": 0.8885135054588318,
+      "loss_ce": 0.0008181866724044085,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 67375220,
+      "step": 1075
+    },
+    {
+      "epoch": 3.5806988352745424,
+      "grad_norm": 13.492464065551758,
+      "learning_rate": 5e-06,
+      "loss": 0.7742,
+      "num_input_tokens_seen": 67436772,
+      "step": 1076
+    },
+    {
+      "epoch": 3.5806988352745424,
+      "loss": 0.775583803653717,
+      "loss_ce": 0.00019318400882184505,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 67436772,
+      "step": 1076
+    },
+    {
+      "epoch": 3.584026622296173,
+      "grad_norm": 13.470839500427246,
+      "learning_rate": 5e-06,
+      "loss": 0.7942,
+      "num_input_tokens_seen": 67498376,
+      "step": 1077
+    },
+    {
+      "epoch": 3.584026622296173,
+      "loss": 0.5680028200149536,
+      "loss_ce": 9.682446034275927e-06,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 67498376,
+      "step": 1077
+    },
+    {
+      "epoch": 3.5873544093178036,
+      "grad_norm": 11.887737274169922,
+      "learning_rate": 5e-06,
+      "loss": 0.7184,
+      "num_input_tokens_seen": 67559900,
+      "step": 1078
+    },
+    {
+      "epoch": 3.5873544093178036,
+      "loss": 0.8478894233703613,
+      "loss_ce": 0.0009656131733208895,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.84765625,
+      "num_input_tokens_seen": 67559900,
+      "step": 1078
+    },
+    {
+      "epoch": 3.590682196339434,
+      "grad_norm": 32.69264602661133,
+      "learning_rate": 5e-06,
+      "loss": 0.7889,
+      "num_input_tokens_seen": 67622724,
+      "step": 1079
+    },
+    {
+      "epoch": 3.590682196339434,
+      "loss": 0.760546863079071,
+      "loss_ce": 4.88169098389335e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.76171875,
+      "num_input_tokens_seen": 67622724,
+      "step": 1079
+    },
+    {
+      "epoch": 3.594009983361065,
+      "grad_norm": 15.543495178222656,
+      "learning_rate": 5e-06,
+      "loss": 0.6654,
+      "num_input_tokens_seen": 67683704,
+      "step": 1080
+    },
+    {
+      "epoch": 3.594009983361065,
+      "loss": 0.6395562291145325,
+      "loss_ce": 2.9867431294405833e-05,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 67683704,
+      "step": 1080
+    },
+    {
+      "epoch": 3.5973377703826954,
+      "grad_norm": 22.80522918701172,
+      "learning_rate": 5e-06,
+      "loss": 0.8463,
+      "num_input_tokens_seen": 67746952,
+      "step": 1081
+    },
+    {
+      "epoch": 3.5973377703826954,
+      "loss": 0.8648264408111572,
+      "loss_ce": 0.000446604797616601,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.86328125,
+      "num_input_tokens_seen": 67746952,
+      "step": 1081
+    },
+    {
+      "epoch": 3.600665557404326,
+      "grad_norm": 18.392467498779297,
+      "learning_rate": 5e-06,
+      "loss": 0.6663,
+      "num_input_tokens_seen": 67809764,
+      "step": 1082
+    },
+    {
+      "epoch": 3.600665557404326,
+      "loss": 0.741300106048584,
+      "loss_ce": 0.0014319655019789934,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.73828125,
+      "num_input_tokens_seen": 67809764,
+      "step": 1082
+    },
+    {
+      "epoch": 3.6039933444259566,
+      "grad_norm": 10.33826732635498,
+      "learning_rate": 5e-06,
+      "loss": 0.6008,
+      "num_input_tokens_seen": 67871348,
+      "step": 1083
+    },
+    {
+      "epoch": 3.6039933444259566,
+      "loss": 0.4612269103527069,
+      "loss_ce": 4.5263179345056415e-05,
+      "loss_iou": 0.11865234375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.4609375,
+      "num_input_tokens_seen": 67871348,
+      "step": 1083
+    },
+    {
+      "epoch": 3.6073211314475873,
+      "grad_norm": 26.75817108154297,
+      "learning_rate": 5e-06,
+      "loss": 0.7098,
+      "num_input_tokens_seen": 67934820,
+      "step": 1084
+    },
+    {
+      "epoch": 3.6073211314475873,
+      "loss": 0.7330468893051147,
+      "loss_ce": 1.4662471585324965e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 67934820,
+      "step": 1084
+    },
+    {
+      "epoch": 3.610648918469218,
+      "grad_norm": 14.098214149475098,
+      "learning_rate": 5e-06,
+      "loss": 0.7828,
+      "num_input_tokens_seen": 67996592,
+      "step": 1085
+    },
+    {
+      "epoch": 3.610648918469218,
+      "loss": 0.6801908016204834,
+      "loss_ce": 1.5059520592330955e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 67996592,
+      "step": 1085
+    },
+    {
+      "epoch": 3.6139767054908485,
+      "grad_norm": 11.56197452545166,
+      "learning_rate": 5e-06,
+      "loss": 0.7931,
+      "num_input_tokens_seen": 68060428,
+      "step": 1086
+    },
+    {
+      "epoch": 3.6139767054908485,
+      "loss": 0.7243208885192871,
+      "loss_ce": 0.0006881133886054158,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 68060428,
+      "step": 1086
+    },
+    {
+      "epoch": 3.617304492512479,
+      "grad_norm": 9.20474910736084,
+      "learning_rate": 5e-06,
+      "loss": 0.7987,
+      "num_input_tokens_seen": 68120428,
+      "step": 1087
+    },
+    {
+      "epoch": 3.617304492512479,
+      "loss": 0.9417086243629456,
+      "loss_ce": 0.00079060293501243,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.07421875,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 68120428,
+      "step": 1087
+    },
+    {
+      "epoch": 3.6206322795341097,
+      "grad_norm": 23.553590774536133,
+      "learning_rate": 5e-06,
+      "loss": 0.7281,
+      "num_input_tokens_seen": 68183092,
+      "step": 1088
+    },
+    {
+      "epoch": 3.6206322795341097,
+      "loss": 0.6970406770706177,
+      "loss_ce": 1.9182030882802792e-05,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 68183092,
+      "step": 1088
+    },
+    {
+      "epoch": 3.6239600665557403,
+      "grad_norm": 16.959150314331055,
+      "learning_rate": 5e-06,
+      "loss": 0.6518,
+      "num_input_tokens_seen": 68245676,
+      "step": 1089
+    },
+    {
+      "epoch": 3.6239600665557403,
+      "loss": 0.6579831838607788,
+      "loss_ce": 2.4194079742301255e-05,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 68245676,
+      "step": 1089
+    },
+    {
+      "epoch": 3.627287853577371,
+      "grad_norm": 47.85329055786133,
+      "learning_rate": 5e-06,
+      "loss": 0.7832,
+      "num_input_tokens_seen": 68308340,
+      "step": 1090
+    },
+    {
+      "epoch": 3.627287853577371,
+      "loss": 0.8862950801849365,
+      "loss_ce": 0.00030875191441737115,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 68308340,
+      "step": 1090
+    },
+    {
+      "epoch": 3.6306156405990015,
+      "grad_norm": 13.920936584472656,
+      "learning_rate": 5e-06,
+      "loss": 0.7249,
+      "num_input_tokens_seen": 68370692,
+      "step": 1091
+    },
+    {
+      "epoch": 3.6306156405990015,
+      "loss": 0.8974728584289551,
+      "loss_ce": 0.0002561031433288008,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 68370692,
+      "step": 1091
+    },
+    {
+      "epoch": 3.633943427620632,
+      "grad_norm": 9.428908348083496,
+      "learning_rate": 5e-06,
+      "loss": 0.6884,
+      "num_input_tokens_seen": 68432480,
+      "step": 1092
+    },
+    {
+      "epoch": 3.633943427620632,
+      "loss": 0.6953365802764893,
+      "loss_ce": 2.4052456865319982e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 68432480,
+      "step": 1092
+    },
+    {
+      "epoch": 3.6372712146422628,
+      "grad_norm": 15.467686653137207,
+      "learning_rate": 5e-06,
+      "loss": 0.6283,
+      "num_input_tokens_seen": 68494344,
+      "step": 1093
+    },
+    {
+      "epoch": 3.6372712146422628,
+      "loss": 0.839697539806366,
+      "loss_ce": 0.0005862598773092031,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.83984375,
+      "num_input_tokens_seen": 68494344,
+      "step": 1093
+    },
+    {
+      "epoch": 3.6405990016638934,
+      "grad_norm": 13.4644775390625,
+      "learning_rate": 5e-06,
+      "loss": 0.9332,
+      "num_input_tokens_seen": 68556756,
+      "step": 1094
+    },
+    {
+      "epoch": 3.6405990016638934,
+      "loss": 1.0598011016845703,
+      "loss_ce": 0.00023084873100742698,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.06591796875,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 68556756,
+      "step": 1094
+    },
+    {
+      "epoch": 3.643926788685524,
+      "grad_norm": 7.175938129425049,
+      "learning_rate": 5e-06,
+      "loss": 0.7075,
+      "num_input_tokens_seen": 68616224,
+      "step": 1095
+    },
+    {
+      "epoch": 3.643926788685524,
+      "loss": 0.7629657983779907,
+      "loss_ce": 2.6325005819671787e-05,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.76171875,
+      "num_input_tokens_seen": 68616224,
+      "step": 1095
+    },
+    {
+      "epoch": 3.6472545757071546,
+      "grad_norm": 18.110021591186523,
+      "learning_rate": 5e-06,
+      "loss": 0.7513,
+      "num_input_tokens_seen": 68679384,
+      "step": 1096
+    },
+    {
+      "epoch": 3.6472545757071546,
+      "loss": 0.542603611946106,
+      "loss_ce": 0.00012313880142755806,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 68679384,
+      "step": 1096
+    },
+    {
+      "epoch": 3.6505823627287852,
+      "grad_norm": 24.87841033935547,
+      "learning_rate": 5e-06,
+      "loss": 0.897,
+      "num_input_tokens_seen": 68740984,
+      "step": 1097
+    },
+    {
+      "epoch": 3.6505823627287852,
+      "loss": 0.7363914251327515,
+      "loss_ce": 0.0005515533266589046,
+      "loss_iou": 0.28125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 68740984,
+      "step": 1097
+    },
+    {
+      "epoch": 3.653910149750416,
+      "grad_norm": 14.83497428894043,
+      "learning_rate": 5e-06,
+      "loss": 0.4878,
+      "num_input_tokens_seen": 68803040,
+      "step": 1098
+    },
+    {
+      "epoch": 3.653910149750416,
+      "loss": 0.4231252670288086,
+      "loss_ce": 2.955349555122666e-05,
+      "loss_iou": 0.1171875,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.423828125,
+      "num_input_tokens_seen": 68803040,
+      "step": 1098
+    },
+    {
+      "epoch": 3.6572379367720464,
+      "grad_norm": 10.35799503326416,
+      "learning_rate": 5e-06,
+      "loss": 0.7078,
+      "num_input_tokens_seen": 68866936,
+      "step": 1099
+    },
+    {
+      "epoch": 3.6572379367720464,
+      "loss": 0.56097412109375,
+      "loss_ce": 0.0014343142975121737,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 68866936,
+      "step": 1099
+    },
+    {
+      "epoch": 3.660565723793677,
+      "grad_norm": 20.12316131591797,
+      "learning_rate": 5e-06,
+      "loss": 0.6702,
+      "num_input_tokens_seen": 68927524,
+      "step": 1100
+    },
+    {
+      "epoch": 3.660565723793677,
+      "loss": 0.9203138947486877,
+      "loss_ce": 0.0006362266722135246,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 68927524,
+      "step": 1100
+    },
+    {
+      "epoch": 3.6638935108153077,
+      "grad_norm": 22.879741668701172,
+      "learning_rate": 5e-06,
+      "loss": 0.8751,
+      "num_input_tokens_seen": 68989620,
+      "step": 1101
+    },
+    {
+      "epoch": 3.6638935108153077,
+      "loss": 0.8814558982849121,
+      "loss_ce": 0.017686322331428528,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.86328125,
+      "num_input_tokens_seen": 68989620,
+      "step": 1101
+    },
+    {
+      "epoch": 3.6672212978369383,
+      "grad_norm": 13.441774368286133,
+      "learning_rate": 5e-06,
+      "loss": 0.7847,
+      "num_input_tokens_seen": 69051672,
+      "step": 1102
+    },
+    {
+      "epoch": 3.6672212978369383,
+      "loss": 0.9437904357910156,
+      "loss_ce": 0.0003089719684794545,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 69051672,
+      "step": 1102
+    },
+    {
+      "epoch": 3.670549084858569,
+      "grad_norm": 15.477404594421387,
+      "learning_rate": 5e-06,
+      "loss": 0.7314,
+      "num_input_tokens_seen": 69115388,
+      "step": 1103
+    },
+    {
+      "epoch": 3.670549084858569,
+      "loss": 0.6018927097320557,
+      "loss_ce": 0.0008184734033420682,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 69115388,
+      "step": 1103
+    },
+    {
+      "epoch": 3.6738768718801995,
+      "grad_norm": 19.463781356811523,
+      "learning_rate": 5e-06,
+      "loss": 0.7772,
+      "num_input_tokens_seen": 69178392,
+      "step": 1104
+    },
+    {
+      "epoch": 3.6738768718801995,
+      "loss": 0.7169560194015503,
+      "loss_ce": 0.00028123901574872434,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 69178392,
+      "step": 1104
+    },
+    {
+      "epoch": 3.67720465890183,
+      "grad_norm": 7.575782775878906,
+      "learning_rate": 5e-06,
+      "loss": 0.5294,
+      "num_input_tokens_seen": 69239676,
+      "step": 1105
+    },
+    {
+      "epoch": 3.67720465890183,
+      "loss": 0.531362771987915,
+      "loss_ce": 0.0006010084762237966,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 69239676,
+      "step": 1105
+    },
+    {
+      "epoch": 3.6805324459234607,
+      "grad_norm": 14.393789291381836,
+      "learning_rate": 5e-06,
+      "loss": 0.7182,
+      "num_input_tokens_seen": 69301396,
+      "step": 1106
+    },
+    {
+      "epoch": 3.6805324459234607,
+      "loss": 0.6370840072631836,
+      "loss_ce": 0.0003652620653156191,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 69301396,
+      "step": 1106
+    },
+    {
+      "epoch": 3.6838602329450914,
+      "grad_norm": 11.867156028747559,
+      "learning_rate": 5e-06,
+      "loss": 0.9087,
+      "num_input_tokens_seen": 69362984,
+      "step": 1107
+    },
+    {
+      "epoch": 3.6838602329450914,
+      "loss": 0.5029775500297546,
+      "loss_ce": 0.00029203054145909846,
+      "loss_iou": 0.1015625,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 69362984,
+      "step": 1107
+    },
+    {
+      "epoch": 3.687188019966722,
+      "grad_norm": 14.382938385009766,
+      "learning_rate": 5e-06,
+      "loss": 0.7059,
+      "num_input_tokens_seen": 69425984,
+      "step": 1108
+    },
+    {
+      "epoch": 3.687188019966722,
+      "loss": 0.7315462827682495,
+      "loss_ce": 0.00034508475800976157,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 69425984,
+      "step": 1108
+    },
+    {
+      "epoch": 3.690515806988353,
+      "grad_norm": 21.43218994140625,
+      "learning_rate": 5e-06,
+      "loss": 0.7946,
+      "num_input_tokens_seen": 69489740,
+      "step": 1109
+    },
+    {
+      "epoch": 3.690515806988353,
+      "loss": 0.7349926829338074,
+      "loss_ce": 0.0006176835740916431,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 69489740,
+      "step": 1109
+    },
+    {
+      "epoch": 3.6938435940099836,
+      "grad_norm": 16.769193649291992,
+      "learning_rate": 5e-06,
+      "loss": 0.6175,
+      "num_input_tokens_seen": 69550704,
+      "step": 1110
+    },
+    {
+      "epoch": 3.6938435940099836,
+      "loss": 0.634501576423645,
+      "loss_ce": 0.0007125326665118337,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 69550704,
+      "step": 1110
+    },
+    {
+      "epoch": 3.6971713810316142,
+      "grad_norm": 32.770267486572266,
+      "learning_rate": 5e-06,
+      "loss": 0.9464,
+      "num_input_tokens_seen": 69614880,
+      "step": 1111
+    },
+    {
+      "epoch": 3.6971713810316142,
+      "loss": 1.0868799686431885,
+      "loss_ce": 0.0009425695170648396,
+      "loss_iou": 0.390625,
+      "loss_num": 0.061279296875,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 69614880,
+      "step": 1111
+    },
+    {
+      "epoch": 3.700499168053245,
+      "grad_norm": 43.10527420043945,
+      "learning_rate": 5e-06,
+      "loss": 0.9164,
+      "num_input_tokens_seen": 69677600,
+      "step": 1112
+    },
+    {
+      "epoch": 3.700499168053245,
+      "loss": 0.7824146151542664,
+      "loss_ce": 4.9382551878807135e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 69677600,
+      "step": 1112
+    },
+    {
+      "epoch": 3.7038269550748755,
+      "grad_norm": 22.692411422729492,
+      "learning_rate": 5e-06,
+      "loss": 0.974,
+      "num_input_tokens_seen": 69739600,
+      "step": 1113
+    },
+    {
+      "epoch": 3.7038269550748755,
+      "loss": 0.9298710823059082,
+      "loss_ce": 0.0009159321198239923,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 69739600,
+      "step": 1113
+    },
+    {
+      "epoch": 3.707154742096506,
+      "grad_norm": 10.36915111541748,
+      "learning_rate": 5e-06,
+      "loss": 0.6903,
+      "num_input_tokens_seen": 69800996,
+      "step": 1114
+    },
+    {
+      "epoch": 3.707154742096506,
+      "loss": 0.6277588605880737,
+      "loss_ce": 0.000439511495642364,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 69800996,
+      "step": 1114
+    },
+    {
+      "epoch": 3.7104825291181367,
+      "grad_norm": 15.28381061553955,
+      "learning_rate": 5e-06,
+      "loss": 0.8741,
+      "num_input_tokens_seen": 69863736,
+      "step": 1115
+    },
+    {
+      "epoch": 3.7104825291181367,
+      "loss": 0.9558818340301514,
+      "loss_ce": 0.0011699418537318707,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 69863736,
+      "step": 1115
+    },
+    {
+      "epoch": 3.7138103161397673,
+      "grad_norm": 11.837628364562988,
+      "learning_rate": 5e-06,
+      "loss": 0.8104,
+      "num_input_tokens_seen": 69927776,
+      "step": 1116
+    },
+    {
+      "epoch": 3.7138103161397673,
+      "loss": 0.8550588488578796,
+      "loss_ce": 0.0003225263499189168,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 69927776,
+      "step": 1116
+    },
+    {
+      "epoch": 3.717138103161398,
+      "grad_norm": 20.32373046875,
+      "learning_rate": 5e-06,
+      "loss": 0.8018,
+      "num_input_tokens_seen": 69991752,
+      "step": 1117
+    },
+    {
+      "epoch": 3.717138103161398,
+      "loss": 0.7404411435127258,
+      "loss_ce": 0.0006950714159756899,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.73828125,
+      "num_input_tokens_seen": 69991752,
+      "step": 1117
+    },
+    {
+      "epoch": 3.7204658901830285,
+      "grad_norm": 14.117117881774902,
+      "learning_rate": 5e-06,
+      "loss": 0.7821,
+      "num_input_tokens_seen": 70055220,
+      "step": 1118
+    },
+    {
+      "epoch": 3.7204658901830285,
+      "loss": 1.1473777294158936,
+      "loss_ce": 0.0010153691982850432,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.058837890625,
+      "loss_xval": 1.1484375,
+      "num_input_tokens_seen": 70055220,
+      "step": 1118
+    },
+    {
+      "epoch": 3.723793677204659,
+      "grad_norm": 11.328316688537598,
+      "learning_rate": 5e-06,
+      "loss": 0.6542,
+      "num_input_tokens_seen": 70117140,
+      "step": 1119
+    },
+    {
+      "epoch": 3.723793677204659,
+      "loss": 0.5794090032577515,
+      "loss_ce": 0.00030745528056286275,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 70117140,
+      "step": 1119
+    },
+    {
+      "epoch": 3.7271214642262898,
+      "grad_norm": 9.209413528442383,
+      "learning_rate": 5e-06,
+      "loss": 0.6797,
+      "num_input_tokens_seen": 70179304,
+      "step": 1120
+    },
+    {
+      "epoch": 3.7271214642262898,
+      "loss": 0.6281410455703735,
+      "loss_ce": 0.0010658506071195006,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 70179304,
+      "step": 1120
+    },
+    {
+      "epoch": 3.7304492512479204,
+      "grad_norm": 12.064332008361816,
+      "learning_rate": 5e-06,
+      "loss": 0.682,
+      "num_input_tokens_seen": 70242640,
+      "step": 1121
+    },
+    {
+      "epoch": 3.7304492512479204,
+      "loss": 0.6883260011672974,
+      "loss_ce": 0.00033775315387174487,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 70242640,
+      "step": 1121
+    },
+    {
+      "epoch": 3.733777038269551,
+      "grad_norm": 21.927749633789062,
+      "learning_rate": 5e-06,
+      "loss": 0.9156,
+      "num_input_tokens_seen": 70306468,
+      "step": 1122
+    },
+    {
+      "epoch": 3.733777038269551,
+      "loss": 0.9951090216636658,
+      "loss_ce": 0.00048012335901148617,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.99609375,
+      "num_input_tokens_seen": 70306468,
+      "step": 1122
+    },
+    {
+      "epoch": 3.7371048252911816,
+      "grad_norm": 43.249393463134766,
+      "learning_rate": 5e-06,
+      "loss": 1.0435,
+      "num_input_tokens_seen": 70369504,
+      "step": 1123
+    },
+    {
+      "epoch": 3.7371048252911816,
+      "loss": 1.0370984077453613,
+      "loss_ce": 0.0014538050163537264,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.08447265625,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 70369504,
+      "step": 1123
+    },
+    {
+      "epoch": 3.740432612312812,
+      "grad_norm": 17.981069564819336,
+      "learning_rate": 5e-06,
+      "loss": 0.6572,
+      "num_input_tokens_seen": 70432836,
+      "step": 1124
+    },
+    {
+      "epoch": 3.740432612312812,
+      "loss": 0.7932811975479126,
+      "loss_ce": 6.828659388702363e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 70432836,
+      "step": 1124
+    },
+    {
+      "epoch": 3.743760399334443,
+      "grad_norm": 13.31531047821045,
+      "learning_rate": 5e-06,
+      "loss": 0.7378,
+      "num_input_tokens_seen": 70495112,
+      "step": 1125
+    },
+    {
+      "epoch": 3.743760399334443,
+      "loss": 0.9534047245979309,
+      "loss_ce": 0.001012163469567895,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 70495112,
+      "step": 1125
+    },
+    {
+      "epoch": 3.7470881863560734,
+      "grad_norm": 13.23071002960205,
+      "learning_rate": 5e-06,
+      "loss": 1.0306,
+      "num_input_tokens_seen": 70557332,
+      "step": 1126
+    },
+    {
+      "epoch": 3.7470881863560734,
+      "loss": 1.243412971496582,
+      "loss_ce": 0.0004931364674121141,
+      "loss_iou": 0.4375,
+      "loss_num": 0.07373046875,
+      "loss_xval": 1.2421875,
+      "num_input_tokens_seen": 70557332,
+      "step": 1126
+    },
+    {
+      "epoch": 3.750415973377704,
+      "grad_norm": 11.754276275634766,
+      "learning_rate": 5e-06,
+      "loss": 0.7345,
+      "num_input_tokens_seen": 70619872,
+      "step": 1127
+    },
+    {
+      "epoch": 3.750415973377704,
+      "loss": 0.9770193099975586,
+      "loss_ce": 0.0011892361799255013,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 70619872,
+      "step": 1127
+    },
+    {
+      "epoch": 3.7537437603993347,
+      "grad_norm": 16.051631927490234,
+      "learning_rate": 5e-06,
+      "loss": 1.002,
+      "num_input_tokens_seen": 70681060,
+      "step": 1128
+    },
+    {
+      "epoch": 3.7537437603993347,
+      "loss": 1.000293493270874,
+      "loss_ce": 0.002002517692744732,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.07763671875,
+      "loss_xval": 1.0,
+      "num_input_tokens_seen": 70681060,
+      "step": 1128
+    },
+    {
+      "epoch": 3.7570715474209653,
+      "grad_norm": 12.550768852233887,
+      "learning_rate": 5e-06,
+      "loss": 0.7239,
+      "num_input_tokens_seen": 70742176,
+      "step": 1129
+    },
+    {
+      "epoch": 3.7570715474209653,
+      "loss": 0.8635753393173218,
+      "loss_ce": 0.000538200605660677,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.86328125,
+      "num_input_tokens_seen": 70742176,
+      "step": 1129
+    },
+    {
+      "epoch": 3.760399334442596,
+      "grad_norm": 14.114124298095703,
+      "learning_rate": 5e-06,
+      "loss": 0.6735,
+      "num_input_tokens_seen": 70804608,
+      "step": 1130
+    },
+    {
+      "epoch": 3.760399334442596,
+      "loss": 0.4799818992614746,
+      "loss_ce": 0.0007338316063396633,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.478515625,
+      "num_input_tokens_seen": 70804608,
+      "step": 1130
+    },
+    {
+      "epoch": 3.7637271214642265,
+      "grad_norm": 21.7076473236084,
+      "learning_rate": 5e-06,
+      "loss": 0.7799,
+      "num_input_tokens_seen": 70867088,
+      "step": 1131
+    },
+    {
+      "epoch": 3.7637271214642265,
+      "loss": 0.4688870310783386,
+      "loss_ce": 0.00016756025434006006,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.46875,
+      "num_input_tokens_seen": 70867088,
+      "step": 1131
+    },
+    {
+      "epoch": 3.767054908485857,
+      "grad_norm": 10.689167022705078,
+      "learning_rate": 5e-06,
+      "loss": 0.6789,
+      "num_input_tokens_seen": 70928940,
+      "step": 1132
+    },
+    {
+      "epoch": 3.767054908485857,
+      "loss": 0.9793890714645386,
+      "loss_ce": 0.0003851721412502229,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.98046875,
+      "num_input_tokens_seen": 70928940,
+      "step": 1132
+    },
+    {
+      "epoch": 3.7703826955074877,
+      "grad_norm": 18.24936866760254,
+      "learning_rate": 5e-06,
+      "loss": 0.8163,
+      "num_input_tokens_seen": 70989924,
+      "step": 1133
+    },
+    {
+      "epoch": 3.7703826955074877,
+      "loss": 0.7339310646057129,
+      "loss_ce": 4.4314670958556235e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.08056640625,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 70989924,
+      "step": 1133
+    },
+    {
+      "epoch": 3.7737104825291183,
+      "grad_norm": 22.505451202392578,
+      "learning_rate": 5e-06,
+      "loss": 0.5424,
+      "num_input_tokens_seen": 71051320,
+      "step": 1134
+    },
+    {
+      "epoch": 3.7737104825291183,
+      "loss": 0.7140070199966431,
+      "loss_ce": 0.0006280855741351843,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 71051320,
+      "step": 1134
+    },
+    {
+      "epoch": 3.777038269550749,
+      "grad_norm": 8.693553924560547,
+      "learning_rate": 5e-06,
+      "loss": 0.5119,
+      "num_input_tokens_seen": 71112332,
+      "step": 1135
+    },
+    {
+      "epoch": 3.777038269550749,
+      "loss": 0.5159156322479248,
+      "loss_ce": 3.124218710581772e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 71112332,
+      "step": 1135
+    },
+    {
+      "epoch": 3.7803660565723796,
+      "grad_norm": 12.785008430480957,
+      "learning_rate": 5e-06,
+      "loss": 0.6925,
+      "num_input_tokens_seen": 71175924,
+      "step": 1136
+    },
+    {
+      "epoch": 3.7803660565723796,
+      "loss": 0.6734526753425598,
+      "loss_ce": 0.0007231989875435829,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 71175924,
+      "step": 1136
+    },
+    {
+      "epoch": 3.78369384359401,
+      "grad_norm": 11.817228317260742,
+      "learning_rate": 5e-06,
+      "loss": 0.9368,
+      "num_input_tokens_seen": 71238860,
+      "step": 1137
+    },
+    {
+      "epoch": 3.78369384359401,
+      "loss": 0.8826572895050049,
+      "loss_ce": 0.00021094588737469167,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 71238860,
+      "step": 1137
+    },
+    {
+      "epoch": 3.787021630615641,
+      "grad_norm": 13.229910850524902,
+      "learning_rate": 5e-06,
+      "loss": 0.6284,
+      "num_input_tokens_seen": 71301228,
+      "step": 1138
+    },
+    {
+      "epoch": 3.787021630615641,
+      "loss": 0.7485630512237549,
+      "loss_ce": 0.0002720615011639893,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 71301228,
+      "step": 1138
+    },
+    {
+      "epoch": 3.7903494176372714,
+      "grad_norm": 39.25370788574219,
+      "learning_rate": 5e-06,
+      "loss": 0.7781,
+      "num_input_tokens_seen": 71364208,
+      "step": 1139
+    },
+    {
+      "epoch": 3.7903494176372714,
+      "loss": 0.860191822052002,
+      "loss_ce": 8.443903061561286e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 71364208,
+      "step": 1139
+    },
+    {
+      "epoch": 3.793677204658902,
+      "grad_norm": 8.552867889404297,
+      "learning_rate": 5e-06,
+      "loss": 0.6363,
+      "num_input_tokens_seen": 71425864,
+      "step": 1140
+    },
+    {
+      "epoch": 3.793677204658902,
+      "loss": 0.6165646314620972,
+      "loss_ce": 0.000842000765260309,
+      "loss_iou": 0.140625,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 71425864,
+      "step": 1140
+    },
+    {
+      "epoch": 3.7970049916805326,
+      "grad_norm": 7.005518436431885,
+      "learning_rate": 5e-06,
+      "loss": 0.7277,
+      "num_input_tokens_seen": 71489608,
+      "step": 1141
+    },
+    {
+      "epoch": 3.7970049916805326,
+      "loss": 0.7442701458930969,
+      "loss_ce": 0.0013502332149073482,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 71489608,
+      "step": 1141
+    },
+    {
+      "epoch": 3.8003327787021632,
+      "grad_norm": 7.645864486694336,
+      "learning_rate": 5e-06,
+      "loss": 0.6064,
+      "num_input_tokens_seen": 71551324,
+      "step": 1142
+    },
+    {
+      "epoch": 3.8003327787021632,
+      "loss": 0.48682481050491333,
+      "loss_ce": 0.00025257206289097667,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.486328125,
+      "num_input_tokens_seen": 71551324,
+      "step": 1142
+    },
+    {
+      "epoch": 3.803660565723794,
+      "grad_norm": 11.24929141998291,
+      "learning_rate": 5e-06,
+      "loss": 0.6623,
+      "num_input_tokens_seen": 71614812,
+      "step": 1143
+    },
+    {
+      "epoch": 3.803660565723794,
+      "loss": 0.5440496206283569,
+      "loss_ce": 4.3287254811730236e-05,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 71614812,
+      "step": 1143
+    },
+    {
+      "epoch": 3.8069883527454245,
+      "grad_norm": 14.913579940795898,
+      "learning_rate": 5e-06,
+      "loss": 1.1182,
+      "num_input_tokens_seen": 71678540,
+      "step": 1144
+    },
+    {
+      "epoch": 3.8069883527454245,
+      "loss": 0.9577356576919556,
+      "loss_ce": 0.0007043948862701654,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.95703125,
+      "num_input_tokens_seen": 71678540,
+      "step": 1144
+    },
+    {
+      "epoch": 3.810316139767055,
+      "grad_norm": 9.077905654907227,
+      "learning_rate": 5e-06,
+      "loss": 0.6147,
+      "num_input_tokens_seen": 71739620,
+      "step": 1145
+    },
+    {
+      "epoch": 3.810316139767055,
+      "loss": 0.5191451907157898,
+      "loss_ce": 4.1206356399925426e-05,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 71739620,
+      "step": 1145
+    },
+    {
+      "epoch": 3.8136439267886857,
+      "grad_norm": 19.119752883911133,
+      "learning_rate": 5e-06,
+      "loss": 0.6505,
+      "num_input_tokens_seen": 71801800,
+      "step": 1146
+    },
+    {
+      "epoch": 3.8136439267886857,
+      "loss": 0.6102641820907593,
+      "loss_ce": 0.0006450038054026663,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 71801800,
+      "step": 1146
+    },
+    {
+      "epoch": 3.8169717138103163,
+      "grad_norm": 24.023101806640625,
+      "learning_rate": 5e-06,
+      "loss": 0.6411,
+      "num_input_tokens_seen": 71866032,
+      "step": 1147
+    },
+    {
+      "epoch": 3.8169717138103163,
+      "loss": 0.600353479385376,
+      "loss_ce": 1.1719953363353852e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 71866032,
+      "step": 1147
+    },
+    {
+      "epoch": 3.820299500831947,
+      "grad_norm": 11.998858451843262,
+      "learning_rate": 5e-06,
+      "loss": 0.7952,
+      "num_input_tokens_seen": 71926820,
+      "step": 1148
+    },
+    {
+      "epoch": 3.820299500831947,
+      "loss": 0.8667431473731995,
+      "loss_ce": 0.001142566092312336,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 71926820,
+      "step": 1148
+    },
+    {
+      "epoch": 3.8236272878535775,
+      "grad_norm": 23.826696395874023,
+      "learning_rate": 5e-06,
+      "loss": 0.6687,
+      "num_input_tokens_seen": 71989184,
+      "step": 1149
+    },
+    {
+      "epoch": 3.8236272878535775,
+      "loss": 0.6472935080528259,
+      "loss_ce": 0.0059361206367611885,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 71989184,
+      "step": 1149
+    },
+    {
+      "epoch": 3.826955074875208,
+      "grad_norm": 15.457709312438965,
+      "learning_rate": 5e-06,
+      "loss": 0.6533,
+      "num_input_tokens_seen": 72052768,
+      "step": 1150
+    },
+    {
+      "epoch": 3.826955074875208,
+      "loss": 0.8420194983482361,
+      "loss_ce": 0.0013212742051109672,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.83984375,
+      "num_input_tokens_seen": 72052768,
+      "step": 1150
+    },
+    {
+      "epoch": 3.8302828618968388,
+      "grad_norm": 11.85561466217041,
+      "learning_rate": 5e-06,
+      "loss": 0.5567,
+      "num_input_tokens_seen": 72115176,
+      "step": 1151
+    },
+    {
+      "epoch": 3.8302828618968388,
+      "loss": 0.491475909948349,
+      "loss_ce": 0.00014292128616943955,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 72115176,
+      "step": 1151
+    },
+    {
+      "epoch": 3.8336106489184694,
+      "grad_norm": 14.689626693725586,
+      "learning_rate": 5e-06,
+      "loss": 0.7026,
+      "num_input_tokens_seen": 72176716,
+      "step": 1152
+    },
+    {
+      "epoch": 3.8336106489184694,
+      "loss": 0.9168117046356201,
+      "loss_ce": 0.00030778077780269086,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 72176716,
+      "step": 1152
+    },
+    {
+      "epoch": 3.8369384359401,
+      "grad_norm": 13.6842679977417,
+      "learning_rate": 5e-06,
+      "loss": 0.9622,
+      "num_input_tokens_seen": 72239940,
+      "step": 1153
+    },
+    {
+      "epoch": 3.8369384359401,
+      "loss": 0.9152591824531555,
+      "loss_ce": 0.0009526039939373732,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 72239940,
+      "step": 1153
+    },
+    {
+      "epoch": 3.8402662229617306,
+      "grad_norm": 11.269613265991211,
+      "learning_rate": 5e-06,
+      "loss": 0.7609,
+      "num_input_tokens_seen": 72301720,
+      "step": 1154
+    },
+    {
+      "epoch": 3.8402662229617306,
+      "loss": 0.6302889585494995,
+      "loss_ce": 0.001016515656374395,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 72301720,
+      "step": 1154
+    },
+    {
+      "epoch": 3.843594009983361,
+      "grad_norm": 21.826889038085938,
+      "learning_rate": 5e-06,
+      "loss": 1.0018,
+      "num_input_tokens_seen": 72363092,
+      "step": 1155
+    },
+    {
+      "epoch": 3.843594009983361,
+      "loss": 1.0488756895065308,
+      "loss_ce": 0.0009020923171192408,
+      "loss_iou": 0.34375,
+      "loss_num": 0.072265625,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 72363092,
+      "step": 1155
+    },
+    {
+      "epoch": 3.846921797004992,
+      "grad_norm": 27.468761444091797,
+      "learning_rate": 5e-06,
+      "loss": 0.9151,
+      "num_input_tokens_seen": 72427664,
+      "step": 1156
+    },
+    {
+      "epoch": 3.846921797004992,
+      "loss": 0.9939867258071899,
+      "loss_ce": 0.0002123451413353905,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 72427664,
+      "step": 1156
+    },
+    {
+      "epoch": 3.8502495840266224,
+      "grad_norm": 20.527379989624023,
+      "learning_rate": 5e-06,
+      "loss": 0.9454,
+      "num_input_tokens_seen": 72492032,
+      "step": 1157
+    },
+    {
+      "epoch": 3.8502495840266224,
+      "loss": 0.8400707244873047,
+      "loss_ce": 0.0010204899590462446,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.83984375,
+      "num_input_tokens_seen": 72492032,
+      "step": 1157
+    },
+    {
+      "epoch": 3.853577371048253,
+      "grad_norm": 8.549810409545898,
+      "learning_rate": 5e-06,
+      "loss": 0.7699,
+      "num_input_tokens_seen": 72554224,
+      "step": 1158
+    },
+    {
+      "epoch": 3.853577371048253,
+      "loss": 0.627190351486206,
+      "loss_ce": 0.0021292981691658497,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 72554224,
+      "step": 1158
+    },
+    {
+      "epoch": 3.8569051580698837,
+      "grad_norm": 21.32713508605957,
+      "learning_rate": 5e-06,
+      "loss": 0.6667,
+      "num_input_tokens_seen": 72617400,
+      "step": 1159
+    },
+    {
+      "epoch": 3.8569051580698837,
+      "loss": 0.8027868270874023,
+      "loss_ce": 5.2467978093773127e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 72617400,
+      "step": 1159
+    },
+    {
+      "epoch": 3.8602329450915143,
+      "grad_norm": 12.14395809173584,
+      "learning_rate": 5e-06,
+      "loss": 0.7003,
+      "num_input_tokens_seen": 72679172,
+      "step": 1160
+    },
+    {
+      "epoch": 3.8602329450915143,
+      "loss": 0.6025677919387817,
+      "loss_ce": 2.867916555260308e-05,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 72679172,
+      "step": 1160
+    },
+    {
+      "epoch": 3.863560732113145,
+      "grad_norm": 19.726367950439453,
+      "learning_rate": 5e-06,
+      "loss": 0.6465,
+      "num_input_tokens_seen": 72742260,
+      "step": 1161
+    },
+    {
+      "epoch": 3.863560732113145,
+      "loss": 0.6337836980819702,
+      "loss_ce": 0.00011673450353555381,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 72742260,
+      "step": 1161
+    },
+    {
+      "epoch": 3.8668885191347755,
+      "grad_norm": 15.199066162109375,
+      "learning_rate": 5e-06,
+      "loss": 0.6762,
+      "num_input_tokens_seen": 72803396,
+      "step": 1162
+    },
+    {
+      "epoch": 3.8668885191347755,
+      "loss": 0.8257670998573303,
+      "loss_ce": 8.353222074219957e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 72803396,
+      "step": 1162
+    },
+    {
+      "epoch": 3.870216306156406,
+      "grad_norm": 27.860286712646484,
+      "learning_rate": 5e-06,
+      "loss": 0.7779,
+      "num_input_tokens_seen": 72865844,
+      "step": 1163
+    },
+    {
+      "epoch": 3.870216306156406,
+      "loss": 0.9027190208435059,
+      "loss_ce": 0.0006194374873302877,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.90234375,
+      "num_input_tokens_seen": 72865844,
+      "step": 1163
+    },
+    {
+      "epoch": 3.8735440931780367,
+      "grad_norm": 22.976346969604492,
+      "learning_rate": 5e-06,
+      "loss": 0.8613,
+      "num_input_tokens_seen": 72926168,
+      "step": 1164
+    },
+    {
+      "epoch": 3.8735440931780367,
+      "loss": 1.036806583404541,
+      "loss_ce": 0.0006737524527125061,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.054931640625,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 72926168,
+      "step": 1164
+    },
+    {
+      "epoch": 3.8768718801996673,
+      "grad_norm": 8.25428581237793,
+      "learning_rate": 5e-06,
+      "loss": 0.7253,
+      "num_input_tokens_seen": 72987380,
+      "step": 1165
+    },
+    {
+      "epoch": 3.8768718801996673,
+      "loss": 0.6392136216163635,
+      "loss_ce": 5.344775854609907e-05,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 72987380,
+      "step": 1165
+    },
+    {
+      "epoch": 3.880199667221298,
+      "grad_norm": 16.900022506713867,
+      "learning_rate": 5e-06,
+      "loss": 0.7989,
+      "num_input_tokens_seen": 73051392,
+      "step": 1166
+    },
+    {
+      "epoch": 3.880199667221298,
+      "loss": 0.7662708759307861,
+      "loss_ce": 0.0011341598583385348,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 73051392,
+      "step": 1166
+    },
+    {
+      "epoch": 3.8835274542429286,
+      "grad_norm": 14.377167701721191,
+      "learning_rate": 5e-06,
+      "loss": 0.763,
+      "num_input_tokens_seen": 73113100,
+      "step": 1167
+    },
+    {
+      "epoch": 3.8835274542429286,
+      "loss": 0.5086641311645508,
+      "loss_ce": 0.00024124824267346412,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.5078125,
+      "num_input_tokens_seen": 73113100,
+      "step": 1167
+    },
+    {
+      "epoch": 3.886855241264559,
+      "grad_norm": 9.697162628173828,
+      "learning_rate": 5e-06,
+      "loss": 0.8776,
+      "num_input_tokens_seen": 73176080,
+      "step": 1168
+    },
+    {
+      "epoch": 3.886855241264559,
+      "loss": 0.7365698218345642,
+      "loss_ce": 0.0007299709832295775,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 73176080,
+      "step": 1168
+    },
+    {
+      "epoch": 3.89018302828619,
+      "grad_norm": 7.753446578979492,
+      "learning_rate": 5e-06,
+      "loss": 0.6761,
+      "num_input_tokens_seen": 73239540,
+      "step": 1169
+    },
+    {
+      "epoch": 3.89018302828619,
+      "loss": 0.8952956199645996,
+      "loss_ce": 0.0007643617573194206,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.89453125,
+      "num_input_tokens_seen": 73239540,
+      "step": 1169
+    },
+    {
+      "epoch": 3.8935108153078204,
+      "grad_norm": 38.60930633544922,
+      "learning_rate": 5e-06,
+      "loss": 0.9386,
+      "num_input_tokens_seen": 73301948,
+      "step": 1170
+    },
+    {
+      "epoch": 3.8935108153078204,
+      "loss": 0.8330338001251221,
+      "loss_ce": 0.0005142604932188988,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 73301948,
+      "step": 1170
+    },
+    {
+      "epoch": 3.896838602329451,
+      "grad_norm": 44.411895751953125,
+      "learning_rate": 5e-06,
+      "loss": 0.877,
+      "num_input_tokens_seen": 73364536,
+      "step": 1171
+    },
+    {
+      "epoch": 3.896838602329451,
+      "loss": 0.9410591125488281,
+      "loss_ce": 0.0005683889612555504,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 73364536,
+      "step": 1171
+    },
+    {
+      "epoch": 3.9001663893510816,
+      "grad_norm": 15.136682510375977,
+      "learning_rate": 5e-06,
+      "loss": 0.5564,
+      "num_input_tokens_seen": 73426144,
+      "step": 1172
+    },
+    {
+      "epoch": 3.9001663893510816,
+      "loss": 0.4998212456703186,
+      "loss_ce": 0.00012645090464502573,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 73426144,
+      "step": 1172
+    },
+    {
+      "epoch": 3.9034941763727122,
+      "grad_norm": 31.158117294311523,
+      "learning_rate": 5e-06,
+      "loss": 0.6874,
+      "num_input_tokens_seen": 73488980,
+      "step": 1173
+    },
+    {
+      "epoch": 3.9034941763727122,
+      "loss": 0.6511569023132324,
+      "loss_ce": 3.384947558515705e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 73488980,
+      "step": 1173
+    },
+    {
+      "epoch": 3.906821963394343,
+      "grad_norm": 18.398252487182617,
+      "learning_rate": 5e-06,
+      "loss": 0.7025,
+      "num_input_tokens_seen": 73550568,
+      "step": 1174
+    },
+    {
+      "epoch": 3.906821963394343,
+      "loss": 0.8032513856887817,
+      "loss_ce": 0.0012493999674916267,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 73550568,
+      "step": 1174
+    },
+    {
+      "epoch": 3.9101497504159735,
+      "grad_norm": 14.897561073303223,
+      "learning_rate": 5e-06,
+      "loss": 0.6192,
+      "num_input_tokens_seen": 73614048,
+      "step": 1175
+    },
+    {
+      "epoch": 3.9101497504159735,
+      "loss": 0.6096196174621582,
+      "loss_ce": 0.000488759484142065,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 73614048,
+      "step": 1175
+    },
+    {
+      "epoch": 3.913477537437604,
+      "grad_norm": 12.411521911621094,
+      "learning_rate": 5e-06,
+      "loss": 0.7318,
+      "num_input_tokens_seen": 73677156,
+      "step": 1176
+    },
+    {
+      "epoch": 3.913477537437604,
+      "loss": 0.7434451580047607,
+      "loss_ce": 0.0008914528880268335,
+      "loss_iou": 0.265625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 73677156,
+      "step": 1176
+    },
+    {
+      "epoch": 3.9168053244592347,
+      "grad_norm": 12.991275787353516,
+      "learning_rate": 5e-06,
+      "loss": 0.7032,
+      "num_input_tokens_seen": 73739192,
+      "step": 1177
+    },
+    {
+      "epoch": 3.9168053244592347,
+      "loss": 0.7536220550537109,
+      "loss_ce": 0.00020409880380611867,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 73739192,
+      "step": 1177
+    },
+    {
+      "epoch": 3.9201331114808653,
+      "grad_norm": 12.387129783630371,
+      "learning_rate": 5e-06,
+      "loss": 0.4625,
+      "num_input_tokens_seen": 73801340,
+      "step": 1178
+    },
+    {
+      "epoch": 3.9201331114808653,
+      "loss": 0.4774191081523895,
+      "loss_ce": 0.0004903915687464178,
+      "loss_iou": 0.10791015625,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 73801340,
+      "step": 1178
+    },
+    {
+      "epoch": 3.923460898502496,
+      "grad_norm": 13.816216468811035,
+      "learning_rate": 5e-06,
+      "loss": 0.5731,
+      "num_input_tokens_seen": 73863168,
+      "step": 1179
+    },
+    {
+      "epoch": 3.923460898502496,
+      "loss": 0.4799365699291229,
+      "loss_ce": 1.7145994206657633e-05,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.48046875,
+      "num_input_tokens_seen": 73863168,
+      "step": 1179
+    },
+    {
+      "epoch": 3.9267886855241265,
+      "grad_norm": 13.6268310546875,
+      "learning_rate": 5e-06,
+      "loss": 0.5817,
+      "num_input_tokens_seen": 73925652,
+      "step": 1180
+    },
+    {
+      "epoch": 3.9267886855241265,
+      "loss": 0.46266400814056396,
+      "loss_ce": 1.7535534425405785e-05,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.462890625,
+      "num_input_tokens_seen": 73925652,
+      "step": 1180
+    },
+    {
+      "epoch": 3.930116472545757,
+      "grad_norm": 10.935044288635254,
+      "learning_rate": 5e-06,
+      "loss": 0.7337,
+      "num_input_tokens_seen": 73988132,
+      "step": 1181
+    },
+    {
+      "epoch": 3.930116472545757,
+      "loss": 0.714240312576294,
+      "loss_ce": 0.000251059333095327,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0625,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 73988132,
+      "step": 1181
+    },
+    {
+      "epoch": 3.9334442595673877,
+      "grad_norm": 21.19894790649414,
+      "learning_rate": 5e-06,
+      "loss": 0.7096,
+      "num_input_tokens_seen": 74051356,
+      "step": 1182
+    },
+    {
+      "epoch": 3.9334442595673877,
+      "loss": 0.7160288691520691,
+      "loss_ce": 0.00020855304319411516,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 74051356,
+      "step": 1182
+    },
+    {
+      "epoch": 3.9367720465890184,
+      "grad_norm": 22.308738708496094,
+      "learning_rate": 5e-06,
+      "loss": 1.0778,
+      "num_input_tokens_seen": 74115356,
+      "step": 1183
+    },
+    {
+      "epoch": 3.9367720465890184,
+      "loss": 1.0566372871398926,
+      "loss_ce": 0.0003629096318036318,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0712890625,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 74115356,
+      "step": 1183
+    },
+    {
+      "epoch": 3.940099833610649,
+      "grad_norm": 12.43722152709961,
+      "learning_rate": 5e-06,
+      "loss": 0.7557,
+      "num_input_tokens_seen": 74178668,
+      "step": 1184
+    },
+    {
+      "epoch": 3.940099833610649,
+      "loss": 0.5400795936584473,
+      "loss_ce": 4.0542239730712026e-05,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 74178668,
+      "step": 1184
+    },
+    {
+      "epoch": 3.9434276206322796,
+      "grad_norm": 12.8514404296875,
+      "learning_rate": 5e-06,
+      "loss": 0.706,
+      "num_input_tokens_seen": 74240964,
+      "step": 1185
+    },
+    {
+      "epoch": 3.9434276206322796,
+      "loss": 0.8476996421813965,
+      "loss_ce": 0.0007758617866784334,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.84765625,
+      "num_input_tokens_seen": 74240964,
+      "step": 1185
+    },
+    {
+      "epoch": 3.94675540765391,
+      "grad_norm": 17.675451278686523,
+      "learning_rate": 5e-06,
+      "loss": 0.8197,
+      "num_input_tokens_seen": 74304024,
+      "step": 1186
+    },
+    {
+      "epoch": 3.94675540765391,
+      "loss": 0.9300236701965332,
+      "loss_ce": 9.207585389958695e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 74304024,
+      "step": 1186
+    },
+    {
+      "epoch": 3.950083194675541,
+      "grad_norm": 8.358492851257324,
+      "learning_rate": 5e-06,
+      "loss": 0.7479,
+      "num_input_tokens_seen": 74367408,
+      "step": 1187
+    },
+    {
+      "epoch": 3.950083194675541,
+      "loss": 0.6704340577125549,
+      "loss_ce": 0.0016108033014461398,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 74367408,
+      "step": 1187
+    },
+    {
+      "epoch": 3.9534109816971714,
+      "grad_norm": 10.79404067993164,
+      "learning_rate": 5e-06,
+      "loss": 0.7976,
+      "num_input_tokens_seen": 74430972,
+      "step": 1188
+    },
+    {
+      "epoch": 3.9534109816971714,
+      "loss": 0.8109077215194702,
+      "loss_ce": 0.0008491338812746108,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.80859375,
+      "num_input_tokens_seen": 74430972,
+      "step": 1188
+    },
+    {
+      "epoch": 3.956738768718802,
+      "grad_norm": 10.02695083618164,
+      "learning_rate": 5e-06,
+      "loss": 0.6818,
+      "num_input_tokens_seen": 74493320,
+      "step": 1189
+    },
+    {
+      "epoch": 3.956738768718802,
+      "loss": 0.6926209330558777,
+      "loss_ce": 0.0002991709334310144,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.69140625,
+      "num_input_tokens_seen": 74493320,
+      "step": 1189
+    },
+    {
+      "epoch": 3.9600665557404326,
+      "grad_norm": 15.090518951416016,
+      "learning_rate": 5e-06,
+      "loss": 0.6759,
+      "num_input_tokens_seen": 74556176,
+      "step": 1190
+    },
+    {
+      "epoch": 3.9600665557404326,
+      "loss": 0.37127041816711426,
+      "loss_ce": 0.0005428859149105847,
+      "loss_iou": 0.10888671875,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 74556176,
+      "step": 1190
+    },
+    {
+      "epoch": 3.9633943427620633,
+      "grad_norm": 15.567995071411133,
+      "learning_rate": 5e-06,
+      "loss": 0.8636,
+      "num_input_tokens_seen": 74619904,
+      "step": 1191
+    },
+    {
+      "epoch": 3.9633943427620633,
+      "loss": 1.1187453269958496,
+      "loss_ce": 0.001313702785409987,
+      "loss_iou": 0.421875,
+      "loss_num": 0.05517578125,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 74619904,
+      "step": 1191
+    },
+    {
+      "epoch": 3.966722129783694,
+      "grad_norm": 20.189136505126953,
+      "learning_rate": 5e-06,
+      "loss": 0.899,
+      "num_input_tokens_seen": 74683056,
+      "step": 1192
+    },
+    {
+      "epoch": 3.966722129783694,
+      "loss": 0.6560426950454712,
+      "loss_ce": 0.002234060550108552,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 74683056,
+      "step": 1192
+    },
+    {
+      "epoch": 3.9700499168053245,
+      "grad_norm": 15.30156135559082,
+      "learning_rate": 5e-06,
+      "loss": 0.5481,
+      "num_input_tokens_seen": 74744020,
+      "step": 1193
+    },
+    {
+      "epoch": 3.9700499168053245,
+      "loss": 0.45218271017074585,
+      "loss_ce": 3.4305761801078916e-05,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 74744020,
+      "step": 1193
+    },
+    {
+      "epoch": 3.973377703826955,
+      "grad_norm": 18.261381149291992,
+      "learning_rate": 5e-06,
+      "loss": 0.6689,
+      "num_input_tokens_seen": 74806148,
+      "step": 1194
+    },
+    {
+      "epoch": 3.973377703826955,
+      "loss": 0.6880115866661072,
+      "loss_ce": 2.3293252525036223e-05,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 74806148,
+      "step": 1194
+    },
+    {
+      "epoch": 3.9767054908485857,
+      "grad_norm": 35.95396423339844,
+      "learning_rate": 5e-06,
+      "loss": 0.9737,
+      "num_input_tokens_seen": 74870276,
+      "step": 1195
+    },
+    {
+      "epoch": 3.9767054908485857,
+      "loss": 1.2164342403411865,
+      "loss_ce": 0.003543566446751356,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.07958984375,
+      "loss_xval": 1.2109375,
+      "num_input_tokens_seen": 74870276,
+      "step": 1195
+    },
+    {
+      "epoch": 3.9800332778702163,
+      "grad_norm": 15.76313304901123,
+      "learning_rate": 5e-06,
+      "loss": 0.7284,
+      "num_input_tokens_seen": 74933660,
+      "step": 1196
+    },
+    {
+      "epoch": 3.9800332778702163,
+      "loss": 0.8802682161331177,
+      "loss_ce": 0.0006294772610999644,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 74933660,
+      "step": 1196
+    },
+    {
+      "epoch": 3.983361064891847,
+      "grad_norm": 8.434407234191895,
+      "learning_rate": 5e-06,
+      "loss": 0.6092,
+      "num_input_tokens_seen": 74997052,
+      "step": 1197
+    },
+    {
+      "epoch": 3.983361064891847,
+      "loss": 0.511523962020874,
+      "loss_ce": 0.0004155955102760345,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 74997052,
+      "step": 1197
+    },
+    {
+      "epoch": 3.9866888519134775,
+      "grad_norm": 10.437862396240234,
+      "learning_rate": 5e-06,
+      "loss": 0.7356,
+      "num_input_tokens_seen": 75058944,
+      "step": 1198
+    },
+    {
+      "epoch": 3.9866888519134775,
+      "loss": 0.9156403541564941,
+      "loss_ce": 0.0020660818554461002,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 75058944,
+      "step": 1198
+    },
+    {
+      "epoch": 3.990016638935108,
+      "grad_norm": 22.736345291137695,
+      "learning_rate": 5e-06,
+      "loss": 0.601,
+      "num_input_tokens_seen": 75120440,
+      "step": 1199
+    },
+    {
+      "epoch": 3.990016638935108,
+      "loss": 0.6733030080795288,
+      "loss_ce": 0.0006955533754080534,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 75120440,
+      "step": 1199
+    },
+    {
+      "epoch": 3.9933444259567388,
+      "grad_norm": 19.93775177001953,
+      "learning_rate": 5e-06,
+      "loss": 0.5766,
+      "num_input_tokens_seen": 75180068,
+      "step": 1200
+    },
+    {
+      "epoch": 3.9933444259567388,
+      "loss": 0.502943754196167,
+      "loss_ce": 1.4059570276003797e-05,
+      "loss_iou": 0.1240234375,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 75180068,
+      "step": 1200
+    },
+    {
+      "epoch": 3.9966722129783694,
+      "grad_norm": 7.942922592163086,
+      "learning_rate": 5e-06,
+      "loss": 0.6018,
+      "num_input_tokens_seen": 75242588,
+      "step": 1201
+    },
+    {
+      "epoch": 3.9966722129783694,
+      "loss": 0.5372599363327026,
+      "loss_ce": 0.0003947424702346325,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 75242588,
+      "step": 1201
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 15.248062133789062,
+      "learning_rate": 5e-06,
+      "loss": 0.6613,
+      "num_input_tokens_seen": 75304732,
+      "step": 1202
+    },
+    {
+      "epoch": 4.0,
+      "loss": 0.7052999138832092,
+      "loss_ce": 0.000648995628580451,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 75304732,
+      "step": 1202
+    },
+    {
+      "epoch": 4.003327787021631,
+      "grad_norm": 11.188032150268555,
+      "learning_rate": 5e-06,
+      "loss": 0.7275,
+      "num_input_tokens_seen": 75366540,
+      "step": 1203
+    },
+    {
+      "epoch": 4.003327787021631,
+      "loss": 0.7967070937156677,
+      "loss_ce": 0.0008086481248028576,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 75366540,
+      "step": 1203
+    },
+    {
+      "epoch": 4.006655574043261,
+      "grad_norm": 5.265565872192383,
+      "learning_rate": 5e-06,
+      "loss": 0.3553,
+      "num_input_tokens_seen": 75427768,
+      "step": 1204
+    },
+    {
+      "epoch": 4.006655574043261,
+      "loss": 0.2486070692539215,
+      "loss_ce": 1.0882431524805725e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 75427768,
+      "step": 1204
+    },
+    {
+      "epoch": 4.009983361064892,
+      "grad_norm": 73.91722106933594,
+      "learning_rate": 5e-06,
+      "loss": 0.7568,
+      "num_input_tokens_seen": 75491840,
+      "step": 1205
+    },
+    {
+      "epoch": 4.009983361064892,
+      "loss": 0.8017033338546753,
+      "loss_ce": 0.000922084494959563,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 75491840,
+      "step": 1205
+    },
+    {
+      "epoch": 4.0133111480865225,
+      "grad_norm": 19.91536521911621,
+      "learning_rate": 5e-06,
+      "loss": 0.5606,
+      "num_input_tokens_seen": 75553160,
+      "step": 1206
+    },
+    {
+      "epoch": 4.0133111480865225,
+      "loss": 0.5015987157821655,
+      "loss_ce": 0.00013387855142354965,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 75553160,
+      "step": 1206
+    },
+    {
+      "epoch": 4.016638935108153,
+      "grad_norm": 12.133269309997559,
+      "learning_rate": 5e-06,
+      "loss": 0.8538,
+      "num_input_tokens_seen": 75618412,
+      "step": 1207
+    },
+    {
+      "epoch": 4.016638935108153,
+      "loss": 0.9982300996780396,
+      "loss_ce": 0.0014039267553016543,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.99609375,
+      "num_input_tokens_seen": 75618412,
+      "step": 1207
+    },
+    {
+      "epoch": 4.019966722129784,
+      "grad_norm": 8.875154495239258,
+      "learning_rate": 5e-06,
+      "loss": 0.775,
+      "num_input_tokens_seen": 75682372,
+      "step": 1208
+    },
+    {
+      "epoch": 4.019966722129784,
+      "loss": 0.8194925785064697,
+      "loss_ce": 0.0001566612918395549,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 75682372,
+      "step": 1208
+    },
+    {
+      "epoch": 4.023294509151414,
+      "grad_norm": 10.672651290893555,
+      "learning_rate": 5e-06,
+      "loss": 0.4433,
+      "num_input_tokens_seen": 75742552,
+      "step": 1209
+    },
+    {
+      "epoch": 4.023294509151414,
+      "loss": 0.3726426362991333,
+      "loss_ce": 2.2998472559265792e-05,
+      "loss_iou": 0.05029296875,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 75742552,
+      "step": 1209
+    },
+    {
+      "epoch": 4.026622296173045,
+      "grad_norm": 31.792285919189453,
+      "learning_rate": 5e-06,
+      "loss": 0.7017,
+      "num_input_tokens_seen": 75804336,
+      "step": 1210
+    },
+    {
+      "epoch": 4.026622296173045,
+      "loss": 0.7170919179916382,
+      "loss_ce": 0.0002950564958155155,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 75804336,
+      "step": 1210
+    },
+    {
+      "epoch": 4.0299500831946755,
+      "grad_norm": 14.573902130126953,
+      "learning_rate": 5e-06,
+      "loss": 0.6705,
+      "num_input_tokens_seen": 75865872,
+      "step": 1211
+    },
+    {
+      "epoch": 4.0299500831946755,
+      "loss": 0.7652605175971985,
+      "loss_ce": 0.00030688103288412094,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 75865872,
+      "step": 1211
+    },
+    {
+      "epoch": 4.033277870216306,
+      "grad_norm": 17.535743713378906,
+      "learning_rate": 5e-06,
+      "loss": 0.6471,
+      "num_input_tokens_seen": 75930268,
+      "step": 1212
+    },
+    {
+      "epoch": 4.033277870216306,
+      "loss": 0.7741683721542358,
+      "loss_ce": 0.0007308665663003922,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 75930268,
+      "step": 1212
+    },
+    {
+      "epoch": 4.036605657237937,
+      "grad_norm": 14.236082077026367,
+      "learning_rate": 5e-06,
+      "loss": 0.9975,
+      "num_input_tokens_seen": 75994488,
+      "step": 1213
+    },
+    {
+      "epoch": 4.036605657237937,
+      "loss": 1.1918745040893555,
+      "loss_ce": 0.0007124332478269935,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.07666015625,
+      "loss_xval": 1.1875,
+      "num_input_tokens_seen": 75994488,
+      "step": 1213
+    },
+    {
+      "epoch": 4.039933444259567,
+      "grad_norm": 12.264684677124023,
+      "learning_rate": 5e-06,
+      "loss": 0.8542,
+      "num_input_tokens_seen": 76057384,
+      "step": 1214
+    },
+    {
+      "epoch": 4.039933444259567,
+      "loss": 1.0039153099060059,
+      "loss_ce": 0.0009246146655641496,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.062255859375,
+      "loss_xval": 1.0,
+      "num_input_tokens_seen": 76057384,
+      "step": 1214
+    },
+    {
+      "epoch": 4.043261231281198,
+      "grad_norm": 11.338297843933105,
+      "learning_rate": 5e-06,
+      "loss": 0.6555,
+      "num_input_tokens_seen": 76120200,
+      "step": 1215
+    },
+    {
+      "epoch": 4.043261231281198,
+      "loss": 0.6709122061729431,
+      "loss_ce": 1.3811413737130351e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 76120200,
+      "step": 1215
+    },
+    {
+      "epoch": 4.046589018302829,
+      "grad_norm": 11.017167091369629,
+      "learning_rate": 5e-06,
+      "loss": 0.759,
+      "num_input_tokens_seen": 76184652,
+      "step": 1216
+    },
+    {
+      "epoch": 4.046589018302829,
+      "loss": 0.8067546486854553,
+      "loss_ce": 0.0002360953949391842,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 76184652,
+      "step": 1216
+    },
+    {
+      "epoch": 4.049916805324459,
+      "grad_norm": 9.538055419921875,
+      "learning_rate": 5e-06,
+      "loss": 0.659,
+      "num_input_tokens_seen": 76248504,
+      "step": 1217
+    },
+    {
+      "epoch": 4.049916805324459,
+      "loss": 0.6807469129562378,
+      "loss_ce": 0.00020490906899794936,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 76248504,
+      "step": 1217
+    },
+    {
+      "epoch": 4.05324459234609,
+      "grad_norm": 13.316681861877441,
+      "learning_rate": 5e-06,
+      "loss": 0.7569,
+      "num_input_tokens_seen": 76312336,
+      "step": 1218
+    },
+    {
+      "epoch": 4.05324459234609,
+      "loss": 0.6233140230178833,
+      "loss_ce": 0.00014506097068078816,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 76312336,
+      "step": 1218
+    },
+    {
+      "epoch": 4.05657237936772,
+      "grad_norm": 18.77733039855957,
+      "learning_rate": 5e-06,
+      "loss": 0.7825,
+      "num_input_tokens_seen": 76377484,
+      "step": 1219
+    },
+    {
+      "epoch": 4.05657237936772,
+      "loss": 0.8869447708129883,
+      "loss_ce": 0.001202607061713934,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 76377484,
+      "step": 1219
+    },
+    {
+      "epoch": 4.059900166389351,
+      "grad_norm": 13.015058517456055,
+      "learning_rate": 5e-06,
+      "loss": 0.8573,
+      "num_input_tokens_seen": 76439780,
+      "step": 1220
+    },
+    {
+      "epoch": 4.059900166389351,
+      "loss": 0.7998440265655518,
+      "loss_ce": 0.0010159012163057923,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 76439780,
+      "step": 1220
+    },
+    {
+      "epoch": 4.063227953410982,
+      "grad_norm": 10.296722412109375,
+      "learning_rate": 5e-06,
+      "loss": 1.0222,
+      "num_input_tokens_seen": 76500992,
+      "step": 1221
+    },
+    {
+      "epoch": 4.063227953410982,
+      "loss": 0.8199893236160278,
+      "loss_ce": 0.0005923279095441103,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.062255859375,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 76500992,
+      "step": 1221
+    },
+    {
+      "epoch": 4.066555740432612,
+      "grad_norm": 16.238292694091797,
+      "learning_rate": 5e-06,
+      "loss": 0.971,
+      "num_input_tokens_seen": 76564108,
+      "step": 1222
+    },
+    {
+      "epoch": 4.066555740432612,
+      "loss": 1.0141180753707886,
+      "loss_ce": 0.0008124337764456868,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.04345703125,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 76564108,
+      "step": 1222
+    },
+    {
+      "epoch": 4.069883527454243,
+      "grad_norm": 15.772029876708984,
+      "learning_rate": 5e-06,
+      "loss": 0.804,
+      "num_input_tokens_seen": 76627116,
+      "step": 1223
+    },
+    {
+      "epoch": 4.069883527454243,
+      "loss": 0.8473187685012817,
+      "loss_ce": 2.8751601348631084e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.84765625,
+      "num_input_tokens_seen": 76627116,
+      "step": 1223
+    },
+    {
+      "epoch": 4.0732113144758735,
+      "grad_norm": 21.51706314086914,
+      "learning_rate": 5e-06,
+      "loss": 0.757,
+      "num_input_tokens_seen": 76688968,
+      "step": 1224
+    },
+    {
+      "epoch": 4.0732113144758735,
+      "loss": 0.6889785528182983,
+      "loss_ce": 1.3671827218786348e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 76688968,
+      "step": 1224
+    },
+    {
+      "epoch": 4.076539101497504,
+      "grad_norm": 13.671257019042969,
+      "learning_rate": 5e-06,
+      "loss": 0.6394,
+      "num_input_tokens_seen": 76752252,
+      "step": 1225
+    },
+    {
+      "epoch": 4.076539101497504,
+      "loss": 0.6889891624450684,
+      "loss_ce": 0.00026844226522371173,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 76752252,
+      "step": 1225
+    },
+    {
+      "epoch": 4.079866888519135,
+      "grad_norm": 12.221083641052246,
+      "learning_rate": 5e-06,
+      "loss": 0.5801,
+      "num_input_tokens_seen": 76814152,
+      "step": 1226
+    },
+    {
+      "epoch": 4.079866888519135,
+      "loss": 0.5570855736732483,
+      "loss_ce": 0.0003839263808913529,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 76814152,
+      "step": 1226
+    },
+    {
+      "epoch": 4.083194675540765,
+      "grad_norm": 20.396621704101562,
+      "learning_rate": 5e-06,
+      "loss": 0.773,
+      "num_input_tokens_seen": 76877564,
+      "step": 1227
+    },
+    {
+      "epoch": 4.083194675540765,
+      "loss": 0.6958671808242798,
+      "loss_ce": 6.636592297581956e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 76877564,
+      "step": 1227
+    },
+    {
+      "epoch": 4.086522462562396,
+      "grad_norm": 21.930482864379883,
+      "learning_rate": 5e-06,
+      "loss": 0.9511,
+      "num_input_tokens_seen": 76943260,
+      "step": 1228
+    },
+    {
+      "epoch": 4.086522462562396,
+      "loss": 1.138702630996704,
+      "loss_ce": 0.0022280181292444468,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.051025390625,
+      "loss_xval": 1.1328125,
+      "num_input_tokens_seen": 76943260,
+      "step": 1228
+    },
+    {
+      "epoch": 4.0898502495840265,
+      "grad_norm": 10.00102424621582,
+      "learning_rate": 5e-06,
+      "loss": 0.6424,
+      "num_input_tokens_seen": 77005432,
+      "step": 1229
+    },
+    {
+      "epoch": 4.0898502495840265,
+      "loss": 0.47109830379486084,
+      "loss_ce": 0.0005172473029233515,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.470703125,
+      "num_input_tokens_seen": 77005432,
+      "step": 1229
+    },
+    {
+      "epoch": 4.093178036605657,
+      "grad_norm": 15.279685020446777,
+      "learning_rate": 5e-06,
+      "loss": 0.8382,
+      "num_input_tokens_seen": 77067552,
+      "step": 1230
+    },
+    {
+      "epoch": 4.093178036605657,
+      "loss": 0.7615088820457458,
+      "loss_ce": 3.428545824135654e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.76171875,
+      "num_input_tokens_seen": 77067552,
+      "step": 1230
+    },
+    {
+      "epoch": 4.096505823627288,
+      "grad_norm": 24.48037338256836,
+      "learning_rate": 5e-06,
+      "loss": 0.7931,
+      "num_input_tokens_seen": 77130584,
+      "step": 1231
+    },
+    {
+      "epoch": 4.096505823627288,
+      "loss": 0.9837879538536072,
+      "loss_ce": 2.3307909941649996e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 77130584,
+      "step": 1231
+    },
+    {
+      "epoch": 4.099833610648918,
+      "grad_norm": 35.03629684448242,
+      "learning_rate": 5e-06,
+      "loss": 1.0287,
+      "num_input_tokens_seen": 77194704,
+      "step": 1232
+    },
+    {
+      "epoch": 4.099833610648918,
+      "loss": 1.1050645112991333,
+      "loss_ce": 0.0008164606406353414,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.07470703125,
+      "loss_xval": 1.1015625,
+      "num_input_tokens_seen": 77194704,
+      "step": 1232
+    },
+    {
+      "epoch": 4.103161397670549,
+      "grad_norm": 47.826690673828125,
+      "learning_rate": 5e-06,
+      "loss": 0.7708,
+      "num_input_tokens_seen": 77258880,
+      "step": 1233
+    },
+    {
+      "epoch": 4.103161397670549,
+      "loss": 0.8139855861663818,
+      "loss_ce": 0.0007531545124948025,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 77258880,
+      "step": 1233
+    },
+    {
+      "epoch": 4.10648918469218,
+      "grad_norm": 29.535999298095703,
+      "learning_rate": 5e-06,
+      "loss": 0.7673,
+      "num_input_tokens_seen": 77321428,
+      "step": 1234
+    },
+    {
+      "epoch": 4.10648918469218,
+      "loss": 0.7830252647399902,
+      "loss_ce": 6.632165604969487e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 77321428,
+      "step": 1234
+    },
+    {
+      "epoch": 4.10981697171381,
+      "grad_norm": 21.408437728881836,
+      "learning_rate": 5e-06,
+      "loss": 0.7928,
+      "num_input_tokens_seen": 77383604,
+      "step": 1235
+    },
+    {
+      "epoch": 4.10981697171381,
+      "loss": 0.7510450482368469,
+      "loss_ce": 0.0008008688455447555,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 77383604,
+      "step": 1235
+    },
+    {
+      "epoch": 4.113144758735441,
+      "grad_norm": 15.90986442565918,
+      "learning_rate": 5e-06,
+      "loss": 0.4682,
+      "num_input_tokens_seen": 77446016,
+      "step": 1236
+    },
+    {
+      "epoch": 4.113144758735441,
+      "loss": 0.3090360760688782,
+      "loss_ce": 0.0005643867189064622,
+      "loss_iou": 0.0654296875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 77446016,
+      "step": 1236
+    },
+    {
+      "epoch": 4.116472545757071,
+      "grad_norm": 22.74576187133789,
+      "learning_rate": 5e-06,
+      "loss": 0.9686,
+      "num_input_tokens_seen": 77511136,
+      "step": 1237
+    },
+    {
+      "epoch": 4.116472545757071,
+      "loss": 1.0496145486831665,
+      "loss_ce": 0.00029816440655849874,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0751953125,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 77511136,
+      "step": 1237
+    },
+    {
+      "epoch": 4.119800332778702,
+      "grad_norm": 27.88201332092285,
+      "learning_rate": 5e-06,
+      "loss": 0.5876,
+      "num_input_tokens_seen": 77574216,
+      "step": 1238
+    },
+    {
+      "epoch": 4.119800332778702,
+      "loss": 0.6051973700523376,
+      "loss_ce": 0.00046104774810373783,
+      "loss_iou": 0.21875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 77574216,
+      "step": 1238
+    },
+    {
+      "epoch": 4.123128119800333,
+      "grad_norm": 11.923408508300781,
+      "learning_rate": 5e-06,
+      "loss": 0.6994,
+      "num_input_tokens_seen": 77636964,
+      "step": 1239
+    },
+    {
+      "epoch": 4.123128119800333,
+      "loss": 0.9356486201286316,
+      "loss_ce": 0.0003458983264863491,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 77636964,
+      "step": 1239
+    },
+    {
+      "epoch": 4.126455906821963,
+      "grad_norm": 14.89736270904541,
+      "learning_rate": 5e-06,
+      "loss": 0.4658,
+      "num_input_tokens_seen": 77697900,
+      "step": 1240
+    },
+    {
+      "epoch": 4.126455906821963,
+      "loss": 0.5524719953536987,
+      "loss_ce": 0.009991498664021492,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 77697900,
+      "step": 1240
+    },
+    {
+      "epoch": 4.129783693843594,
+      "grad_norm": 7.385460376739502,
+      "learning_rate": 5e-06,
+      "loss": 0.7427,
+      "num_input_tokens_seen": 77760148,
+      "step": 1241
+    },
+    {
+      "epoch": 4.129783693843594,
+      "loss": 0.9304953813552856,
+      "loss_ce": 0.0008078092359937727,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 77760148,
+      "step": 1241
+    },
+    {
+      "epoch": 4.1331114808652245,
+      "grad_norm": 11.04951286315918,
+      "learning_rate": 5e-06,
+      "loss": 0.6882,
+      "num_input_tokens_seen": 77821680,
+      "step": 1242
+    },
+    {
+      "epoch": 4.1331114808652245,
+      "loss": 0.5894361734390259,
+      "loss_ce": 1.963950853678398e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 77821680,
+      "step": 1242
+    },
+    {
+      "epoch": 4.136439267886855,
+      "grad_norm": 41.49568176269531,
+      "learning_rate": 5e-06,
+      "loss": 0.9372,
+      "num_input_tokens_seen": 77884232,
+      "step": 1243
+    },
+    {
+      "epoch": 4.136439267886855,
+      "loss": 0.7601950168609619,
+      "loss_ce": 0.0006735285860486329,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 77884232,
+      "step": 1243
+    },
+    {
+      "epoch": 4.139767054908486,
+      "grad_norm": 7.810762405395508,
+      "learning_rate": 5e-06,
+      "loss": 0.6704,
+      "num_input_tokens_seen": 77947048,
+      "step": 1244
+    },
+    {
+      "epoch": 4.139767054908486,
+      "loss": 0.817348062992096,
+      "loss_ce": 0.00057560222921893,
+      "loss_iou": 0.265625,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 77947048,
+      "step": 1244
+    },
+    {
+      "epoch": 4.143094841930116,
+      "grad_norm": 14.463872909545898,
+      "learning_rate": 5e-06,
+      "loss": 0.5967,
+      "num_input_tokens_seen": 78007588,
+      "step": 1245
+    },
+    {
+      "epoch": 4.143094841930116,
+      "loss": 0.4404696822166443,
+      "loss_ce": 3.99709097109735e-05,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 78007588,
+      "step": 1245
+    },
+    {
+      "epoch": 4.146422628951747,
+      "grad_norm": 11.963314056396484,
+      "learning_rate": 5e-06,
+      "loss": 0.7085,
+      "num_input_tokens_seen": 78069416,
+      "step": 1246
+    },
+    {
+      "epoch": 4.146422628951747,
+      "loss": 0.6378343105316162,
+      "loss_ce": 1.6925867384998128e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 78069416,
+      "step": 1246
+    },
+    {
+      "epoch": 4.149750415973378,
+      "grad_norm": 23.427759170532227,
+      "learning_rate": 5e-06,
+      "loss": 0.6977,
+      "num_input_tokens_seen": 78132404,
+      "step": 1247
+    },
+    {
+      "epoch": 4.149750415973378,
+      "loss": 0.7420227527618408,
+      "loss_ce": 0.00038455973844975233,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 78132404,
+      "step": 1247
+    },
+    {
+      "epoch": 4.153078202995008,
+      "grad_norm": 19.418766021728516,
+      "learning_rate": 5e-06,
+      "loss": 0.5784,
+      "num_input_tokens_seen": 78194872,
+      "step": 1248
+    },
+    {
+      "epoch": 4.153078202995008,
+      "loss": 0.4353361427783966,
+      "loss_ce": 0.00021654315060004592,
+      "loss_iou": 0.11181640625,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.435546875,
+      "num_input_tokens_seen": 78194872,
+      "step": 1248
+    },
+    {
+      "epoch": 4.156405990016639,
+      "grad_norm": 12.20608139038086,
+      "learning_rate": 5e-06,
+      "loss": 0.8433,
+      "num_input_tokens_seen": 78258412,
+      "step": 1249
+    },
+    {
+      "epoch": 4.156405990016639,
+      "loss": 0.7217040061950684,
+      "loss_ce": 2.4295461116707884e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 78258412,
+      "step": 1249
+    },
+    {
+      "epoch": 4.159733777038269,
+      "grad_norm": 12.527824401855469,
+      "learning_rate": 5e-06,
+      "loss": 0.7033,
+      "num_input_tokens_seen": 78322204,
+      "step": 1250
+    },
+    {
+      "epoch": 4.159733777038269,
+      "eval_seeclick_CIoU": 0.12239253148436546,
+      "eval_seeclick_GIoU": 0.1511169895529747,
+      "eval_seeclick_IoU": 0.21942590922117233,
+      "eval_seeclick_MAE_all": 0.18183603137731552,
+      "eval_seeclick_MAE_h": 0.05301396735012531,
+      "eval_seeclick_MAE_w": 0.11510565504431725,
+      "eval_seeclick_MAE_x_boxes": 0.26491403579711914,
+      "eval_seeclick_MAE_y_boxes": 0.14476536214351654,
+      "eval_seeclick_NUM_probability": 0.9971157312393188,
+      "eval_seeclick_inside_bbox": 0.2718750014901161,
+      "eval_seeclick_loss": 2.71364426612854,
+      "eval_seeclick_loss_ce": 0.0916629321873188,
+      "eval_seeclick_loss_iou": 0.8514404296875,
+      "eval_seeclick_loss_num": 0.18135833740234375,
+      "eval_seeclick_loss_xval": 2.6103515625,
+      "eval_seeclick_runtime": 65.2703,
+      "eval_seeclick_samples_per_second": 0.72,
+      "eval_seeclick_steps_per_second": 0.031,
+      "num_input_tokens_seen": 78322204,
+      "step": 1250
+    },
+    {
+      "epoch": 4.159733777038269,
+      "eval_icons_CIoU": 0.03201650455594063,
+      "eval_icons_GIoU": 0.14019615948200226,
+      "eval_icons_IoU": 0.1814664751291275,
+      "eval_icons_MAE_all": 0.1366742141544819,
+      "eval_icons_MAE_h": 0.053603796288371086,
+      "eval_icons_MAE_w": 0.13712040334939957,
+      "eval_icons_MAE_x_boxes": 0.16213426738977432,
+      "eval_icons_MAE_y_boxes": 0.042389593087136745,
+      "eval_icons_NUM_probability": 0.9999926686286926,
+      "eval_icons_inside_bbox": 0.3576388955116272,
+      "eval_icons_loss": 2.41593861579895,
+      "eval_icons_loss_ce": 1.236241587321274e-06,
+      "eval_icons_loss_iou": 0.845458984375,
+      "eval_icons_loss_num": 0.13992691040039062,
+      "eval_icons_loss_xval": 2.38916015625,
+      "eval_icons_runtime": 64.8775,
+      "eval_icons_samples_per_second": 0.771,
+      "eval_icons_steps_per_second": 0.031,
+      "num_input_tokens_seen": 78322204,
+      "step": 1250
+    },
+    {
+      "epoch": 4.159733777038269,
+      "eval_screenspot_CIoU": 0.05670065308610598,
+      "eval_screenspot_GIoU": 0.12369907399018605,
+      "eval_screenspot_IoU": 0.20270906388759613,
+      "eval_screenspot_MAE_all": 0.20509299635887146,
+      "eval_screenspot_MAE_h": 0.0750602458914121,
+      "eval_screenspot_MAE_w": 0.1766246110200882,
+      "eval_screenspot_MAE_x_boxes": 0.26625461379686993,
+      "eval_screenspot_MAE_y_boxes": 0.14071540037790933,
+      "eval_screenspot_NUM_probability": 0.9999165137608846,
+      "eval_screenspot_inside_bbox": 0.3387500047683716,
+      "eval_screenspot_loss": 2.7868411540985107,
+      "eval_screenspot_loss_ce": 0.0001397205999940828,
+      "eval_screenspot_loss_iou": 0.8883463541666666,
+      "eval_screenspot_loss_num": 0.21124267578125,
+      "eval_screenspot_loss_xval": 2.8326822916666665,
+      "eval_screenspot_runtime": 112.2559,
+      "eval_screenspot_samples_per_second": 0.793,
+      "eval_screenspot_steps_per_second": 0.027,
+      "num_input_tokens_seen": 78322204,
+      "step": 1250
+    },
+    {
+      "epoch": 4.159733777038269,
+      "eval_compot_CIoU": -0.04526886157691479,
+      "eval_compot_GIoU": 0.07369405962526798,
+      "eval_compot_IoU": 0.12557204440236092,
+      "eval_compot_MAE_all": 0.21759440749883652,
+      "eval_compot_MAE_h": 0.0656399242579937,
+      "eval_compot_MAE_w": 0.24755792319774628,
+      "eval_compot_MAE_x_boxes": 0.19112109392881393,
+      "eval_compot_MAE_y_boxes": 0.13230591267347336,
+      "eval_compot_NUM_probability": 0.9999650120735168,
+      "eval_compot_inside_bbox": 0.2395833358168602,
+      "eval_compot_loss": 2.8879714012145996,
+      "eval_compot_loss_ce": 0.005962656578049064,
+      "eval_compot_loss_iou": 0.923095703125,
+      "eval_compot_loss_num": 0.205718994140625,
+      "eval_compot_loss_xval": 2.873046875,
+      "eval_compot_runtime": 74.5769,
+      "eval_compot_samples_per_second": 0.67,
+      "eval_compot_steps_per_second": 0.027,
+      "num_input_tokens_seen": 78322204,
+      "step": 1250
+    },
+    {
+      "epoch": 4.159733777038269,
+      "eval_custom_ui_MAE_all": 0.07984927669167519,
+      "eval_custom_ui_MAE_x": 0.07491207309067249,
+      "eval_custom_ui_MAE_y": 0.08478647843003273,
+      "eval_custom_ui_NUM_probability": 0.9999922215938568,
+      "eval_custom_ui_loss": 0.38728201389312744,
+      "eval_custom_ui_loss_ce": 1.5295036746465485e-05,
+      "eval_custom_ui_loss_num": 0.0752716064453125,
+      "eval_custom_ui_loss_xval": 0.37652587890625,
+      "eval_custom_ui_runtime": 58.5282,
+      "eval_custom_ui_samples_per_second": 0.854,
+      "eval_custom_ui_steps_per_second": 0.034,
+      "num_input_tokens_seen": 78322204,
+      "step": 1250
+    },
+    {
+      "epoch": 4.159733777038269,
+      "loss": 0.3760015368461609,
+      "loss_ce": 2.4953300453489646e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 78322204,
+      "step": 1250
+    },
+    {
+      "epoch": 4.1630615640599,
+      "grad_norm": 8.160618782043457,
+      "learning_rate": 5e-06,
+      "loss": 0.5385,
+      "num_input_tokens_seen": 78384488,
+      "step": 1251
+    },
+    {
+      "epoch": 4.1630615640599,
+      "loss": 0.5602893829345703,
+      "loss_ce": 0.0011768110562115908,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 78384488,
+      "step": 1251
+    },
+    {
+      "epoch": 4.166389351081531,
+      "grad_norm": 36.30681610107422,
+      "learning_rate": 5e-06,
+      "loss": 0.9031,
+      "num_input_tokens_seen": 78447932,
+      "step": 1252
+    },
+    {
+      "epoch": 4.166389351081531,
+      "loss": 0.8445026874542236,
+      "loss_ce": 0.00026445844559930265,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 78447932,
+      "step": 1252
+    },
+    {
+      "epoch": 4.169717138103161,
+      "grad_norm": 49.61948776245117,
+      "learning_rate": 5e-06,
+      "loss": 0.9137,
+      "num_input_tokens_seen": 78513152,
+      "step": 1253
+    },
+    {
+      "epoch": 4.169717138103161,
+      "loss": 0.8838903903961182,
+      "loss_ce": 0.0010779636213555932,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 78513152,
+      "step": 1253
+    },
+    {
+      "epoch": 4.173044925124792,
+      "grad_norm": 29.25813102722168,
+      "learning_rate": 5e-06,
+      "loss": 0.4711,
+      "num_input_tokens_seen": 78574284,
+      "step": 1254
+    },
+    {
+      "epoch": 4.173044925124792,
+      "loss": 0.4923175275325775,
+      "loss_ce": 0.0004657130048144609,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 78574284,
+      "step": 1254
+    },
+    {
+      "epoch": 4.1763727121464225,
+      "grad_norm": 20.463504791259766,
+      "learning_rate": 5e-06,
+      "loss": 0.6105,
+      "num_input_tokens_seen": 78635596,
+      "step": 1255
+    },
+    {
+      "epoch": 4.1763727121464225,
+      "loss": 0.667670488357544,
+      "loss_ce": 0.0014106841990724206,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 78635596,
+      "step": 1255
+    },
+    {
+      "epoch": 4.179700499168053,
+      "grad_norm": 18.627574920654297,
+      "learning_rate": 5e-06,
+      "loss": 0.6598,
+      "num_input_tokens_seen": 78697748,
+      "step": 1256
+    },
+    {
+      "epoch": 4.179700499168053,
+      "loss": 0.6276933550834656,
+      "loss_ce": 0.00021382489649113268,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 78697748,
+      "step": 1256
+    },
+    {
+      "epoch": 4.183028286189684,
+      "grad_norm": 8.834410667419434,
+      "learning_rate": 5e-06,
+      "loss": 0.5655,
+      "num_input_tokens_seen": 78759760,
+      "step": 1257
+    },
+    {
+      "epoch": 4.183028286189684,
+      "loss": 0.6625989079475403,
+      "loss_ce": 0.0004895288147963583,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 78759760,
+      "step": 1257
+    },
+    {
+      "epoch": 4.186356073211314,
+      "grad_norm": 14.236786842346191,
+      "learning_rate": 5e-06,
+      "loss": 0.6359,
+      "num_input_tokens_seen": 78823028,
+      "step": 1258
+    },
+    {
+      "epoch": 4.186356073211314,
+      "loss": 0.6722902059555054,
+      "loss_ce": 4.9012665840564296e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 78823028,
+      "step": 1258
+    },
+    {
+      "epoch": 4.189683860232945,
+      "grad_norm": 83.77409362792969,
+      "learning_rate": 5e-06,
+      "loss": 0.5601,
+      "num_input_tokens_seen": 78885504,
+      "step": 1259
+    },
+    {
+      "epoch": 4.189683860232945,
+      "loss": 0.4171263575553894,
+      "loss_ce": 0.0001341925235465169,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 78885504,
+      "step": 1259
+    },
+    {
+      "epoch": 4.1930116472545755,
+      "grad_norm": 12.761661529541016,
+      "learning_rate": 5e-06,
+      "loss": 0.8203,
+      "num_input_tokens_seen": 78947504,
+      "step": 1260
+    },
+    {
+      "epoch": 4.1930116472545755,
+      "loss": 0.7472410202026367,
+      "loss_ce": 4.86068420286756e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.062255859375,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 78947504,
+      "step": 1260
+    },
+    {
+      "epoch": 4.196339434276206,
+      "grad_norm": 12.434175491333008,
+      "learning_rate": 5e-06,
+      "loss": 0.6015,
+      "num_input_tokens_seen": 79006968,
+      "step": 1261
+    },
+    {
+      "epoch": 4.196339434276206,
+      "loss": 0.5894126296043396,
+      "loss_ce": 0.0007895919261500239,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 79006968,
+      "step": 1261
+    },
+    {
+      "epoch": 4.199667221297837,
+      "grad_norm": 10.954483985900879,
+      "learning_rate": 5e-06,
+      "loss": 0.9019,
+      "num_input_tokens_seen": 79068724,
+      "step": 1262
+    },
+    {
+      "epoch": 4.199667221297837,
+      "loss": 0.8818979859352112,
+      "loss_ce": 0.0005503545398823917,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 79068724,
+      "step": 1262
+    },
+    {
+      "epoch": 4.202995008319467,
+      "grad_norm": 10.366689682006836,
+      "learning_rate": 5e-06,
+      "loss": 0.5979,
+      "num_input_tokens_seen": 79127728,
+      "step": 1263
+    },
+    {
+      "epoch": 4.202995008319467,
+      "loss": 0.7296203970909119,
+      "loss_ce": 6.130624115030514e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 79127728,
+      "step": 1263
+    },
+    {
+      "epoch": 4.206322795341098,
+      "grad_norm": 9.267184257507324,
+      "learning_rate": 5e-06,
+      "loss": 0.7216,
+      "num_input_tokens_seen": 79190896,
+      "step": 1264
+    },
+    {
+      "epoch": 4.206322795341098,
+      "loss": 0.5524642467498779,
+      "loss_ce": 0.00021812476916238666,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 79190896,
+      "step": 1264
+    },
+    {
+      "epoch": 4.209650582362729,
+      "grad_norm": 9.680049896240234,
+      "learning_rate": 5e-06,
+      "loss": 0.7248,
+      "num_input_tokens_seen": 79253700,
+      "step": 1265
+    },
+    {
+      "epoch": 4.209650582362729,
+      "loss": 0.6744685173034668,
+      "loss_ce": 0.00015209820412565023,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 79253700,
+      "step": 1265
+    },
+    {
+      "epoch": 4.212978369384359,
+      "grad_norm": 13.558832168579102,
+      "learning_rate": 5e-06,
+      "loss": 0.8204,
+      "num_input_tokens_seen": 79316324,
+      "step": 1266
+    },
+    {
+      "epoch": 4.212978369384359,
+      "loss": 0.8318663835525513,
+      "loss_ce": 0.00020139635307714343,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 79316324,
+      "step": 1266
+    },
+    {
+      "epoch": 4.21630615640599,
+      "grad_norm": 34.612335205078125,
+      "learning_rate": 5e-06,
+      "loss": 0.6904,
+      "num_input_tokens_seen": 79379112,
+      "step": 1267
+    },
+    {
+      "epoch": 4.21630615640599,
+      "loss": 0.7487878799438477,
+      "loss_ce": 8.608042662672233e-06,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 79379112,
+      "step": 1267
+    },
+    {
+      "epoch": 4.21963394342762,
+      "grad_norm": 13.187520027160645,
+      "learning_rate": 5e-06,
+      "loss": 0.7209,
+      "num_input_tokens_seen": 79441800,
+      "step": 1268
+    },
+    {
+      "epoch": 4.21963394342762,
+      "loss": 0.6884113550186157,
+      "loss_ce": 0.0010334014659747481,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 79441800,
+      "step": 1268
+    },
+    {
+      "epoch": 4.222961730449251,
+      "grad_norm": 29.02869415283203,
+      "learning_rate": 5e-06,
+      "loss": 0.7215,
+      "num_input_tokens_seen": 79503748,
+      "step": 1269
+    },
+    {
+      "epoch": 4.222961730449251,
+      "loss": 0.7619832754135132,
+      "loss_ce": 2.0363593648653477e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.76171875,
+      "num_input_tokens_seen": 79503748,
+      "step": 1269
+    },
+    {
+      "epoch": 4.226289517470882,
+      "grad_norm": 56.70915603637695,
+      "learning_rate": 5e-06,
+      "loss": 0.6429,
+      "num_input_tokens_seen": 79564680,
+      "step": 1270
+    },
+    {
+      "epoch": 4.226289517470882,
+      "loss": 0.7590841054916382,
+      "loss_ce": 5.0929971621371806e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 79564680,
+      "step": 1270
+    },
+    {
+      "epoch": 4.229617304492512,
+      "grad_norm": 16.210630416870117,
+      "learning_rate": 5e-06,
+      "loss": 0.7066,
+      "num_input_tokens_seen": 79626448,
+      "step": 1271
+    },
+    {
+      "epoch": 4.229617304492512,
+      "loss": 0.6228231191635132,
+      "loss_ce": 0.001607277779839933,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 79626448,
+      "step": 1271
+    },
+    {
+      "epoch": 4.232945091514143,
+      "grad_norm": 26.057252883911133,
+      "learning_rate": 5e-06,
+      "loss": 0.6158,
+      "num_input_tokens_seen": 79689628,
+      "step": 1272
+    },
+    {
+      "epoch": 4.232945091514143,
+      "loss": 0.6330506801605225,
+      "loss_ce": 0.00023816811153665185,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 79689628,
+      "step": 1272
+    },
+    {
+      "epoch": 4.2362728785357735,
+      "grad_norm": 8.330951690673828,
+      "learning_rate": 5e-06,
+      "loss": 0.75,
+      "num_input_tokens_seen": 79752856,
+      "step": 1273
+    },
+    {
+      "epoch": 4.2362728785357735,
+      "loss": 0.9343417286872864,
+      "loss_ce": 1.557775613036938e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 79752856,
+      "step": 1273
+    },
+    {
+      "epoch": 4.239600665557404,
+      "grad_norm": 28.429784774780273,
+      "learning_rate": 5e-06,
+      "loss": 0.7537,
+      "num_input_tokens_seen": 79813784,
+      "step": 1274
+    },
+    {
+      "epoch": 4.239600665557404,
+      "loss": 0.5956960916519165,
+      "loss_ce": 0.0004812246188521385,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 79813784,
+      "step": 1274
+    },
+    {
+      "epoch": 4.242928452579035,
+      "grad_norm": 14.092952728271484,
+      "learning_rate": 5e-06,
+      "loss": 0.8371,
+      "num_input_tokens_seen": 79878556,
+      "step": 1275
+    },
+    {
+      "epoch": 4.242928452579035,
+      "loss": 0.9099881052970886,
+      "loss_ce": 7.601917604915798e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.91015625,
+      "num_input_tokens_seen": 79878556,
+      "step": 1275
+    },
+    {
+      "epoch": 4.246256239600665,
+      "grad_norm": 25.563735961914062,
+      "learning_rate": 5e-06,
+      "loss": 0.7424,
+      "num_input_tokens_seen": 79940316,
+      "step": 1276
+    },
+    {
+      "epoch": 4.246256239600665,
+      "loss": 0.7878633737564087,
+      "loss_ce": 2.1567015210166574e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 79940316,
+      "step": 1276
+    },
+    {
+      "epoch": 4.249584026622296,
+      "grad_norm": 16.13930892944336,
+      "learning_rate": 5e-06,
+      "loss": 0.6181,
+      "num_input_tokens_seen": 80002848,
+      "step": 1277
+    },
+    {
+      "epoch": 4.249584026622296,
+      "loss": 0.6337130069732666,
+      "loss_ce": 0.00016809771477710456,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 80002848,
+      "step": 1277
+    },
+    {
+      "epoch": 4.252911813643927,
+      "grad_norm": 13.609673500061035,
+      "learning_rate": 5e-06,
+      "loss": 0.8332,
+      "num_input_tokens_seen": 80065168,
+      "step": 1278
+    },
+    {
+      "epoch": 4.252911813643927,
+      "loss": 0.688213586807251,
+      "loss_ce": 0.00022533058654516935,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 80065168,
+      "step": 1278
+    },
+    {
+      "epoch": 4.256239600665557,
+      "grad_norm": 13.89177417755127,
+      "learning_rate": 5e-06,
+      "loss": 0.7432,
+      "num_input_tokens_seen": 80126552,
+      "step": 1279
+    },
+    {
+      "epoch": 4.256239600665557,
+      "loss": 0.7881263494491577,
+      "loss_ce": 4.041851207148284e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 80126552,
+      "step": 1279
+    },
+    {
+      "epoch": 4.259567387687188,
+      "grad_norm": 9.584406852722168,
+      "learning_rate": 5e-06,
+      "loss": 0.7185,
+      "num_input_tokens_seen": 80189328,
+      "step": 1280
+    },
+    {
+      "epoch": 4.259567387687188,
+      "loss": 0.6353198289871216,
+      "loss_ce": 0.0009204413508996367,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 80189328,
+      "step": 1280
+    },
+    {
+      "epoch": 4.262895174708818,
+      "grad_norm": 8.969870567321777,
+      "learning_rate": 5e-06,
+      "loss": 0.783,
+      "num_input_tokens_seen": 80252808,
+      "step": 1281
+    },
+    {
+      "epoch": 4.262895174708818,
+      "loss": 0.9145519137382507,
+      "loss_ce": 0.0019542670343071222,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 80252808,
+      "step": 1281
+    },
+    {
+      "epoch": 4.266222961730449,
+      "grad_norm": 19.581363677978516,
+      "learning_rate": 5e-06,
+      "loss": 0.5597,
+      "num_input_tokens_seen": 80315084,
+      "step": 1282
+    },
+    {
+      "epoch": 4.266222961730449,
+      "loss": 0.49742501974105835,
+      "loss_ce": 1.9012808479601517e-05,
+      "loss_iou": 0.0751953125,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.498046875,
+      "num_input_tokens_seen": 80315084,
+      "step": 1282
+    },
+    {
+      "epoch": 4.26955074875208,
+      "grad_norm": 18.102149963378906,
+      "learning_rate": 5e-06,
+      "loss": 0.6311,
+      "num_input_tokens_seen": 80378400,
+      "step": 1283
+    },
+    {
+      "epoch": 4.26955074875208,
+      "loss": 0.47788214683532715,
+      "loss_ce": 9.894504910334945e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.478515625,
+      "num_input_tokens_seen": 80378400,
+      "step": 1283
+    },
+    {
+      "epoch": 4.27287853577371,
+      "grad_norm": 14.705218315124512,
+      "learning_rate": 5e-06,
+      "loss": 0.7301,
+      "num_input_tokens_seen": 80440412,
+      "step": 1284
+    },
+    {
+      "epoch": 4.27287853577371,
+      "loss": 0.6811676025390625,
+      "loss_ce": 1.5225450624711812e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 80440412,
+      "step": 1284
+    },
+    {
+      "epoch": 4.276206322795341,
+      "grad_norm": 11.81163501739502,
+      "learning_rate": 5e-06,
+      "loss": 0.7499,
+      "num_input_tokens_seen": 80503416,
+      "step": 1285
+    },
+    {
+      "epoch": 4.276206322795341,
+      "loss": 0.718535304069519,
+      "loss_ce": 2.9458708013407886e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 80503416,
+      "step": 1285
+    },
+    {
+      "epoch": 4.2795341098169715,
+      "grad_norm": 11.332803726196289,
+      "learning_rate": 5e-06,
+      "loss": 0.6372,
+      "num_input_tokens_seen": 80566256,
+      "step": 1286
+    },
+    {
+      "epoch": 4.2795341098169715,
+      "loss": 0.6951990127563477,
+      "loss_ce": 0.0007410263060592115,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 80566256,
+      "step": 1286
+    },
+    {
+      "epoch": 4.282861896838602,
+      "grad_norm": 12.372754096984863,
+      "learning_rate": 5e-06,
+      "loss": 0.6649,
+      "num_input_tokens_seen": 80628540,
+      "step": 1287
+    },
+    {
+      "epoch": 4.282861896838602,
+      "loss": 0.8295961618423462,
+      "loss_ce": 0.0007388241938315332,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 80628540,
+      "step": 1287
+    },
+    {
+      "epoch": 4.286189683860233,
+      "grad_norm": 7.756584167480469,
+      "learning_rate": 5e-06,
+      "loss": 0.6643,
+      "num_input_tokens_seen": 80690812,
+      "step": 1288
+    },
+    {
+      "epoch": 4.286189683860233,
+      "loss": 0.6631507277488708,
+      "loss_ce": 0.001529662637040019,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 80690812,
+      "step": 1288
+    },
+    {
+      "epoch": 4.289517470881863,
+      "grad_norm": 12.011344909667969,
+      "learning_rate": 5e-06,
+      "loss": 0.8091,
+      "num_input_tokens_seen": 80754852,
+      "step": 1289
+    },
+    {
+      "epoch": 4.289517470881863,
+      "loss": 0.5517739057540894,
+      "loss_ce": 0.00026023387908935547,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 80754852,
+      "step": 1289
+    },
+    {
+      "epoch": 4.292845257903494,
+      "grad_norm": 14.856667518615723,
+      "learning_rate": 5e-06,
+      "loss": 0.7544,
+      "num_input_tokens_seen": 80816168,
+      "step": 1290
+    },
+    {
+      "epoch": 4.292845257903494,
+      "loss": 0.5227793455123901,
+      "loss_ce": 0.000226860967813991,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 80816168,
+      "step": 1290
+    },
+    {
+      "epoch": 4.2961730449251245,
+      "grad_norm": 10.020024299621582,
+      "learning_rate": 5e-06,
+      "loss": 0.7188,
+      "num_input_tokens_seen": 80880640,
+      "step": 1291
+    },
+    {
+      "epoch": 4.2961730449251245,
+      "loss": 0.760665237903595,
+      "loss_ce": 0.0022423923946917057,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 80880640,
+      "step": 1291
+    },
+    {
+      "epoch": 4.299500831946755,
+      "grad_norm": 11.497454643249512,
+      "learning_rate": 5e-06,
+      "loss": 0.5628,
+      "num_input_tokens_seen": 80943176,
+      "step": 1292
+    },
+    {
+      "epoch": 4.299500831946755,
+      "loss": 0.687080979347229,
+      "loss_ce": 6.930784729775041e-05,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 80943176,
+      "step": 1292
+    },
+    {
+      "epoch": 4.302828618968386,
+      "grad_norm": 12.758271217346191,
+      "learning_rate": 5e-06,
+      "loss": 0.8416,
+      "num_input_tokens_seen": 81007208,
+      "step": 1293
+    },
+    {
+      "epoch": 4.302828618968386,
+      "loss": 0.7446075677871704,
+      "loss_ce": 0.0004669097252190113,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 81007208,
+      "step": 1293
+    },
+    {
+      "epoch": 4.306156405990016,
+      "grad_norm": 52.66012954711914,
+      "learning_rate": 5e-06,
+      "loss": 0.5894,
+      "num_input_tokens_seen": 81067976,
+      "step": 1294
+    },
+    {
+      "epoch": 4.306156405990016,
+      "loss": 0.37765905261039734,
+      "loss_ce": 0.0007974714390002191,
+      "loss_iou": 0.10693359375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.376953125,
+      "num_input_tokens_seen": 81067976,
+      "step": 1294
+    },
+    {
+      "epoch": 4.309484193011647,
+      "grad_norm": 12.44371509552002,
+      "learning_rate": 5e-06,
+      "loss": 0.5179,
+      "num_input_tokens_seen": 81128904,
+      "step": 1295
+    },
+    {
+      "epoch": 4.309484193011647,
+      "loss": 0.6360405683517456,
+      "loss_ce": 5.427756696008146e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 81128904,
+      "step": 1295
+    },
+    {
+      "epoch": 4.312811980033278,
+      "grad_norm": 24.886842727661133,
+      "learning_rate": 5e-06,
+      "loss": 0.705,
+      "num_input_tokens_seen": 81191592,
+      "step": 1296
+    },
+    {
+      "epoch": 4.312811980033278,
+      "loss": 0.7338913083076477,
+      "loss_ce": 4.5570727706945036e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 81191592,
+      "step": 1296
+    },
+    {
+      "epoch": 4.316139767054908,
+      "grad_norm": 30.665220260620117,
+      "learning_rate": 5e-06,
+      "loss": 0.7969,
+      "num_input_tokens_seen": 81255356,
+      "step": 1297
+    },
+    {
+      "epoch": 4.316139767054908,
+      "loss": 0.7743918895721436,
+      "loss_ce": 0.001564753009006381,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 81255356,
+      "step": 1297
+    },
+    {
+      "epoch": 4.319467554076539,
+      "grad_norm": 29.902305603027344,
+      "learning_rate": 5e-06,
+      "loss": 0.8438,
+      "num_input_tokens_seen": 81318216,
+      "step": 1298
+    },
+    {
+      "epoch": 4.319467554076539,
+      "loss": 0.7068020105361938,
+      "loss_ce": 1.4880570233799517e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.70703125,
+      "num_input_tokens_seen": 81318216,
+      "step": 1298
+    },
+    {
+      "epoch": 4.322795341098169,
+      "grad_norm": 13.050071716308594,
+      "learning_rate": 5e-06,
+      "loss": 0.6442,
+      "num_input_tokens_seen": 81381692,
+      "step": 1299
+    },
+    {
+      "epoch": 4.322795341098169,
+      "loss": 0.3645215630531311,
+      "loss_ce": 1.962293026736006e-05,
+      "loss_iou": 0.1298828125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 81381692,
+      "step": 1299
+    },
+    {
+      "epoch": 4.3261231281198,
+      "grad_norm": 14.20059871673584,
+      "learning_rate": 5e-06,
+      "loss": 0.6968,
+      "num_input_tokens_seen": 81445260,
+      "step": 1300
+    },
+    {
+      "epoch": 4.3261231281198,
+      "loss": 0.9003438353538513,
+      "loss_ce": 0.0004414235008880496,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 81445260,
+      "step": 1300
+    },
+    {
+      "epoch": 4.329450915141431,
+      "grad_norm": 10.232033729553223,
+      "learning_rate": 5e-06,
+      "loss": 0.6684,
+      "num_input_tokens_seen": 81507220,
+      "step": 1301
+    },
+    {
+      "epoch": 4.329450915141431,
+      "loss": 0.922446608543396,
+      "loss_ce": 0.00044959314982406795,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 81507220,
+      "step": 1301
+    },
+    {
+      "epoch": 4.332778702163061,
+      "grad_norm": 8.437857627868652,
+      "learning_rate": 5e-06,
+      "loss": 0.5038,
+      "num_input_tokens_seen": 81570812,
+      "step": 1302
+    },
+    {
+      "epoch": 4.332778702163061,
+      "loss": 0.6514133214950562,
+      "loss_ce": 4.610713222064078e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 81570812,
+      "step": 1302
+    },
+    {
+      "epoch": 4.336106489184692,
+      "grad_norm": 9.5712308883667,
+      "learning_rate": 5e-06,
+      "loss": 0.9257,
+      "num_input_tokens_seen": 81634604,
+      "step": 1303
+    },
+    {
+      "epoch": 4.336106489184692,
+      "loss": 1.1645619869232178,
+      "loss_ce": 0.0007436010637320578,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.08203125,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 81634604,
+      "step": 1303
+    },
+    {
+      "epoch": 4.3394342762063225,
+      "grad_norm": 24.645009994506836,
+      "learning_rate": 5e-06,
+      "loss": 0.8503,
+      "num_input_tokens_seen": 81698128,
+      "step": 1304
+    },
+    {
+      "epoch": 4.3394342762063225,
+      "loss": 0.7591370940208435,
+      "loss_ce": 0.0015687549021095037,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 81698128,
+      "step": 1304
+    },
+    {
+      "epoch": 4.342762063227953,
+      "grad_norm": 18.755535125732422,
+      "learning_rate": 5e-06,
+      "loss": 0.8487,
+      "num_input_tokens_seen": 81760788,
+      "step": 1305
+    },
+    {
+      "epoch": 4.342762063227953,
+      "loss": 0.7894872426986694,
+      "loss_ce": 0.001157181104645133,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 81760788,
+      "step": 1305
+    },
+    {
+      "epoch": 4.346089850249584,
+      "grad_norm": 11.36084270477295,
+      "learning_rate": 5e-06,
+      "loss": 0.8839,
+      "num_input_tokens_seen": 81823160,
+      "step": 1306
+    },
+    {
+      "epoch": 4.346089850249584,
+      "loss": 0.6701819896697998,
+      "loss_ce": 1.5993844499462284e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 81823160,
+      "step": 1306
+    },
+    {
+      "epoch": 4.349417637271214,
+      "grad_norm": 14.648335456848145,
+      "learning_rate": 5e-06,
+      "loss": 0.96,
+      "num_input_tokens_seen": 81885560,
+      "step": 1307
+    },
+    {
+      "epoch": 4.349417637271214,
+      "loss": 1.0643436908721924,
+      "loss_ce": 0.001599527895450592,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.083984375,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 81885560,
+      "step": 1307
+    },
+    {
+      "epoch": 4.352745424292845,
+      "grad_norm": 15.327980995178223,
+      "learning_rate": 5e-06,
+      "loss": 0.5124,
+      "num_input_tokens_seen": 81947560,
+      "step": 1308
+    },
+    {
+      "epoch": 4.352745424292845,
+      "loss": 0.28150349855422974,
+      "loss_ce": 0.0004366114444565028,
+      "loss_iou": 0.04931640625,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 81947560,
+      "step": 1308
+    },
+    {
+      "epoch": 4.356073211314476,
+      "grad_norm": 12.198101043701172,
+      "learning_rate": 5e-06,
+      "loss": 0.6174,
+      "num_input_tokens_seen": 82008732,
+      "step": 1309
+    },
+    {
+      "epoch": 4.356073211314476,
+      "loss": 0.6670141816139221,
+      "loss_ce": 0.001486798282712698,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 82008732,
+      "step": 1309
+    },
+    {
+      "epoch": 4.359400998336106,
+      "grad_norm": 12.514129638671875,
+      "learning_rate": 5e-06,
+      "loss": 0.6429,
+      "num_input_tokens_seen": 82071748,
+      "step": 1310
+    },
+    {
+      "epoch": 4.359400998336106,
+      "loss": 0.5801935195922852,
+      "loss_ce": 5.4305866797221825e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 82071748,
+      "step": 1310
+    },
+    {
+      "epoch": 4.362728785357737,
+      "grad_norm": 12.292180061340332,
+      "learning_rate": 5e-06,
+      "loss": 0.7611,
+      "num_input_tokens_seen": 82135220,
+      "step": 1311
+    },
+    {
+      "epoch": 4.362728785357737,
+      "loss": 0.6303189992904663,
+      "loss_ce": 0.00043624168029055,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 82135220,
+      "step": 1311
+    },
+    {
+      "epoch": 4.366056572379367,
+      "grad_norm": 19.755245208740234,
+      "learning_rate": 5e-06,
+      "loss": 0.9127,
+      "num_input_tokens_seen": 82198120,
+      "step": 1312
+    },
+    {
+      "epoch": 4.366056572379367,
+      "loss": 0.985588788986206,
+      "loss_ce": 0.0004814087296836078,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 82198120,
+      "step": 1312
+    },
+    {
+      "epoch": 4.369384359400998,
+      "grad_norm": 16.86263656616211,
+      "learning_rate": 5e-06,
+      "loss": 0.6468,
+      "num_input_tokens_seen": 82260076,
+      "step": 1313
+    },
+    {
+      "epoch": 4.369384359400998,
+      "loss": 0.682517409324646,
+      "loss_ce": 0.000388490705518052,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.68359375,
+      "num_input_tokens_seen": 82260076,
+      "step": 1313
+    },
+    {
+      "epoch": 4.372712146422629,
+      "grad_norm": 9.756426811218262,
+      "learning_rate": 5e-06,
+      "loss": 0.9654,
+      "num_input_tokens_seen": 82322316,
+      "step": 1314
+    },
+    {
+      "epoch": 4.372712146422629,
+      "loss": 0.9413772821426392,
+      "loss_ce": 0.00021517441200558096,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 82322316,
+      "step": 1314
+    },
+    {
+      "epoch": 4.376039933444259,
+      "grad_norm": 11.446276664733887,
+      "learning_rate": 5e-06,
+      "loss": 0.6246,
+      "num_input_tokens_seen": 82386100,
+      "step": 1315
+    },
+    {
+      "epoch": 4.376039933444259,
+      "loss": 0.7032071352005005,
+      "loss_ce": 0.0006314606289379299,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 82386100,
+      "step": 1315
+    },
+    {
+      "epoch": 4.37936772046589,
+      "grad_norm": 11.805602073669434,
+      "learning_rate": 5e-06,
+      "loss": 0.8133,
+      "num_input_tokens_seen": 82448276,
+      "step": 1316
+    },
+    {
+      "epoch": 4.37936772046589,
+      "loss": 0.5263853669166565,
+      "loss_ce": 1.8209135305369273e-05,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 82448276,
+      "step": 1316
+    },
+    {
+      "epoch": 4.3826955074875205,
+      "grad_norm": 14.017489433288574,
+      "learning_rate": 5e-06,
+      "loss": 0.7262,
+      "num_input_tokens_seen": 82509572,
+      "step": 1317
+    },
+    {
+      "epoch": 4.3826955074875205,
+      "loss": 0.7585628032684326,
+      "loss_ce": 0.0005061531555838883,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 82509572,
+      "step": 1317
+    },
+    {
+      "epoch": 4.386023294509151,
+      "grad_norm": 9.340023040771484,
+      "learning_rate": 5e-06,
+      "loss": 0.9733,
+      "num_input_tokens_seen": 82572988,
+      "step": 1318
+    },
+    {
+      "epoch": 4.386023294509151,
+      "loss": 0.9580419063568115,
+      "loss_ce": 0.00015617434110026807,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.95703125,
+      "num_input_tokens_seen": 82572988,
+      "step": 1318
+    },
+    {
+      "epoch": 4.389351081530782,
+      "grad_norm": 23.777090072631836,
+      "learning_rate": 5e-06,
+      "loss": 0.6908,
+      "num_input_tokens_seen": 82635976,
+      "step": 1319
+    },
+    {
+      "epoch": 4.389351081530782,
+      "loss": 0.8914474248886108,
+      "loss_ce": 0.0010665201116353273,
+      "loss_iou": 0.34375,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.890625,
+      "num_input_tokens_seen": 82635976,
+      "step": 1319
+    },
+    {
+      "epoch": 4.392678868552412,
+      "grad_norm": 22.563859939575195,
+      "learning_rate": 5e-06,
+      "loss": 0.6377,
+      "num_input_tokens_seen": 82700040,
+      "step": 1320
+    },
+    {
+      "epoch": 4.392678868552412,
+      "loss": 0.6494787931442261,
+      "loss_ce": 0.002262030728161335,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 82700040,
+      "step": 1320
+    },
+    {
+      "epoch": 4.396006655574043,
+      "grad_norm": 16.087520599365234,
+      "learning_rate": 5e-06,
+      "loss": 0.677,
+      "num_input_tokens_seen": 82763308,
+      "step": 1321
+    },
+    {
+      "epoch": 4.396006655574043,
+      "loss": 0.8195084929466248,
+      "loss_ce": 0.0006608610274270177,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 82763308,
+      "step": 1321
+    },
+    {
+      "epoch": 4.3993344425956735,
+      "grad_norm": 15.090987205505371,
+      "learning_rate": 5e-06,
+      "loss": 0.6979,
+      "num_input_tokens_seen": 82826744,
+      "step": 1322
+    },
+    {
+      "epoch": 4.3993344425956735,
+      "loss": 0.4208059310913086,
+      "loss_ce": 2.9584578442154452e-05,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 82826744,
+      "step": 1322
+    },
+    {
+      "epoch": 4.402662229617304,
+      "grad_norm": 14.373215675354004,
+      "learning_rate": 5e-06,
+      "loss": 0.7911,
+      "num_input_tokens_seen": 82889604,
+      "step": 1323
+    },
+    {
+      "epoch": 4.402662229617304,
+      "loss": 0.6858798265457153,
+      "loss_ce": 0.0013094794703647494,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.68359375,
+      "num_input_tokens_seen": 82889604,
+      "step": 1323
+    },
+    {
+      "epoch": 4.405990016638935,
+      "grad_norm": 29.24193572998047,
+      "learning_rate": 5e-06,
+      "loss": 0.9223,
+      "num_input_tokens_seen": 82952864,
+      "step": 1324
+    },
+    {
+      "epoch": 4.405990016638935,
+      "loss": 1.03691828250885,
+      "loss_ce": 5.295296432450414e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 82952864,
+      "step": 1324
+    },
+    {
+      "epoch": 4.409317803660565,
+      "grad_norm": 14.150858879089355,
+      "learning_rate": 5e-06,
+      "loss": 0.7532,
+      "num_input_tokens_seen": 83015712,
+      "step": 1325
+    },
+    {
+      "epoch": 4.409317803660565,
+      "loss": 0.8280280232429504,
+      "loss_ce": 0.00020820634381379932,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 83015712,
+      "step": 1325
+    },
+    {
+      "epoch": 4.412645590682196,
+      "grad_norm": 29.858686447143555,
+      "learning_rate": 5e-06,
+      "loss": 0.6685,
+      "num_input_tokens_seen": 83078676,
+      "step": 1326
+    },
+    {
+      "epoch": 4.412645590682196,
+      "loss": 0.7052803039550781,
+      "loss_ce": 0.00044628471368923783,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 83078676,
+      "step": 1326
+    },
+    {
+      "epoch": 4.415973377703827,
+      "grad_norm": 15.250243186950684,
+      "learning_rate": 5e-06,
+      "loss": 0.6323,
+      "num_input_tokens_seen": 83141596,
+      "step": 1327
+    },
+    {
+      "epoch": 4.415973377703827,
+      "loss": 0.8748631477355957,
+      "loss_ce": 0.00022933242144063115,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.875,
+      "num_input_tokens_seen": 83141596,
+      "step": 1327
+    },
+    {
+      "epoch": 4.419301164725457,
+      "grad_norm": 17.645246505737305,
+      "learning_rate": 5e-06,
+      "loss": 0.7055,
+      "num_input_tokens_seen": 83202652,
+      "step": 1328
+    },
+    {
+      "epoch": 4.419301164725457,
+      "loss": 0.8196154832839966,
+      "loss_ce": 3.535650466801599e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 83202652,
+      "step": 1328
+    },
+    {
+      "epoch": 4.422628951747088,
+      "grad_norm": 10.930619239807129,
+      "learning_rate": 5e-06,
+      "loss": 0.5952,
+      "num_input_tokens_seen": 83266684,
+      "step": 1329
+    },
+    {
+      "epoch": 4.422628951747088,
+      "loss": 0.5401097536087036,
+      "loss_ce": 7.067422848194838e-05,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 83266684,
+      "step": 1329
+    },
+    {
+      "epoch": 4.425956738768718,
+      "grad_norm": 21.14279556274414,
+      "learning_rate": 5e-06,
+      "loss": 0.7999,
+      "num_input_tokens_seen": 83330424,
+      "step": 1330
+    },
+    {
+      "epoch": 4.425956738768718,
+      "loss": 0.5837583541870117,
+      "loss_ce": 1.807482840376906e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 83330424,
+      "step": 1330
+    },
+    {
+      "epoch": 4.429284525790349,
+      "grad_norm": 18.78822135925293,
+      "learning_rate": 5e-06,
+      "loss": 0.7177,
+      "num_input_tokens_seen": 83394464,
+      "step": 1331
+    },
+    {
+      "epoch": 4.429284525790349,
+      "loss": 0.7250956296920776,
+      "loss_ce": 0.00024215054872911423,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 83394464,
+      "step": 1331
+    },
+    {
+      "epoch": 4.43261231281198,
+      "grad_norm": 17.859045028686523,
+      "learning_rate": 5e-06,
+      "loss": 0.5746,
+      "num_input_tokens_seen": 83456320,
+      "step": 1332
+    },
+    {
+      "epoch": 4.43261231281198,
+      "loss": 0.5926786065101624,
+      "loss_ce": 2.7240082999924198e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 83456320,
+      "step": 1332
+    },
+    {
+      "epoch": 4.43594009983361,
+      "grad_norm": 28.78057861328125,
+      "learning_rate": 5e-06,
+      "loss": 0.725,
+      "num_input_tokens_seen": 83519880,
+      "step": 1333
+    },
+    {
+      "epoch": 4.43594009983361,
+      "loss": 0.6718736886978149,
+      "loss_ce": 0.0007310921791940928,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 83519880,
+      "step": 1333
+    },
+    {
+      "epoch": 4.439267886855241,
+      "grad_norm": 13.343917846679688,
+      "learning_rate": 5e-06,
+      "loss": 0.6955,
+      "num_input_tokens_seen": 83581528,
+      "step": 1334
+    },
+    {
+      "epoch": 4.439267886855241,
+      "loss": 0.6046224236488342,
+      "loss_ce": 0.00013025110820308328,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 83581528,
+      "step": 1334
+    },
+    {
+      "epoch": 4.4425956738768715,
+      "grad_norm": 24.18421745300293,
+      "learning_rate": 5e-06,
+      "loss": 0.9336,
+      "num_input_tokens_seen": 83645572,
+      "step": 1335
+    },
+    {
+      "epoch": 4.4425956738768715,
+      "loss": 0.9804202318191528,
+      "loss_ce": 0.00019555243488866836,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.98046875,
+      "num_input_tokens_seen": 83645572,
+      "step": 1335
+    },
+    {
+      "epoch": 4.445923460898502,
+      "grad_norm": 19.95655632019043,
+      "learning_rate": 5e-06,
+      "loss": 0.6435,
+      "num_input_tokens_seen": 83708536,
+      "step": 1336
+    },
+    {
+      "epoch": 4.445923460898502,
+      "loss": 0.7947807312011719,
+      "loss_ce": 0.00046916649444028735,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 83708536,
+      "step": 1336
+    },
+    {
+      "epoch": 4.449251247920133,
+      "grad_norm": 13.971914291381836,
+      "learning_rate": 5e-06,
+      "loss": 0.6339,
+      "num_input_tokens_seen": 83771940,
+      "step": 1337
+    },
+    {
+      "epoch": 4.449251247920133,
+      "loss": 0.627832293510437,
+      "loss_ce": 2.4645694793434814e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 83771940,
+      "step": 1337
+    },
+    {
+      "epoch": 4.452579034941763,
+      "grad_norm": 28.279212951660156,
+      "learning_rate": 5e-06,
+      "loss": 0.5821,
+      "num_input_tokens_seen": 83833204,
+      "step": 1338
+    },
+    {
+      "epoch": 4.452579034941763,
+      "loss": 0.6125773191452026,
+      "loss_ce": 0.00015060522127896547,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 83833204,
+      "step": 1338
+    },
+    {
+      "epoch": 4.455906821963394,
+      "grad_norm": 24.268396377563477,
+      "learning_rate": 5e-06,
+      "loss": 0.82,
+      "num_input_tokens_seen": 83895096,
+      "step": 1339
+    },
+    {
+      "epoch": 4.455906821963394,
+      "loss": 0.8381412029266357,
+      "loss_ce": 6.506794306915253e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.83984375,
+      "num_input_tokens_seen": 83895096,
+      "step": 1339
+    },
+    {
+      "epoch": 4.4592346089850246,
+      "grad_norm": 8.350409507751465,
+      "learning_rate": 5e-06,
+      "loss": 0.8535,
+      "num_input_tokens_seen": 83957376,
+      "step": 1340
+    },
+    {
+      "epoch": 4.4592346089850246,
+      "loss": 1.058552861213684,
+      "loss_ce": 0.0008135715615935624,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.060546875,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 83957376,
+      "step": 1340
+    },
+    {
+      "epoch": 4.462562396006655,
+      "grad_norm": 17.044376373291016,
+      "learning_rate": 5e-06,
+      "loss": 0.767,
+      "num_input_tokens_seen": 84020204,
+      "step": 1341
+    },
+    {
+      "epoch": 4.462562396006655,
+      "loss": 0.725034236907959,
+      "loss_ce": 0.0011573644587770104,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 84020204,
+      "step": 1341
+    },
+    {
+      "epoch": 4.465890183028286,
+      "grad_norm": 11.960426330566406,
+      "learning_rate": 5e-06,
+      "loss": 0.9019,
+      "num_input_tokens_seen": 84085084,
+      "step": 1342
+    },
+    {
+      "epoch": 4.465890183028286,
+      "loss": 0.8089953660964966,
+      "loss_ce": 0.0013781640445813537,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.80859375,
+      "num_input_tokens_seen": 84085084,
+      "step": 1342
+    },
+    {
+      "epoch": 4.469217970049916,
+      "grad_norm": 13.375765800476074,
+      "learning_rate": 5e-06,
+      "loss": 0.6083,
+      "num_input_tokens_seen": 84147448,
+      "step": 1343
+    },
+    {
+      "epoch": 4.469217970049916,
+      "loss": 0.7444367408752441,
+      "loss_ce": 0.0001740536536090076,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 84147448,
+      "step": 1343
+    },
+    {
+      "epoch": 4.472545757071547,
+      "grad_norm": 24.798673629760742,
+      "learning_rate": 5e-06,
+      "loss": 1.0596,
+      "num_input_tokens_seen": 84211792,
+      "step": 1344
+    },
+    {
+      "epoch": 4.472545757071547,
+      "loss": 1.0900293588638306,
+      "loss_ce": 0.001162198605015874,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.07080078125,
+      "loss_xval": 1.0859375,
+      "num_input_tokens_seen": 84211792,
+      "step": 1344
+    },
+    {
+      "epoch": 4.475873544093178,
+      "grad_norm": 46.3792610168457,
+      "learning_rate": 5e-06,
+      "loss": 0.7852,
+      "num_input_tokens_seen": 84274836,
+      "step": 1345
+    },
+    {
+      "epoch": 4.475873544093178,
+      "loss": 0.5791232585906982,
+      "loss_ce": 2.174517430830747e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 84274836,
+      "step": 1345
+    },
+    {
+      "epoch": 4.479201331114808,
+      "grad_norm": 19.27472496032715,
+      "learning_rate": 5e-06,
+      "loss": 0.5985,
+      "num_input_tokens_seen": 84338324,
+      "step": 1346
+    },
+    {
+      "epoch": 4.479201331114808,
+      "loss": 0.7217234373092651,
+      "loss_ce": 0.0005319793708622456,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 84338324,
+      "step": 1346
+    },
+    {
+      "epoch": 4.482529118136439,
+      "grad_norm": 20.793636322021484,
+      "learning_rate": 5e-06,
+      "loss": 0.7234,
+      "num_input_tokens_seen": 84400104,
+      "step": 1347
+    },
+    {
+      "epoch": 4.482529118136439,
+      "loss": 0.6533622741699219,
+      "loss_ce": 0.00040814554085955024,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 84400104,
+      "step": 1347
+    },
+    {
+      "epoch": 4.4858569051580695,
+      "grad_norm": 12.040908813476562,
+      "learning_rate": 5e-06,
+      "loss": 0.815,
+      "num_input_tokens_seen": 84462732,
+      "step": 1348
+    },
+    {
+      "epoch": 4.4858569051580695,
+      "loss": 0.7836884260177612,
+      "loss_ce": 0.00048530942876823246,
+      "loss_iou": 0.265625,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 84462732,
+      "step": 1348
+    },
+    {
+      "epoch": 4.4891846921797,
+      "grad_norm": 18.848194122314453,
+      "learning_rate": 5e-06,
+      "loss": 0.8167,
+      "num_input_tokens_seen": 84526312,
+      "step": 1349
+    },
+    {
+      "epoch": 4.4891846921797,
+      "loss": 0.8186991214752197,
+      "loss_ce": 0.0005839316290803254,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 84526312,
+      "step": 1349
+    },
+    {
+      "epoch": 4.492512479201331,
+      "grad_norm": 16.940589904785156,
+      "learning_rate": 5e-06,
+      "loss": 0.8814,
+      "num_input_tokens_seen": 84589304,
+      "step": 1350
+    },
+    {
+      "epoch": 4.492512479201331,
+      "loss": 0.7323105335235596,
+      "loss_ce": 0.0004990027518942952,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 84589304,
+      "step": 1350
+    },
+    {
+      "epoch": 4.495840266222961,
+      "grad_norm": 12.424930572509766,
+      "learning_rate": 5e-06,
+      "loss": 0.6786,
+      "num_input_tokens_seen": 84653056,
+      "step": 1351
+    },
+    {
+      "epoch": 4.495840266222961,
+      "loss": 0.594731330871582,
+      "loss_ce": 4.724020072899293e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 84653056,
+      "step": 1351
+    },
+    {
+      "epoch": 4.499168053244592,
+      "grad_norm": 8.971871376037598,
+      "learning_rate": 5e-06,
+      "loss": 0.719,
+      "num_input_tokens_seen": 84715604,
+      "step": 1352
+    },
+    {
+      "epoch": 4.499168053244592,
+      "loss": 0.7548971176147461,
+      "loss_ce": 0.00025852146791294217,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 84715604,
+      "step": 1352
+    },
+    {
+      "epoch": 4.5024958402662225,
+      "grad_norm": 25.51188087463379,
+      "learning_rate": 5e-06,
+      "loss": 0.5393,
+      "num_input_tokens_seen": 84778444,
+      "step": 1353
+    },
+    {
+      "epoch": 4.5024958402662225,
+      "loss": 0.4582330584526062,
+      "loss_ce": 4.212010753690265e-05,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.458984375,
+      "num_input_tokens_seen": 84778444,
+      "step": 1353
+    },
+    {
+      "epoch": 4.505823627287853,
+      "grad_norm": 28.863351821899414,
+      "learning_rate": 5e-06,
+      "loss": 0.586,
+      "num_input_tokens_seen": 84841420,
+      "step": 1354
+    },
+    {
+      "epoch": 4.505823627287853,
+      "loss": 0.7277034521102905,
+      "loss_ce": 0.0015376773662865162,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 84841420,
+      "step": 1354
+    },
+    {
+      "epoch": 4.509151414309484,
+      "grad_norm": 11.419045448303223,
+      "learning_rate": 5e-06,
+      "loss": 0.9489,
+      "num_input_tokens_seen": 84905164,
+      "step": 1355
+    },
+    {
+      "epoch": 4.509151414309484,
+      "loss": 0.9230769872665405,
+      "loss_ce": 0.00022541567159350961,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 84905164,
+      "step": 1355
+    },
+    {
+      "epoch": 4.512479201331114,
+      "grad_norm": 8.96274471282959,
+      "learning_rate": 5e-06,
+      "loss": 0.7418,
+      "num_input_tokens_seen": 84969012,
+      "step": 1356
+    },
+    {
+      "epoch": 4.512479201331114,
+      "loss": 0.7852834463119507,
+      "loss_ce": 0.0008596146362833679,
+      "loss_iou": 0.25,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 84969012,
+      "step": 1356
+    },
+    {
+      "epoch": 4.515806988352745,
+      "grad_norm": 8.02625560760498,
+      "learning_rate": 5e-06,
+      "loss": 0.6062,
+      "num_input_tokens_seen": 85031800,
+      "step": 1357
+    },
+    {
+      "epoch": 4.515806988352745,
+      "loss": 0.5827827453613281,
+      "loss_ce": 1.9044146029045805e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 85031800,
+      "step": 1357
+    },
+    {
+      "epoch": 4.519134775374376,
+      "grad_norm": 46.9227409362793,
+      "learning_rate": 5e-06,
+      "loss": 0.6798,
+      "num_input_tokens_seen": 85095404,
+      "step": 1358
+    },
+    {
+      "epoch": 4.519134775374376,
+      "loss": 0.7685559988021851,
+      "loss_ce": 0.00012336293002590537,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 85095404,
+      "step": 1358
+    },
+    {
+      "epoch": 4.522462562396006,
+      "grad_norm": 34.97386932373047,
+      "learning_rate": 5e-06,
+      "loss": 0.6895,
+      "num_input_tokens_seen": 85158752,
+      "step": 1359
+    },
+    {
+      "epoch": 4.522462562396006,
+      "loss": 0.8442423343658447,
+      "loss_ce": 4.0554509723733645e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 85158752,
+      "step": 1359
+    },
+    {
+      "epoch": 4.525790349417637,
+      "grad_norm": 44.28080749511719,
+      "learning_rate": 5e-06,
+      "loss": 1.0204,
+      "num_input_tokens_seen": 85222508,
+      "step": 1360
+    },
+    {
+      "epoch": 4.525790349417637,
+      "loss": 0.9339932203292847,
+      "loss_ce": 3.327502417960204e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.93359375,
+      "num_input_tokens_seen": 85222508,
+      "step": 1360
+    },
+    {
+      "epoch": 4.529118136439267,
+      "grad_norm": 6.506110191345215,
+      "learning_rate": 5e-06,
+      "loss": 0.6121,
+      "num_input_tokens_seen": 85283528,
+      "step": 1361
+    },
+    {
+      "epoch": 4.529118136439267,
+      "loss": 0.5493087768554688,
+      "loss_ce": 0.00023652684467379004,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 85283528,
+      "step": 1361
+    },
+    {
+      "epoch": 4.532445923460898,
+      "grad_norm": 10.900887489318848,
+      "learning_rate": 5e-06,
+      "loss": 0.7382,
+      "num_input_tokens_seen": 85347012,
+      "step": 1362
+    },
+    {
+      "epoch": 4.532445923460898,
+      "loss": 0.7557834386825562,
+      "loss_ce": 0.000931144692003727,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 85347012,
+      "step": 1362
+    },
+    {
+      "epoch": 4.535773710482529,
+      "grad_norm": 35.10896301269531,
+      "learning_rate": 5e-06,
+      "loss": 0.766,
+      "num_input_tokens_seen": 85409800,
+      "step": 1363
+    },
+    {
+      "epoch": 4.535773710482529,
+      "loss": 0.899711549282074,
+      "loss_ce": 0.0012740622041746974,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 85409800,
+      "step": 1363
+    },
+    {
+      "epoch": 4.539101497504159,
+      "grad_norm": 34.48576354980469,
+      "learning_rate": 5e-06,
+      "loss": 0.8018,
+      "num_input_tokens_seen": 85471848,
+      "step": 1364
+    },
+    {
+      "epoch": 4.539101497504159,
+      "loss": 0.7342795133590698,
+      "loss_ce": 0.0008810532744973898,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 85471848,
+      "step": 1364
+    },
+    {
+      "epoch": 4.54242928452579,
+      "grad_norm": 8.720282554626465,
+      "learning_rate": 5e-06,
+      "loss": 0.6187,
+      "num_input_tokens_seen": 85535028,
+      "step": 1365
+    },
+    {
+      "epoch": 4.54242928452579,
+      "loss": 0.6450221538543701,
+      "loss_ce": 2.5852832550299354e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 85535028,
+      "step": 1365
+    },
+    {
+      "epoch": 4.5457570715474205,
+      "grad_norm": 31.563127517700195,
+      "learning_rate": 5e-06,
+      "loss": 0.6799,
+      "num_input_tokens_seen": 85598856,
+      "step": 1366
+    },
+    {
+      "epoch": 4.5457570715474205,
+      "loss": 0.5118007659912109,
+      "loss_ce": 0.0013026673113927245,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 85598856,
+      "step": 1366
+    },
+    {
+      "epoch": 4.549084858569051,
+      "grad_norm": 16.306110382080078,
+      "learning_rate": 5e-06,
+      "loss": 0.4548,
+      "num_input_tokens_seen": 85661384,
+      "step": 1367
+    },
+    {
+      "epoch": 4.549084858569051,
+      "loss": 0.43176573514938354,
+      "loss_ce": 3.0542746571882162e-06,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 85661384,
+      "step": 1367
+    },
+    {
+      "epoch": 4.552412645590682,
+      "grad_norm": 11.156771659851074,
+      "learning_rate": 5e-06,
+      "loss": 0.6683,
+      "num_input_tokens_seen": 85724720,
+      "step": 1368
+    },
+    {
+      "epoch": 4.552412645590682,
+      "loss": 0.8609325289726257,
+      "loss_ce": 0.0008250840473920107,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 85724720,
+      "step": 1368
+    },
+    {
+      "epoch": 4.555740432612312,
+      "grad_norm": 11.832956314086914,
+      "learning_rate": 5e-06,
+      "loss": 0.4961,
+      "num_input_tokens_seen": 85786512,
+      "step": 1369
+    },
+    {
+      "epoch": 4.555740432612312,
+      "loss": 0.27058541774749756,
+      "loss_ce": 1.6586316633038223e-05,
+      "loss_iou": 0.0810546875,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 85786512,
+      "step": 1369
+    },
+    {
+      "epoch": 4.559068219633943,
+      "grad_norm": 10.155105590820312,
+      "learning_rate": 5e-06,
+      "loss": 0.7228,
+      "num_input_tokens_seen": 85849112,
+      "step": 1370
+    },
+    {
+      "epoch": 4.559068219633943,
+      "loss": 0.6971813440322876,
+      "loss_ce": 0.0015025895554572344,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 85849112,
+      "step": 1370
+    },
+    {
+      "epoch": 4.5623960066555735,
+      "grad_norm": 13.69803524017334,
+      "learning_rate": 5e-06,
+      "loss": 0.6738,
+      "num_input_tokens_seen": 85912496,
+      "step": 1371
+    },
+    {
+      "epoch": 4.5623960066555735,
+      "loss": 0.6464178562164307,
+      "loss_ce": 0.00042175239650532603,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 85912496,
+      "step": 1371
+    },
+    {
+      "epoch": 4.565723793677205,
+      "grad_norm": 14.567577362060547,
+      "learning_rate": 5e-06,
+      "loss": 0.4595,
+      "num_input_tokens_seen": 85974008,
+      "step": 1372
+    },
+    {
+      "epoch": 4.565723793677205,
+      "loss": 0.5371776223182678,
+      "loss_ce": 0.00019030201656278223,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 85974008,
+      "step": 1372
+    },
+    {
+      "epoch": 4.569051580698836,
+      "grad_norm": 11.320833206176758,
+      "learning_rate": 5e-06,
+      "loss": 0.4038,
+      "num_input_tokens_seen": 86036360,
+      "step": 1373
+    },
+    {
+      "epoch": 4.569051580698836,
+      "loss": 0.4081704616546631,
+      "loss_ce": 0.00021149149688426405,
+      "loss_iou": 0.1162109375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 86036360,
+      "step": 1373
+    },
+    {
+      "epoch": 4.572379367720466,
+      "grad_norm": 8.902009010314941,
+      "learning_rate": 5e-06,
+      "loss": 0.6014,
+      "num_input_tokens_seen": 86098024,
+      "step": 1374
+    },
+    {
+      "epoch": 4.572379367720466,
+      "loss": 0.531557559967041,
+      "loss_ce": 2.3850705019867746e-06,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 86098024,
+      "step": 1374
+    },
+    {
+      "epoch": 4.575707154742097,
+      "grad_norm": 8.554510116577148,
+      "learning_rate": 5e-06,
+      "loss": 0.8192,
+      "num_input_tokens_seen": 86160932,
+      "step": 1375
+    },
+    {
+      "epoch": 4.575707154742097,
+      "loss": 1.052072525024414,
+      "loss_ce": 0.0005588348722085357,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.057861328125,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 86160932,
+      "step": 1375
+    },
+    {
+      "epoch": 4.5790349417637275,
+      "grad_norm": 13.125633239746094,
+      "learning_rate": 5e-06,
+      "loss": 0.6643,
+      "num_input_tokens_seen": 86223500,
+      "step": 1376
+    },
+    {
+      "epoch": 4.5790349417637275,
+      "loss": 0.6385508179664612,
+      "loss_ce": 0.00015362344856839627,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 86223500,
+      "step": 1376
+    },
+    {
+      "epoch": 4.582362728785358,
+      "grad_norm": 8.183598518371582,
+      "learning_rate": 5e-06,
+      "loss": 0.5159,
+      "num_input_tokens_seen": 86284812,
+      "step": 1377
+    },
+    {
+      "epoch": 4.582362728785358,
+      "loss": 0.6321967840194702,
+      "loss_ce": 0.00036084238672629,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 86284812,
+      "step": 1377
+    },
+    {
+      "epoch": 4.585690515806989,
+      "grad_norm": 12.763681411743164,
+      "learning_rate": 5e-06,
+      "loss": 0.6332,
+      "num_input_tokens_seen": 86348508,
+      "step": 1378
+    },
+    {
+      "epoch": 4.585690515806989,
+      "loss": 0.5611585378646851,
+      "loss_ce": 0.00012342022091615945,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 86348508,
+      "step": 1378
+    },
+    {
+      "epoch": 4.589018302828619,
+      "grad_norm": 6.641970634460449,
+      "learning_rate": 5e-06,
+      "loss": 0.5477,
+      "num_input_tokens_seen": 86409256,
+      "step": 1379
+    },
+    {
+      "epoch": 4.589018302828619,
+      "loss": 0.5002583265304565,
+      "loss_ce": 1.4131117495708168e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 86409256,
+      "step": 1379
+    },
+    {
+      "epoch": 4.59234608985025,
+      "grad_norm": 26.70740509033203,
+      "learning_rate": 5e-06,
+      "loss": 0.9453,
+      "num_input_tokens_seen": 86473068,
+      "step": 1380
+    },
+    {
+      "epoch": 4.59234608985025,
+      "loss": 1.0360589027404785,
+      "loss_ce": 4.8118923587026075e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.059814453125,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 86473068,
+      "step": 1380
+    },
+    {
+      "epoch": 4.595673876871881,
+      "grad_norm": 8.058368682861328,
+      "learning_rate": 5e-06,
+      "loss": 0.7622,
+      "num_input_tokens_seen": 86535376,
+      "step": 1381
+    },
+    {
+      "epoch": 4.595673876871881,
+      "loss": 0.6889675259590149,
+      "loss_ce": 2.6555962904240005e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 86535376,
+      "step": 1381
+    },
+    {
+      "epoch": 4.599001663893511,
+      "grad_norm": 18.06012725830078,
+      "learning_rate": 5e-06,
+      "loss": 0.7193,
+      "num_input_tokens_seen": 86598728,
+      "step": 1382
+    },
+    {
+      "epoch": 4.599001663893511,
+      "loss": 0.5724185705184937,
+      "loss_ce": 0.00015296557103283703,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 86598728,
+      "step": 1382
+    },
+    {
+      "epoch": 4.602329450915142,
+      "grad_norm": 6.92100715637207,
+      "learning_rate": 5e-06,
+      "loss": 0.743,
+      "num_input_tokens_seen": 86661740,
+      "step": 1383
+    },
+    {
+      "epoch": 4.602329450915142,
+      "loss": 0.6485388875007629,
+      "loss_ce": 0.000711729924660176,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 86661740,
+      "step": 1383
+    },
+    {
+      "epoch": 4.605657237936772,
+      "grad_norm": 14.679880142211914,
+      "learning_rate": 5e-06,
+      "loss": 0.7799,
+      "num_input_tokens_seen": 86726256,
+      "step": 1384
+    },
+    {
+      "epoch": 4.605657237936772,
+      "loss": 0.7865468859672546,
+      "loss_ce": 0.0009023505263030529,
+      "loss_iou": 0.3125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 86726256,
+      "step": 1384
+    },
+    {
+      "epoch": 4.608985024958403,
+      "grad_norm": 15.619338989257812,
+      "learning_rate": 5e-06,
+      "loss": 0.5165,
+      "num_input_tokens_seen": 86787164,
+      "step": 1385
+    },
+    {
+      "epoch": 4.608985024958403,
+      "loss": 0.5341886281967163,
+      "loss_ce": 8.948968570621219e-06,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 86787164,
+      "step": 1385
+    },
+    {
+      "epoch": 4.612312811980034,
+      "grad_norm": 11.672150611877441,
+      "learning_rate": 5e-06,
+      "loss": 0.6922,
+      "num_input_tokens_seen": 86849700,
+      "step": 1386
+    },
+    {
+      "epoch": 4.612312811980034,
+      "loss": 0.7309243679046631,
+      "loss_ce": 0.00045559878344647586,
+      "loss_iou": 0.21875,
+      "loss_num": 0.058837890625,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 86849700,
+      "step": 1386
+    },
+    {
+      "epoch": 4.615640599001664,
+      "grad_norm": 10.545302391052246,
+      "learning_rate": 5e-06,
+      "loss": 0.9422,
+      "num_input_tokens_seen": 86911848,
+      "step": 1387
+    },
+    {
+      "epoch": 4.615640599001664,
+      "loss": 1.0675830841064453,
+      "loss_ce": 0.00026132259517908096,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0517578125,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 86911848,
+      "step": 1387
+    },
+    {
+      "epoch": 4.618968386023295,
+      "grad_norm": 18.017080307006836,
+      "learning_rate": 5e-06,
+      "loss": 0.6596,
+      "num_input_tokens_seen": 86974584,
+      "step": 1388
+    },
+    {
+      "epoch": 4.618968386023295,
+      "loss": 0.7650818228721619,
+      "loss_ce": 0.0007995798951014876,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 86974584,
+      "step": 1388
+    },
+    {
+      "epoch": 4.6222961730449255,
+      "grad_norm": 7.790538787841797,
+      "learning_rate": 5e-06,
+      "loss": 0.7499,
+      "num_input_tokens_seen": 87035060,
+      "step": 1389
+    },
+    {
+      "epoch": 4.6222961730449255,
+      "loss": 0.7779638171195984,
+      "loss_ce": 9.695755579741672e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 87035060,
+      "step": 1389
+    },
+    {
+      "epoch": 4.625623960066556,
+      "grad_norm": 23.20652198791504,
+      "learning_rate": 5e-06,
+      "loss": 0.7455,
+      "num_input_tokens_seen": 87095904,
+      "step": 1390
+    },
+    {
+      "epoch": 4.625623960066556,
+      "loss": 0.7988479733467102,
+      "loss_ce": 0.0002639756421558559,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 87095904,
+      "step": 1390
+    },
+    {
+      "epoch": 4.628951747088187,
+      "grad_norm": 17.283559799194336,
+      "learning_rate": 5e-06,
+      "loss": 0.6634,
+      "num_input_tokens_seen": 87158536,
+      "step": 1391
+    },
+    {
+      "epoch": 4.628951747088187,
+      "loss": 0.8823537230491638,
+      "loss_ce": 0.0007619569660164416,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 87158536,
+      "step": 1391
+    },
+    {
+      "epoch": 4.632279534109817,
+      "grad_norm": 12.51486587524414,
+      "learning_rate": 5e-06,
+      "loss": 0.7721,
+      "num_input_tokens_seen": 87221724,
+      "step": 1392
+    },
+    {
+      "epoch": 4.632279534109817,
+      "loss": 0.759742021560669,
+      "loss_ce": 9.84323924058117e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.076171875,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 87221724,
+      "step": 1392
+    },
+    {
+      "epoch": 4.635607321131448,
+      "grad_norm": 8.030847549438477,
+      "learning_rate": 5e-06,
+      "loss": 0.8579,
+      "num_input_tokens_seen": 87284540,
+      "step": 1393
+    },
+    {
+      "epoch": 4.635607321131448,
+      "loss": 0.8183420896530151,
+      "loss_ce": 0.0014476042706519365,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 87284540,
+      "step": 1393
+    },
+    {
+      "epoch": 4.6389351081530785,
+      "grad_norm": 19.066869735717773,
+      "learning_rate": 5e-06,
+      "loss": 0.6132,
+      "num_input_tokens_seen": 87345048,
+      "step": 1394
+    },
+    {
+      "epoch": 4.6389351081530785,
+      "loss": 0.7153163552284241,
+      "loss_ce": 0.00047258762060664594,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 87345048,
+      "step": 1394
+    },
+    {
+      "epoch": 4.642262895174709,
+      "grad_norm": 7.339348316192627,
+      "learning_rate": 5e-06,
+      "loss": 0.647,
+      "num_input_tokens_seen": 87408380,
+      "step": 1395
+    },
+    {
+      "epoch": 4.642262895174709,
+      "loss": 0.777653694152832,
+      "loss_ce": 0.0004319966828916222,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 87408380,
+      "step": 1395
+    },
+    {
+      "epoch": 4.64559068219634,
+      "grad_norm": 11.174449920654297,
+      "learning_rate": 5e-06,
+      "loss": 0.8674,
+      "num_input_tokens_seen": 87471892,
+      "step": 1396
+    },
+    {
+      "epoch": 4.64559068219634,
+      "loss": 0.796759307384491,
+      "loss_ce": 0.0002505228912923485,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 87471892,
+      "step": 1396
+    },
+    {
+      "epoch": 4.64891846921797,
+      "grad_norm": 11.844344139099121,
+      "learning_rate": 5e-06,
+      "loss": 0.6369,
+      "num_input_tokens_seen": 87534040,
+      "step": 1397
+    },
+    {
+      "epoch": 4.64891846921797,
+      "loss": 0.6389685869216919,
+      "loss_ce": 0.0004188179736956954,
+      "loss_iou": 0.21875,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 87534040,
+      "step": 1397
+    },
+    {
+      "epoch": 4.652246256239601,
+      "grad_norm": 15.831585884094238,
+      "learning_rate": 5e-06,
+      "loss": 0.9046,
+      "num_input_tokens_seen": 87596240,
+      "step": 1398
+    },
+    {
+      "epoch": 4.652246256239601,
+      "loss": 0.7054827213287354,
+      "loss_ce": 0.0008928956813178957,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 87596240,
+      "step": 1398
+    },
+    {
+      "epoch": 4.655574043261232,
+      "grad_norm": 6.208771705627441,
+      "learning_rate": 5e-06,
+      "loss": 0.6544,
+      "num_input_tokens_seen": 87657412,
+      "step": 1399
+    },
+    {
+      "epoch": 4.655574043261232,
+      "loss": 0.7047736644744873,
+      "loss_ce": 0.00042791658779606223,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 87657412,
+      "step": 1399
+    },
+    {
+      "epoch": 4.658901830282862,
+      "grad_norm": 15.71760082244873,
+      "learning_rate": 5e-06,
+      "loss": 0.5694,
+      "num_input_tokens_seen": 87721156,
+      "step": 1400
+    },
+    {
+      "epoch": 4.658901830282862,
+      "loss": 0.4697321057319641,
+      "loss_ce": 0.0004938616184517741,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.46875,
+      "num_input_tokens_seen": 87721156,
+      "step": 1400
+    },
+    {
+      "epoch": 4.662229617304493,
+      "grad_norm": 6.701347351074219,
+      "learning_rate": 5e-06,
+      "loss": 0.8324,
+      "num_input_tokens_seen": 87784928,
+      "step": 1401
+    },
+    {
+      "epoch": 4.662229617304493,
+      "loss": 0.851381778717041,
+      "loss_ce": 0.0012840689159929752,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.8515625,
+      "num_input_tokens_seen": 87784928,
+      "step": 1401
+    },
+    {
+      "epoch": 4.665557404326123,
+      "grad_norm": 9.888707160949707,
+      "learning_rate": 5e-06,
+      "loss": 0.6811,
+      "num_input_tokens_seen": 87846888,
+      "step": 1402
+    },
+    {
+      "epoch": 4.665557404326123,
+      "loss": 0.545545220375061,
+      "loss_ce": 0.0025764962192624807,
+      "loss_iou": 0.15625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 87846888,
+      "step": 1402
+    },
+    {
+      "epoch": 4.668885191347754,
+      "grad_norm": 12.931994438171387,
+      "learning_rate": 5e-06,
+      "loss": 0.6414,
+      "num_input_tokens_seen": 87909236,
+      "step": 1403
+    },
+    {
+      "epoch": 4.668885191347754,
+      "loss": 0.5946015119552612,
+      "loss_ce": 0.000485275435494259,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 87909236,
+      "step": 1403
+    },
+    {
+      "epoch": 4.672212978369385,
+      "grad_norm": 10.527796745300293,
+      "learning_rate": 5e-06,
+      "loss": 0.7106,
+      "num_input_tokens_seen": 87971780,
+      "step": 1404
+    },
+    {
+      "epoch": 4.672212978369385,
+      "loss": 0.6194738149642944,
+      "loss_ce": 0.002530462807044387,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 87971780,
+      "step": 1404
+    },
+    {
+      "epoch": 4.675540765391015,
+      "grad_norm": 7.018352031707764,
+      "learning_rate": 5e-06,
+      "loss": 0.8546,
+      "num_input_tokens_seen": 88034908,
+      "step": 1405
+    },
+    {
+      "epoch": 4.675540765391015,
+      "loss": 0.9526970386505127,
+      "loss_ce": 6.0325139202177525e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 88034908,
+      "step": 1405
+    },
+    {
+      "epoch": 4.678868552412646,
+      "grad_norm": 7.258470058441162,
+      "learning_rate": 5e-06,
+      "loss": 0.5011,
+      "num_input_tokens_seen": 88096500,
+      "step": 1406
+    },
+    {
+      "epoch": 4.678868552412646,
+      "loss": 0.65594881772995,
+      "loss_ce": 0.00018714150064624846,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 88096500,
+      "step": 1406
+    },
+    {
+      "epoch": 4.6821963394342765,
+      "grad_norm": 12.497472763061523,
+      "learning_rate": 5e-06,
+      "loss": 0.5734,
+      "num_input_tokens_seen": 88157424,
+      "step": 1407
+    },
+    {
+      "epoch": 4.6821963394342765,
+      "loss": 0.5287968516349792,
+      "loss_ce": 0.0008427201537415385,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 88157424,
+      "step": 1407
+    },
+    {
+      "epoch": 4.685524126455907,
+      "grad_norm": 13.06940746307373,
+      "learning_rate": 5e-06,
+      "loss": 0.6403,
+      "num_input_tokens_seen": 88219076,
+      "step": 1408
+    },
+    {
+      "epoch": 4.685524126455907,
+      "loss": 0.5052746534347534,
+      "loss_ce": 2.560833490861114e-05,
+      "loss_iou": 0.140625,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 88219076,
+      "step": 1408
+    },
+    {
+      "epoch": 4.688851913477538,
+      "grad_norm": 8.848599433898926,
+      "learning_rate": 5e-06,
+      "loss": 0.5396,
+      "num_input_tokens_seen": 88281864,
+      "step": 1409
+    },
+    {
+      "epoch": 4.688851913477538,
+      "loss": 0.6765046119689941,
+      "loss_ce": 0.00011294549040030688,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 88281864,
+      "step": 1409
+    },
+    {
+      "epoch": 4.692179700499168,
+      "grad_norm": 8.952995300292969,
+      "learning_rate": 5e-06,
+      "loss": 0.4109,
+      "num_input_tokens_seen": 88342824,
+      "step": 1410
+    },
+    {
+      "epoch": 4.692179700499168,
+      "loss": 0.5194376707077026,
+      "loss_ce": 0.0006388693582266569,
+      "loss_iou": 0.1103515625,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 88342824,
+      "step": 1410
+    },
+    {
+      "epoch": 4.695507487520799,
+      "grad_norm": 11.815835952758789,
+      "learning_rate": 5e-06,
+      "loss": 0.7675,
+      "num_input_tokens_seen": 88405772,
+      "step": 1411
+    },
+    {
+      "epoch": 4.695507487520799,
+      "loss": 0.6887185573577881,
+      "loss_ce": 0.00018099366570822895,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 88405772,
+      "step": 1411
+    },
+    {
+      "epoch": 4.6988352745424296,
+      "grad_norm": 8.99443531036377,
+      "learning_rate": 5e-06,
+      "loss": 0.6304,
+      "num_input_tokens_seen": 88468084,
+      "step": 1412
+    },
+    {
+      "epoch": 4.6988352745424296,
+      "loss": 0.8284512758255005,
+      "loss_ce": 8.219605660997331e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 88468084,
+      "step": 1412
+    },
+    {
+      "epoch": 4.70216306156406,
+      "grad_norm": 12.500134468078613,
+      "learning_rate": 5e-06,
+      "loss": 0.808,
+      "num_input_tokens_seen": 88528824,
+      "step": 1413
+    },
+    {
+      "epoch": 4.70216306156406,
+      "loss": 0.9188534021377563,
+      "loss_ce": 3.0147095458232798e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 88528824,
+      "step": 1413
+    },
+    {
+      "epoch": 4.705490848585691,
+      "grad_norm": 8.149920463562012,
+      "learning_rate": 5e-06,
+      "loss": 0.5456,
+      "num_input_tokens_seen": 88591064,
+      "step": 1414
+    },
+    {
+      "epoch": 4.705490848585691,
+      "loss": 0.4510113000869751,
+      "loss_ce": 2.2540578356711194e-05,
+      "loss_iou": 0.10595703125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.451171875,
+      "num_input_tokens_seen": 88591064,
+      "step": 1414
+    },
+    {
+      "epoch": 4.708818635607321,
+      "grad_norm": 14.00053882598877,
+      "learning_rate": 5e-06,
+      "loss": 0.995,
+      "num_input_tokens_seen": 88653992,
+      "step": 1415
+    },
+    {
+      "epoch": 4.708818635607321,
+      "loss": 1.069594144821167,
+      "loss_ce": 1.4102173736318946e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 88653992,
+      "step": 1415
+    },
+    {
+      "epoch": 4.712146422628952,
+      "grad_norm": 15.54645824432373,
+      "learning_rate": 5e-06,
+      "loss": 0.6461,
+      "num_input_tokens_seen": 88716116,
+      "step": 1416
+    },
+    {
+      "epoch": 4.712146422628952,
+      "loss": 0.41860735416412354,
+      "loss_ce": 0.0001808122469810769,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 88716116,
+      "step": 1416
+    },
+    {
+      "epoch": 4.715474209650583,
+      "grad_norm": 13.604162216186523,
+      "learning_rate": 5e-06,
+      "loss": 0.5831,
+      "num_input_tokens_seen": 88778872,
+      "step": 1417
+    },
+    {
+      "epoch": 4.715474209650583,
+      "loss": 0.6808904409408569,
+      "loss_ce": 0.00022634794004261494,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 88778872,
+      "step": 1417
+    },
+    {
+      "epoch": 4.718801996672213,
+      "grad_norm": 22.623563766479492,
+      "learning_rate": 5e-06,
+      "loss": 0.5977,
+      "num_input_tokens_seen": 88840616,
+      "step": 1418
+    },
+    {
+      "epoch": 4.718801996672213,
+      "loss": 0.7182178497314453,
+      "loss_ce": 0.0010547826532274485,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 88840616,
+      "step": 1418
+    },
+    {
+      "epoch": 4.722129783693844,
+      "grad_norm": 25.521087646484375,
+      "learning_rate": 5e-06,
+      "loss": 0.9424,
+      "num_input_tokens_seen": 88904500,
+      "step": 1419
+    },
+    {
+      "epoch": 4.722129783693844,
+      "loss": 1.0386605262756348,
+      "loss_ce": 0.0005440718960016966,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0615234375,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 88904500,
+      "step": 1419
+    },
+    {
+      "epoch": 4.7254575707154745,
+      "grad_norm": 32.94253921508789,
+      "learning_rate": 5e-06,
+      "loss": 0.6881,
+      "num_input_tokens_seen": 88966924,
+      "step": 1420
+    },
+    {
+      "epoch": 4.7254575707154745,
+      "loss": 0.8263764381408691,
+      "loss_ce": 0.000936914118938148,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 88966924,
+      "step": 1420
+    },
+    {
+      "epoch": 4.728785357737105,
+      "grad_norm": 12.381750106811523,
+      "learning_rate": 5e-06,
+      "loss": 0.6323,
+      "num_input_tokens_seen": 89030864,
+      "step": 1421
+    },
+    {
+      "epoch": 4.728785357737105,
+      "loss": 0.665398359298706,
+      "loss_ce": 0.00011519945110194385,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 89030864,
+      "step": 1421
+    },
+    {
+      "epoch": 4.732113144758736,
+      "grad_norm": 12.26732349395752,
+      "learning_rate": 5e-06,
+      "loss": 0.5234,
+      "num_input_tokens_seen": 89092332,
+      "step": 1422
+    },
+    {
+      "epoch": 4.732113144758736,
+      "loss": 0.2868744730949402,
+      "loss_ce": 9.247070920537226e-06,
+      "loss_iou": 0.0380859375,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 89092332,
+      "step": 1422
+    },
+    {
+      "epoch": 4.735440931780366,
+      "grad_norm": 7.801537990570068,
+      "learning_rate": 5e-06,
+      "loss": 0.8458,
+      "num_input_tokens_seen": 89155040,
+      "step": 1423
+    },
+    {
+      "epoch": 4.735440931780366,
+      "loss": 0.6569440364837646,
+      "loss_ce": 0.00038879140629433095,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 89155040,
+      "step": 1423
+    },
+    {
+      "epoch": 4.738768718801997,
+      "grad_norm": 17.682525634765625,
+      "learning_rate": 5e-06,
+      "loss": 0.7594,
+      "num_input_tokens_seen": 89217588,
+      "step": 1424
+    },
+    {
+      "epoch": 4.738768718801997,
+      "loss": 0.6541336178779602,
+      "loss_ce": 0.0006301991525106132,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 89217588,
+      "step": 1424
+    },
+    {
+      "epoch": 4.7420965058236275,
+      "grad_norm": 19.01534080505371,
+      "learning_rate": 5e-06,
+      "loss": 0.7232,
+      "num_input_tokens_seen": 89281484,
+      "step": 1425
+    },
+    {
+      "epoch": 4.7420965058236275,
+      "loss": 0.5709552764892578,
+      "loss_ce": 0.0006427803309634328,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 89281484,
+      "step": 1425
+    },
+    {
+      "epoch": 4.745424292845258,
+      "grad_norm": 14.757177352905273,
+      "learning_rate": 5e-06,
+      "loss": 0.6833,
+      "num_input_tokens_seen": 89344164,
+      "step": 1426
+    },
+    {
+      "epoch": 4.745424292845258,
+      "loss": 0.8227342367172241,
+      "loss_ce": 0.00010239605035167187,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 89344164,
+      "step": 1426
+    },
+    {
+      "epoch": 4.748752079866889,
+      "grad_norm": 16.073335647583008,
+      "learning_rate": 5e-06,
+      "loss": 0.7343,
+      "num_input_tokens_seen": 89407480,
+      "step": 1427
+    },
+    {
+      "epoch": 4.748752079866889,
+      "loss": 0.6605852842330933,
+      "loss_ce": 6.27873232588172e-05,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 89407480,
+      "step": 1427
+    },
+    {
+      "epoch": 4.752079866888519,
+      "grad_norm": 12.472293853759766,
+      "learning_rate": 5e-06,
+      "loss": 0.7983,
+      "num_input_tokens_seen": 89470712,
+      "step": 1428
+    },
+    {
+      "epoch": 4.752079866888519,
+      "loss": 0.6767415404319763,
+      "loss_ce": 0.0004719903226941824,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 89470712,
+      "step": 1428
+    },
+    {
+      "epoch": 4.75540765391015,
+      "grad_norm": 19.07227325439453,
+      "learning_rate": 5e-06,
+      "loss": 0.8962,
+      "num_input_tokens_seen": 89533400,
+      "step": 1429
+    },
+    {
+      "epoch": 4.75540765391015,
+      "loss": 1.0112426280975342,
+      "loss_ce": 0.0002563234302215278,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.08251953125,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 89533400,
+      "step": 1429
+    },
+    {
+      "epoch": 4.758735440931781,
+      "grad_norm": 11.964755058288574,
+      "learning_rate": 5e-06,
+      "loss": 0.7413,
+      "num_input_tokens_seen": 89597008,
+      "step": 1430
+    },
+    {
+      "epoch": 4.758735440931781,
+      "loss": 0.42090633511543274,
+      "loss_ce": 7.881029887357727e-06,
+      "loss_iou": 0.12451171875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 89597008,
+      "step": 1430
+    },
+    {
+      "epoch": 4.762063227953411,
+      "grad_norm": 25.646238327026367,
+      "learning_rate": 5e-06,
+      "loss": 0.788,
+      "num_input_tokens_seen": 89660676,
+      "step": 1431
+    },
+    {
+      "epoch": 4.762063227953411,
+      "loss": 0.9184248447418213,
+      "loss_ce": 0.0015548146329820156,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.91796875,
+      "num_input_tokens_seen": 89660676,
+      "step": 1431
+    },
+    {
+      "epoch": 4.765391014975042,
+      "grad_norm": 13.840594291687012,
+      "learning_rate": 5e-06,
+      "loss": 0.7977,
+      "num_input_tokens_seen": 89724804,
+      "step": 1432
+    },
+    {
+      "epoch": 4.765391014975042,
+      "loss": 0.9030267596244812,
+      "loss_ce": 0.00019475248700473458,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.90234375,
+      "num_input_tokens_seen": 89724804,
+      "step": 1432
+    },
+    {
+      "epoch": 4.768718801996672,
+      "grad_norm": 7.904409885406494,
+      "learning_rate": 5e-06,
+      "loss": 0.6688,
+      "num_input_tokens_seen": 89787772,
+      "step": 1433
+    },
+    {
+      "epoch": 4.768718801996672,
+      "loss": 0.8160985708236694,
+      "loss_ce": 0.000180591203388758,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 89787772,
+      "step": 1433
+    },
+    {
+      "epoch": 4.772046589018303,
+      "grad_norm": 24.053165435791016,
+      "learning_rate": 5e-06,
+      "loss": 0.871,
+      "num_input_tokens_seen": 89851700,
+      "step": 1434
+    },
+    {
+      "epoch": 4.772046589018303,
+      "loss": 0.7983778715133667,
+      "loss_ce": 0.000526325951796025,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 89851700,
+      "step": 1434
+    },
+    {
+      "epoch": 4.775374376039934,
+      "grad_norm": 17.900188446044922,
+      "learning_rate": 5e-06,
+      "loss": 0.9376,
+      "num_input_tokens_seen": 89914596,
+      "step": 1435
+    },
+    {
+      "epoch": 4.775374376039934,
+      "loss": 1.1730791330337524,
+      "loss_ce": 0.00022753766097594053,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0791015625,
+      "loss_xval": 1.171875,
+      "num_input_tokens_seen": 89914596,
+      "step": 1435
+    },
+    {
+      "epoch": 4.778702163061564,
+      "grad_norm": 12.668126106262207,
+      "learning_rate": 5e-06,
+      "loss": 0.7976,
+      "num_input_tokens_seen": 89976928,
+      "step": 1436
+    },
+    {
+      "epoch": 4.778702163061564,
+      "loss": 0.8083900809288025,
+      "loss_ce": 0.0007728736381977797,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.80859375,
+      "num_input_tokens_seen": 89976928,
+      "step": 1436
+    },
+    {
+      "epoch": 4.782029950083195,
+      "grad_norm": 13.394881248474121,
+      "learning_rate": 5e-06,
+      "loss": 0.7966,
+      "num_input_tokens_seen": 90040060,
+      "step": 1437
+    },
+    {
+      "epoch": 4.782029950083195,
+      "loss": 1.0120275020599365,
+      "loss_ce": 6.466710328822955e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.07470703125,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 90040060,
+      "step": 1437
+    },
+    {
+      "epoch": 4.7853577371048255,
+      "grad_norm": 11.221780776977539,
+      "learning_rate": 5e-06,
+      "loss": 0.7538,
+      "num_input_tokens_seen": 90103360,
+      "step": 1438
+    },
+    {
+      "epoch": 4.7853577371048255,
+      "loss": 0.7139836549758911,
+      "loss_ce": 0.0008489080937579274,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 90103360,
+      "step": 1438
+    },
+    {
+      "epoch": 4.788685524126456,
+      "grad_norm": 12.567187309265137,
+      "learning_rate": 5e-06,
+      "loss": 0.8503,
+      "num_input_tokens_seen": 90167208,
+      "step": 1439
+    },
+    {
+      "epoch": 4.788685524126456,
+      "loss": 0.9829078316688538,
+      "loss_ce": 0.00024178545572794974,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 90167208,
+      "step": 1439
+    },
+    {
+      "epoch": 4.792013311148087,
+      "grad_norm": 12.651561737060547,
+      "learning_rate": 5e-06,
+      "loss": 0.5847,
+      "num_input_tokens_seen": 90229644,
+      "step": 1440
+    },
+    {
+      "epoch": 4.792013311148087,
+      "loss": 0.4671025276184082,
+      "loss_ce": 0.0011601548176258802,
+      "loss_iou": 0.1279296875,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.466796875,
+      "num_input_tokens_seen": 90229644,
+      "step": 1440
+    },
+    {
+      "epoch": 4.795341098169717,
+      "grad_norm": 16.65874671936035,
+      "learning_rate": 5e-06,
+      "loss": 0.7644,
+      "num_input_tokens_seen": 90291568,
+      "step": 1441
+    },
+    {
+      "epoch": 4.795341098169717,
+      "loss": 0.5984110236167908,
+      "loss_ce": 0.0002665033971425146,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 90291568,
+      "step": 1441
+    },
+    {
+      "epoch": 4.798668885191348,
+      "grad_norm": 25.318418502807617,
+      "learning_rate": 5e-06,
+      "loss": 0.6216,
+      "num_input_tokens_seen": 90355464,
+      "step": 1442
+    },
+    {
+      "epoch": 4.798668885191348,
+      "loss": 0.681909441947937,
+      "loss_ce": 8.57599443406798e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.68359375,
+      "num_input_tokens_seen": 90355464,
+      "step": 1442
+    },
+    {
+      "epoch": 4.8019966722129785,
+      "grad_norm": 9.664568901062012,
+      "learning_rate": 5e-06,
+      "loss": 0.6261,
+      "num_input_tokens_seen": 90418132,
+      "step": 1443
+    },
+    {
+      "epoch": 4.8019966722129785,
+      "loss": 0.6628125309944153,
+      "loss_ce": 0.001679704524576664,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 90418132,
+      "step": 1443
+    },
+    {
+      "epoch": 4.805324459234609,
+      "grad_norm": 6.967924118041992,
+      "learning_rate": 5e-06,
+      "loss": 0.5027,
+      "num_input_tokens_seen": 90481356,
+      "step": 1444
+    },
+    {
+      "epoch": 4.805324459234609,
+      "loss": 0.3537171483039856,
+      "loss_ce": 7.945661491248757e-05,
+      "loss_iou": 0.10693359375,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 90481356,
+      "step": 1444
+    },
+    {
+      "epoch": 4.80865224625624,
+      "grad_norm": 24.378341674804688,
+      "learning_rate": 5e-06,
+      "loss": 0.7154,
+      "num_input_tokens_seen": 90544404,
+      "step": 1445
+    },
+    {
+      "epoch": 4.80865224625624,
+      "loss": 0.7716976404190063,
+      "loss_ce": 0.00045743631199002266,
+      "loss_iou": 0.28125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 90544404,
+      "step": 1445
+    },
+    {
+      "epoch": 4.81198003327787,
+      "grad_norm": 33.42325973510742,
+      "learning_rate": 5e-06,
+      "loss": 0.7386,
+      "num_input_tokens_seen": 90607728,
+      "step": 1446
+    },
+    {
+      "epoch": 4.81198003327787,
+      "loss": 0.9378975033760071,
+      "loss_ce": 3.1257190130418167e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0625,
+      "loss_xval": 0.9375,
+      "num_input_tokens_seen": 90607728,
+      "step": 1446
+    },
+    {
+      "epoch": 4.815307820299501,
+      "grad_norm": 10.282571792602539,
+      "learning_rate": 5e-06,
+      "loss": 0.6003,
+      "num_input_tokens_seen": 90670312,
+      "step": 1447
+    },
+    {
+      "epoch": 4.815307820299501,
+      "loss": 0.44258010387420654,
+      "loss_ce": 7.520718645537272e-05,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.443359375,
+      "num_input_tokens_seen": 90670312,
+      "step": 1447
+    },
+    {
+      "epoch": 4.818635607321132,
+      "grad_norm": 9.647857666015625,
+      "learning_rate": 5e-06,
+      "loss": 0.8134,
+      "num_input_tokens_seen": 90734576,
+      "step": 1448
+    },
+    {
+      "epoch": 4.818635607321132,
+      "loss": 0.7873010635375977,
+      "loss_ce": 0.0010461233323439956,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 90734576,
+      "step": 1448
+    },
+    {
+      "epoch": 4.821963394342762,
+      "grad_norm": 16.907331466674805,
+      "learning_rate": 5e-06,
+      "loss": 0.669,
+      "num_input_tokens_seen": 90797412,
+      "step": 1449
+    },
+    {
+      "epoch": 4.821963394342762,
+      "loss": 0.623419463634491,
+      "loss_ce": 6.367723472067155e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 90797412,
+      "step": 1449
+    },
+    {
+      "epoch": 4.825291181364393,
+      "grad_norm": 24.30735969543457,
+      "learning_rate": 5e-06,
+      "loss": 0.756,
+      "num_input_tokens_seen": 90858524,
+      "step": 1450
+    },
+    {
+      "epoch": 4.825291181364393,
+      "loss": 0.6738710999488831,
+      "loss_ce": 0.0005312650464475155,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 90858524,
+      "step": 1450
+    },
+    {
+      "epoch": 4.8286189683860234,
+      "grad_norm": 11.708526611328125,
+      "learning_rate": 5e-06,
+      "loss": 0.5307,
+      "num_input_tokens_seen": 90921152,
+      "step": 1451
+    },
+    {
+      "epoch": 4.8286189683860234,
+      "loss": 0.584744393825531,
+      "loss_ce": 0.000760018068831414,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 90921152,
+      "step": 1451
+    },
+    {
+      "epoch": 4.831946755407654,
+      "grad_norm": 22.66852378845215,
+      "learning_rate": 5e-06,
+      "loss": 0.806,
+      "num_input_tokens_seen": 90984628,
+      "step": 1452
+    },
+    {
+      "epoch": 4.831946755407654,
+      "loss": 0.6450674533843994,
+      "loss_ce": 0.0005361848743632436,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 90984628,
+      "step": 1452
+    },
+    {
+      "epoch": 4.835274542429285,
+      "grad_norm": 34.99443817138672,
+      "learning_rate": 5e-06,
+      "loss": 0.5531,
+      "num_input_tokens_seen": 91045592,
+      "step": 1453
+    },
+    {
+      "epoch": 4.835274542429285,
+      "loss": 0.597947895526886,
+      "loss_ce": 4.751010055770166e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 91045592,
+      "step": 1453
+    },
+    {
+      "epoch": 4.838602329450915,
+      "grad_norm": 31.513866424560547,
+      "learning_rate": 5e-06,
+      "loss": 0.7581,
+      "num_input_tokens_seen": 91108276,
+      "step": 1454
+    },
+    {
+      "epoch": 4.838602329450915,
+      "loss": 0.5201447010040283,
+      "loss_ce": 3.078287591051776e-06,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 91108276,
+      "step": 1454
+    },
+    {
+      "epoch": 4.841930116472546,
+      "grad_norm": 26.065690994262695,
+      "learning_rate": 5e-06,
+      "loss": 0.6533,
+      "num_input_tokens_seen": 91171596,
+      "step": 1455
+    },
+    {
+      "epoch": 4.841930116472546,
+      "loss": 0.7839912176132202,
+      "loss_ce": 0.0002998454438056797,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 91171596,
+      "step": 1455
+    },
+    {
+      "epoch": 4.8452579034941765,
+      "grad_norm": 9.963427543640137,
+      "learning_rate": 5e-06,
+      "loss": 0.9933,
+      "num_input_tokens_seen": 91234328,
+      "step": 1456
+    },
+    {
+      "epoch": 4.8452579034941765,
+      "loss": 1.1222238540649414,
+      "loss_ce": 0.0006418825942091644,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0703125,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 91234328,
+      "step": 1456
+    },
+    {
+      "epoch": 4.848585690515807,
+      "grad_norm": 17.40574836730957,
+      "learning_rate": 5e-06,
+      "loss": 0.7954,
+      "num_input_tokens_seen": 91298288,
+      "step": 1457
+    },
+    {
+      "epoch": 4.848585690515807,
+      "loss": 0.5493208765983582,
+      "loss_ce": 4.492412699619308e-06,
+      "loss_iou": 0.21875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 91298288,
+      "step": 1457
+    },
+    {
+      "epoch": 4.851913477537438,
+      "grad_norm": 36.004634857177734,
+      "learning_rate": 5e-06,
+      "loss": 0.507,
+      "num_input_tokens_seen": 91360532,
+      "step": 1458
+    },
+    {
+      "epoch": 4.851913477537438,
+      "loss": 0.29166799783706665,
+      "loss_ce": 0.0007134195184335113,
+      "loss_iou": 0.0390625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 91360532,
+      "step": 1458
+    },
+    {
+      "epoch": 4.855241264559068,
+      "grad_norm": 24.00579071044922,
+      "learning_rate": 5e-06,
+      "loss": 0.6614,
+      "num_input_tokens_seen": 91423224,
+      "step": 1459
+    },
+    {
+      "epoch": 4.855241264559068,
+      "loss": 0.8000708222389221,
+      "loss_ce": 0.0012427503243088722,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 91423224,
+      "step": 1459
+    },
+    {
+      "epoch": 4.858569051580699,
+      "grad_norm": 7.054749011993408,
+      "learning_rate": 5e-06,
+      "loss": 0.7916,
+      "num_input_tokens_seen": 91486976,
+      "step": 1460
+    },
+    {
+      "epoch": 4.858569051580699,
+      "loss": 0.6811645030975342,
+      "loss_ce": 0.00037837924901396036,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 91486976,
+      "step": 1460
+    },
+    {
+      "epoch": 4.86189683860233,
+      "grad_norm": 15.921218872070312,
+      "learning_rate": 5e-06,
+      "loss": 0.8073,
+      "num_input_tokens_seen": 91549948,
+      "step": 1461
+    },
+    {
+      "epoch": 4.86189683860233,
+      "loss": 1.1966166496276855,
+      "loss_ce": 0.0005717898602597415,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.07568359375,
+      "loss_xval": 1.1953125,
+      "num_input_tokens_seen": 91549948,
+      "step": 1461
+    },
+    {
+      "epoch": 4.86522462562396,
+      "grad_norm": 15.619834899902344,
+      "learning_rate": 5e-06,
+      "loss": 0.7384,
+      "num_input_tokens_seen": 91612528,
+      "step": 1462
+    },
+    {
+      "epoch": 4.86522462562396,
+      "loss": 0.8267630338668823,
+      "loss_ce": 0.0005911277839913964,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 91612528,
+      "step": 1462
+    },
+    {
+      "epoch": 4.868552412645591,
+      "grad_norm": 10.34393310546875,
+      "learning_rate": 5e-06,
+      "loss": 0.7152,
+      "num_input_tokens_seen": 91674264,
+      "step": 1463
+    },
+    {
+      "epoch": 4.868552412645591,
+      "loss": 0.6649792790412903,
+      "loss_ce": 0.0007947350386530161,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 91674264,
+      "step": 1463
+    },
+    {
+      "epoch": 4.871880199667221,
+      "grad_norm": 8.09216022491455,
+      "learning_rate": 5e-06,
+      "loss": 0.471,
+      "num_input_tokens_seen": 91737116,
+      "step": 1464
+    },
+    {
+      "epoch": 4.871880199667221,
+      "loss": 0.45063310861587524,
+      "loss_ce": 0.00019367330241948366,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.451171875,
+      "num_input_tokens_seen": 91737116,
+      "step": 1464
+    },
+    {
+      "epoch": 4.875207986688852,
+      "grad_norm": 17.690427780151367,
+      "learning_rate": 5e-06,
+      "loss": 0.7025,
+      "num_input_tokens_seen": 91799080,
+      "step": 1465
+    },
+    {
+      "epoch": 4.875207986688852,
+      "loss": 0.6187844276428223,
+      "loss_ce": 1.0004376235883683e-05,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 91799080,
+      "step": 1465
+    },
+    {
+      "epoch": 4.878535773710483,
+      "grad_norm": 19.25313377380371,
+      "learning_rate": 5e-06,
+      "loss": 0.6483,
+      "num_input_tokens_seen": 91860988,
+      "step": 1466
+    },
+    {
+      "epoch": 4.878535773710483,
+      "loss": 0.559953510761261,
+      "loss_ce": 0.00020009189029224217,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 91860988,
+      "step": 1466
+    },
+    {
+      "epoch": 4.881863560732113,
+      "grad_norm": 10.655545234680176,
+      "learning_rate": 5e-06,
+      "loss": 0.7747,
+      "num_input_tokens_seen": 91924968,
+      "step": 1467
+    },
+    {
+      "epoch": 4.881863560732113,
+      "loss": 0.636091947555542,
+      "loss_ce": 0.00034974643494933844,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 91924968,
+      "step": 1467
+    },
+    {
+      "epoch": 4.885191347753744,
+      "grad_norm": 20.099782943725586,
+      "learning_rate": 5e-06,
+      "loss": 0.7424,
+      "num_input_tokens_seen": 91988252,
+      "step": 1468
+    },
+    {
+      "epoch": 4.885191347753744,
+      "loss": 0.5978015661239624,
+      "loss_ce": 0.0008777122711762786,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 91988252,
+      "step": 1468
+    },
+    {
+      "epoch": 4.8885191347753745,
+      "grad_norm": 17.030847549438477,
+      "learning_rate": 5e-06,
+      "loss": 0.6069,
+      "num_input_tokens_seen": 92049740,
+      "step": 1469
+    },
+    {
+      "epoch": 4.8885191347753745,
+      "loss": 0.7965136766433716,
+      "loss_ce": 4.913673365081195e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 92049740,
+      "step": 1469
+    },
+    {
+      "epoch": 4.891846921797005,
+      "grad_norm": 17.971389770507812,
+      "learning_rate": 5e-06,
+      "loss": 0.7334,
+      "num_input_tokens_seen": 92112536,
+      "step": 1470
+    },
+    {
+      "epoch": 4.891846921797005,
+      "loss": 0.6343286037445068,
+      "loss_ce": 0.0005395347252488136,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 92112536,
+      "step": 1470
+    },
+    {
+      "epoch": 4.895174708818636,
+      "grad_norm": 23.792451858520508,
+      "learning_rate": 5e-06,
+      "loss": 0.8228,
+      "num_input_tokens_seen": 92176628,
+      "step": 1471
+    },
+    {
+      "epoch": 4.895174708818636,
+      "loss": 0.6928781270980835,
+      "loss_ce": 7.047529834380839e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.69140625,
+      "num_input_tokens_seen": 92176628,
+      "step": 1471
+    },
+    {
+      "epoch": 4.898502495840266,
+      "grad_norm": 10.210338592529297,
+      "learning_rate": 5e-06,
+      "loss": 0.5654,
+      "num_input_tokens_seen": 92238496,
+      "step": 1472
+    },
+    {
+      "epoch": 4.898502495840266,
+      "loss": 0.5062616467475891,
+      "loss_ce": 0.0018671302823349833,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 92238496,
+      "step": 1472
+    },
+    {
+      "epoch": 4.901830282861897,
+      "grad_norm": 27.42046546936035,
+      "learning_rate": 5e-06,
+      "loss": 0.7748,
+      "num_input_tokens_seen": 92301304,
+      "step": 1473
+    },
+    {
+      "epoch": 4.901830282861897,
+      "loss": 0.6527865529060364,
+      "loss_ce": 0.00019865072681568563,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 92301304,
+      "step": 1473
+    },
+    {
+      "epoch": 4.9051580698835275,
+      "grad_norm": 24.284852981567383,
+      "learning_rate": 5e-06,
+      "loss": 0.6917,
+      "num_input_tokens_seen": 92363496,
+      "step": 1474
+    },
+    {
+      "epoch": 4.9051580698835275,
+      "loss": 0.5194106101989746,
+      "loss_ce": 0.00018454447854310274,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 92363496,
+      "step": 1474
+    },
+    {
+      "epoch": 4.908485856905158,
+      "grad_norm": 9.942500114440918,
+      "learning_rate": 5e-06,
+      "loss": 0.5869,
+      "num_input_tokens_seen": 92426528,
+      "step": 1475
+    },
+    {
+      "epoch": 4.908485856905158,
+      "loss": 0.43600332736968994,
+      "loss_ce": 0.0009447159245610237,
+      "loss_iou": 0.0732421875,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.435546875,
+      "num_input_tokens_seen": 92426528,
+      "step": 1475
+    },
+    {
+      "epoch": 4.911813643926789,
+      "grad_norm": 12.120987892150879,
+      "learning_rate": 5e-06,
+      "loss": 0.5949,
+      "num_input_tokens_seen": 92488064,
+      "step": 1476
+    },
+    {
+      "epoch": 4.911813643926789,
+      "loss": 0.6362853050231934,
+      "loss_ce": 0.0012755857314914465,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 92488064,
+      "step": 1476
+    },
+    {
+      "epoch": 4.915141430948419,
+      "grad_norm": 21.779373168945312,
+      "learning_rate": 5e-06,
+      "loss": 0.5642,
+      "num_input_tokens_seen": 92550804,
+      "step": 1477
+    },
+    {
+      "epoch": 4.915141430948419,
+      "loss": 0.5219756960868835,
+      "loss_ce": 3.0526402952091303e-06,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 92550804,
+      "step": 1477
+    },
+    {
+      "epoch": 4.91846921797005,
+      "grad_norm": 41.83907699584961,
+      "learning_rate": 5e-06,
+      "loss": 0.8394,
+      "num_input_tokens_seen": 92614052,
+      "step": 1478
+    },
+    {
+      "epoch": 4.91846921797005,
+      "loss": 0.7585424780845642,
+      "loss_ce": 0.0014623773749917746,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 92614052,
+      "step": 1478
+    },
+    {
+      "epoch": 4.921797004991681,
+      "grad_norm": 23.484262466430664,
+      "learning_rate": 5e-06,
+      "loss": 0.7083,
+      "num_input_tokens_seen": 92676992,
+      "step": 1479
+    },
+    {
+      "epoch": 4.921797004991681,
+      "loss": 0.45134830474853516,
+      "loss_ce": 0.000298495520837605,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.451171875,
+      "num_input_tokens_seen": 92676992,
+      "step": 1479
+    },
+    {
+      "epoch": 4.925124792013311,
+      "grad_norm": 16.01560401916504,
+      "learning_rate": 5e-06,
+      "loss": 0.7104,
+      "num_input_tokens_seen": 92739476,
+      "step": 1480
+    },
+    {
+      "epoch": 4.925124792013311,
+      "loss": 0.7750359773635864,
+      "loss_ce": 1.1565551176317967e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 92739476,
+      "step": 1480
+    },
+    {
+      "epoch": 4.928452579034942,
+      "grad_norm": 14.046684265136719,
+      "learning_rate": 5e-06,
+      "loss": 0.5505,
+      "num_input_tokens_seen": 92801844,
+      "step": 1481
+    },
+    {
+      "epoch": 4.928452579034942,
+      "loss": 0.6174169778823853,
+      "loss_ce": 0.000961879501119256,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 92801844,
+      "step": 1481
+    },
+    {
+      "epoch": 4.931780366056572,
+      "grad_norm": 24.918357849121094,
+      "learning_rate": 5e-06,
+      "loss": 0.6422,
+      "num_input_tokens_seen": 92864440,
+      "step": 1482
+    },
+    {
+      "epoch": 4.931780366056572,
+      "loss": 0.5476626753807068,
+      "loss_ce": 0.003900481853634119,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 92864440,
+      "step": 1482
+    },
+    {
+      "epoch": 4.935108153078203,
+      "grad_norm": 12.272085189819336,
+      "learning_rate": 5e-06,
+      "loss": 0.5356,
+      "num_input_tokens_seen": 92926160,
+      "step": 1483
+    },
+    {
+      "epoch": 4.935108153078203,
+      "loss": 0.41046756505966187,
+      "loss_ce": 0.0007995814085006714,
+      "loss_iou": 0.1298828125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.41015625,
+      "num_input_tokens_seen": 92926160,
+      "step": 1483
+    },
+    {
+      "epoch": 4.938435940099834,
+      "grad_norm": 14.432737350463867,
+      "learning_rate": 5e-06,
+      "loss": 0.8823,
+      "num_input_tokens_seen": 92990268,
+      "step": 1484
+    },
+    {
+      "epoch": 4.938435940099834,
+      "loss": 0.904925525188446,
+      "loss_ce": 0.0006286612479016185,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 92990268,
+      "step": 1484
+    },
+    {
+      "epoch": 4.941763727121464,
+      "grad_norm": 28.021799087524414,
+      "learning_rate": 5e-06,
+      "loss": 0.6162,
+      "num_input_tokens_seen": 93053936,
+      "step": 1485
+    },
+    {
+      "epoch": 4.941763727121464,
+      "loss": 0.40271496772766113,
+      "loss_ce": 5.004140803066548e-06,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.40234375,
+      "num_input_tokens_seen": 93053936,
+      "step": 1485
+    },
+    {
+      "epoch": 4.945091514143095,
+      "grad_norm": 36.70539093017578,
+      "learning_rate": 5e-06,
+      "loss": 0.7026,
+      "num_input_tokens_seen": 93117408,
+      "step": 1486
+    },
+    {
+      "epoch": 4.945091514143095,
+      "loss": 0.7068368196487427,
+      "loss_ce": 0.0009041887824423611,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.70703125,
+      "num_input_tokens_seen": 93117408,
+      "step": 1486
+    },
+    {
+      "epoch": 4.9484193011647255,
+      "grad_norm": 34.976402282714844,
+      "learning_rate": 5e-06,
+      "loss": 0.6626,
+      "num_input_tokens_seen": 93179780,
+      "step": 1487
+    },
+    {
+      "epoch": 4.9484193011647255,
+      "loss": 0.5312550067901611,
+      "loss_ce": 5.038962626713328e-06,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 93179780,
+      "step": 1487
+    },
+    {
+      "epoch": 4.951747088186356,
+      "grad_norm": 14.874862670898438,
+      "learning_rate": 5e-06,
+      "loss": 0.583,
+      "num_input_tokens_seen": 93242380,
+      "step": 1488
+    },
+    {
+      "epoch": 4.951747088186356,
+      "loss": 0.5796291828155518,
+      "loss_ce": 3.933107655029744e-05,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 93242380,
+      "step": 1488
+    },
+    {
+      "epoch": 4.955074875207987,
+      "grad_norm": 14.622214317321777,
+      "learning_rate": 5e-06,
+      "loss": 0.5826,
+      "num_input_tokens_seen": 93305320,
+      "step": 1489
+    },
+    {
+      "epoch": 4.955074875207987,
+      "loss": 0.746558666229248,
+      "loss_ce": 3.767916496144608e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 93305320,
+      "step": 1489
+    },
+    {
+      "epoch": 4.958402662229617,
+      "grad_norm": 20.690074920654297,
+      "learning_rate": 5e-06,
+      "loss": 0.8156,
+      "num_input_tokens_seen": 93367680,
+      "step": 1490
+    },
+    {
+      "epoch": 4.958402662229617,
+      "loss": 0.8623336553573608,
+      "loss_ce": 0.0005172833916731179,
+      "loss_iou": 0.28125,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.86328125,
+      "num_input_tokens_seen": 93367680,
+      "step": 1490
+    },
+    {
+      "epoch": 4.961730449251248,
+      "grad_norm": 19.866193771362305,
+      "learning_rate": 5e-06,
+      "loss": 0.6329,
+      "num_input_tokens_seen": 93429344,
+      "step": 1491
+    },
+    {
+      "epoch": 4.961730449251248,
+      "loss": 0.5215956568717957,
+      "loss_ce": 0.00011129368795081973,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 93429344,
+      "step": 1491
+    },
+    {
+      "epoch": 4.965058236272879,
+      "grad_norm": 14.48011589050293,
+      "learning_rate": 5e-06,
+      "loss": 0.9734,
+      "num_input_tokens_seen": 93493372,
+      "step": 1492
+    },
+    {
+      "epoch": 4.965058236272879,
+      "loss": 0.8466031551361084,
+      "loss_ce": 0.0009000533609651029,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.058837890625,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 93493372,
+      "step": 1492
+    },
+    {
+      "epoch": 4.968386023294509,
+      "grad_norm": 12.389137268066406,
+      "learning_rate": 5e-06,
+      "loss": 0.8139,
+      "num_input_tokens_seen": 93556592,
+      "step": 1493
+    },
+    {
+      "epoch": 4.968386023294509,
+      "loss": 0.672704815864563,
+      "loss_ce": 0.0007077268091961741,
+      "loss_iou": 0.21875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 93556592,
+      "step": 1493
+    },
+    {
+      "epoch": 4.97171381031614,
+      "grad_norm": 14.919951438903809,
+      "learning_rate": 5e-06,
+      "loss": 0.6283,
+      "num_input_tokens_seen": 93619256,
+      "step": 1494
+    },
+    {
+      "epoch": 4.97171381031614,
+      "loss": 0.4168055057525635,
+      "loss_ce": 0.00017951334302779287,
+      "loss_iou": 0.1123046875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 93619256,
+      "step": 1494
+    },
+    {
+      "epoch": 4.97504159733777,
+      "grad_norm": 34.406837463378906,
+      "learning_rate": 5e-06,
+      "loss": 0.7431,
+      "num_input_tokens_seen": 93681420,
+      "step": 1495
+    },
+    {
+      "epoch": 4.97504159733777,
+      "loss": 0.7975531220436096,
+      "loss_ce": 0.0037908926606178284,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 93681420,
+      "step": 1495
+    },
+    {
+      "epoch": 4.978369384359401,
+      "grad_norm": 16.960229873657227,
+      "learning_rate": 5e-06,
+      "loss": 0.8396,
+      "num_input_tokens_seen": 93744272,
+      "step": 1496
+    },
+    {
+      "epoch": 4.978369384359401,
+      "loss": 1.021630048751831,
+      "loss_ce": 0.00038986955769360065,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 93744272,
+      "step": 1496
+    },
+    {
+      "epoch": 4.981697171381032,
+      "grad_norm": 12.859408378601074,
+      "learning_rate": 5e-06,
+      "loss": 0.7763,
+      "num_input_tokens_seen": 93806256,
+      "step": 1497
+    },
+    {
+      "epoch": 4.981697171381032,
+      "loss": 0.5506644248962402,
+      "loss_ce": 0.001958362990990281,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 93806256,
+      "step": 1497
+    },
+    {
+      "epoch": 4.985024958402662,
+      "grad_norm": 31.596189498901367,
+      "learning_rate": 5e-06,
+      "loss": 0.69,
+      "num_input_tokens_seen": 93869616,
+      "step": 1498
+    },
+    {
+      "epoch": 4.985024958402662,
+      "loss": 0.746276319026947,
+      "loss_ce": 0.00018256355542689562,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 93869616,
+      "step": 1498
+    },
+    {
+      "epoch": 4.988352745424293,
+      "grad_norm": 24.65460777282715,
+      "learning_rate": 5e-06,
+      "loss": 0.7051,
+      "num_input_tokens_seen": 93931152,
+      "step": 1499
+    },
+    {
+      "epoch": 4.988352745424293,
+      "loss": 0.7398415207862854,
+      "loss_ce": 0.00033956102561205626,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.73828125,
+      "num_input_tokens_seen": 93931152,
+      "step": 1499
+    },
+    {
+      "epoch": 4.9916805324459235,
+      "grad_norm": 10.513225555419922,
+      "learning_rate": 5e-06,
+      "loss": 0.5617,
+      "num_input_tokens_seen": 93994508,
+      "step": 1500
+    },
+    {
+      "epoch": 4.9916805324459235,
+      "eval_seeclick_CIoU": 0.12539703398942947,
+      "eval_seeclick_GIoU": 0.1438598707318306,
+      "eval_seeclick_IoU": 0.22447702288627625,
+      "eval_seeclick_MAE_all": 0.1706928089261055,
+      "eval_seeclick_MAE_h": 0.05414240434765816,
+      "eval_seeclick_MAE_w": 0.1219484768807888,
+      "eval_seeclick_MAE_x_boxes": 0.24619252979755402,
+      "eval_seeclick_MAE_y_boxes": 0.1315431222319603,
+      "eval_seeclick_NUM_probability": 0.9974748492240906,
+      "eval_seeclick_inside_bbox": 0.30520834028720856,
+      "eval_seeclick_loss": 2.665666341781616,
+      "eval_seeclick_loss_ce": 0.10174492001533508,
+      "eval_seeclick_loss_iou": 0.855712890625,
+      "eval_seeclick_loss_num": 0.16909027099609375,
+      "eval_seeclick_loss_xval": 2.556640625,
+      "eval_seeclick_runtime": 65.5562,
+      "eval_seeclick_samples_per_second": 0.717,
+      "eval_seeclick_steps_per_second": 0.031,
+      "num_input_tokens_seen": 93994508,
+      "step": 1500
+    },
+    {
+      "epoch": 4.9916805324459235,
+      "eval_icons_CIoU": 0.04876134404912591,
+      "eval_icons_GIoU": 0.15376271307468414,
+      "eval_icons_IoU": 0.20101643353700638,
+      "eval_icons_MAE_all": 0.13954433798789978,
+      "eval_icons_MAE_h": 0.061773356050252914,
+      "eval_icons_MAE_w": 0.15250807255506516,
+      "eval_icons_MAE_x_boxes": 0.1526419073343277,
+      "eval_icons_MAE_y_boxes": 0.03896789811551571,
+      "eval_icons_NUM_probability": 0.9999941885471344,
+      "eval_icons_inside_bbox": 0.3263888955116272,
+      "eval_icons_loss": 2.3821887969970703,
+      "eval_icons_loss_ce": 1.977284341592167e-06,
+      "eval_icons_loss_iou": 0.833984375,
+      "eval_icons_loss_num": 0.14191246032714844,
+      "eval_icons_loss_xval": 2.3779296875,
+      "eval_icons_runtime": 76.6569,
+      "eval_icons_samples_per_second": 0.652,
+      "eval_icons_steps_per_second": 0.026,
+      "num_input_tokens_seen": 93994508,
+      "step": 1500
+    },
+    {
+      "epoch": 4.9916805324459235,
+      "eval_screenspot_CIoU": 0.05785238494475683,
+      "eval_screenspot_GIoU": 0.11830167099833488,
+      "eval_screenspot_IoU": 0.20486934979756674,
+      "eval_screenspot_MAE_all": 0.1998411019643148,
+      "eval_screenspot_MAE_h": 0.07882813364267349,
+      "eval_screenspot_MAE_w": 0.1679569110274315,
+      "eval_screenspot_MAE_x_boxes": 0.26309264699618023,
+      "eval_screenspot_MAE_y_boxes": 0.14158939321835837,
+      "eval_screenspot_NUM_probability": 0.999947190284729,
+      "eval_screenspot_inside_bbox": 0.34166666865348816,
+      "eval_screenspot_loss": 2.7864136695861816,
+      "eval_screenspot_loss_ce": 7.061872323295877e-05,
+      "eval_screenspot_loss_iou": 0.8956705729166666,
+      "eval_screenspot_loss_num": 0.204803466796875,
+      "eval_screenspot_loss_xval": 2.8151041666666665,
+      "eval_screenspot_runtime": 118.0213,
+      "eval_screenspot_samples_per_second": 0.754,
+      "eval_screenspot_steps_per_second": 0.025,
+      "num_input_tokens_seen": 93994508,
+      "step": 1500
+    },
+    {
+      "epoch": 4.9916805324459235,
+      "eval_compot_CIoU": -0.00016517378389835358,
+      "eval_compot_GIoU": 0.10418009385466576,
+      "eval_compot_IoU": 0.16831143200397491,
+      "eval_compot_MAE_all": 0.18598458915948868,
+      "eval_compot_MAE_h": 0.07219501584768295,
+      "eval_compot_MAE_w": 0.24491457641124725,
+      "eval_compot_MAE_x_boxes": 0.182175874710083,
+      "eval_compot_MAE_y_boxes": 0.06786558777093887,
+      "eval_compot_NUM_probability": 0.9999881684780121,
+      "eval_compot_inside_bbox": 0.2986111119389534,
+      "eval_compot_loss": 2.7441301345825195,
+      "eval_compot_loss_ce": 0.004642534884624183,
+      "eval_compot_loss_iou": 0.90576171875,
+      "eval_compot_loss_num": 0.1902923583984375,
+      "eval_compot_loss_xval": 2.7626953125,
+      "eval_compot_runtime": 67.269,
+      "eval_compot_samples_per_second": 0.743,
+      "eval_compot_steps_per_second": 0.03,
+      "num_input_tokens_seen": 93994508,
+      "step": 1500
+    },
+    {
+      "epoch": 4.9916805324459235,
+      "eval_custom_ui_MAE_all": 0.07861483469605446,
+      "eval_custom_ui_MAE_x": 0.07045421935617924,
+      "eval_custom_ui_MAE_y": 0.08677546679973602,
+      "eval_custom_ui_NUM_probability": 0.9999936819076538,
+      "eval_custom_ui_loss": 0.37946105003356934,
+      "eval_custom_ui_loss_ce": 9.375693934998708e-06,
+      "eval_custom_ui_loss_num": 0.0741729736328125,
+      "eval_custom_ui_loss_xval": 0.37078857421875,
+      "eval_custom_ui_runtime": 51.6085,
+      "eval_custom_ui_samples_per_second": 0.969,
+      "eval_custom_ui_steps_per_second": 0.039,
+      "num_input_tokens_seen": 93994508,
+      "step": 1500
+    },
+    {
+      "epoch": 4.9916805324459235,
+      "loss": 0.3743995428085327,
+      "loss_ce": 9.905661499942653e-06,
+      "loss_iou": 0.0,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 93994508,
+      "step": 1500
+    },
+    {
+      "epoch": 4.995008319467554,
+      "grad_norm": 24.133996963500977,
+      "learning_rate": 5e-06,
+      "loss": 0.5173,
+      "num_input_tokens_seen": 94056528,
+      "step": 1501
+    },
+    {
+      "epoch": 4.995008319467554,
+      "loss": 0.4280321002006531,
+      "loss_ce": 5.3581257816404104e-05,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 94056528,
+      "step": 1501
+    },
+    {
+      "epoch": 4.998336106489185,
+      "grad_norm": 13.59617805480957,
+      "learning_rate": 5e-06,
+      "loss": 0.7287,
+      "num_input_tokens_seen": 94120456,
+      "step": 1502
+    },
+    {
+      "epoch": 4.998336106489185,
+      "loss": 0.8299544453620911,
+      "loss_ce": 0.0006087642977945507,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 94120456,
+      "step": 1502
+    },
+    {
+      "epoch": 4.998336106489185,
+      "loss": 0.8235313296318054,
+      "loss_ce": 4.499550414038822e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 94152608,
+      "step": 1502
+    },
+    {
+      "epoch": 5.001663893510815,
+      "grad_norm": 14.112279891967773,
+      "learning_rate": 5e-06,
+      "loss": 0.7359,
+      "num_input_tokens_seen": 94184688,
+      "step": 1503
+    },
+    {
+      "epoch": 5.001663893510815,
+      "loss": 0.6481797695159912,
+      "loss_ce": 0.0002305309899384156,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 94184688,
+      "step": 1503
+    },
+    {
+      "epoch": 5.004991680532446,
+      "grad_norm": 26.775850296020508,
+      "learning_rate": 5e-06,
+      "loss": 0.8273,
+      "num_input_tokens_seen": 94250036,
+      "step": 1504
+    },
+    {
+      "epoch": 5.004991680532446,
+      "loss": 0.9145175814628601,
+      "loss_ce": 0.0004551278834696859,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 94250036,
+      "step": 1504
+    },
+    {
+      "epoch": 5.0083194675540765,
+      "grad_norm": 11.916851997375488,
+      "learning_rate": 5e-06,
+      "loss": 0.6772,
+      "num_input_tokens_seen": 94313792,
+      "step": 1505
+    },
+    {
+      "epoch": 5.0083194675540765,
+      "loss": 0.7098668217658997,
+      "loss_ce": 0.0008824262768030167,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 94313792,
+      "step": 1505
+    },
+    {
+      "epoch": 5.011647254575707,
+      "grad_norm": 8.042437553405762,
+      "learning_rate": 5e-06,
+      "loss": 0.7836,
+      "num_input_tokens_seen": 94375952,
+      "step": 1506
+    },
+    {
+      "epoch": 5.011647254575707,
+      "loss": 0.8639097809791565,
+      "loss_ce": 0.000384394807042554,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.86328125,
+      "num_input_tokens_seen": 94375952,
+      "step": 1506
+    },
+    {
+      "epoch": 5.014975041597338,
+      "grad_norm": 15.622085571289062,
+      "learning_rate": 5e-06,
+      "loss": 0.8583,
+      "num_input_tokens_seen": 94440520,
+      "step": 1507
+    },
+    {
+      "epoch": 5.014975041597338,
+      "loss": 1.238585114479065,
+      "loss_ce": 0.0032335498835891485,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.06103515625,
+      "loss_xval": 1.234375,
+      "num_input_tokens_seen": 94440520,
+      "step": 1507
+    },
+    {
+      "epoch": 5.018302828618968,
+      "grad_norm": 31.38747787475586,
+      "learning_rate": 5e-06,
+      "loss": 0.7085,
+      "num_input_tokens_seen": 94500620,
+      "step": 1508
+    },
+    {
+      "epoch": 5.018302828618968,
+      "loss": 0.5411911010742188,
+      "loss_ce": 0.00041960144881159067,
+      "loss_iou": 0.1875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 94500620,
+      "step": 1508
+    },
+    {
+      "epoch": 5.021630615640599,
+      "grad_norm": 24.40467643737793,
+      "learning_rate": 5e-06,
+      "loss": 0.755,
+      "num_input_tokens_seen": 94563172,
+      "step": 1509
+    },
+    {
+      "epoch": 5.021630615640599,
+      "loss": 0.886728048324585,
+      "loss_ce": 0.0011079427786171436,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 94563172,
+      "step": 1509
+    },
+    {
+      "epoch": 5.02495840266223,
+      "grad_norm": 28.08826446533203,
+      "learning_rate": 5e-06,
+      "loss": 0.7189,
+      "num_input_tokens_seen": 94626456,
+      "step": 1510
+    },
+    {
+      "epoch": 5.02495840266223,
+      "loss": 0.5464110374450684,
+      "loss_ce": 0.0007567618158645928,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 94626456,
+      "step": 1510
+    },
+    {
+      "epoch": 5.02828618968386,
+      "grad_norm": 32.060081481933594,
+      "learning_rate": 5e-06,
+      "loss": 0.6385,
+      "num_input_tokens_seen": 94689628,
+      "step": 1511
+    },
+    {
+      "epoch": 5.02828618968386,
+      "loss": 0.6361727714538574,
+      "loss_ce": 0.00018641223141457886,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 94689628,
+      "step": 1511
+    },
+    {
+      "epoch": 5.031613976705491,
+      "grad_norm": 28.850717544555664,
+      "learning_rate": 5e-06,
+      "loss": 0.6486,
+      "num_input_tokens_seen": 94752952,
+      "step": 1512
+    },
+    {
+      "epoch": 5.031613976705491,
+      "loss": 0.6386792659759521,
+      "loss_ce": 7.388218818960013e-06,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 94752952,
+      "step": 1512
+    },
+    {
+      "epoch": 5.034941763727121,
+      "grad_norm": 26.582441329956055,
+      "learning_rate": 5e-06,
+      "loss": 0.7219,
+      "num_input_tokens_seen": 94816220,
+      "step": 1513
+    },
+    {
+      "epoch": 5.034941763727121,
+      "loss": 0.8012294173240662,
+      "loss_ce": 0.0014247273793444037,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 94816220,
+      "step": 1513
+    },
+    {
+      "epoch": 5.038269550748752,
+      "grad_norm": 13.041583061218262,
+      "learning_rate": 5e-06,
+      "loss": 0.7812,
+      "num_input_tokens_seen": 94878556,
+      "step": 1514
+    },
+    {
+      "epoch": 5.038269550748752,
+      "loss": 0.8982586860656738,
+      "loss_ce": 6.535501597682014e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 94878556,
+      "step": 1514
+    },
+    {
+      "epoch": 5.041597337770383,
+      "grad_norm": 41.85260009765625,
+      "learning_rate": 5e-06,
+      "loss": 0.6035,
+      "num_input_tokens_seen": 94940876,
+      "step": 1515
+    },
+    {
+      "epoch": 5.041597337770383,
+      "loss": 0.6592752933502197,
+      "loss_ce": 0.001194267300888896,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 94940876,
+      "step": 1515
+    },
+    {
+      "epoch": 5.044925124792013,
+      "grad_norm": 33.620452880859375,
+      "learning_rate": 5e-06,
+      "loss": 0.8496,
+      "num_input_tokens_seen": 95004780,
+      "step": 1516
+    },
+    {
+      "epoch": 5.044925124792013,
+      "loss": 1.0527219772338867,
+      "loss_ce": 0.000597875623498112,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.07275390625,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 95004780,
+      "step": 1516
+    },
+    {
+      "epoch": 5.048252911813644,
+      "grad_norm": 14.673892974853516,
+      "learning_rate": 5e-06,
+      "loss": 0.5555,
+      "num_input_tokens_seen": 95067216,
+      "step": 1517
+    },
+    {
+      "epoch": 5.048252911813644,
+      "loss": 0.41764336824417114,
+      "loss_ce": 0.0001018413677229546,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 95067216,
+      "step": 1517
+    },
+    {
+      "epoch": 5.0515806988352745,
+      "grad_norm": 10.968241691589355,
+      "learning_rate": 5e-06,
+      "loss": 0.6427,
+      "num_input_tokens_seen": 95129008,
+      "step": 1518
+    },
+    {
+      "epoch": 5.0515806988352745,
+      "loss": 0.5915021896362305,
+      "loss_ce": 0.0006818820256739855,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 95129008,
+      "step": 1518
+    },
+    {
+      "epoch": 5.054908485856905,
+      "grad_norm": 26.52812957763672,
+      "learning_rate": 5e-06,
+      "loss": 0.7537,
+      "num_input_tokens_seen": 95192232,
+      "step": 1519
+    },
+    {
+      "epoch": 5.054908485856905,
+      "loss": 0.903398871421814,
+      "loss_ce": 0.0005668269586749375,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.90234375,
+      "num_input_tokens_seen": 95192232,
+      "step": 1519
+    },
+    {
+      "epoch": 5.058236272878536,
+      "grad_norm": 30.482860565185547,
+      "learning_rate": 5e-06,
+      "loss": 0.8095,
+      "num_input_tokens_seen": 95256580,
+      "step": 1520
+    },
+    {
+      "epoch": 5.058236272878536,
+      "loss": 0.7557247877120972,
+      "loss_ce": 0.0013303017476573586,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 95256580,
+      "step": 1520
+    },
+    {
+      "epoch": 5.061564059900166,
+      "grad_norm": 39.979736328125,
+      "learning_rate": 5e-06,
+      "loss": 0.6808,
+      "num_input_tokens_seen": 95318464,
+      "step": 1521
+    },
+    {
+      "epoch": 5.061564059900166,
+      "loss": 0.7826014161109924,
+      "loss_ce": 8.651560165162664e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 95318464,
+      "step": 1521
+    },
+    {
+      "epoch": 5.064891846921797,
+      "grad_norm": 14.766712188720703,
+      "learning_rate": 5e-06,
+      "loss": 0.5086,
+      "num_input_tokens_seen": 95379756,
+      "step": 1522
+    },
+    {
+      "epoch": 5.064891846921797,
+      "loss": 0.5393363237380981,
+      "loss_ce": 0.0009147044620476663,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 95379756,
+      "step": 1522
+    },
+    {
+      "epoch": 5.068219633943428,
+      "grad_norm": 7.251678466796875,
+      "learning_rate": 5e-06,
+      "loss": 0.4442,
+      "num_input_tokens_seen": 95441884,
+      "step": 1523
+    },
+    {
+      "epoch": 5.068219633943428,
+      "loss": 0.4630827307701111,
+      "loss_ce": 0.00031416656565852463,
+      "loss_iou": 0.10546875,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.462890625,
+      "num_input_tokens_seen": 95441884,
+      "step": 1523
+    },
+    {
+      "epoch": 5.071547420965058,
+      "grad_norm": 6.830024242401123,
+      "learning_rate": 5e-06,
+      "loss": 0.961,
+      "num_input_tokens_seen": 95505028,
+      "step": 1524
+    },
+    {
+      "epoch": 5.071547420965058,
+      "loss": 0.8202039003372192,
+      "loss_ce": 0.0006237871712073684,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 95505028,
+      "step": 1524
+    },
+    {
+      "epoch": 5.074875207986689,
+      "grad_norm": 21.969810485839844,
+      "learning_rate": 5e-06,
+      "loss": 1.14,
+      "num_input_tokens_seen": 95568468,
+      "step": 1525
+    },
+    {
+      "epoch": 5.074875207986689,
+      "loss": 0.9151379466056824,
+      "loss_ce": 0.0004040507774334401,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 95568468,
+      "step": 1525
+    },
+    {
+      "epoch": 5.078202995008319,
+      "grad_norm": 57.692718505859375,
+      "learning_rate": 5e-06,
+      "loss": 0.642,
+      "num_input_tokens_seen": 95631624,
+      "step": 1526
+    },
+    {
+      "epoch": 5.078202995008319,
+      "loss": 0.7038395404815674,
+      "loss_ce": 0.00010419228055980057,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 95631624,
+      "step": 1526
+    },
+    {
+      "epoch": 5.08153078202995,
+      "grad_norm": 10.670079231262207,
+      "learning_rate": 5e-06,
+      "loss": 0.8521,
+      "num_input_tokens_seen": 95694840,
+      "step": 1527
+    },
+    {
+      "epoch": 5.08153078202995,
+      "loss": 0.6557031273841858,
+      "loss_ce": 0.0011621195590123534,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 95694840,
+      "step": 1527
+    },
+    {
+      "epoch": 5.084858569051581,
+      "grad_norm": 14.62645435333252,
+      "learning_rate": 5e-06,
+      "loss": 0.734,
+      "num_input_tokens_seen": 95757612,
+      "step": 1528
+    },
+    {
+      "epoch": 5.084858569051581,
+      "loss": 0.6704707145690918,
+      "loss_ce": 0.0006709391018375754,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 95757612,
+      "step": 1528
+    },
+    {
+      "epoch": 5.088186356073211,
+      "grad_norm": 16.821863174438477,
+      "learning_rate": 5e-06,
+      "loss": 0.9306,
+      "num_input_tokens_seen": 95821208,
+      "step": 1529
+    },
+    {
+      "epoch": 5.088186356073211,
+      "loss": 1.184844970703125,
+      "loss_ce": 3.0552084353985265e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0810546875,
+      "loss_xval": 1.1875,
+      "num_input_tokens_seen": 95821208,
+      "step": 1529
+    },
+    {
+      "epoch": 5.091514143094842,
+      "grad_norm": 79.41419982910156,
+      "learning_rate": 5e-06,
+      "loss": 0.7217,
+      "num_input_tokens_seen": 95883876,
+      "step": 1530
+    },
+    {
+      "epoch": 5.091514143094842,
+      "loss": 0.8382284641265869,
+      "loss_ce": 0.0009482197929173708,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 95883876,
+      "step": 1530
+    },
+    {
+      "epoch": 5.0948419301164725,
+      "grad_norm": 25.368114471435547,
+      "learning_rate": 5e-06,
+      "loss": 0.6514,
+      "num_input_tokens_seen": 95947076,
+      "step": 1531
+    },
+    {
+      "epoch": 5.0948419301164725,
+      "loss": 0.7189469337463379,
+      "loss_ce": 1.380766843794845e-05,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 95947076,
+      "step": 1531
+    },
+    {
+      "epoch": 5.098169717138103,
+      "grad_norm": 13.117644309997559,
+      "learning_rate": 5e-06,
+      "loss": 0.8008,
+      "num_input_tokens_seen": 96009676,
+      "step": 1532
+    },
+    {
+      "epoch": 5.098169717138103,
+      "loss": 0.8573014736175537,
+      "loss_ce": 0.00012374338984955102,
+      "loss_iou": 0.234375,
+      "loss_num": 0.07763671875,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 96009676,
+      "step": 1532
+    },
+    {
+      "epoch": 5.101497504159734,
+      "grad_norm": 10.874475479125977,
+      "learning_rate": 5e-06,
+      "loss": 0.376,
+      "num_input_tokens_seen": 96071412,
+      "step": 1533
+    },
+    {
+      "epoch": 5.101497504159734,
+      "loss": 0.4256836473941803,
+      "loss_ce": 2.4467857656418346e-05,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 96071412,
+      "step": 1533
+    },
+    {
+      "epoch": 5.104825291181364,
+      "grad_norm": 15.079007148742676,
+      "learning_rate": 5e-06,
+      "loss": 0.5986,
+      "num_input_tokens_seen": 96133696,
+      "step": 1534
+    },
+    {
+      "epoch": 5.104825291181364,
+      "loss": 0.5522220134735107,
+      "loss_ce": 0.001440738094970584,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 96133696,
+      "step": 1534
+    },
+    {
+      "epoch": 5.108153078202995,
+      "grad_norm": 12.708340644836426,
+      "learning_rate": 5e-06,
+      "loss": 0.7919,
+      "num_input_tokens_seen": 96197404,
+      "step": 1535
+    },
+    {
+      "epoch": 5.108153078202995,
+      "loss": 0.9537239670753479,
+      "loss_ce": 0.001331356936134398,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 96197404,
+      "step": 1535
+    },
+    {
+      "epoch": 5.1114808652246255,
+      "grad_norm": 9.230545043945312,
+      "learning_rate": 5e-06,
+      "loss": 0.534,
+      "num_input_tokens_seen": 96260424,
+      "step": 1536
+    },
+    {
+      "epoch": 5.1114808652246255,
+      "loss": 0.5380446314811707,
+      "loss_ce": 0.0006911091622896492,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 96260424,
+      "step": 1536
+    },
+    {
+      "epoch": 5.114808652246256,
+      "grad_norm": 7.485297203063965,
+      "learning_rate": 5e-06,
+      "loss": 0.6481,
+      "num_input_tokens_seen": 96324052,
+      "step": 1537
+    },
+    {
+      "epoch": 5.114808652246256,
+      "loss": 0.8440206050872803,
+      "loss_ce": 0.00014855540939606726,
+      "loss_iou": 0.296875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 96324052,
+      "step": 1537
+    },
+    {
+      "epoch": 5.118136439267887,
+      "grad_norm": 12.548988342285156,
+      "learning_rate": 5e-06,
+      "loss": 0.7396,
+      "num_input_tokens_seen": 96386140,
+      "step": 1538
+    },
+    {
+      "epoch": 5.118136439267887,
+      "loss": 0.907487154006958,
+      "loss_ce": 1.645735392230563e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 96386140,
+      "step": 1538
+    },
+    {
+      "epoch": 5.121464226289517,
+      "grad_norm": 9.125192642211914,
+      "learning_rate": 5e-06,
+      "loss": 0.5798,
+      "num_input_tokens_seen": 96445552,
+      "step": 1539
+    },
+    {
+      "epoch": 5.121464226289517,
+      "loss": 0.5199623703956604,
+      "loss_ce": 3.884044417645782e-06,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 96445552,
+      "step": 1539
+    },
+    {
+      "epoch": 5.124792013311148,
+      "grad_norm": 9.516877174377441,
+      "learning_rate": 5e-06,
+      "loss": 0.4506,
+      "num_input_tokens_seen": 96508364,
+      "step": 1540
+    },
+    {
+      "epoch": 5.124792013311148,
+      "loss": 0.5682762265205383,
+      "loss_ce": 0.00016097842308226973,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 96508364,
+      "step": 1540
+    },
+    {
+      "epoch": 5.128119800332779,
+      "grad_norm": 13.660243034362793,
+      "learning_rate": 5e-06,
+      "loss": 0.7161,
+      "num_input_tokens_seen": 96571852,
+      "step": 1541
+    },
+    {
+      "epoch": 5.128119800332779,
+      "loss": 0.8170957565307617,
+      "loss_ce": 0.002154330490157008,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 96571852,
+      "step": 1541
+    },
+    {
+      "epoch": 5.131447587354409,
+      "grad_norm": 12.244997024536133,
+      "learning_rate": 5e-06,
+      "loss": 0.8757,
+      "num_input_tokens_seen": 96637356,
+      "step": 1542
+    },
+    {
+      "epoch": 5.131447587354409,
+      "loss": 0.7130774259567261,
+      "loss_ce": 0.001651703380048275,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 96637356,
+      "step": 1542
+    },
+    {
+      "epoch": 5.13477537437604,
+      "grad_norm": 10.925747871398926,
+      "learning_rate": 5e-06,
+      "loss": 0.6949,
+      "num_input_tokens_seen": 96698348,
+      "step": 1543
+    },
+    {
+      "epoch": 5.13477537437604,
+      "loss": 0.516825795173645,
+      "loss_ce": 0.000468335987534374,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 96698348,
+      "step": 1543
+    },
+    {
+      "epoch": 5.13810316139767,
+      "grad_norm": 16.20490074157715,
+      "learning_rate": 5e-06,
+      "loss": 0.8087,
+      "num_input_tokens_seen": 96759328,
+      "step": 1544
+    },
+    {
+      "epoch": 5.13810316139767,
+      "loss": 0.8088467121124268,
+      "loss_ce": 8.813009117147885e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.80859375,
+      "num_input_tokens_seen": 96759328,
+      "step": 1544
+    },
+    {
+      "epoch": 5.141430948419301,
+      "grad_norm": 13.553998947143555,
+      "learning_rate": 5e-06,
+      "loss": 0.9186,
+      "num_input_tokens_seen": 96823196,
+      "step": 1545
+    },
+    {
+      "epoch": 5.141430948419301,
+      "loss": 0.8833928108215332,
+      "loss_ce": 0.0019231259357184172,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 96823196,
+      "step": 1545
+    },
+    {
+      "epoch": 5.144758735440932,
+      "grad_norm": 16.64641571044922,
+      "learning_rate": 5e-06,
+      "loss": 0.5477,
+      "num_input_tokens_seen": 96886160,
+      "step": 1546
+    },
+    {
+      "epoch": 5.144758735440932,
+      "loss": 0.5232193470001221,
+      "loss_ce": 2.6001296646427363e-05,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 96886160,
+      "step": 1546
+    },
+    {
+      "epoch": 5.148086522462562,
+      "grad_norm": 12.320746421813965,
+      "learning_rate": 5e-06,
+      "loss": 0.8047,
+      "num_input_tokens_seen": 96949040,
+      "step": 1547
+    },
+    {
+      "epoch": 5.148086522462562,
+      "loss": 0.7521053552627563,
+      "loss_ce": 3.014335743500851e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 96949040,
+      "step": 1547
+    },
+    {
+      "epoch": 5.151414309484193,
+      "grad_norm": 8.744440078735352,
+      "learning_rate": 5e-06,
+      "loss": 0.8747,
+      "num_input_tokens_seen": 97011576,
+      "step": 1548
+    },
+    {
+      "epoch": 5.151414309484193,
+      "loss": 0.8263041973114014,
+      "loss_ce": 0.0001323087781202048,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 97011576,
+      "step": 1548
+    },
+    {
+      "epoch": 5.1547420965058235,
+      "grad_norm": 8.398408889770508,
+      "learning_rate": 5e-06,
+      "loss": 0.6888,
+      "num_input_tokens_seen": 97073700,
+      "step": 1549
+    },
+    {
+      "epoch": 5.1547420965058235,
+      "loss": 0.6298340559005737,
+      "loss_ce": 0.0016601935494691133,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 97073700,
+      "step": 1549
+    },
+    {
+      "epoch": 5.158069883527454,
+      "grad_norm": 13.537564277648926,
+      "learning_rate": 5e-06,
+      "loss": 0.703,
+      "num_input_tokens_seen": 97137488,
+      "step": 1550
+    },
+    {
+      "epoch": 5.158069883527454,
+      "loss": 0.791496992111206,
+      "loss_ce": 0.0006034679245203733,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 97137488,
+      "step": 1550
+    },
+    {
+      "epoch": 5.161397670549085,
+      "grad_norm": 35.76241683959961,
+      "learning_rate": 5e-06,
+      "loss": 0.7945,
+      "num_input_tokens_seen": 97198004,
+      "step": 1551
+    },
+    {
+      "epoch": 5.161397670549085,
+      "loss": 0.7911405563354492,
+      "loss_ce": 2.8721469789161347e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 97198004,
+      "step": 1551
+    },
+    {
+      "epoch": 5.164725457570715,
+      "grad_norm": 14.66862964630127,
+      "learning_rate": 5e-06,
+      "loss": 0.6592,
+      "num_input_tokens_seen": 97260336,
+      "step": 1552
+    },
+    {
+      "epoch": 5.164725457570715,
+      "loss": 0.5582370162010193,
+      "loss_ce": 0.00025361799634993076,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 97260336,
+      "step": 1552
+    },
+    {
+      "epoch": 5.168053244592346,
+      "grad_norm": 8.499744415283203,
+      "learning_rate": 5e-06,
+      "loss": 0.5118,
+      "num_input_tokens_seen": 97322608,
+      "step": 1553
+    },
+    {
+      "epoch": 5.168053244592346,
+      "loss": 0.5915481448173523,
+      "loss_ce": 0.0005447297007776797,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 97322608,
+      "step": 1553
+    },
+    {
+      "epoch": 5.1713810316139766,
+      "grad_norm": 13.193866729736328,
+      "learning_rate": 5e-06,
+      "loss": 0.5383,
+      "num_input_tokens_seen": 97385600,
+      "step": 1554
+    },
+    {
+      "epoch": 5.1713810316139766,
+      "loss": 0.35793358087539673,
+      "loss_ce": 2.3437820345861837e-05,
+      "loss_iou": 0.095703125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 97385600,
+      "step": 1554
+    },
+    {
+      "epoch": 5.174708818635607,
+      "grad_norm": 19.962928771972656,
+      "learning_rate": 5e-06,
+      "loss": 0.6914,
+      "num_input_tokens_seen": 97447764,
+      "step": 1555
+    },
+    {
+      "epoch": 5.174708818635607,
+      "loss": 0.6495640277862549,
+      "loss_ce": 0.0005161368753761053,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 97447764,
+      "step": 1555
+    },
+    {
+      "epoch": 5.178036605657238,
+      "grad_norm": 15.584821701049805,
+      "learning_rate": 5e-06,
+      "loss": 0.737,
+      "num_input_tokens_seen": 97510692,
+      "step": 1556
+    },
+    {
+      "epoch": 5.178036605657238,
+      "loss": 0.7587304711341858,
+      "loss_ce": 0.0005517329555004835,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 97510692,
+      "step": 1556
+    },
+    {
+      "epoch": 5.181364392678868,
+      "grad_norm": 16.34670066833496,
+      "learning_rate": 5e-06,
+      "loss": 1.0405,
+      "num_input_tokens_seen": 97573088,
+      "step": 1557
+    },
+    {
+      "epoch": 5.181364392678868,
+      "loss": 1.0706849098205566,
+      "loss_ce": 0.00012822201824747026,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.08154296875,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 97573088,
+      "step": 1557
+    },
+    {
+      "epoch": 5.184692179700499,
+      "grad_norm": 24.74776840209961,
+      "learning_rate": 5e-06,
+      "loss": 0.7951,
+      "num_input_tokens_seen": 97636512,
+      "step": 1558
+    },
+    {
+      "epoch": 5.184692179700499,
+      "loss": 0.9686456918716431,
+      "loss_ce": 1.7786769603844732e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.96875,
+      "num_input_tokens_seen": 97636512,
+      "step": 1558
+    },
+    {
+      "epoch": 5.18801996672213,
+      "grad_norm": 8.181451797485352,
+      "learning_rate": 5e-06,
+      "loss": 0.7342,
+      "num_input_tokens_seen": 97699408,
+      "step": 1559
+    },
+    {
+      "epoch": 5.18801996672213,
+      "loss": 0.6913556456565857,
+      "loss_ce": 0.0006818252149969339,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.69140625,
+      "num_input_tokens_seen": 97699408,
+      "step": 1559
+    },
+    {
+      "epoch": 5.19134775374376,
+      "grad_norm": 9.973201751708984,
+      "learning_rate": 5e-06,
+      "loss": 0.7311,
+      "num_input_tokens_seen": 97761664,
+      "step": 1560
+    },
+    {
+      "epoch": 5.19134775374376,
+      "loss": 0.5677616596221924,
+      "loss_ce": 1.2609091754711699e-05,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 97761664,
+      "step": 1560
+    },
+    {
+      "epoch": 5.194675540765391,
+      "grad_norm": 13.940975189208984,
+      "learning_rate": 5e-06,
+      "loss": 0.5023,
+      "num_input_tokens_seen": 97823884,
+      "step": 1561
+    },
+    {
+      "epoch": 5.194675540765391,
+      "loss": 0.5250530242919922,
+      "loss_ce": 2.8577667762874626e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 97823884,
+      "step": 1561
+    },
+    {
+      "epoch": 5.1980033277870215,
+      "grad_norm": 10.562603950500488,
+      "learning_rate": 5e-06,
+      "loss": 0.6535,
+      "num_input_tokens_seen": 97886020,
+      "step": 1562
+    },
+    {
+      "epoch": 5.1980033277870215,
+      "loss": 0.6427997350692749,
+      "loss_ce": 0.0007098691421560943,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 97886020,
+      "step": 1562
+    },
+    {
+      "epoch": 5.201331114808652,
+      "grad_norm": 20.083114624023438,
+      "learning_rate": 5e-06,
+      "loss": 0.6058,
+      "num_input_tokens_seen": 97947632,
+      "step": 1563
+    },
+    {
+      "epoch": 5.201331114808652,
+      "loss": 0.564403772354126,
+      "loss_ce": 0.0003778576210606843,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 97947632,
+      "step": 1563
+    },
+    {
+      "epoch": 5.204658901830283,
+      "grad_norm": 34.775455474853516,
+      "learning_rate": 5e-06,
+      "loss": 0.8295,
+      "num_input_tokens_seen": 98010084,
+      "step": 1564
+    },
+    {
+      "epoch": 5.204658901830283,
+      "loss": 0.667966902256012,
+      "loss_ce": 5.918982424191199e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 98010084,
+      "step": 1564
+    },
+    {
+      "epoch": 5.207986688851913,
+      "grad_norm": 14.078563690185547,
+      "learning_rate": 5e-06,
+      "loss": 0.5324,
+      "num_input_tokens_seen": 98072816,
+      "step": 1565
+    },
+    {
+      "epoch": 5.207986688851913,
+      "loss": 0.29302480816841125,
+      "loss_ce": 0.00048331127618439496,
+      "loss_iou": 0.0693359375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 98072816,
+      "step": 1565
+    },
+    {
+      "epoch": 5.211314475873544,
+      "grad_norm": 12.865399360656738,
+      "learning_rate": 5e-06,
+      "loss": 0.6898,
+      "num_input_tokens_seen": 98135732,
+      "step": 1566
+    },
+    {
+      "epoch": 5.211314475873544,
+      "loss": 0.6527900695800781,
+      "loss_ce": 0.0006904720212332904,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 98135732,
+      "step": 1566
+    },
+    {
+      "epoch": 5.2146422628951745,
+      "grad_norm": 30.85588836669922,
+      "learning_rate": 5e-06,
+      "loss": 0.7293,
+      "num_input_tokens_seen": 98200008,
+      "step": 1567
+    },
+    {
+      "epoch": 5.2146422628951745,
+      "loss": 0.7736085057258606,
+      "loss_ce": 0.00041512242751196027,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 98200008,
+      "step": 1567
+    },
+    {
+      "epoch": 5.217970049916805,
+      "grad_norm": 29.402868270874023,
+      "learning_rate": 5e-06,
+      "loss": 0.8229,
+      "num_input_tokens_seen": 98264616,
+      "step": 1568
+    },
+    {
+      "epoch": 5.217970049916805,
+      "loss": 0.63280189037323,
+      "loss_ce": 0.00072178163100034,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 98264616,
+      "step": 1568
+    },
+    {
+      "epoch": 5.221297836938436,
+      "grad_norm": 20.66412925720215,
+      "learning_rate": 5e-06,
+      "loss": 0.6474,
+      "num_input_tokens_seen": 98327500,
+      "step": 1569
+    },
+    {
+      "epoch": 5.221297836938436,
+      "loss": 0.8841537833213806,
+      "loss_ce": 0.00024261744692921638,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 98327500,
+      "step": 1569
+    },
+    {
+      "epoch": 5.224625623960066,
+      "grad_norm": 7.796050071716309,
+      "learning_rate": 5e-06,
+      "loss": 0.6683,
+      "num_input_tokens_seen": 98390560,
+      "step": 1570
+    },
+    {
+      "epoch": 5.224625623960066,
+      "loss": 0.6554022431373596,
+      "loss_ce": 6.712816229992313e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 98390560,
+      "step": 1570
+    },
+    {
+      "epoch": 5.227953410981697,
+      "grad_norm": 22.678808212280273,
+      "learning_rate": 5e-06,
+      "loss": 0.743,
+      "num_input_tokens_seen": 98451556,
+      "step": 1571
+    },
+    {
+      "epoch": 5.227953410981697,
+      "loss": 0.8813857436180115,
+      "loss_ce": 3.806907989201136e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 98451556,
+      "step": 1571
+    },
+    {
+      "epoch": 5.231281198003328,
+      "grad_norm": 41.750709533691406,
+      "learning_rate": 5e-06,
+      "loss": 1.0732,
+      "num_input_tokens_seen": 98514720,
+      "step": 1572
+    },
+    {
+      "epoch": 5.231281198003328,
+      "loss": 1.1592261791229248,
+      "loss_ce": 0.0010230218758806586,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0908203125,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 98514720,
+      "step": 1572
+    },
+    {
+      "epoch": 5.234608985024958,
+      "grad_norm": 45.86041259765625,
+      "learning_rate": 5e-06,
+      "loss": 0.7024,
+      "num_input_tokens_seen": 98578124,
+      "step": 1573
+    },
+    {
+      "epoch": 5.234608985024958,
+      "loss": 0.7642949819564819,
+      "loss_ce": 0.0006078842561691999,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 98578124,
+      "step": 1573
+    },
+    {
+      "epoch": 5.237936772046589,
+      "grad_norm": 13.958196640014648,
+      "learning_rate": 5e-06,
+      "loss": 0.8009,
+      "num_input_tokens_seen": 98642292,
+      "step": 1574
+    },
+    {
+      "epoch": 5.237936772046589,
+      "loss": 0.6747276782989502,
+      "loss_ce": 0.0016319530550390482,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 98642292,
+      "step": 1574
+    },
+    {
+      "epoch": 5.241264559068219,
+      "grad_norm": 1243.618896484375,
+      "learning_rate": 5e-06,
+      "loss": 0.774,
+      "num_input_tokens_seen": 98704420,
+      "step": 1575
+    },
+    {
+      "epoch": 5.241264559068219,
+      "loss": 0.8179287314414978,
+      "loss_ce": 0.0005458915256895125,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 98704420,
+      "step": 1575
+    },
+    {
+      "epoch": 5.24459234608985,
+      "grad_norm": 10.360309600830078,
+      "learning_rate": 5e-06,
+      "loss": 0.7435,
+      "num_input_tokens_seen": 98769476,
+      "step": 1576
+    },
+    {
+      "epoch": 5.24459234608985,
+      "loss": 0.7698988914489746,
+      "loss_ce": 0.0002455537032801658,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 98769476,
+      "step": 1576
+    },
+    {
+      "epoch": 5.247920133111481,
+      "grad_norm": 7.8587565422058105,
+      "learning_rate": 5e-06,
+      "loss": 0.4282,
+      "num_input_tokens_seen": 98830068,
+      "step": 1577
+    },
+    {
+      "epoch": 5.247920133111481,
+      "loss": 0.5417072772979736,
+      "loss_ce": 2.0255403796909377e-05,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 98830068,
+      "step": 1577
+    },
+    {
+      "epoch": 5.251247920133111,
+      "grad_norm": 8.830144882202148,
+      "learning_rate": 5e-06,
+      "loss": 0.7409,
+      "num_input_tokens_seen": 98892736,
+      "step": 1578
+    },
+    {
+      "epoch": 5.251247920133111,
+      "loss": 0.6595664620399475,
+      "loss_ce": 2.0516697986749932e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 98892736,
+      "step": 1578
+    },
+    {
+      "epoch": 5.254575707154742,
+      "grad_norm": 15.969278335571289,
+      "learning_rate": 5e-06,
+      "loss": 0.6557,
+      "num_input_tokens_seen": 98956260,
+      "step": 1579
+    },
+    {
+      "epoch": 5.254575707154742,
+      "loss": 0.4240087568759918,
+      "loss_ce": 0.0003026947088073939,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.423828125,
+      "num_input_tokens_seen": 98956260,
+      "step": 1579
+    },
+    {
+      "epoch": 5.2579034941763725,
+      "grad_norm": 12.759567260742188,
+      "learning_rate": 5e-06,
+      "loss": 0.5799,
+      "num_input_tokens_seen": 99017636,
+      "step": 1580
+    },
+    {
+      "epoch": 5.2579034941763725,
+      "loss": 0.4847814440727234,
+      "loss_ce": 0.001138861756771803,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.484375,
+      "num_input_tokens_seen": 99017636,
+      "step": 1580
+    },
+    {
+      "epoch": 5.261231281198003,
+      "grad_norm": 7.065427780151367,
+      "learning_rate": 5e-06,
+      "loss": 0.6962,
+      "num_input_tokens_seen": 99081064,
+      "step": 1581
+    },
+    {
+      "epoch": 5.261231281198003,
+      "loss": 0.7337771654129028,
+      "loss_ce": 0.0007449376862496138,
+      "loss_iou": 0.25,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 99081064,
+      "step": 1581
+    },
+    {
+      "epoch": 5.264559068219634,
+      "grad_norm": 12.841547012329102,
+      "learning_rate": 5e-06,
+      "loss": 0.7063,
+      "num_input_tokens_seen": 99143616,
+      "step": 1582
+    },
+    {
+      "epoch": 5.264559068219634,
+      "loss": 0.6888269186019897,
+      "loss_ce": 0.00022827113571111113,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 99143616,
+      "step": 1582
+    },
+    {
+      "epoch": 5.267886855241264,
+      "grad_norm": 21.898542404174805,
+      "learning_rate": 5e-06,
+      "loss": 0.7441,
+      "num_input_tokens_seen": 99206620,
+      "step": 1583
+    },
+    {
+      "epoch": 5.267886855241264,
+      "loss": 1.0425233840942383,
+      "loss_ce": 0.00040918824379332364,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.044921875,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 99206620,
+      "step": 1583
+    },
+    {
+      "epoch": 5.271214642262895,
+      "grad_norm": 18.596426010131836,
+      "learning_rate": 5e-06,
+      "loss": 0.7307,
+      "num_input_tokens_seen": 99270064,
+      "step": 1584
+    },
+    {
+      "epoch": 5.271214642262895,
+      "loss": 0.7801717519760132,
+      "loss_ce": 0.00038661700091324747,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 99270064,
+      "step": 1584
+    },
+    {
+      "epoch": 5.2745424292845255,
+      "grad_norm": 23.57819366455078,
+      "learning_rate": 5e-06,
+      "loss": 0.8169,
+      "num_input_tokens_seen": 99333904,
+      "step": 1585
+    },
+    {
+      "epoch": 5.2745424292845255,
+      "loss": 0.5337511897087097,
+      "loss_ce": 0.0007311389781534672,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 99333904,
+      "step": 1585
+    },
+    {
+      "epoch": 5.277870216306156,
+      "grad_norm": 37.10131072998047,
+      "learning_rate": 5e-06,
+      "loss": 0.7919,
+      "num_input_tokens_seen": 99398088,
+      "step": 1586
+    },
+    {
+      "epoch": 5.277870216306156,
+      "loss": 0.803447961807251,
+      "loss_ce": 0.0004694595991168171,
+      "loss_iou": 0.328125,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 99398088,
+      "step": 1586
+    },
+    {
+      "epoch": 5.281198003327787,
+      "grad_norm": 19.156675338745117,
+      "learning_rate": 5e-06,
+      "loss": 0.6396,
+      "num_input_tokens_seen": 99459368,
+      "step": 1587
+    },
+    {
+      "epoch": 5.281198003327787,
+      "loss": 0.9905081391334534,
+      "loss_ce": 2.9618571716127917e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.9921875,
+      "num_input_tokens_seen": 99459368,
+      "step": 1587
+    },
+    {
+      "epoch": 5.284525790349417,
+      "grad_norm": 8.31318473815918,
+      "learning_rate": 5e-06,
+      "loss": 0.4815,
+      "num_input_tokens_seen": 99520824,
+      "step": 1588
+    },
+    {
+      "epoch": 5.284525790349417,
+      "loss": 0.33191707730293274,
+      "loss_ce": 0.0004961899248883128,
+      "loss_iou": 0.09375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 99520824,
+      "step": 1588
+    },
+    {
+      "epoch": 5.287853577371048,
+      "grad_norm": 9.956439971923828,
+      "learning_rate": 5e-06,
+      "loss": 0.6581,
+      "num_input_tokens_seen": 99581932,
+      "step": 1589
+    },
+    {
+      "epoch": 5.287853577371048,
+      "loss": 0.7131649255752563,
+      "loss_ce": 3.017848575836979e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 99581932,
+      "step": 1589
+    },
+    {
+      "epoch": 5.291181364392679,
+      "grad_norm": 7.714879512786865,
+      "learning_rate": 5e-06,
+      "loss": 0.7462,
+      "num_input_tokens_seen": 99644652,
+      "step": 1590
+    },
+    {
+      "epoch": 5.291181364392679,
+      "loss": 0.7490200996398926,
+      "loss_ce": 0.0003017770650330931,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 99644652,
+      "step": 1590
+    },
+    {
+      "epoch": 5.294509151414309,
+      "grad_norm": 28.146652221679688,
+      "learning_rate": 5e-06,
+      "loss": 0.6564,
+      "num_input_tokens_seen": 99708500,
+      "step": 1591
+    },
+    {
+      "epoch": 5.294509151414309,
+      "loss": 0.8539837598800659,
+      "loss_ce": 0.0013836309080943465,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.8515625,
+      "num_input_tokens_seen": 99708500,
+      "step": 1591
+    },
+    {
+      "epoch": 5.29783693843594,
+      "grad_norm": 19.461881637573242,
+      "learning_rate": 5e-06,
+      "loss": 0.6438,
+      "num_input_tokens_seen": 99771208,
+      "step": 1592
+    },
+    {
+      "epoch": 5.29783693843594,
+      "loss": 0.6133031845092773,
+      "loss_ce": 2.1985513740219176e-05,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 99771208,
+      "step": 1592
+    },
+    {
+      "epoch": 5.3011647254575704,
+      "grad_norm": 26.864931106567383,
+      "learning_rate": 5e-06,
+      "loss": 0.5504,
+      "num_input_tokens_seen": 99833760,
+      "step": 1593
+    },
+    {
+      "epoch": 5.3011647254575704,
+      "loss": 0.5138239860534668,
+      "loss_ce": 0.0006403637235052884,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 99833760,
+      "step": 1593
+    },
+    {
+      "epoch": 5.304492512479201,
+      "grad_norm": 8.363646507263184,
+      "learning_rate": 5e-06,
+      "loss": 0.7576,
+      "num_input_tokens_seen": 99897804,
+      "step": 1594
+    },
+    {
+      "epoch": 5.304492512479201,
+      "loss": 0.7315871715545654,
+      "loss_ce": 0.0003860450815409422,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 99897804,
+      "step": 1594
+    },
+    {
+      "epoch": 5.307820299500832,
+      "grad_norm": 69.57489013671875,
+      "learning_rate": 5e-06,
+      "loss": 0.5305,
+      "num_input_tokens_seen": 99960196,
+      "step": 1595
+    },
+    {
+      "epoch": 5.307820299500832,
+      "loss": 0.6942192912101746,
+      "loss_ce": 0.00018853397341445088,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 99960196,
+      "step": 1595
+    },
+    {
+      "epoch": 5.311148086522462,
+      "grad_norm": 11.756884574890137,
+      "learning_rate": 5e-06,
+      "loss": 0.6475,
+      "num_input_tokens_seen": 100022584,
+      "step": 1596
+    },
+    {
+      "epoch": 5.311148086522462,
+      "loss": 0.7136390805244446,
+      "loss_ce": 1.6043655705288984e-05,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.058837890625,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 100022584,
+      "step": 1596
+    },
+    {
+      "epoch": 5.314475873544093,
+      "grad_norm": 8.837007522583008,
+      "learning_rate": 5e-06,
+      "loss": 0.6514,
+      "num_input_tokens_seen": 100086124,
+      "step": 1597
+    },
+    {
+      "epoch": 5.314475873544093,
+      "loss": 0.6199404001235962,
+      "loss_ce": 0.0003114750434178859,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 100086124,
+      "step": 1597
+    },
+    {
+      "epoch": 5.3178036605657235,
+      "grad_norm": 8.616264343261719,
+      "learning_rate": 5e-06,
+      "loss": 0.7488,
+      "num_input_tokens_seen": 100148952,
+      "step": 1598
+    },
+    {
+      "epoch": 5.3178036605657235,
+      "loss": 0.45825719833374023,
+      "loss_ce": 5.2268542276578955e-06,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.458984375,
+      "num_input_tokens_seen": 100148952,
+      "step": 1598
+    },
+    {
+      "epoch": 5.321131447587354,
+      "grad_norm": 12.141873359680176,
+      "learning_rate": 5e-06,
+      "loss": 0.5747,
+      "num_input_tokens_seen": 100210792,
+      "step": 1599
+    },
+    {
+      "epoch": 5.321131447587354,
+      "loss": 0.4171640872955322,
+      "loss_ce": 0.0004160656244494021,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 100210792,
+      "step": 1599
+    },
+    {
+      "epoch": 5.324459234608985,
+      "grad_norm": 17.452781677246094,
+      "learning_rate": 5e-06,
+      "loss": 0.5113,
+      "num_input_tokens_seen": 100272944,
+      "step": 1600
+    },
+    {
+      "epoch": 5.324459234608985,
+      "loss": 0.5932729840278625,
+      "loss_ce": 0.0009878204436972737,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 100272944,
+      "step": 1600
+    },
+    {
+      "epoch": 5.327787021630615,
+      "grad_norm": 13.12947940826416,
+      "learning_rate": 5e-06,
+      "loss": 0.4211,
+      "num_input_tokens_seen": 100336000,
+      "step": 1601
+    },
+    {
+      "epoch": 5.327787021630615,
+      "loss": 0.34248068928718567,
+      "loss_ce": 0.00013448798563331366,
+      "loss_iou": 0.12255859375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 100336000,
+      "step": 1601
+    },
+    {
+      "epoch": 5.331114808652246,
+      "grad_norm": 16.778797149658203,
+      "learning_rate": 5e-06,
+      "loss": 0.9173,
+      "num_input_tokens_seen": 100398688,
+      "step": 1602
+    },
+    {
+      "epoch": 5.331114808652246,
+      "loss": 0.812654972076416,
+      "loss_ce": 0.00015499009168706834,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 100398688,
+      "step": 1602
+    },
+    {
+      "epoch": 5.334442595673877,
+      "grad_norm": 41.81510543823242,
+      "learning_rate": 5e-06,
+      "loss": 0.7542,
+      "num_input_tokens_seen": 100461044,
+      "step": 1603
+    },
+    {
+      "epoch": 5.334442595673877,
+      "loss": 0.9035730361938477,
+      "loss_ce": 8.593749953433871e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.90234375,
+      "num_input_tokens_seen": 100461044,
+      "step": 1603
+    },
+    {
+      "epoch": 5.337770382695507,
+      "grad_norm": 27.792001724243164,
+      "learning_rate": 5e-06,
+      "loss": 0.6624,
+      "num_input_tokens_seen": 100524272,
+      "step": 1604
+    },
+    {
+      "epoch": 5.337770382695507,
+      "loss": 0.4387521743774414,
+      "loss_ce": 3.14875396725256e-05,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.439453125,
+      "num_input_tokens_seen": 100524272,
+      "step": 1604
+    },
+    {
+      "epoch": 5.341098169717138,
+      "grad_norm": 10.232015609741211,
+      "learning_rate": 5e-06,
+      "loss": 0.6419,
+      "num_input_tokens_seen": 100586896,
+      "step": 1605
+    },
+    {
+      "epoch": 5.341098169717138,
+      "loss": 0.7033101320266724,
+      "loss_ce": 0.00018513888062443584,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 100586896,
+      "step": 1605
+    },
+    {
+      "epoch": 5.344425956738768,
+      "grad_norm": 9.116806983947754,
+      "learning_rate": 5e-06,
+      "loss": 0.5147,
+      "num_input_tokens_seen": 100649132,
+      "step": 1606
+    },
+    {
+      "epoch": 5.344425956738768,
+      "loss": 0.42553964257240295,
+      "loss_ce": 2.5271147023886442e-06,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 100649132,
+      "step": 1606
+    },
+    {
+      "epoch": 5.347753743760399,
+      "grad_norm": 24.01177978515625,
+      "learning_rate": 5e-06,
+      "loss": 0.7176,
+      "num_input_tokens_seen": 100712640,
+      "step": 1607
+    },
+    {
+      "epoch": 5.347753743760399,
+      "loss": 0.8379149436950684,
+      "loss_ce": 0.0006346513982862234,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 100712640,
+      "step": 1607
+    },
+    {
+      "epoch": 5.35108153078203,
+      "grad_norm": 17.368818283081055,
+      "learning_rate": 5e-06,
+      "loss": 0.6702,
+      "num_input_tokens_seen": 100775344,
+      "step": 1608
+    },
+    {
+      "epoch": 5.35108153078203,
+      "loss": 0.8547390699386597,
+      "loss_ce": 2.7412108920543687e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 100775344,
+      "step": 1608
+    },
+    {
+      "epoch": 5.35440931780366,
+      "grad_norm": 19.727764129638672,
+      "learning_rate": 5e-06,
+      "loss": 0.7125,
+      "num_input_tokens_seen": 100839532,
+      "step": 1609
+    },
+    {
+      "epoch": 5.35440931780366,
+      "loss": 0.8060716390609741,
+      "loss_ce": 0.0013841248583048582,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 100839532,
+      "step": 1609
+    },
+    {
+      "epoch": 5.357737104825291,
+      "grad_norm": 14.09434700012207,
+      "learning_rate": 5e-06,
+      "loss": 0.6749,
+      "num_input_tokens_seen": 100903348,
+      "step": 1610
+    },
+    {
+      "epoch": 5.357737104825291,
+      "loss": 0.7217329740524292,
+      "loss_ce": 0.0001753973338054493,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 100903348,
+      "step": 1610
+    },
+    {
+      "epoch": 5.3610648918469215,
+      "grad_norm": 36.83820343017578,
+      "learning_rate": 5e-06,
+      "loss": 0.5817,
+      "num_input_tokens_seen": 100964912,
+      "step": 1611
+    },
+    {
+      "epoch": 5.3610648918469215,
+      "loss": 0.6560474634170532,
+      "loss_ce": 4.160653043072671e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 100964912,
+      "step": 1611
+    },
+    {
+      "epoch": 5.364392678868552,
+      "grad_norm": 11.305191993713379,
+      "learning_rate": 5e-06,
+      "loss": 0.4619,
+      "num_input_tokens_seen": 101026424,
+      "step": 1612
+    },
+    {
+      "epoch": 5.364392678868552,
+      "loss": 0.6465030908584595,
+      "loss_ce": 1.8739890947472304e-05,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 101026424,
+      "step": 1612
+    },
+    {
+      "epoch": 5.367720465890183,
+      "grad_norm": 11.862913131713867,
+      "learning_rate": 5e-06,
+      "loss": 0.8137,
+      "num_input_tokens_seen": 101090924,
+      "step": 1613
+    },
+    {
+      "epoch": 5.367720465890183,
+      "loss": 0.8782895803451538,
+      "loss_ce": 0.00023779345792718232,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 101090924,
+      "step": 1613
+    },
+    {
+      "epoch": 5.371048252911813,
+      "grad_norm": 18.821861267089844,
+      "learning_rate": 5e-06,
+      "loss": 0.3774,
+      "num_input_tokens_seen": 101153632,
+      "step": 1614
+    },
+    {
+      "epoch": 5.371048252911813,
+      "loss": 0.4713307321071625,
+      "loss_ce": 1.724835419736337e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.470703125,
+      "num_input_tokens_seen": 101153632,
+      "step": 1614
+    },
+    {
+      "epoch": 5.374376039933444,
+      "grad_norm": 18.81785774230957,
+      "learning_rate": 5e-06,
+      "loss": 0.4723,
+      "num_input_tokens_seen": 101216952,
+      "step": 1615
+    },
+    {
+      "epoch": 5.374376039933444,
+      "loss": 0.5333555936813354,
+      "loss_ce": 0.0013731429353356361,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 101216952,
+      "step": 1615
+    },
+    {
+      "epoch": 5.3777038269550745,
+      "grad_norm": 8.743084907531738,
+      "learning_rate": 5e-06,
+      "loss": 0.5347,
+      "num_input_tokens_seen": 101278632,
+      "step": 1616
+    },
+    {
+      "epoch": 5.3777038269550745,
+      "loss": 0.5145453214645386,
+      "loss_ce": 0.00014100028784014285,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 101278632,
+      "step": 1616
+    },
+    {
+      "epoch": 5.381031613976705,
+      "grad_norm": 15.498371124267578,
+      "learning_rate": 5e-06,
+      "loss": 0.8427,
+      "num_input_tokens_seen": 101341908,
+      "step": 1617
+    },
+    {
+      "epoch": 5.381031613976705,
+      "loss": 0.5787339210510254,
+      "loss_ce": 0.00012060911831213161,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 101341908,
+      "step": 1617
+    },
+    {
+      "epoch": 5.384359400998336,
+      "grad_norm": 25.164121627807617,
+      "learning_rate": 5e-06,
+      "loss": 0.5582,
+      "num_input_tokens_seen": 101403536,
+      "step": 1618
+    },
+    {
+      "epoch": 5.384359400998336,
+      "loss": 0.4660763144493103,
+      "loss_ce": 1.1841300874948502e-05,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.466796875,
+      "num_input_tokens_seen": 101403536,
+      "step": 1618
+    },
+    {
+      "epoch": 5.387687188019966,
+      "grad_norm": 10.2332763671875,
+      "learning_rate": 5e-06,
+      "loss": 0.6704,
+      "num_input_tokens_seen": 101467088,
+      "step": 1619
+    },
+    {
+      "epoch": 5.387687188019966,
+      "loss": 0.490899920463562,
+      "loss_ce": 0.00042139904689975083,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.490234375,
+      "num_input_tokens_seen": 101467088,
+      "step": 1619
+    },
+    {
+      "epoch": 5.391014975041597,
+      "grad_norm": 19.87059783935547,
+      "learning_rate": 5e-06,
+      "loss": 0.5354,
+      "num_input_tokens_seen": 101529148,
+      "step": 1620
+    },
+    {
+      "epoch": 5.391014975041597,
+      "loss": 0.5794316530227661,
+      "loss_ce": 0.0011235260171815753,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 101529148,
+      "step": 1620
+    },
+    {
+      "epoch": 5.394342762063228,
+      "grad_norm": 11.307812690734863,
+      "learning_rate": 5e-06,
+      "loss": 0.6216,
+      "num_input_tokens_seen": 101592284,
+      "step": 1621
+    },
+    {
+      "epoch": 5.394342762063228,
+      "loss": 0.6232434511184692,
+      "loss_ce": 0.0006848212797194719,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 101592284,
+      "step": 1621
+    },
+    {
+      "epoch": 5.397670549084858,
+      "grad_norm": 22.18927001953125,
+      "learning_rate": 5e-06,
+      "loss": 0.6218,
+      "num_input_tokens_seen": 101654680,
+      "step": 1622
+    },
+    {
+      "epoch": 5.397670549084858,
+      "loss": 0.717785120010376,
+      "loss_ce": 0.0004999724333174527,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 101654680,
+      "step": 1622
+    },
+    {
+      "epoch": 5.400998336106489,
+      "grad_norm": 14.127787590026855,
+      "learning_rate": 5e-06,
+      "loss": 0.7036,
+      "num_input_tokens_seen": 101717188,
+      "step": 1623
+    },
+    {
+      "epoch": 5.400998336106489,
+      "loss": 0.8181177377700806,
+      "loss_ce": 0.00036871584597975016,
+      "loss_iou": 0.28125,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 101717188,
+      "step": 1623
+    },
+    {
+      "epoch": 5.404326123128119,
+      "grad_norm": 15.851911544799805,
+      "learning_rate": 5e-06,
+      "loss": 0.7727,
+      "num_input_tokens_seen": 101780468,
+      "step": 1624
+    },
+    {
+      "epoch": 5.404326123128119,
+      "loss": 0.8187301158905029,
+      "loss_ce": 0.00018765513959806412,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 101780468,
+      "step": 1624
+    },
+    {
+      "epoch": 5.40765391014975,
+      "grad_norm": 7.549576759338379,
+      "learning_rate": 5e-06,
+      "loss": 0.4095,
+      "num_input_tokens_seen": 101842308,
+      "step": 1625
+    },
+    {
+      "epoch": 5.40765391014975,
+      "loss": 0.38657066226005554,
+      "loss_ce": 0.0008284934447146952,
+      "loss_iou": 0.08984375,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 101842308,
+      "step": 1625
+    },
+    {
+      "epoch": 5.410981697171381,
+      "grad_norm": 9.882723808288574,
+      "learning_rate": 5e-06,
+      "loss": 0.6379,
+      "num_input_tokens_seen": 101903744,
+      "step": 1626
+    },
+    {
+      "epoch": 5.410981697171381,
+      "loss": 0.6620615720748901,
+      "loss_ce": 0.0009287626016885042,
+      "loss_iou": 0.25,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 101903744,
+      "step": 1626
+    },
+    {
+      "epoch": 5.414309484193011,
+      "grad_norm": 14.373669624328613,
+      "learning_rate": 5e-06,
+      "loss": 0.8154,
+      "num_input_tokens_seen": 101967284,
+      "step": 1627
+    },
+    {
+      "epoch": 5.414309484193011,
+      "loss": 0.6620774865150452,
+      "loss_ce": 0.0007005495717749,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 101967284,
+      "step": 1627
+    },
+    {
+      "epoch": 5.417637271214642,
+      "grad_norm": 18.5496768951416,
+      "learning_rate": 5e-06,
+      "loss": 0.6312,
+      "num_input_tokens_seen": 102029524,
+      "step": 1628
+    },
+    {
+      "epoch": 5.417637271214642,
+      "loss": 0.7796862721443176,
+      "loss_ce": 0.0016101216897368431,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 102029524,
+      "step": 1628
+    },
+    {
+      "epoch": 5.4209650582362725,
+      "grad_norm": 14.205424308776855,
+      "learning_rate": 5e-06,
+      "loss": 0.5817,
+      "num_input_tokens_seen": 102093696,
+      "step": 1629
+    },
+    {
+      "epoch": 5.4209650582362725,
+      "loss": 0.5077414512634277,
+      "loss_ce": 0.00017305587243754417,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.5078125,
+      "num_input_tokens_seen": 102093696,
+      "step": 1629
+    },
+    {
+      "epoch": 5.424292845257903,
+      "grad_norm": 9.155461311340332,
+      "learning_rate": 5e-06,
+      "loss": 0.5884,
+      "num_input_tokens_seen": 102155176,
+      "step": 1630
+    },
+    {
+      "epoch": 5.424292845257903,
+      "loss": 0.5432171821594238,
+      "loss_ce": 0.0003704904520418495,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 102155176,
+      "step": 1630
+    },
+    {
+      "epoch": 5.427620632279534,
+      "grad_norm": 8.620030403137207,
+      "learning_rate": 5e-06,
+      "loss": 0.6747,
+      "num_input_tokens_seen": 102217376,
+      "step": 1631
+    },
+    {
+      "epoch": 5.427620632279534,
+      "loss": 0.6632170081138611,
+      "loss_ce": 9.006463187688496e-06,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 102217376,
+      "step": 1631
+    },
+    {
+      "epoch": 5.430948419301164,
+      "grad_norm": 10.027673721313477,
+      "learning_rate": 5e-06,
+      "loss": 0.5378,
+      "num_input_tokens_seen": 102281180,
+      "step": 1632
+    },
+    {
+      "epoch": 5.430948419301164,
+      "loss": 0.47565510869026184,
+      "loss_ce": 0.00019121626974083483,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.474609375,
+      "num_input_tokens_seen": 102281180,
+      "step": 1632
+    },
+    {
+      "epoch": 5.434276206322795,
+      "grad_norm": 52.82054901123047,
+      "learning_rate": 5e-06,
+      "loss": 0.8202,
+      "num_input_tokens_seen": 102343984,
+      "step": 1633
+    },
+    {
+      "epoch": 5.434276206322795,
+      "loss": 0.6174899935722351,
+      "loss_ce": 0.0004245634190738201,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 102343984,
+      "step": 1633
+    },
+    {
+      "epoch": 5.437603993344426,
+      "grad_norm": 13.176911354064941,
+      "learning_rate": 5e-06,
+      "loss": 0.9236,
+      "num_input_tokens_seen": 102407440,
+      "step": 1634
+    },
+    {
+      "epoch": 5.437603993344426,
+      "loss": 0.7702251076698303,
+      "loss_ce": 0.0005717647145502269,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 102407440,
+      "step": 1634
+    },
+    {
+      "epoch": 5.440931780366056,
+      "grad_norm": 10.518148422241211,
+      "learning_rate": 5e-06,
+      "loss": 0.7139,
+      "num_input_tokens_seen": 102470212,
+      "step": 1635
+    },
+    {
+      "epoch": 5.440931780366056,
+      "loss": 0.7326650619506836,
+      "loss_ce": 0.0007314276299439371,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 102470212,
+      "step": 1635
+    },
+    {
+      "epoch": 5.444259567387687,
+      "grad_norm": 15.578624725341797,
+      "learning_rate": 5e-06,
+      "loss": 0.8894,
+      "num_input_tokens_seen": 102534180,
+      "step": 1636
+    },
+    {
+      "epoch": 5.444259567387687,
+      "loss": 1.045713186264038,
+      "loss_ce": 0.0006692318129353225,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.07275390625,
+      "loss_xval": 1.046875,
+      "num_input_tokens_seen": 102534180,
+      "step": 1636
+    },
+    {
+      "epoch": 5.447587354409317,
+      "grad_norm": 11.064337730407715,
+      "learning_rate": 5e-06,
+      "loss": 0.6611,
+      "num_input_tokens_seen": 102596244,
+      "step": 1637
+    },
+    {
+      "epoch": 5.447587354409317,
+      "loss": 0.8054574728012085,
+      "loss_ce": 0.00015959216398186982,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 102596244,
+      "step": 1637
+    },
+    {
+      "epoch": 5.450915141430948,
+      "grad_norm": 33.97342300415039,
+      "learning_rate": 5e-06,
+      "loss": 0.6283,
+      "num_input_tokens_seen": 102658532,
+      "step": 1638
+    },
+    {
+      "epoch": 5.450915141430948,
+      "loss": 0.6672518849372864,
+      "loss_ce": 0.0009921238524839282,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 102658532,
+      "step": 1638
+    },
+    {
+      "epoch": 5.454242928452579,
+      "grad_norm": 8.071794509887695,
+      "learning_rate": 5e-06,
+      "loss": 0.6823,
+      "num_input_tokens_seen": 102720076,
+      "step": 1639
+    },
+    {
+      "epoch": 5.454242928452579,
+      "loss": 0.8720945715904236,
+      "loss_ce": 2.4301192752318457e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 102720076,
+      "step": 1639
+    },
+    {
+      "epoch": 5.457570715474209,
+      "grad_norm": 9.842574119567871,
+      "learning_rate": 5e-06,
+      "loss": 0.7563,
+      "num_input_tokens_seen": 102782416,
+      "step": 1640
+    },
+    {
+      "epoch": 5.457570715474209,
+      "loss": 0.6718653440475464,
+      "loss_ce": 0.0009668688289821148,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 102782416,
+      "step": 1640
+    },
+    {
+      "epoch": 5.46089850249584,
+      "grad_norm": 16.268909454345703,
+      "learning_rate": 5e-06,
+      "loss": 0.7966,
+      "num_input_tokens_seen": 102845652,
+      "step": 1641
+    },
+    {
+      "epoch": 5.46089850249584,
+      "loss": 0.8888282775878906,
+      "loss_ce": 0.0006446776678785682,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 102845652,
+      "step": 1641
+    },
+    {
+      "epoch": 5.4642262895174705,
+      "grad_norm": 14.76564884185791,
+      "learning_rate": 5e-06,
+      "loss": 0.7013,
+      "num_input_tokens_seen": 102906500,
+      "step": 1642
+    },
+    {
+      "epoch": 5.4642262895174705,
+      "loss": 0.7110059261322021,
+      "loss_ce": 0.0001294732792302966,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 102906500,
+      "step": 1642
+    },
+    {
+      "epoch": 5.467554076539101,
+      "grad_norm": 19.5927791595459,
+      "learning_rate": 5e-06,
+      "loss": 0.5605,
+      "num_input_tokens_seen": 102969944,
+      "step": 1643
+    },
+    {
+      "epoch": 5.467554076539101,
+      "loss": 0.7769869565963745,
+      "loss_ce": 9.381456948176492e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 102969944,
+      "step": 1643
+    },
+    {
+      "epoch": 5.470881863560733,
+      "grad_norm": 16.91885757446289,
+      "learning_rate": 5e-06,
+      "loss": 0.7235,
+      "num_input_tokens_seen": 103033352,
+      "step": 1644
+    },
+    {
+      "epoch": 5.470881863560733,
+      "loss": 0.531306803226471,
+      "loss_ce": 0.0009113232372328639,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 103033352,
+      "step": 1644
+    },
+    {
+      "epoch": 5.474209650582363,
+      "grad_norm": 16.343990325927734,
+      "learning_rate": 5e-06,
+      "loss": 0.7818,
+      "num_input_tokens_seen": 103097296,
+      "step": 1645
+    },
+    {
+      "epoch": 5.474209650582363,
+      "loss": 1.0094261169433594,
+      "loss_ce": 0.0006371331983245909,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.058837890625,
+      "loss_xval": 1.0078125,
+      "num_input_tokens_seen": 103097296,
+      "step": 1645
+    },
+    {
+      "epoch": 5.477537437603994,
+      "grad_norm": 7.790471076965332,
+      "learning_rate": 5e-06,
+      "loss": 0.5469,
+      "num_input_tokens_seen": 103159720,
+      "step": 1646
+    },
+    {
+      "epoch": 5.477537437603994,
+      "loss": 0.3706333041191101,
+      "loss_ce": 2.7853264327859506e-05,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 103159720,
+      "step": 1646
+    },
+    {
+      "epoch": 5.480865224625624,
+      "grad_norm": 37.15055847167969,
+      "learning_rate": 5e-06,
+      "loss": 0.7048,
+      "num_input_tokens_seen": 103222660,
+      "step": 1647
+    },
+    {
+      "epoch": 5.480865224625624,
+      "loss": 0.6658989787101746,
+      "loss_ce": 0.0004937172634527087,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 103222660,
+      "step": 1647
+    },
+    {
+      "epoch": 5.484193011647255,
+      "grad_norm": 27.953083038330078,
+      "learning_rate": 5e-06,
+      "loss": 0.8436,
+      "num_input_tokens_seen": 103286316,
+      "step": 1648
+    },
+    {
+      "epoch": 5.484193011647255,
+      "loss": 0.8316751718521118,
+      "loss_ce": 1.0130894224857911e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0625,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 103286316,
+      "step": 1648
+    },
+    {
+      "epoch": 5.487520798668886,
+      "grad_norm": 20.30833625793457,
+      "learning_rate": 5e-06,
+      "loss": 0.6182,
+      "num_input_tokens_seen": 103349840,
+      "step": 1649
+    },
+    {
+      "epoch": 5.487520798668886,
+      "loss": 0.5982181429862976,
+      "loss_ce": 1.259712917089928e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 103349840,
+      "step": 1649
+    },
+    {
+      "epoch": 5.490848585690516,
+      "grad_norm": 20.08171272277832,
+      "learning_rate": 5e-06,
+      "loss": 0.5541,
+      "num_input_tokens_seen": 103413016,
+      "step": 1650
+    },
+    {
+      "epoch": 5.490848585690516,
+      "loss": 0.7060139775276184,
+      "loss_ce": 0.0004475625464692712,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.70703125,
+      "num_input_tokens_seen": 103413016,
+      "step": 1650
+    },
+    {
+      "epoch": 5.494176372712147,
+      "grad_norm": 8.847830772399902,
+      "learning_rate": 5e-06,
+      "loss": 0.7636,
+      "num_input_tokens_seen": 103477308,
+      "step": 1651
+    },
+    {
+      "epoch": 5.494176372712147,
+      "loss": 0.9405819773674011,
+      "loss_ce": 0.0003964377101510763,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 103477308,
+      "step": 1651
+    },
+    {
+      "epoch": 5.4975041597337775,
+      "grad_norm": 13.779448509216309,
+      "learning_rate": 5e-06,
+      "loss": 0.835,
+      "num_input_tokens_seen": 103539928,
+      "step": 1652
+    },
+    {
+      "epoch": 5.4975041597337775,
+      "loss": 1.0582094192504883,
+      "loss_ce": 0.0008363999659195542,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0810546875,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 103539928,
+      "step": 1652
+    },
+    {
+      "epoch": 5.500831946755408,
+      "grad_norm": 21.17045021057129,
+      "learning_rate": 5e-06,
+      "loss": 0.6832,
+      "num_input_tokens_seen": 103602172,
+      "step": 1653
+    },
+    {
+      "epoch": 5.500831946755408,
+      "loss": 0.42666590213775635,
+      "loss_ce": 0.00039636611472815275,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 103602172,
+      "step": 1653
+    },
+    {
+      "epoch": 5.504159733777039,
+      "grad_norm": 20.944114685058594,
+      "learning_rate": 5e-06,
+      "loss": 0.6127,
+      "num_input_tokens_seen": 103663200,
+      "step": 1654
+    },
+    {
+      "epoch": 5.504159733777039,
+      "loss": 0.7828108668327332,
+      "loss_ce": 9.606579260434955e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 103663200,
+      "step": 1654
+    },
+    {
+      "epoch": 5.507487520798669,
+      "grad_norm": 8.944890022277832,
+      "learning_rate": 5e-06,
+      "loss": 0.6528,
+      "num_input_tokens_seen": 103726100,
+      "step": 1655
+    },
+    {
+      "epoch": 5.507487520798669,
+      "loss": 0.6580873727798462,
+      "loss_ce": 0.00012836034875363111,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 103726100,
+      "step": 1655
+    },
+    {
+      "epoch": 5.5108153078203,
+      "grad_norm": 11.251640319824219,
+      "learning_rate": 5e-06,
+      "loss": 0.7174,
+      "num_input_tokens_seen": 103788604,
+      "step": 1656
+    },
+    {
+      "epoch": 5.5108153078203,
+      "loss": 0.6210998296737671,
+      "loss_ce": 6.068248694646172e-06,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 103788604,
+      "step": 1656
+    },
+    {
+      "epoch": 5.5141430948419305,
+      "grad_norm": 18.31918716430664,
+      "learning_rate": 5e-06,
+      "loss": 0.6866,
+      "num_input_tokens_seen": 103851268,
+      "step": 1657
+    },
+    {
+      "epoch": 5.5141430948419305,
+      "loss": 0.780825138092041,
+      "loss_ce": 0.0007958478527143598,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 103851268,
+      "step": 1657
+    },
+    {
+      "epoch": 5.517470881863561,
+      "grad_norm": 17.294458389282227,
+      "learning_rate": 5e-06,
+      "loss": 0.4707,
+      "num_input_tokens_seen": 103912924,
+      "step": 1658
+    },
+    {
+      "epoch": 5.517470881863561,
+      "loss": 0.45224249362945557,
+      "loss_ce": 0.0002466700680088252,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.451171875,
+      "num_input_tokens_seen": 103912924,
+      "step": 1658
+    },
+    {
+      "epoch": 5.520798668885192,
+      "grad_norm": 8.51252269744873,
+      "learning_rate": 5e-06,
+      "loss": 0.6149,
+      "num_input_tokens_seen": 103975188,
+      "step": 1659
+    },
+    {
+      "epoch": 5.520798668885192,
+      "loss": 0.6649562120437622,
+      "loss_ce": 0.00040545634692534804,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 103975188,
+      "step": 1659
+    },
+    {
+      "epoch": 5.524126455906822,
+      "grad_norm": 15.996707916259766,
+      "learning_rate": 5e-06,
+      "loss": 0.7851,
+      "num_input_tokens_seen": 104037616,
+      "step": 1660
+    },
+    {
+      "epoch": 5.524126455906822,
+      "loss": 0.8272863626480103,
+      "loss_ce": 1.5841425920370966e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.828125,
+      "num_input_tokens_seen": 104037616,
+      "step": 1660
+    },
+    {
+      "epoch": 5.527454242928453,
+      "grad_norm": 12.95364761352539,
+      "learning_rate": 5e-06,
+      "loss": 0.5927,
+      "num_input_tokens_seen": 104099448,
+      "step": 1661
+    },
+    {
+      "epoch": 5.527454242928453,
+      "loss": 0.4973302185535431,
+      "loss_ce": 0.0005956004024483263,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.49609375,
+      "num_input_tokens_seen": 104099448,
+      "step": 1661
+    },
+    {
+      "epoch": 5.530782029950084,
+      "grad_norm": 11.21883773803711,
+      "learning_rate": 5e-06,
+      "loss": 0.6192,
+      "num_input_tokens_seen": 104161152,
+      "step": 1662
+    },
+    {
+      "epoch": 5.530782029950084,
+      "loss": 0.8068508505821228,
+      "loss_ce": 0.0006985386717133224,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 104161152,
+      "step": 1662
+    },
+    {
+      "epoch": 5.534109816971714,
+      "grad_norm": 13.665806770324707,
+      "learning_rate": 5e-06,
+      "loss": 0.7915,
+      "num_input_tokens_seen": 104224144,
+      "step": 1663
+    },
+    {
+      "epoch": 5.534109816971714,
+      "loss": 0.9511141180992126,
+      "loss_ce": 0.00043049981468357146,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.94921875,
+      "num_input_tokens_seen": 104224144,
+      "step": 1663
+    },
+    {
+      "epoch": 5.537437603993345,
+      "grad_norm": 8.892619132995605,
+      "learning_rate": 5e-06,
+      "loss": 0.8575,
+      "num_input_tokens_seen": 104287548,
+      "step": 1664
+    },
+    {
+      "epoch": 5.537437603993345,
+      "loss": 0.8013952970504761,
+      "loss_ce": 0.001346514793112874,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 104287548,
+      "step": 1664
+    },
+    {
+      "epoch": 5.5407653910149754,
+      "grad_norm": 8.533040046691895,
+      "learning_rate": 5e-06,
+      "loss": 0.6726,
+      "num_input_tokens_seen": 104348436,
+      "step": 1665
+    },
+    {
+      "epoch": 5.5407653910149754,
+      "loss": 0.4962202310562134,
+      "loss_ce": 4.4194543988851365e-06,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.49609375,
+      "num_input_tokens_seen": 104348436,
+      "step": 1665
+    },
+    {
+      "epoch": 5.544093178036606,
+      "grad_norm": 15.838689804077148,
+      "learning_rate": 5e-06,
+      "loss": 0.7047,
+      "num_input_tokens_seen": 104411136,
+      "step": 1666
+    },
+    {
+      "epoch": 5.544093178036606,
+      "loss": 0.7474393844604492,
+      "loss_ce": 2.877732640627073e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 104411136,
+      "step": 1666
+    },
+    {
+      "epoch": 5.547420965058237,
+      "grad_norm": 13.48023796081543,
+      "learning_rate": 5e-06,
+      "loss": 0.54,
+      "num_input_tokens_seen": 104473460,
+      "step": 1667
+    },
+    {
+      "epoch": 5.547420965058237,
+      "loss": 0.4172057807445526,
+      "loss_ce": 0.0005798295023851097,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 104473460,
+      "step": 1667
+    },
+    {
+      "epoch": 5.550748752079867,
+      "grad_norm": 13.00550651550293,
+      "learning_rate": 5e-06,
+      "loss": 0.5776,
+      "num_input_tokens_seen": 104535100,
+      "step": 1668
+    },
+    {
+      "epoch": 5.550748752079867,
+      "loss": 0.5370232462882996,
+      "loss_ce": 0.00028008728986606,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 104535100,
+      "step": 1668
+    },
+    {
+      "epoch": 5.554076539101498,
+      "grad_norm": 25.64752769470215,
+      "learning_rate": 5e-06,
+      "loss": 0.5463,
+      "num_input_tokens_seen": 104597888,
+      "step": 1669
+    },
+    {
+      "epoch": 5.554076539101498,
+      "loss": 0.4867459237575531,
+      "loss_ce": 0.0007840264588594437,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.486328125,
+      "num_input_tokens_seen": 104597888,
+      "step": 1669
+    },
+    {
+      "epoch": 5.5574043261231285,
+      "grad_norm": 11.490925788879395,
+      "learning_rate": 5e-06,
+      "loss": 0.6501,
+      "num_input_tokens_seen": 104659476,
+      "step": 1670
+    },
+    {
+      "epoch": 5.5574043261231285,
+      "loss": 0.45913708209991455,
+      "loss_ce": 3.0645875085610896e-05,
+      "loss_iou": 0.1064453125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.458984375,
+      "num_input_tokens_seen": 104659476,
+      "step": 1670
+    },
+    {
+      "epoch": 5.560732113144759,
+      "grad_norm": 11.503711700439453,
+      "learning_rate": 5e-06,
+      "loss": 0.7049,
+      "num_input_tokens_seen": 104723016,
+      "step": 1671
+    },
+    {
+      "epoch": 5.560732113144759,
+      "loss": 0.8446140289306641,
+      "loss_ce": 0.0008640355663374066,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 104723016,
+      "step": 1671
+    },
+    {
+      "epoch": 5.56405990016639,
+      "grad_norm": 8.733305931091309,
+      "learning_rate": 5e-06,
+      "loss": 0.5664,
+      "num_input_tokens_seen": 104784856,
+      "step": 1672
+    },
+    {
+      "epoch": 5.56405990016639,
+      "loss": 0.4931206703186035,
+      "loss_ce": 0.0008110986091196537,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 104784856,
+      "step": 1672
+    },
+    {
+      "epoch": 5.56738768718802,
+      "grad_norm": 9.786175727844238,
+      "learning_rate": 5e-06,
+      "loss": 0.846,
+      "num_input_tokens_seen": 104848380,
+      "step": 1673
+    },
+    {
+      "epoch": 5.56738768718802,
+      "loss": 0.7731995582580566,
+      "loss_ce": 6.221240255399607e-06,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 104848380,
+      "step": 1673
+    },
+    {
+      "epoch": 5.570715474209651,
+      "grad_norm": 13.724798202514648,
+      "learning_rate": 5e-06,
+      "loss": 0.6908,
+      "num_input_tokens_seen": 104910960,
+      "step": 1674
+    },
+    {
+      "epoch": 5.570715474209651,
+      "loss": 0.691491961479187,
+      "loss_ce": 8.570987120037898e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.69140625,
+      "num_input_tokens_seen": 104910960,
+      "step": 1674
+    },
+    {
+      "epoch": 5.574043261231282,
+      "grad_norm": 12.126524925231934,
+      "learning_rate": 5e-06,
+      "loss": 0.7477,
+      "num_input_tokens_seen": 104974332,
+      "step": 1675
+    },
+    {
+      "epoch": 5.574043261231282,
+      "loss": 0.45954495668411255,
+      "loss_ce": 7.229376933537424e-05,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.458984375,
+      "num_input_tokens_seen": 104974332,
+      "step": 1675
+    },
+    {
+      "epoch": 5.577371048252912,
+      "grad_norm": 21.586347579956055,
+      "learning_rate": 5e-06,
+      "loss": 0.6319,
+      "num_input_tokens_seen": 105036216,
+      "step": 1676
+    },
+    {
+      "epoch": 5.577371048252912,
+      "loss": 0.858528733253479,
+      "loss_ce": 6.919073348399252e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 105036216,
+      "step": 1676
+    },
+    {
+      "epoch": 5.580698835274543,
+      "grad_norm": 6.689985275268555,
+      "learning_rate": 5e-06,
+      "loss": 0.5895,
+      "num_input_tokens_seen": 105098888,
+      "step": 1677
+    },
+    {
+      "epoch": 5.580698835274543,
+      "loss": 0.5708335041999817,
+      "loss_ce": 0.0006430556531995535,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 105098888,
+      "step": 1677
+    },
+    {
+      "epoch": 5.584026622296173,
+      "grad_norm": 23.635780334472656,
+      "learning_rate": 5e-06,
+      "loss": 0.6477,
+      "num_input_tokens_seen": 105160692,
+      "step": 1678
+    },
+    {
+      "epoch": 5.584026622296173,
+      "loss": 0.745747447013855,
+      "loss_ce": 0.0005082354764454067,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 105160692,
+      "step": 1678
+    },
+    {
+      "epoch": 5.587354409317804,
+      "grad_norm": 17.890554428100586,
+      "learning_rate": 5e-06,
+      "loss": 0.8851,
+      "num_input_tokens_seen": 105224932,
+      "step": 1679
+    },
+    {
+      "epoch": 5.587354409317804,
+      "loss": 0.6944231986999512,
+      "loss_ce": 0.0005755070014856756,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 105224932,
+      "step": 1679
+    },
+    {
+      "epoch": 5.590682196339435,
+      "grad_norm": 36.01473617553711,
+      "learning_rate": 5e-06,
+      "loss": 0.5336,
+      "num_input_tokens_seen": 105287516,
+      "step": 1680
+    },
+    {
+      "epoch": 5.590682196339435,
+      "loss": 0.4492305517196655,
+      "loss_ce": 1.179968148790067e-05,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 105287516,
+      "step": 1680
+    },
+    {
+      "epoch": 5.594009983361065,
+      "grad_norm": 15.766006469726562,
+      "learning_rate": 5e-06,
+      "loss": 0.8177,
+      "num_input_tokens_seen": 105349820,
+      "step": 1681
+    },
+    {
+      "epoch": 5.594009983361065,
+      "loss": 0.9034368991851807,
+      "loss_ce": 0.00011656155402306467,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.90234375,
+      "num_input_tokens_seen": 105349820,
+      "step": 1681
+    },
+    {
+      "epoch": 5.597337770382696,
+      "grad_norm": 10.683774948120117,
+      "learning_rate": 5e-06,
+      "loss": 0.596,
+      "num_input_tokens_seen": 105411476,
+      "step": 1682
+    },
+    {
+      "epoch": 5.597337770382696,
+      "loss": 0.8543940186500549,
+      "loss_ce": 0.000634245399851352,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 105411476,
+      "step": 1682
+    },
+    {
+      "epoch": 5.6006655574043265,
+      "grad_norm": 15.730971336364746,
+      "learning_rate": 5e-06,
+      "loss": 0.7665,
+      "num_input_tokens_seen": 105473888,
+      "step": 1683
+    },
+    {
+      "epoch": 5.6006655574043265,
+      "loss": 0.8576712012290955,
+      "loss_ce": 5.173462795937667e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 105473888,
+      "step": 1683
+    },
+    {
+      "epoch": 5.603993344425957,
+      "grad_norm": 17.79537582397461,
+      "learning_rate": 5e-06,
+      "loss": 0.7979,
+      "num_input_tokens_seen": 105537972,
+      "step": 1684
+    },
+    {
+      "epoch": 5.603993344425957,
+      "loss": 0.8166719675064087,
+      "loss_ce": 0.00026570551563054323,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.81640625,
+      "num_input_tokens_seen": 105537972,
+      "step": 1684
+    },
+    {
+      "epoch": 5.607321131447588,
+      "grad_norm": 15.09589672088623,
+      "learning_rate": 5e-06,
+      "loss": 0.6113,
+      "num_input_tokens_seen": 105600116,
+      "step": 1685
+    },
+    {
+      "epoch": 5.607321131447588,
+      "loss": 0.7013496160507202,
+      "loss_ce": 5.5670123401796445e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 105600116,
+      "step": 1685
+    },
+    {
+      "epoch": 5.610648918469218,
+      "grad_norm": 9.822373390197754,
+      "learning_rate": 5e-06,
+      "loss": 0.5208,
+      "num_input_tokens_seen": 105661696,
+      "step": 1686
+    },
+    {
+      "epoch": 5.610648918469218,
+      "loss": 0.4002895951271057,
+      "loss_ce": 0.00044828990940004587,
+      "loss_iou": 0.1044921875,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 105661696,
+      "step": 1686
+    },
+    {
+      "epoch": 5.613976705490849,
+      "grad_norm": 12.23663330078125,
+      "learning_rate": 5e-06,
+      "loss": 0.8393,
+      "num_input_tokens_seen": 105724768,
+      "step": 1687
+    },
+    {
+      "epoch": 5.613976705490849,
+      "loss": 0.684592604637146,
+      "loss_ce": 2.223967749159783e-05,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.68359375,
+      "num_input_tokens_seen": 105724768,
+      "step": 1687
+    },
+    {
+      "epoch": 5.6173044925124795,
+      "grad_norm": 38.26483917236328,
+      "learning_rate": 5e-06,
+      "loss": 0.5627,
+      "num_input_tokens_seen": 105787560,
+      "step": 1688
+    },
+    {
+      "epoch": 5.6173044925124795,
+      "loss": 0.44944727420806885,
+      "loss_ce": 0.0004726658225990832,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 105787560,
+      "step": 1688
+    },
+    {
+      "epoch": 5.62063227953411,
+      "grad_norm": 17.305635452270508,
+      "learning_rate": 5e-06,
+      "loss": 0.6798,
+      "num_input_tokens_seen": 105851016,
+      "step": 1689
+    },
+    {
+      "epoch": 5.62063227953411,
+      "loss": 0.553048849105835,
+      "loss_ce": 9.341451004729606e-06,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.5546875,
+      "num_input_tokens_seen": 105851016,
+      "step": 1689
+    },
+    {
+      "epoch": 5.623960066555741,
+      "grad_norm": 19.972763061523438,
+      "learning_rate": 5e-06,
+      "loss": 0.7356,
+      "num_input_tokens_seen": 105912112,
+      "step": 1690
+    },
+    {
+      "epoch": 5.623960066555741,
+      "loss": 0.8830522298812866,
+      "loss_ce": 0.00023976643569767475,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 105912112,
+      "step": 1690
+    },
+    {
+      "epoch": 5.627287853577371,
+      "grad_norm": 66.9946060180664,
+      "learning_rate": 5e-06,
+      "loss": 0.8263,
+      "num_input_tokens_seen": 105975356,
+      "step": 1691
+    },
+    {
+      "epoch": 5.627287853577371,
+      "loss": 0.643821120262146,
+      "loss_ce": 2.2260259356698953e-05,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 105975356,
+      "step": 1691
+    },
+    {
+      "epoch": 5.630615640599002,
+      "grad_norm": 33.13805389404297,
+      "learning_rate": 5e-06,
+      "loss": 0.6637,
+      "num_input_tokens_seen": 106037320,
+      "step": 1692
+    },
+    {
+      "epoch": 5.630615640599002,
+      "loss": 0.5694460272789001,
+      "loss_ce": 0.00035422114888206124,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 106037320,
+      "step": 1692
+    },
+    {
+      "epoch": 5.633943427620633,
+      "grad_norm": 15.999832153320312,
+      "learning_rate": 5e-06,
+      "loss": 0.7618,
+      "num_input_tokens_seen": 106100968,
+      "step": 1693
+    },
+    {
+      "epoch": 5.633943427620633,
+      "loss": 0.5870428085327148,
+      "loss_ce": 0.0003728655865415931,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 106100968,
+      "step": 1693
+    },
+    {
+      "epoch": 5.637271214642263,
+      "grad_norm": 22.809717178344727,
+      "learning_rate": 5e-06,
+      "loss": 0.5141,
+      "num_input_tokens_seen": 106163984,
+      "step": 1694
+    },
+    {
+      "epoch": 5.637271214642263,
+      "loss": 0.5070875883102417,
+      "loss_ce": 7.488038590963697e-06,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.5078125,
+      "num_input_tokens_seen": 106163984,
+      "step": 1694
+    },
+    {
+      "epoch": 5.640599001663894,
+      "grad_norm": 10.32789134979248,
+      "learning_rate": 5e-06,
+      "loss": 0.4488,
+      "num_input_tokens_seen": 106225564,
+      "step": 1695
+    },
+    {
+      "epoch": 5.640599001663894,
+      "loss": 0.332084059715271,
+      "loss_ce": 7.034345344436588e-06,
+      "loss_iou": 0.1103515625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 106225564,
+      "step": 1695
+    },
+    {
+      "epoch": 5.643926788685524,
+      "grad_norm": 7.523070335388184,
+      "learning_rate": 5e-06,
+      "loss": 0.6172,
+      "num_input_tokens_seen": 106287900,
+      "step": 1696
+    },
+    {
+      "epoch": 5.643926788685524,
+      "loss": 0.775996208190918,
+      "loss_ce": 0.00023933117336127907,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 106287900,
+      "step": 1696
+    },
+    {
+      "epoch": 5.647254575707155,
+      "grad_norm": 15.663951873779297,
+      "learning_rate": 5e-06,
+      "loss": 0.4675,
+      "num_input_tokens_seen": 106348884,
+      "step": 1697
+    },
+    {
+      "epoch": 5.647254575707155,
+      "loss": 0.4975832402706146,
+      "loss_ce": 2.462265911162831e-05,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.498046875,
+      "num_input_tokens_seen": 106348884,
+      "step": 1697
+    },
+    {
+      "epoch": 5.650582362728786,
+      "grad_norm": 9.897237777709961,
+      "learning_rate": 5e-06,
+      "loss": 0.7378,
+      "num_input_tokens_seen": 106411368,
+      "step": 1698
+    },
+    {
+      "epoch": 5.650582362728786,
+      "loss": 0.7313656806945801,
+      "loss_ce": 0.0006527729565277696,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 106411368,
+      "step": 1698
+    },
+    {
+      "epoch": 5.653910149750416,
+      "grad_norm": 27.847606658935547,
+      "learning_rate": 5e-06,
+      "loss": 0.7811,
+      "num_input_tokens_seen": 106473388,
+      "step": 1699
+    },
+    {
+      "epoch": 5.653910149750416,
+      "loss": 1.0008139610290527,
+      "loss_ce": 0.0003256350464653224,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.07080078125,
+      "loss_xval": 1.0,
+      "num_input_tokens_seen": 106473388,
+      "step": 1699
+    },
+    {
+      "epoch": 5.657237936772047,
+      "grad_norm": 11.24642276763916,
+      "learning_rate": 5e-06,
+      "loss": 0.8014,
+      "num_input_tokens_seen": 106537688,
+      "step": 1700
+    },
+    {
+      "epoch": 5.657237936772047,
+      "loss": 0.9711390137672424,
+      "loss_ce": 0.0006800297414883971,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.96875,
+      "num_input_tokens_seen": 106537688,
+      "step": 1700
+    },
+    {
+      "epoch": 5.6605657237936775,
+      "grad_norm": 14.969793319702148,
+      "learning_rate": 5e-06,
+      "loss": 0.7234,
+      "num_input_tokens_seen": 106600456,
+      "step": 1701
+    },
+    {
+      "epoch": 5.6605657237936775,
+      "loss": 0.5358263254165649,
+      "loss_ce": 0.000670083099976182,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 106600456,
+      "step": 1701
+    },
+    {
+      "epoch": 5.663893510815308,
+      "grad_norm": 12.321464538574219,
+      "learning_rate": 5e-06,
+      "loss": 0.5965,
+      "num_input_tokens_seen": 106662656,
+      "step": 1702
+    },
+    {
+      "epoch": 5.663893510815308,
+      "loss": 0.7050811052322388,
+      "loss_ce": 0.000857477483805269,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.058837890625,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 106662656,
+      "step": 1702
+    },
+    {
+      "epoch": 5.667221297836939,
+      "grad_norm": 17.309473037719727,
+      "learning_rate": 5e-06,
+      "loss": 0.8267,
+      "num_input_tokens_seen": 106726504,
+      "step": 1703
+    },
+    {
+      "epoch": 5.667221297836939,
+      "loss": 0.8586465120315552,
+      "loss_ce": 3.938740519515704e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 106726504,
+      "step": 1703
+    },
+    {
+      "epoch": 5.670549084858569,
+      "grad_norm": 21.560821533203125,
+      "learning_rate": 5e-06,
+      "loss": 0.6404,
+      "num_input_tokens_seen": 106789148,
+      "step": 1704
+    },
+    {
+      "epoch": 5.670549084858569,
+      "loss": 0.4272652268409729,
+      "loss_ce": 1.9131181034026667e-05,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 106789148,
+      "step": 1704
+    },
+    {
+      "epoch": 5.6738768718802,
+      "grad_norm": 68.77386474609375,
+      "learning_rate": 5e-06,
+      "loss": 0.7511,
+      "num_input_tokens_seen": 106851232,
+      "step": 1705
+    },
+    {
+      "epoch": 5.6738768718802,
+      "loss": 0.5564388632774353,
+      "loss_ce": 4.237954271957278e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.5546875,
+      "num_input_tokens_seen": 106851232,
+      "step": 1705
+    },
+    {
+      "epoch": 5.677204658901831,
+      "grad_norm": 9.309255599975586,
+      "learning_rate": 5e-06,
+      "loss": 0.6206,
+      "num_input_tokens_seen": 106914752,
+      "step": 1706
+    },
+    {
+      "epoch": 5.677204658901831,
+      "loss": 0.5360674858093262,
+      "loss_ce": 5.677162334905006e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 106914752,
+      "step": 1706
+    },
+    {
+      "epoch": 5.680532445923461,
+      "grad_norm": 9.004875183105469,
+      "learning_rate": 5e-06,
+      "loss": 0.5238,
+      "num_input_tokens_seen": 106976952,
+      "step": 1707
+    },
+    {
+      "epoch": 5.680532445923461,
+      "loss": 0.45058244466781616,
+      "loss_ce": 0.00014298340829554945,
+      "loss_iou": 0.1279296875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.451171875,
+      "num_input_tokens_seen": 106976952,
+      "step": 1707
+    },
+    {
+      "epoch": 5.683860232945092,
+      "grad_norm": 12.503425598144531,
+      "learning_rate": 5e-06,
+      "loss": 0.6518,
+      "num_input_tokens_seen": 107039528,
+      "step": 1708
+    },
+    {
+      "epoch": 5.683860232945092,
+      "loss": 0.7054813504219055,
+      "loss_ce": 0.000281153799733147,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.70703125,
+      "num_input_tokens_seen": 107039528,
+      "step": 1708
+    },
+    {
+      "epoch": 5.687188019966722,
+      "grad_norm": 123.16822052001953,
+      "learning_rate": 5e-06,
+      "loss": 0.6203,
+      "num_input_tokens_seen": 107103448,
+      "step": 1709
+    },
+    {
+      "epoch": 5.687188019966722,
+      "loss": 0.49697205424308777,
+      "loss_ce": 0.00014590048522222787,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.49609375,
+      "num_input_tokens_seen": 107103448,
+      "step": 1709
+    },
+    {
+      "epoch": 5.690515806988353,
+      "grad_norm": 19.90471839904785,
+      "learning_rate": 5e-06,
+      "loss": 0.6721,
+      "num_input_tokens_seen": 107165844,
+      "step": 1710
+    },
+    {
+      "epoch": 5.690515806988353,
+      "loss": 0.703316330909729,
+      "loss_ce": 0.00019130959117319435,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 107165844,
+      "step": 1710
+    },
+    {
+      "epoch": 5.693843594009984,
+      "grad_norm": 26.4517765045166,
+      "learning_rate": 5e-06,
+      "loss": 0.7115,
+      "num_input_tokens_seen": 107228880,
+      "step": 1711
+    },
+    {
+      "epoch": 5.693843594009984,
+      "loss": 0.5219588279724121,
+      "loss_ce": 0.0009016690892167389,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 107228880,
+      "step": 1711
+    },
+    {
+      "epoch": 5.697171381031614,
+      "grad_norm": 24.016637802124023,
+      "learning_rate": 5e-06,
+      "loss": 0.6032,
+      "num_input_tokens_seen": 107291652,
+      "step": 1712
+    },
+    {
+      "epoch": 5.697171381031614,
+      "loss": 0.6287029385566711,
+      "loss_ce": 0.0006511926185339689,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 107291652,
+      "step": 1712
+    },
+    {
+      "epoch": 5.700499168053245,
+      "grad_norm": 12.63536548614502,
+      "learning_rate": 5e-06,
+      "loss": 0.5321,
+      "num_input_tokens_seen": 107353984,
+      "step": 1713
+    },
+    {
+      "epoch": 5.700499168053245,
+      "loss": 0.6348440647125244,
+      "loss_ce": 0.001055013621225953,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 107353984,
+      "step": 1713
+    },
+    {
+      "epoch": 5.7038269550748755,
+      "grad_norm": 28.98735237121582,
+      "learning_rate": 5e-06,
+      "loss": 0.8102,
+      "num_input_tokens_seen": 107417396,
+      "step": 1714
+    },
+    {
+      "epoch": 5.7038269550748755,
+      "loss": 0.7450417280197144,
+      "loss_ce": 0.00016867804515641183,
+      "loss_iou": 0.265625,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 107417396,
+      "step": 1714
+    },
+    {
+      "epoch": 5.707154742096506,
+      "grad_norm": 12.304791450500488,
+      "learning_rate": 5e-06,
+      "loss": 0.4742,
+      "num_input_tokens_seen": 107480376,
+      "step": 1715
+    },
+    {
+      "epoch": 5.707154742096506,
+      "loss": 0.4631630778312683,
+      "loss_ce": 8.936785161495209e-05,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.462890625,
+      "num_input_tokens_seen": 107480376,
+      "step": 1715
+    },
+    {
+      "epoch": 5.710482529118137,
+      "grad_norm": 14.445513725280762,
+      "learning_rate": 5e-06,
+      "loss": 0.8061,
+      "num_input_tokens_seen": 107544188,
+      "step": 1716
+    },
+    {
+      "epoch": 5.710482529118137,
+      "loss": 0.8209518194198608,
+      "loss_ce": 0.0010055896127596498,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 107544188,
+      "step": 1716
+    },
+    {
+      "epoch": 5.713810316139767,
+      "grad_norm": 8.823054313659668,
+      "learning_rate": 5e-06,
+      "loss": 0.4795,
+      "num_input_tokens_seen": 107605148,
+      "step": 1717
+    },
+    {
+      "epoch": 5.713810316139767,
+      "loss": 0.4528508484363556,
+      "loss_ce": 0.0004582681867759675,
+      "loss_iou": 0.078125,
+      "loss_num": 0.059326171875,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 107605148,
+      "step": 1717
+    },
+    {
+      "epoch": 5.717138103161398,
+      "grad_norm": 16.8961181640625,
+      "learning_rate": 5e-06,
+      "loss": 0.4838,
+      "num_input_tokens_seen": 107668080,
+      "step": 1718
+    },
+    {
+      "epoch": 5.717138103161398,
+      "loss": 0.4450841546058655,
+      "loss_ce": 0.0009313120390288532,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.443359375,
+      "num_input_tokens_seen": 107668080,
+      "step": 1718
+    },
+    {
+      "epoch": 5.7204658901830285,
+      "grad_norm": 9.961505889892578,
+      "learning_rate": 5e-06,
+      "loss": 0.341,
+      "num_input_tokens_seen": 107729940,
+      "step": 1719
+    },
+    {
+      "epoch": 5.7204658901830285,
+      "loss": 0.3053891658782959,
+      "loss_ce": 9.12985997274518e-05,
+      "loss_iou": 0.078125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 107729940,
+      "step": 1719
+    },
+    {
+      "epoch": 5.723793677204659,
+      "grad_norm": 11.72249984741211,
+      "learning_rate": 5e-06,
+      "loss": 0.6196,
+      "num_input_tokens_seen": 107791916,
+      "step": 1720
+    },
+    {
+      "epoch": 5.723793677204659,
+      "loss": 0.8364963531494141,
+      "loss_ce": 0.0005588560015894473,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 107791916,
+      "step": 1720
+    },
+    {
+      "epoch": 5.72712146422629,
+      "grad_norm": 22.064348220825195,
+      "learning_rate": 5e-06,
+      "loss": 0.6433,
+      "num_input_tokens_seen": 107855036,
+      "step": 1721
+    },
+    {
+      "epoch": 5.72712146422629,
+      "loss": 0.5307466983795166,
+      "loss_ce": 0.0015108881052583456,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.05126953125,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 107855036,
+      "step": 1721
+    },
+    {
+      "epoch": 5.73044925124792,
+      "grad_norm": 9.683394432067871,
+      "learning_rate": 5e-06,
+      "loss": 0.6907,
+      "num_input_tokens_seen": 107918084,
+      "step": 1722
+    },
+    {
+      "epoch": 5.73044925124792,
+      "loss": 0.5834569931030273,
+      "loss_ce": 0.00038815615698695183,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 107918084,
+      "step": 1722
+    },
+    {
+      "epoch": 5.733777038269551,
+      "grad_norm": 6.485795974731445,
+      "learning_rate": 5e-06,
+      "loss": 0.4933,
+      "num_input_tokens_seen": 107980216,
+      "step": 1723
+    },
+    {
+      "epoch": 5.733777038269551,
+      "loss": 0.6450976133346558,
+      "loss_ce": 0.0006884537287987769,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 107980216,
+      "step": 1723
+    },
+    {
+      "epoch": 5.737104825291182,
+      "grad_norm": 17.993772506713867,
+      "learning_rate": 5e-06,
+      "loss": 0.7217,
+      "num_input_tokens_seen": 108041460,
+      "step": 1724
+    },
+    {
+      "epoch": 5.737104825291182,
+      "loss": 0.8249725103378296,
+      "loss_ce": 8.237032307079062e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 108041460,
+      "step": 1724
+    },
+    {
+      "epoch": 5.740432612312812,
+      "grad_norm": 23.316713333129883,
+      "learning_rate": 5e-06,
+      "loss": 0.624,
+      "num_input_tokens_seen": 108104100,
+      "step": 1725
+    },
+    {
+      "epoch": 5.740432612312812,
+      "loss": 0.7806700468063354,
+      "loss_ce": 0.002044552704319358,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 108104100,
+      "step": 1725
+    },
+    {
+      "epoch": 5.743760399334443,
+      "grad_norm": 14.220916748046875,
+      "learning_rate": 5e-06,
+      "loss": 0.8526,
+      "num_input_tokens_seen": 108166016,
+      "step": 1726
+    },
+    {
+      "epoch": 5.743760399334443,
+      "loss": 0.9530130624771118,
+      "loss_ce": 1.0050905984826386e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 108166016,
+      "step": 1726
+    },
+    {
+      "epoch": 5.747088186356073,
+      "grad_norm": 19.730939865112305,
+      "learning_rate": 5e-06,
+      "loss": 0.7547,
+      "num_input_tokens_seen": 108228780,
+      "step": 1727
+    },
+    {
+      "epoch": 5.747088186356073,
+      "loss": 1.039328932762146,
+      "loss_ce": 0.0002664237981662154,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0791015625,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 108228780,
+      "step": 1727
+    },
+    {
+      "epoch": 5.750415973377704,
+      "grad_norm": 12.41170597076416,
+      "learning_rate": 5e-06,
+      "loss": 0.7916,
+      "num_input_tokens_seen": 108292028,
+      "step": 1728
+    },
+    {
+      "epoch": 5.750415973377704,
+      "loss": 0.9140459299087524,
+      "loss_ce": 0.0005938184331171215,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 108292028,
+      "step": 1728
+    },
+    {
+      "epoch": 5.753743760399335,
+      "grad_norm": 13.194222450256348,
+      "learning_rate": 5e-06,
+      "loss": 0.8602,
+      "num_input_tokens_seen": 108355096,
+      "step": 1729
+    },
+    {
+      "epoch": 5.753743760399335,
+      "loss": 0.9234409928321838,
+      "loss_ce": 0.00010111027950188145,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 108355096,
+      "step": 1729
+    },
+    {
+      "epoch": 5.757071547420965,
+      "grad_norm": 20.04388427734375,
+      "learning_rate": 5e-06,
+      "loss": 0.7181,
+      "num_input_tokens_seen": 108417048,
+      "step": 1730
+    },
+    {
+      "epoch": 5.757071547420965,
+      "loss": 0.9073438048362732,
+      "loss_ce": 0.0006054791738279164,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 108417048,
+      "step": 1730
+    },
+    {
+      "epoch": 5.760399334442596,
+      "grad_norm": 17.463380813598633,
+      "learning_rate": 5e-06,
+      "loss": 0.741,
+      "num_input_tokens_seen": 108480692,
+      "step": 1731
+    },
+    {
+      "epoch": 5.760399334442596,
+      "loss": 0.6669446229934692,
+      "loss_ce": 7.449155964422971e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 108480692,
+      "step": 1731
+    },
+    {
+      "epoch": 5.7637271214642265,
+      "grad_norm": 22.218618392944336,
+      "learning_rate": 5e-06,
+      "loss": 0.5948,
+      "num_input_tokens_seen": 108542816,
+      "step": 1732
+    },
+    {
+      "epoch": 5.7637271214642265,
+      "loss": 0.4774358868598938,
+      "loss_ce": 0.000995467184111476,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 108542816,
+      "step": 1732
+    },
+    {
+      "epoch": 5.767054908485857,
+      "grad_norm": 12.377209663391113,
+      "learning_rate": 5e-06,
+      "loss": 0.4993,
+      "num_input_tokens_seen": 108606396,
+      "step": 1733
+    },
+    {
+      "epoch": 5.767054908485857,
+      "loss": 0.5402237176895142,
+      "loss_ce": 0.00019988913845736533,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 108606396,
+      "step": 1733
+    },
+    {
+      "epoch": 5.770382695507488,
+      "grad_norm": 9.325462341308594,
+      "learning_rate": 5e-06,
+      "loss": 0.8247,
+      "num_input_tokens_seen": 108668468,
+      "step": 1734
+    },
+    {
+      "epoch": 5.770382695507488,
+      "loss": 0.9547851085662842,
+      "loss_ce": 0.00397943751886487,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.94921875,
+      "num_input_tokens_seen": 108668468,
+      "step": 1734
+    },
+    {
+      "epoch": 5.773710482529118,
+      "grad_norm": 12.919450759887695,
+      "learning_rate": 5e-06,
+      "loss": 0.6592,
+      "num_input_tokens_seen": 108731260,
+      "step": 1735
+    },
+    {
+      "epoch": 5.773710482529118,
+      "loss": 0.5413807034492493,
+      "loss_ce": 0.0009144245414063334,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 108731260,
+      "step": 1735
+    },
+    {
+      "epoch": 5.777038269550749,
+      "grad_norm": 11.477002143859863,
+      "learning_rate": 5e-06,
+      "loss": 0.8761,
+      "num_input_tokens_seen": 108795324,
+      "step": 1736
+    },
+    {
+      "epoch": 5.777038269550749,
+      "loss": 0.9495025873184204,
+      "loss_ce": 0.0010161716490983963,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.94921875,
+      "num_input_tokens_seen": 108795324,
+      "step": 1736
+    },
+    {
+      "epoch": 5.78036605657238,
+      "grad_norm": 9.099837303161621,
+      "learning_rate": 5e-06,
+      "loss": 0.7722,
+      "num_input_tokens_seen": 108857560,
+      "step": 1737
+    },
+    {
+      "epoch": 5.78036605657238,
+      "loss": 0.7459558248519897,
+      "loss_ce": 0.000472470186650753,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 108857560,
+      "step": 1737
+    },
+    {
+      "epoch": 5.78369384359401,
+      "grad_norm": 14.444173812866211,
+      "learning_rate": 5e-06,
+      "loss": 0.4758,
+      "num_input_tokens_seen": 108918780,
+      "step": 1738
+    },
+    {
+      "epoch": 5.78369384359401,
+      "loss": 0.4299461543560028,
+      "loss_ce": 1.4529123291140422e-05,
+      "loss_iou": 0.12353515625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 108918780,
+      "step": 1738
+    },
+    {
+      "epoch": 5.787021630615641,
+      "grad_norm": 17.906532287597656,
+      "learning_rate": 5e-06,
+      "loss": 0.6279,
+      "num_input_tokens_seen": 108980980,
+      "step": 1739
+    },
+    {
+      "epoch": 5.787021630615641,
+      "loss": 0.6438461542129517,
+      "loss_ce": 0.0005356057081371546,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 108980980,
+      "step": 1739
+    },
+    {
+      "epoch": 5.790349417637271,
+      "grad_norm": 13.573742866516113,
+      "learning_rate": 5e-06,
+      "loss": 0.7242,
+      "num_input_tokens_seen": 109043940,
+      "step": 1740
+    },
+    {
+      "epoch": 5.790349417637271,
+      "loss": 0.9529647827148438,
+      "loss_ce": 0.0005722053465433419,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 109043940,
+      "step": 1740
+    },
+    {
+      "epoch": 5.793677204658902,
+      "grad_norm": 13.186030387878418,
+      "learning_rate": 5e-06,
+      "loss": 0.6727,
+      "num_input_tokens_seen": 109106144,
+      "step": 1741
+    },
+    {
+      "epoch": 5.793677204658902,
+      "loss": 0.6280597448348999,
+      "loss_ce": 0.000374168943380937,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 109106144,
+      "step": 1741
+    },
+    {
+      "epoch": 5.797004991680533,
+      "grad_norm": 11.846780776977539,
+      "learning_rate": 5e-06,
+      "loss": 0.5714,
+      "num_input_tokens_seen": 109169260,
+      "step": 1742
+    },
+    {
+      "epoch": 5.797004991680533,
+      "loss": 0.6322175860404968,
+      "loss_ce": 1.547130523249507e-05,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 109169260,
+      "step": 1742
+    },
+    {
+      "epoch": 5.800332778702163,
+      "grad_norm": 17.20931053161621,
+      "learning_rate": 5e-06,
+      "loss": 0.6911,
+      "num_input_tokens_seen": 109232900,
+      "step": 1743
+    },
+    {
+      "epoch": 5.800332778702163,
+      "loss": 0.5807763338088989,
+      "loss_ce": 0.00020993230282329023,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 109232900,
+      "step": 1743
+    },
+    {
+      "epoch": 5.803660565723794,
+      "grad_norm": 17.745994567871094,
+      "learning_rate": 5e-06,
+      "loss": 0.7031,
+      "num_input_tokens_seen": 109295748,
+      "step": 1744
+    },
+    {
+      "epoch": 5.803660565723794,
+      "loss": 0.8015212416648865,
+      "loss_ce": 7.5699854278354906e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 109295748,
+      "step": 1744
+    },
+    {
+      "epoch": 5.8069883527454245,
+      "grad_norm": 8.905264854431152,
+      "learning_rate": 5e-06,
+      "loss": 0.4016,
+      "num_input_tokens_seen": 109358312,
+      "step": 1745
+    },
+    {
+      "epoch": 5.8069883527454245,
+      "loss": 0.39209282398223877,
+      "loss_ce": 2.9924158297944814e-06,
+      "loss_iou": 0.109375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 109358312,
+      "step": 1745
+    },
+    {
+      "epoch": 5.810316139767055,
+      "grad_norm": 5.492832660675049,
+      "learning_rate": 5e-06,
+      "loss": 0.593,
+      "num_input_tokens_seen": 109421796,
+      "step": 1746
+    },
+    {
+      "epoch": 5.810316139767055,
+      "loss": 0.4544283151626587,
+      "loss_ce": 0.00014364512753672898,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 109421796,
+      "step": 1746
+    },
+    {
+      "epoch": 5.813643926788686,
+      "grad_norm": 13.455323219299316,
+      "learning_rate": 5e-06,
+      "loss": 0.8716,
+      "num_input_tokens_seen": 109485112,
+      "step": 1747
+    },
+    {
+      "epoch": 5.813643926788686,
+      "loss": 0.9304600954055786,
+      "loss_ce": 0.0002843089459929615,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 109485112,
+      "step": 1747
+    },
+    {
+      "epoch": 5.816971713810316,
+      "grad_norm": 17.337217330932617,
+      "learning_rate": 5e-06,
+      "loss": 0.6384,
+      "num_input_tokens_seen": 109547828,
+      "step": 1748
+    },
+    {
+      "epoch": 5.816971713810316,
+      "loss": 0.8259046673774719,
+      "loss_ce": 0.00046522487537004054,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 109547828,
+      "step": 1748
+    },
+    {
+      "epoch": 5.820299500831947,
+      "grad_norm": 8.980031967163086,
+      "learning_rate": 5e-06,
+      "loss": 0.5732,
+      "num_input_tokens_seen": 109610380,
+      "step": 1749
+    },
+    {
+      "epoch": 5.820299500831947,
+      "loss": 0.4640018939971924,
+      "loss_ce": 1.261913166672457e-05,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.46484375,
+      "num_input_tokens_seen": 109610380,
+      "step": 1749
+    },
+    {
+      "epoch": 5.8236272878535775,
+      "grad_norm": 13.945286750793457,
+      "learning_rate": 5e-06,
+      "loss": 0.5297,
+      "num_input_tokens_seen": 109672416,
+      "step": 1750
+    },
+    {
+      "epoch": 5.8236272878535775,
+      "eval_seeclick_CIoU": 0.10236945003271103,
+      "eval_seeclick_GIoU": 0.11712978780269623,
+      "eval_seeclick_IoU": 0.2036900669336319,
+      "eval_seeclick_MAE_all": 0.16942255944013596,
+      "eval_seeclick_MAE_h": 0.049304405227303505,
+      "eval_seeclick_MAE_w": 0.12896203249692917,
+      "eval_seeclick_MAE_x_boxes": 0.23738989233970642,
+      "eval_seeclick_MAE_y_boxes": 0.1408754587173462,
+      "eval_seeclick_NUM_probability": 0.9998235106468201,
+      "eval_seeclick_inside_bbox": 0.20937500149011612,
+      "eval_seeclick_loss": 2.7325868606567383,
+      "eval_seeclick_loss_ce": 0.11027439311146736,
+      "eval_seeclick_loss_iou": 0.889892578125,
+      "eval_seeclick_loss_num": 0.16834259033203125,
+      "eval_seeclick_loss_xval": 2.6220703125,
+      "eval_seeclick_runtime": 67.9027,
+      "eval_seeclick_samples_per_second": 0.692,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 109672416,
+      "step": 1750
+    },
+    {
+      "epoch": 5.8236272878535775,
+      "eval_icons_CIoU": 0.04104041517712176,
+      "eval_icons_GIoU": 0.1604958102107048,
+      "eval_icons_IoU": 0.19673043489456177,
+      "eval_icons_MAE_all": 0.14254845678806305,
+      "eval_icons_MAE_h": 0.05420855712145567,
+      "eval_icons_MAE_w": 0.15784180164337158,
+      "eval_icons_MAE_x_boxes": 0.15393128246068954,
+      "eval_icons_MAE_y_boxes": 0.047592333517968655,
+      "eval_icons_NUM_probability": 0.9999956786632538,
+      "eval_icons_inside_bbox": 0.3420138955116272,
+      "eval_icons_loss": 2.3909385204315186,
+      "eval_icons_loss_ce": 9.011300790007226e-07,
+      "eval_icons_loss_iou": 0.824462890625,
+      "eval_icons_loss_num": 0.14644622802734375,
+      "eval_icons_loss_xval": 2.3818359375,
+      "eval_icons_runtime": 68.4022,
+      "eval_icons_samples_per_second": 0.731,
+      "eval_icons_steps_per_second": 0.029,
+      "num_input_tokens_seen": 109672416,
+      "step": 1750
+    },
+    {
+      "epoch": 5.8236272878535775,
+      "eval_screenspot_CIoU": 0.06936451761672895,
+      "eval_screenspot_GIoU": 0.13237376511096954,
+      "eval_screenspot_IoU": 0.21043485403060913,
+      "eval_screenspot_MAE_all": 0.18726058304309845,
+      "eval_screenspot_MAE_h": 0.07297260562578838,
+      "eval_screenspot_MAE_w": 0.14615009228388467,
+      "eval_screenspot_MAE_x_boxes": 0.25001437962055206,
+      "eval_screenspot_MAE_y_boxes": 0.13508301973342896,
+      "eval_screenspot_NUM_probability": 0.9999534090360006,
+      "eval_screenspot_inside_bbox": 0.3625000019868215,
+      "eval_screenspot_loss": 2.6945841312408447,
+      "eval_screenspot_loss_ce": 5.838420596167756e-05,
+      "eval_screenspot_loss_iou": 0.88134765625,
+      "eval_screenspot_loss_num": 0.194793701171875,
+      "eval_screenspot_loss_xval": 2.7360026041666665,
+      "eval_screenspot_runtime": 119.8843,
+      "eval_screenspot_samples_per_second": 0.742,
+      "eval_screenspot_steps_per_second": 0.025,
+      "num_input_tokens_seen": 109672416,
+      "step": 1750
+    },
+    {
+      "epoch": 5.8236272878535775,
+      "eval_compot_CIoU": 0.003425128757953644,
+      "eval_compot_GIoU": 0.11101753637194633,
+      "eval_compot_IoU": 0.17780261486768723,
+      "eval_compot_MAE_all": 0.2019421085715294,
+      "eval_compot_MAE_h": 0.07107486762106419,
+      "eval_compot_MAE_w": 0.2705545723438263,
+      "eval_compot_MAE_x_boxes": 0.20419960469007492,
+      "eval_compot_MAE_y_boxes": 0.08390221372246742,
+      "eval_compot_NUM_probability": 0.9999839067459106,
+      "eval_compot_inside_bbox": 0.3107638955116272,
+      "eval_compot_loss": 2.85368275642395,
+      "eval_compot_loss_ce": 0.004224188975058496,
+      "eval_compot_loss_iou": 0.915283203125,
+      "eval_compot_loss_num": 0.2141571044921875,
+      "eval_compot_loss_xval": 2.9033203125,
+      "eval_compot_runtime": 72.3216,
+      "eval_compot_samples_per_second": 0.691,
+      "eval_compot_steps_per_second": 0.028,
+      "num_input_tokens_seen": 109672416,
+      "step": 1750
+    },
+    {
+      "epoch": 5.8236272878535775,
+      "eval_custom_ui_MAE_all": 0.08030644804239273,
+      "eval_custom_ui_MAE_x": 0.07421864569187164,
+      "eval_custom_ui_MAE_y": 0.08639425411820412,
+      "eval_custom_ui_NUM_probability": 0.999993622303009,
+      "eval_custom_ui_loss": 0.3886740803718567,
+      "eval_custom_ui_loss_ce": 2.7836218805532553e-06,
+      "eval_custom_ui_loss_num": 0.074920654296875,
+      "eval_custom_ui_loss_xval": 0.3746337890625,
+      "eval_custom_ui_runtime": 53.634,
+      "eval_custom_ui_samples_per_second": 0.932,
+      "eval_custom_ui_steps_per_second": 0.037,
+      "num_input_tokens_seen": 109672416,
+      "step": 1750
+    },
+    {
+      "epoch": 5.8236272878535775,
+      "loss": 0.3675568401813507,
+      "loss_ce": 3.1467793633055408e-06,
+      "loss_iou": 0.0,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.3671875,
+      "num_input_tokens_seen": 109672416,
+      "step": 1750
+    },
+    {
+      "epoch": 5.826955074875208,
+      "grad_norm": 9.24660873413086,
+      "learning_rate": 5e-06,
+      "loss": 0.8293,
+      "num_input_tokens_seen": 109735632,
+      "step": 1751
+    },
+    {
+      "epoch": 5.826955074875208,
+      "loss": 0.6466149091720581,
+      "loss_ce": 0.0006188148399814963,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 109735632,
+      "step": 1751
+    },
+    {
+      "epoch": 5.830282861896839,
+      "grad_norm": 8.810318946838379,
+      "learning_rate": 5e-06,
+      "loss": 0.7299,
+      "num_input_tokens_seen": 109799360,
+      "step": 1752
+    },
+    {
+      "epoch": 5.830282861896839,
+      "loss": 0.7848711013793945,
+      "loss_ce": 0.0011797224869951606,
+      "loss_iou": 0.3125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 109799360,
+      "step": 1752
+    },
+    {
+      "epoch": 5.833610648918469,
+      "grad_norm": 10.353952407836914,
+      "learning_rate": 5e-06,
+      "loss": 0.6097,
+      "num_input_tokens_seen": 109860312,
+      "step": 1753
+    },
+    {
+      "epoch": 5.833610648918469,
+      "loss": 0.6687078475952148,
+      "loss_ce": 6.68507709633559e-06,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 109860312,
+      "step": 1753
+    },
+    {
+      "epoch": 5.8369384359401,
+      "grad_norm": 37.4395866394043,
+      "learning_rate": 5e-06,
+      "loss": 0.6346,
+      "num_input_tokens_seen": 109921664,
+      "step": 1754
+    },
+    {
+      "epoch": 5.8369384359401,
+      "loss": 0.6058779358863831,
+      "loss_ce": 0.0012636694591492414,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 109921664,
+      "step": 1754
+    },
+    {
+      "epoch": 5.840266222961731,
+      "grad_norm": 18.188020706176758,
+      "learning_rate": 5e-06,
+      "loss": 0.5968,
+      "num_input_tokens_seen": 109983676,
+      "step": 1755
+    },
+    {
+      "epoch": 5.840266222961731,
+      "loss": 0.5723915100097656,
+      "loss_ce": 0.003986367955803871,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 109983676,
+      "step": 1755
+    },
+    {
+      "epoch": 5.843594009983361,
+      "grad_norm": 17.082475662231445,
+      "learning_rate": 5e-06,
+      "loss": 0.6797,
+      "num_input_tokens_seen": 110046380,
+      "step": 1756
+    },
+    {
+      "epoch": 5.843594009983361,
+      "loss": 0.4576748311519623,
+      "loss_ce": 0.0001553079637233168,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.45703125,
+      "num_input_tokens_seen": 110046380,
+      "step": 1756
+    },
+    {
+      "epoch": 5.846921797004992,
+      "grad_norm": 9.511809349060059,
+      "learning_rate": 5e-06,
+      "loss": 0.525,
+      "num_input_tokens_seen": 110109532,
+      "step": 1757
+    },
+    {
+      "epoch": 5.846921797004992,
+      "loss": 0.4188133776187897,
+      "loss_ce": 0.00011222133616684005,
+      "loss_iou": 0.10888671875,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 110109532,
+      "step": 1757
+    },
+    {
+      "epoch": 5.850249584026622,
+      "grad_norm": 11.285202026367188,
+      "learning_rate": 5e-06,
+      "loss": 0.5777,
+      "num_input_tokens_seen": 110172300,
+      "step": 1758
+    },
+    {
+      "epoch": 5.850249584026622,
+      "loss": 0.657119631767273,
+      "loss_ce": 1.5172716302913614e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 110172300,
+      "step": 1758
+    },
+    {
+      "epoch": 5.853577371048253,
+      "grad_norm": 13.522523880004883,
+      "learning_rate": 5e-06,
+      "loss": 0.4297,
+      "num_input_tokens_seen": 110235616,
+      "step": 1759
+    },
+    {
+      "epoch": 5.853577371048253,
+      "loss": 0.5291931629180908,
+      "loss_ce": 1.8332864783587866e-05,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 110235616,
+      "step": 1759
+    },
+    {
+      "epoch": 5.856905158069884,
+      "grad_norm": 10.221686363220215,
+      "learning_rate": 5e-06,
+      "loss": 0.3553,
+      "num_input_tokens_seen": 110296840,
+      "step": 1760
+    },
+    {
+      "epoch": 5.856905158069884,
+      "loss": 0.4893026351928711,
+      "loss_ce": 0.00016688673349563032,
+      "loss_iou": 0.123046875,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.48828125,
+      "num_input_tokens_seen": 110296840,
+      "step": 1760
+    },
+    {
+      "epoch": 5.860232945091514,
+      "grad_norm": 21.090612411499023,
+      "learning_rate": 5e-06,
+      "loss": 0.6718,
+      "num_input_tokens_seen": 110359480,
+      "step": 1761
+    },
+    {
+      "epoch": 5.860232945091514,
+      "loss": 0.7827401161193848,
+      "loss_ce": 2.530876372475177e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 110359480,
+      "step": 1761
+    },
+    {
+      "epoch": 5.863560732113145,
+      "grad_norm": 25.800317764282227,
+      "learning_rate": 5e-06,
+      "loss": 0.764,
+      "num_input_tokens_seen": 110421012,
+      "step": 1762
+    },
+    {
+      "epoch": 5.863560732113145,
+      "loss": 0.8109021186828613,
+      "loss_ce": 0.0003552451671566814,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.05126953125,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 110421012,
+      "step": 1762
+    },
+    {
+      "epoch": 5.8668885191347755,
+      "grad_norm": 13.869083404541016,
+      "learning_rate": 5e-06,
+      "loss": 0.6566,
+      "num_input_tokens_seen": 110484348,
+      "step": 1763
+    },
+    {
+      "epoch": 5.8668885191347755,
+      "loss": 0.7028859257698059,
+      "loss_ce": 5.098950623505516e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 110484348,
+      "step": 1763
+    },
+    {
+      "epoch": 5.870216306156406,
+      "grad_norm": 11.48287296295166,
+      "learning_rate": 5e-06,
+      "loss": 0.7074,
+      "num_input_tokens_seen": 110547872,
+      "step": 1764
+    },
+    {
+      "epoch": 5.870216306156406,
+      "loss": 0.8577039241790771,
+      "loss_ce": 0.0011365680256858468,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 110547872,
+      "step": 1764
+    },
+    {
+      "epoch": 5.873544093178037,
+      "grad_norm": 14.099336624145508,
+      "learning_rate": 5e-06,
+      "loss": 0.6525,
+      "num_input_tokens_seen": 110611396,
+      "step": 1765
+    },
+    {
+      "epoch": 5.873544093178037,
+      "loss": 0.7531905174255371,
+      "loss_ce": 0.0022139903157949448,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 110611396,
+      "step": 1765
+    },
+    {
+      "epoch": 5.876871880199667,
+      "grad_norm": 17.61534309387207,
+      "learning_rate": 5e-06,
+      "loss": 0.8037,
+      "num_input_tokens_seen": 110673936,
+      "step": 1766
+    },
+    {
+      "epoch": 5.876871880199667,
+      "loss": 0.6117116808891296,
+      "loss_ce": 1.73293192347046e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 110673936,
+      "step": 1766
+    },
+    {
+      "epoch": 5.880199667221298,
+      "grad_norm": 31.511371612548828,
+      "learning_rate": 5e-06,
+      "loss": 0.7188,
+      "num_input_tokens_seen": 110737000,
+      "step": 1767
+    },
+    {
+      "epoch": 5.880199667221298,
+      "loss": 1.0573480129241943,
+      "loss_ce": 0.0015619065379723907,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.08154296875,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 110737000,
+      "step": 1767
+    },
+    {
+      "epoch": 5.883527454242929,
+      "grad_norm": 35.806556701660156,
+      "learning_rate": 5e-06,
+      "loss": 0.505,
+      "num_input_tokens_seen": 110800016,
+      "step": 1768
+    },
+    {
+      "epoch": 5.883527454242929,
+      "loss": 0.48276591300964355,
+      "loss_ce": 0.0013205973664298654,
+      "loss_iou": 0.15625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.48046875,
+      "num_input_tokens_seen": 110800016,
+      "step": 1768
+    },
+    {
+      "epoch": 5.886855241264559,
+      "grad_norm": 25.79120635986328,
+      "learning_rate": 5e-06,
+      "loss": 0.5601,
+      "num_input_tokens_seen": 110864032,
+      "step": 1769
+    },
+    {
+      "epoch": 5.886855241264559,
+      "loss": 0.5003718137741089,
+      "loss_ce": 0.0004939304199069738,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 110864032,
+      "step": 1769
+    },
+    {
+      "epoch": 5.89018302828619,
+      "grad_norm": 9.691274642944336,
+      "learning_rate": 5e-06,
+      "loss": 0.5045,
+      "num_input_tokens_seen": 110925816,
+      "step": 1770
+    },
+    {
+      "epoch": 5.89018302828619,
+      "loss": 0.4838283658027649,
+      "loss_ce": 3.321943586342968e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.484375,
+      "num_input_tokens_seen": 110925816,
+      "step": 1770
+    },
+    {
+      "epoch": 5.89351081530782,
+      "grad_norm": 12.821023941040039,
+      "learning_rate": 5e-06,
+      "loss": 0.6885,
+      "num_input_tokens_seen": 110989460,
+      "step": 1771
+    },
+    {
+      "epoch": 5.89351081530782,
+      "loss": 0.5637482404708862,
+      "loss_ce": 0.0002716918243095279,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 110989460,
+      "step": 1771
+    },
+    {
+      "epoch": 5.896838602329451,
+      "grad_norm": 9.636116981506348,
+      "learning_rate": 5e-06,
+      "loss": 0.444,
+      "num_input_tokens_seen": 111052032,
+      "step": 1772
+    },
+    {
+      "epoch": 5.896838602329451,
+      "loss": 0.4734896123409271,
+      "loss_ce": 0.0005892139161005616,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.47265625,
+      "num_input_tokens_seen": 111052032,
+      "step": 1772
+    },
+    {
+      "epoch": 5.900166389351082,
+      "grad_norm": 18.86623764038086,
+      "learning_rate": 5e-06,
+      "loss": 0.7722,
+      "num_input_tokens_seen": 111115768,
+      "step": 1773
+    },
+    {
+      "epoch": 5.900166389351082,
+      "loss": 0.7136220932006836,
+      "loss_ce": 0.0007315138936974108,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 111115768,
+      "step": 1773
+    },
+    {
+      "epoch": 5.903494176372712,
+      "grad_norm": 13.020090103149414,
+      "learning_rate": 5e-06,
+      "loss": 0.6273,
+      "num_input_tokens_seen": 111179252,
+      "step": 1774
+    },
+    {
+      "epoch": 5.903494176372712,
+      "loss": 0.5287013053894043,
+      "loss_ce": 1.4801251381868497e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 111179252,
+      "step": 1774
+    },
+    {
+      "epoch": 5.906821963394343,
+      "grad_norm": 23.916812896728516,
+      "learning_rate": 5e-06,
+      "loss": 0.5319,
+      "num_input_tokens_seen": 111241816,
+      "step": 1775
+    },
+    {
+      "epoch": 5.906821963394343,
+      "loss": 0.443467378616333,
+      "loss_ce": 0.0006573410355485976,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.443359375,
+      "num_input_tokens_seen": 111241816,
+      "step": 1775
+    },
+    {
+      "epoch": 5.9101497504159735,
+      "grad_norm": 19.583410263061523,
+      "learning_rate": 5e-06,
+      "loss": 0.5933,
+      "num_input_tokens_seen": 111304188,
+      "step": 1776
+    },
+    {
+      "epoch": 5.9101497504159735,
+      "loss": 0.5054777264595032,
+      "loss_ce": 0.00010664336150512099,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 111304188,
+      "step": 1776
+    },
+    {
+      "epoch": 5.913477537437604,
+      "grad_norm": 13.995909690856934,
+      "learning_rate": 5e-06,
+      "loss": 0.7816,
+      "num_input_tokens_seen": 111364944,
+      "step": 1777
+    },
+    {
+      "epoch": 5.913477537437604,
+      "loss": 1.0646007061004639,
+      "loss_ce": 0.0003917526628356427,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 1.0625,
+      "num_input_tokens_seen": 111364944,
+      "step": 1777
+    },
+    {
+      "epoch": 5.916805324459235,
+      "grad_norm": 9.358750343322754,
+      "learning_rate": 5e-06,
+      "loss": 0.5087,
+      "num_input_tokens_seen": 111427100,
+      "step": 1778
+    },
+    {
+      "epoch": 5.916805324459235,
+      "loss": 0.7140218615531921,
+      "loss_ce": 0.0008870940655469894,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 111427100,
+      "step": 1778
+    },
+    {
+      "epoch": 5.920133111480865,
+      "grad_norm": 13.724045753479004,
+      "learning_rate": 5e-06,
+      "loss": 0.7266,
+      "num_input_tokens_seen": 111490448,
+      "step": 1779
+    },
+    {
+      "epoch": 5.920133111480865,
+      "loss": 0.6259258985519409,
+      "loss_ce": 1.0390387615188956e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 111490448,
+      "step": 1779
+    },
+    {
+      "epoch": 5.923460898502496,
+      "grad_norm": 10.216179847717285,
+      "learning_rate": 5e-06,
+      "loss": 0.6562,
+      "num_input_tokens_seen": 111554604,
+      "step": 1780
+    },
+    {
+      "epoch": 5.923460898502496,
+      "loss": 0.7411832809448242,
+      "loss_ce": 0.00046060560271143913,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 111554604,
+      "step": 1780
+    },
+    {
+      "epoch": 5.9267886855241265,
+      "grad_norm": 17.054357528686523,
+      "learning_rate": 5e-06,
+      "loss": 0.7283,
+      "num_input_tokens_seen": 111617796,
+      "step": 1781
+    },
+    {
+      "epoch": 5.9267886855241265,
+      "loss": 0.9874345064163208,
+      "loss_ce": 0.0007401675684377551,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.98828125,
+      "num_input_tokens_seen": 111617796,
+      "step": 1781
+    },
+    {
+      "epoch": 5.930116472545757,
+      "grad_norm": 14.132950782775879,
+      "learning_rate": 5e-06,
+      "loss": 0.8466,
+      "num_input_tokens_seen": 111679904,
+      "step": 1782
+    },
+    {
+      "epoch": 5.930116472545757,
+      "loss": 0.9425029754638672,
+      "loss_ce": 0.0010967559646815062,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 111679904,
+      "step": 1782
+    },
+    {
+      "epoch": 5.933444259567388,
+      "grad_norm": 18.246789932250977,
+      "learning_rate": 5e-06,
+      "loss": 0.7549,
+      "num_input_tokens_seen": 111743964,
+      "step": 1783
+    },
+    {
+      "epoch": 5.933444259567388,
+      "loss": 0.9670125842094421,
+      "loss_ce": 0.00021572083642240614,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.96875,
+      "num_input_tokens_seen": 111743964,
+      "step": 1783
+    },
+    {
+      "epoch": 5.936772046589018,
+      "grad_norm": 18.004018783569336,
+      "learning_rate": 5e-06,
+      "loss": 0.7019,
+      "num_input_tokens_seen": 111805552,
+      "step": 1784
+    },
+    {
+      "epoch": 5.936772046589018,
+      "loss": 0.8806197643280029,
+      "loss_ce": 4.619309947884176e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 111805552,
+      "step": 1784
+    },
+    {
+      "epoch": 5.940099833610649,
+      "grad_norm": 10.934527397155762,
+      "learning_rate": 5e-06,
+      "loss": 0.5895,
+      "num_input_tokens_seen": 111869060,
+      "step": 1785
+    },
+    {
+      "epoch": 5.940099833610649,
+      "loss": 0.5993823409080505,
+      "loss_ce": 1.712212906568311e-05,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 111869060,
+      "step": 1785
+    },
+    {
+      "epoch": 5.94342762063228,
+      "grad_norm": 11.646177291870117,
+      "learning_rate": 5e-06,
+      "loss": 0.7022,
+      "num_input_tokens_seen": 111932476,
+      "step": 1786
+    },
+    {
+      "epoch": 5.94342762063228,
+      "loss": 0.7437913417816162,
+      "loss_ce": 1.6888494428712875e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 111932476,
+      "step": 1786
+    },
+    {
+      "epoch": 5.94675540765391,
+      "grad_norm": 9.77434253692627,
+      "learning_rate": 5e-06,
+      "loss": 0.3784,
+      "num_input_tokens_seen": 111995188,
+      "step": 1787
+    },
+    {
+      "epoch": 5.94675540765391,
+      "loss": 0.3339497745037079,
+      "loss_ce": 0.00014850017032586038,
+      "loss_iou": 0.1240234375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 111995188,
+      "step": 1787
+    },
+    {
+      "epoch": 5.950083194675541,
+      "grad_norm": 9.4547700881958,
+      "learning_rate": 5e-06,
+      "loss": 0.6886,
+      "num_input_tokens_seen": 112059688,
+      "step": 1788
+    },
+    {
+      "epoch": 5.950083194675541,
+      "loss": 0.7533557415008545,
+      "loss_ce": 0.0010364109184592962,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 112059688,
+      "step": 1788
+    },
+    {
+      "epoch": 5.953410981697171,
+      "grad_norm": 10.41918659210205,
+      "learning_rate": 5e-06,
+      "loss": 0.4527,
+      "num_input_tokens_seen": 112122556,
+      "step": 1789
+    },
+    {
+      "epoch": 5.953410981697171,
+      "loss": 0.4796237349510193,
+      "loss_ce": 9.511190000921488e-06,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.48046875,
+      "num_input_tokens_seen": 112122556,
+      "step": 1789
+    },
+    {
+      "epoch": 5.956738768718802,
+      "grad_norm": 7.984902381896973,
+      "learning_rate": 5e-06,
+      "loss": 0.6002,
+      "num_input_tokens_seen": 112184820,
+      "step": 1790
+    },
+    {
+      "epoch": 5.956738768718802,
+      "loss": 0.5412213206291199,
+      "loss_ce": 0.0007550385198555887,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 112184820,
+      "step": 1790
+    },
+    {
+      "epoch": 5.960066555740433,
+      "grad_norm": 10.358345031738281,
+      "learning_rate": 5e-06,
+      "loss": 0.6859,
+      "num_input_tokens_seen": 112246916,
+      "step": 1791
+    },
+    {
+      "epoch": 5.960066555740433,
+      "loss": 0.5734908580780029,
+      "loss_ce": 4.511678980634315e-06,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.058837890625,
+      "loss_xval": 0.57421875,
+      "num_input_tokens_seen": 112246916,
+      "step": 1791
+    },
+    {
+      "epoch": 5.963394342762063,
+      "grad_norm": 9.835028648376465,
+      "learning_rate": 5e-06,
+      "loss": 0.7321,
+      "num_input_tokens_seen": 112309796,
+      "step": 1792
+    },
+    {
+      "epoch": 5.963394342762063,
+      "loss": 0.707068681716919,
+      "loss_ce": 3.736479629878886e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.70703125,
+      "num_input_tokens_seen": 112309796,
+      "step": 1792
+    },
+    {
+      "epoch": 5.966722129783694,
+      "grad_norm": 8.542290687561035,
+      "learning_rate": 5e-06,
+      "loss": 0.771,
+      "num_input_tokens_seen": 112372752,
+      "step": 1793
+    },
+    {
+      "epoch": 5.966722129783694,
+      "loss": 0.784670352935791,
+      "loss_ce": 0.00036859206738881767,
+      "loss_iou": 0.28125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 112372752,
+      "step": 1793
+    },
+    {
+      "epoch": 5.9700499168053245,
+      "grad_norm": 10.128270149230957,
+      "learning_rate": 5e-06,
+      "loss": 0.6849,
+      "num_input_tokens_seen": 112435784,
+      "step": 1794
+    },
+    {
+      "epoch": 5.9700499168053245,
+      "loss": 0.7285070419311523,
+      "loss_ce": 5.2509160013869405e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 112435784,
+      "step": 1794
+    },
+    {
+      "epoch": 5.973377703826955,
+      "grad_norm": 21.481040954589844,
+      "learning_rate": 5e-06,
+      "loss": 0.7537,
+      "num_input_tokens_seen": 112499088,
+      "step": 1795
+    },
+    {
+      "epoch": 5.973377703826955,
+      "loss": 0.4841929078102112,
+      "loss_ce": 0.000794428342487663,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.484375,
+      "num_input_tokens_seen": 112499088,
+      "step": 1795
+    },
+    {
+      "epoch": 5.976705490848586,
+      "grad_norm": 17.77248764038086,
+      "learning_rate": 5e-06,
+      "loss": 0.5665,
+      "num_input_tokens_seen": 112562248,
+      "step": 1796
+    },
+    {
+      "epoch": 5.976705490848586,
+      "loss": 0.34294500946998596,
+      "loss_ce": 0.0002020865649683401,
+      "loss_iou": 0.1142578125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 112562248,
+      "step": 1796
+    },
+    {
+      "epoch": 5.980033277870216,
+      "grad_norm": 9.828653335571289,
+      "learning_rate": 5e-06,
+      "loss": 0.615,
+      "num_input_tokens_seen": 112624292,
+      "step": 1797
+    },
+    {
+      "epoch": 5.980033277870216,
+      "loss": 0.8139055371284485,
+      "loss_ce": 0.0006731028552167118,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 112624292,
+      "step": 1797
+    },
+    {
+      "epoch": 5.983361064891847,
+      "grad_norm": 10.219780921936035,
+      "learning_rate": 5e-06,
+      "loss": 0.5873,
+      "num_input_tokens_seen": 112687100,
+      "step": 1798
+    },
+    {
+      "epoch": 5.983361064891847,
+      "loss": 0.6688265204429626,
+      "loss_ce": 3.2744080726843094e-06,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 112687100,
+      "step": 1798
+    },
+    {
+      "epoch": 5.9866888519134775,
+      "grad_norm": 17.44321632385254,
+      "learning_rate": 5e-06,
+      "loss": 0.6239,
+      "num_input_tokens_seen": 112748904,
+      "step": 1799
+    },
+    {
+      "epoch": 5.9866888519134775,
+      "loss": 0.7041700482368469,
+      "loss_ce": 0.0011061305413022637,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 112748904,
+      "step": 1799
+    },
+    {
+      "epoch": 5.990016638935108,
+      "grad_norm": 20.630294799804688,
+      "learning_rate": 5e-06,
+      "loss": 0.7101,
+      "num_input_tokens_seen": 112810924,
+      "step": 1800
+    },
+    {
+      "epoch": 5.990016638935108,
+      "loss": 0.6245334148406982,
+      "loss_ce": 0.00014378642663359642,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 112810924,
+      "step": 1800
+    },
+    {
+      "epoch": 5.993344425956739,
+      "grad_norm": 17.083229064941406,
+      "learning_rate": 5e-06,
+      "loss": 0.686,
+      "num_input_tokens_seen": 112875104,
+      "step": 1801
+    },
+    {
+      "epoch": 5.993344425956739,
+      "loss": 0.7393364310264587,
+      "loss_ce": 0.0003227871493436396,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.73828125,
+      "num_input_tokens_seen": 112875104,
+      "step": 1801
+    },
+    {
+      "epoch": 5.996672212978369,
+      "grad_norm": 5.339547157287598,
+      "learning_rate": 5e-06,
+      "loss": 0.4334,
+      "num_input_tokens_seen": 112937884,
+      "step": 1802
+    },
+    {
+      "epoch": 5.996672212978369,
+      "loss": 0.4524919390678406,
+      "loss_ce": 0.00022140935470815748,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 112937884,
+      "step": 1802
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 12.052851676940918,
+      "learning_rate": 5e-06,
+      "loss": 0.5541,
+      "num_input_tokens_seen": 113000836,
+      "step": 1803
+    },
+    {
+      "epoch": 6.0,
+      "loss": 0.6314157247543335,
+      "loss_ce": 0.00012909869838040322,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 113000836,
+      "step": 1803
+    },
+    {
+      "epoch": 6.003327787021631,
+      "grad_norm": 17.944114685058594,
+      "learning_rate": 5e-06,
+      "loss": 0.6348,
+      "num_input_tokens_seen": 113064196,
+      "step": 1804
+    },
+    {
+      "epoch": 6.003327787021631,
+      "loss": 0.6592069864273071,
+      "loss_ce": 2.7252650397713296e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 113064196,
+      "step": 1804
+    },
+    {
+      "epoch": 6.006655574043261,
+      "grad_norm": 8.979909896850586,
+      "learning_rate": 5e-06,
+      "loss": 0.4955,
+      "num_input_tokens_seen": 113125704,
+      "step": 1805
+    },
+    {
+      "epoch": 6.006655574043261,
+      "loss": 0.5311020612716675,
+      "loss_ce": 0.0003403459268156439,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 113125704,
+      "step": 1805
+    },
+    {
+      "epoch": 6.009983361064892,
+      "grad_norm": 23.03765869140625,
+      "learning_rate": 5e-06,
+      "loss": 0.5458,
+      "num_input_tokens_seen": 113189092,
+      "step": 1806
+    },
+    {
+      "epoch": 6.009983361064892,
+      "loss": 0.6970461010932922,
+      "loss_ce": 0.00026878988137468696,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 113189092,
+      "step": 1806
+    },
+    {
+      "epoch": 6.0133111480865225,
+      "grad_norm": 40.743404388427734,
+      "learning_rate": 5e-06,
+      "loss": 0.7297,
+      "num_input_tokens_seen": 113252312,
+      "step": 1807
+    },
+    {
+      "epoch": 6.0133111480865225,
+      "loss": 0.9480460286140442,
+      "loss_ce": 0.0007804275956004858,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.05126953125,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 113252312,
+      "step": 1807
+    },
+    {
+      "epoch": 6.016638935108153,
+      "grad_norm": 24.678939819335938,
+      "learning_rate": 5e-06,
+      "loss": 0.4995,
+      "num_input_tokens_seen": 113315144,
+      "step": 1808
+    },
+    {
+      "epoch": 6.016638935108153,
+      "loss": 0.6151464581489563,
+      "loss_ce": 0.00015620810154359788,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 113315144,
+      "step": 1808
+    },
+    {
+      "epoch": 6.019966722129784,
+      "grad_norm": 8.93970775604248,
+      "learning_rate": 5e-06,
+      "loss": 0.6358,
+      "num_input_tokens_seen": 113378196,
+      "step": 1809
+    },
+    {
+      "epoch": 6.019966722129784,
+      "loss": 0.4480162262916565,
+      "loss_ce": 0.0001402330381097272,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.447265625,
+      "num_input_tokens_seen": 113378196,
+      "step": 1809
+    },
+    {
+      "epoch": 6.023294509151414,
+      "grad_norm": 6.139615058898926,
+      "learning_rate": 5e-06,
+      "loss": 0.3159,
+      "num_input_tokens_seen": 113438784,
+      "step": 1810
+    },
+    {
+      "epoch": 6.023294509151414,
+      "loss": 0.3206220269203186,
+      "loss_ce": 4.330090632720385e-06,
+      "loss_iou": 0.07421875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 113438784,
+      "step": 1810
+    },
+    {
+      "epoch": 6.026622296173045,
+      "grad_norm": 25.28673553466797,
+      "learning_rate": 5e-06,
+      "loss": 0.5646,
+      "num_input_tokens_seen": 113501584,
+      "step": 1811
+    },
+    {
+      "epoch": 6.026622296173045,
+      "loss": 0.47726649045944214,
+      "loss_ce": 0.0009481181623414159,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 113501584,
+      "step": 1811
+    },
+    {
+      "epoch": 6.0299500831946755,
+      "grad_norm": 23.174034118652344,
+      "learning_rate": 5e-06,
+      "loss": 0.6892,
+      "num_input_tokens_seen": 113564772,
+      "step": 1812
+    },
+    {
+      "epoch": 6.0299500831946755,
+      "loss": 0.638823390007019,
+      "loss_ce": 0.0003956563596148044,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 113564772,
+      "step": 1812
+    },
+    {
+      "epoch": 6.033277870216306,
+      "grad_norm": 9.570660591125488,
+      "learning_rate": 5e-06,
+      "loss": 0.482,
+      "num_input_tokens_seen": 113627164,
+      "step": 1813
+    },
+    {
+      "epoch": 6.033277870216306,
+      "loss": 0.5443622469902039,
+      "loss_ce": 0.00017276505241170526,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 113627164,
+      "step": 1813
+    },
+    {
+      "epoch": 6.036605657237937,
+      "grad_norm": 15.14444351196289,
+      "learning_rate": 5e-06,
+      "loss": 0.567,
+      "num_input_tokens_seen": 113690112,
+      "step": 1814
+    },
+    {
+      "epoch": 6.036605657237937,
+      "loss": 0.4782255291938782,
+      "loss_ce": 1.5094023183337413e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.478515625,
+      "num_input_tokens_seen": 113690112,
+      "step": 1814
+    },
+    {
+      "epoch": 6.039933444259567,
+      "grad_norm": 13.660589218139648,
+      "learning_rate": 5e-06,
+      "loss": 0.617,
+      "num_input_tokens_seen": 113753248,
+      "step": 1815
+    },
+    {
+      "epoch": 6.039933444259567,
+      "loss": 0.7184145450592041,
+      "loss_ce": 3.077809378737584e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 113753248,
+      "step": 1815
+    },
+    {
+      "epoch": 6.043261231281198,
+      "grad_norm": 9.653757095336914,
+      "learning_rate": 5e-06,
+      "loss": 0.9097,
+      "num_input_tokens_seen": 113817632,
+      "step": 1816
+    },
+    {
+      "epoch": 6.043261231281198,
+      "loss": 1.0042712688446045,
+      "loss_ce": 0.0004870828415732831,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.064453125,
+      "loss_xval": 1.0,
+      "num_input_tokens_seen": 113817632,
+      "step": 1816
+    },
+    {
+      "epoch": 6.046589018302829,
+      "grad_norm": 10.754157066345215,
+      "learning_rate": 5e-06,
+      "loss": 0.791,
+      "num_input_tokens_seen": 113880676,
+      "step": 1817
+    },
+    {
+      "epoch": 6.046589018302829,
+      "loss": 0.6697897911071777,
+      "loss_ce": 0.00011203553731320426,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 113880676,
+      "step": 1817
+    },
+    {
+      "epoch": 6.049916805324459,
+      "grad_norm": 12.63485336303711,
+      "learning_rate": 5e-06,
+      "loss": 0.5059,
+      "num_input_tokens_seen": 113941992,
+      "step": 1818
+    },
+    {
+      "epoch": 6.049916805324459,
+      "loss": 0.4937788248062134,
+      "loss_ce": 4.377495315566193e-06,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.494140625,
+      "num_input_tokens_seen": 113941992,
+      "step": 1818
+    },
+    {
+      "epoch": 6.05324459234609,
+      "grad_norm": 6.852193832397461,
+      "learning_rate": 5e-06,
+      "loss": 0.5946,
+      "num_input_tokens_seen": 114003684,
+      "step": 1819
+    },
+    {
+      "epoch": 6.05324459234609,
+      "loss": 0.6906334161758423,
+      "loss_ce": 0.0008140857680700719,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.69140625,
+      "num_input_tokens_seen": 114003684,
+      "step": 1819
+    },
+    {
+      "epoch": 6.05657237936772,
+      "grad_norm": 14.351150512695312,
+      "learning_rate": 5e-06,
+      "loss": 0.7158,
+      "num_input_tokens_seen": 114065336,
+      "step": 1820
+    },
+    {
+      "epoch": 6.05657237936772,
+      "loss": 0.7381648421287537,
+      "loss_ce": 5.69224130231305e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.73828125,
+      "num_input_tokens_seen": 114065336,
+      "step": 1820
+    },
+    {
+      "epoch": 6.059900166389351,
+      "grad_norm": 30.204614639282227,
+      "learning_rate": 5e-06,
+      "loss": 0.6653,
+      "num_input_tokens_seen": 114127604,
+      "step": 1821
+    },
+    {
+      "epoch": 6.059900166389351,
+      "loss": 0.6157270073890686,
+      "loss_ce": 0.0010419311001896858,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 114127604,
+      "step": 1821
+    },
+    {
+      "epoch": 6.063227953410982,
+      "grad_norm": 9.658365249633789,
+      "learning_rate": 5e-06,
+      "loss": 0.5936,
+      "num_input_tokens_seen": 114190672,
+      "step": 1822
+    },
+    {
+      "epoch": 6.063227953410982,
+      "loss": 0.6472856402397156,
+      "loss_ce": 0.0008012944017536938,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 114190672,
+      "step": 1822
+    },
+    {
+      "epoch": 6.066555740432612,
+      "grad_norm": 18.5205135345459,
+      "learning_rate": 5e-06,
+      "loss": 0.6124,
+      "num_input_tokens_seen": 114253368,
+      "step": 1823
+    },
+    {
+      "epoch": 6.066555740432612,
+      "loss": 0.735488772392273,
+      "loss_ce": 0.0006254613981582224,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 114253368,
+      "step": 1823
+    },
+    {
+      "epoch": 6.069883527454243,
+      "grad_norm": 9.273344993591309,
+      "learning_rate": 5e-06,
+      "loss": 0.8266,
+      "num_input_tokens_seen": 114314064,
+      "step": 1824
+    },
+    {
+      "epoch": 6.069883527454243,
+      "loss": 0.7096292972564697,
+      "loss_ce": 0.0001566863793414086,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 114314064,
+      "step": 1824
+    },
+    {
+      "epoch": 6.0732113144758735,
+      "grad_norm": 10.49995231628418,
+      "learning_rate": 5e-06,
+      "loss": 0.4339,
+      "num_input_tokens_seen": 114374024,
+      "step": 1825
+    },
+    {
+      "epoch": 6.0732113144758735,
+      "loss": 0.4444683790206909,
+      "loss_ce": 1.035763489198871e-05,
+      "loss_iou": 0.10302734375,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.4453125,
+      "num_input_tokens_seen": 114374024,
+      "step": 1825
+    },
+    {
+      "epoch": 6.076539101497504,
+      "grad_norm": 30.338027954101562,
+      "learning_rate": 5e-06,
+      "loss": 0.6147,
+      "num_input_tokens_seen": 114437368,
+      "step": 1826
+    },
+    {
+      "epoch": 6.076539101497504,
+      "loss": 0.7704422473907471,
+      "loss_ce": 0.0011551063507795334,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 114437368,
+      "step": 1826
+    },
+    {
+      "epoch": 6.079866888519135,
+      "grad_norm": 21.672454833984375,
+      "learning_rate": 5e-06,
+      "loss": 0.69,
+      "num_input_tokens_seen": 114500248,
+      "step": 1827
+    },
+    {
+      "epoch": 6.079866888519135,
+      "loss": 0.42282021045684814,
+      "loss_ce": 0.0004569635493680835,
+      "loss_iou": 0.140625,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 114500248,
+      "step": 1827
+    },
+    {
+      "epoch": 6.083194675540765,
+      "grad_norm": 12.467771530151367,
+      "learning_rate": 5e-06,
+      "loss": 0.5934,
+      "num_input_tokens_seen": 114562540,
+      "step": 1828
+    },
+    {
+      "epoch": 6.083194675540765,
+      "loss": 0.4341202676296234,
+      "loss_ce": 0.0006791083724237978,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.43359375,
+      "num_input_tokens_seen": 114562540,
+      "step": 1828
+    },
+    {
+      "epoch": 6.086522462562396,
+      "grad_norm": 11.479426383972168,
+      "learning_rate": 5e-06,
+      "loss": 0.6028,
+      "num_input_tokens_seen": 114624740,
+      "step": 1829
+    },
+    {
+      "epoch": 6.086522462562396,
+      "loss": 0.5540913343429565,
+      "loss_ce": 1.4216830095392652e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.5546875,
+      "num_input_tokens_seen": 114624740,
+      "step": 1829
+    },
+    {
+      "epoch": 6.0898502495840265,
+      "grad_norm": 14.962776184082031,
+      "learning_rate": 5e-06,
+      "loss": 0.7424,
+      "num_input_tokens_seen": 114689968,
+      "step": 1830
+    },
+    {
+      "epoch": 6.0898502495840265,
+      "loss": 0.5499890446662903,
+      "loss_ce": 0.000794704130385071,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 114689968,
+      "step": 1830
+    },
+    {
+      "epoch": 6.093178036605657,
+      "grad_norm": 46.97150421142578,
+      "learning_rate": 5e-06,
+      "loss": 0.8063,
+      "num_input_tokens_seen": 114755060,
+      "step": 1831
+    },
+    {
+      "epoch": 6.093178036605657,
+      "loss": 0.8358315825462341,
+      "loss_ce": 0.0001382491027470678,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 114755060,
+      "step": 1831
+    },
+    {
+      "epoch": 6.096505823627288,
+      "grad_norm": 24.191722869873047,
+      "learning_rate": 5e-06,
+      "loss": 0.5722,
+      "num_input_tokens_seen": 114817492,
+      "step": 1832
+    },
+    {
+      "epoch": 6.096505823627288,
+      "loss": 0.5860408544540405,
+      "loss_ce": 0.00022541148064192384,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 114817492,
+      "step": 1832
+    },
+    {
+      "epoch": 6.099833610648918,
+      "grad_norm": 12.793700218200684,
+      "learning_rate": 5e-06,
+      "loss": 0.7779,
+      "num_input_tokens_seen": 114881384,
+      "step": 1833
+    },
+    {
+      "epoch": 6.099833610648918,
+      "loss": 0.7877440452575684,
+      "loss_ce": 0.0002685006766114384,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 114881384,
+      "step": 1833
+    },
+    {
+      "epoch": 6.103161397670549,
+      "grad_norm": 11.71887493133545,
+      "learning_rate": 5e-06,
+      "loss": 0.581,
+      "num_input_tokens_seen": 114943416,
+      "step": 1834
+    },
+    {
+      "epoch": 6.103161397670549,
+      "loss": 0.3188535273075104,
+      "loss_ce": 5.8923419601342175e-06,
+      "loss_iou": 0.0625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 114943416,
+      "step": 1834
+    },
+    {
+      "epoch": 6.10648918469218,
+      "grad_norm": 8.89264965057373,
+      "learning_rate": 5e-06,
+      "loss": 0.5685,
+      "num_input_tokens_seen": 115005544,
+      "step": 1835
+    },
+    {
+      "epoch": 6.10648918469218,
+      "loss": 0.5760512351989746,
+      "loss_ce": 1.4534259662468685e-06,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.57421875,
+      "num_input_tokens_seen": 115005544,
+      "step": 1835
+    },
+    {
+      "epoch": 6.10981697171381,
+      "grad_norm": 23.207515716552734,
+      "learning_rate": 5e-06,
+      "loss": 0.6417,
+      "num_input_tokens_seen": 115066228,
+      "step": 1836
+    },
+    {
+      "epoch": 6.10981697171381,
+      "loss": 0.5505459308624268,
+      "loss_ce": 0.001595688983798027,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 115066228,
+      "step": 1836
+    },
+    {
+      "epoch": 6.113144758735441,
+      "grad_norm": 44.30556106567383,
+      "learning_rate": 5e-06,
+      "loss": 0.8094,
+      "num_input_tokens_seen": 115130620,
+      "step": 1837
+    },
+    {
+      "epoch": 6.113144758735441,
+      "loss": 0.7348802089691162,
+      "loss_ce": 0.0002610796072985977,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 115130620,
+      "step": 1837
+    },
+    {
+      "epoch": 6.116472545757071,
+      "grad_norm": 24.386632919311523,
+      "learning_rate": 5e-06,
+      "loss": 0.6976,
+      "num_input_tokens_seen": 115192732,
+      "step": 1838
+    },
+    {
+      "epoch": 6.116472545757071,
+      "loss": 0.7894414663314819,
+      "loss_ce": 0.0007451603887602687,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 115192732,
+      "step": 1838
+    },
+    {
+      "epoch": 6.119800332778702,
+      "grad_norm": 13.562687873840332,
+      "learning_rate": 5e-06,
+      "loss": 0.6654,
+      "num_input_tokens_seen": 115257288,
+      "step": 1839
+    },
+    {
+      "epoch": 6.119800332778702,
+      "loss": 0.803615152835846,
+      "loss_ce": 0.0008807668928056955,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 115257288,
+      "step": 1839
+    },
+    {
+      "epoch": 6.123128119800333,
+      "grad_norm": 15.4951753616333,
+      "learning_rate": 5e-06,
+      "loss": 0.7436,
+      "num_input_tokens_seen": 115321772,
+      "step": 1840
+    },
+    {
+      "epoch": 6.123128119800333,
+      "loss": 0.7024877071380615,
+      "loss_ce": 0.0003393086954019964,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 115321772,
+      "step": 1840
+    },
+    {
+      "epoch": 6.126455906821963,
+      "grad_norm": 16.896167755126953,
+      "learning_rate": 5e-06,
+      "loss": 0.5496,
+      "num_input_tokens_seen": 115383640,
+      "step": 1841
+    },
+    {
+      "epoch": 6.126455906821963,
+      "loss": 0.6101754903793335,
+      "loss_ce": 0.0001900848001241684,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 115383640,
+      "step": 1841
+    },
+    {
+      "epoch": 6.129783693843594,
+      "grad_norm": 25.171308517456055,
+      "learning_rate": 5e-06,
+      "loss": 0.5562,
+      "num_input_tokens_seen": 115445032,
+      "step": 1842
+    },
+    {
+      "epoch": 6.129783693843594,
+      "loss": 0.6438261270523071,
+      "loss_ce": 0.0011869219597429037,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 115445032,
+      "step": 1842
+    },
+    {
+      "epoch": 6.1331114808652245,
+      "grad_norm": 17.913129806518555,
+      "learning_rate": 5e-06,
+      "loss": 0.3981,
+      "num_input_tokens_seen": 115506208,
+      "step": 1843
+    },
+    {
+      "epoch": 6.1331114808652245,
+      "loss": 0.3706634044647217,
+      "loss_ce": 5.7946737797465175e-05,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 115506208,
+      "step": 1843
+    },
+    {
+      "epoch": 6.136439267886855,
+      "grad_norm": 11.869161605834961,
+      "learning_rate": 5e-06,
+      "loss": 0.5245,
+      "num_input_tokens_seen": 115568952,
+      "step": 1844
+    },
+    {
+      "epoch": 6.136439267886855,
+      "loss": 0.6334899663925171,
+      "loss_ce": 0.0007995049236342311,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 115568952,
+      "step": 1844
+    },
+    {
+      "epoch": 6.139767054908486,
+      "grad_norm": 19.262910842895508,
+      "learning_rate": 5e-06,
+      "loss": 0.6218,
+      "num_input_tokens_seen": 115631376,
+      "step": 1845
+    },
+    {
+      "epoch": 6.139767054908486,
+      "loss": 0.6551563143730164,
+      "loss_ce": 4.9998088798020035e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 115631376,
+      "step": 1845
+    },
+    {
+      "epoch": 6.143094841930116,
+      "grad_norm": 16.876794815063477,
+      "learning_rate": 5e-06,
+      "loss": 0.5707,
+      "num_input_tokens_seen": 115694676,
+      "step": 1846
+    },
+    {
+      "epoch": 6.143094841930116,
+      "loss": 0.5590164661407471,
+      "loss_ce": 0.002131724264472723,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 115694676,
+      "step": 1846
+    },
+    {
+      "epoch": 6.146422628951747,
+      "grad_norm": 13.508707046508789,
+      "learning_rate": 5e-06,
+      "loss": 0.4522,
+      "num_input_tokens_seen": 115757384,
+      "step": 1847
+    },
+    {
+      "epoch": 6.146422628951747,
+      "loss": 0.39521026611328125,
+      "loss_ce": 0.00019072243594564497,
+      "loss_iou": 0.119140625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 115757384,
+      "step": 1847
+    },
+    {
+      "epoch": 6.149750415973378,
+      "grad_norm": 24.865703582763672,
+      "learning_rate": 5e-06,
+      "loss": 0.6482,
+      "num_input_tokens_seen": 115820368,
+      "step": 1848
+    },
+    {
+      "epoch": 6.149750415973378,
+      "loss": 0.8225171566009521,
+      "loss_ce": 7.408522833429743e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 115820368,
+      "step": 1848
+    },
+    {
+      "epoch": 6.153078202995008,
+      "grad_norm": 10.617776870727539,
+      "learning_rate": 5e-06,
+      "loss": 0.7869,
+      "num_input_tokens_seen": 115883888,
+      "step": 1849
+    },
+    {
+      "epoch": 6.153078202995008,
+      "loss": 0.7970017194747925,
+      "loss_ce": 4.6364102672669105e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 115883888,
+      "step": 1849
+    },
+    {
+      "epoch": 6.156405990016639,
+      "grad_norm": 13.21096134185791,
+      "learning_rate": 5e-06,
+      "loss": 0.6439,
+      "num_input_tokens_seen": 115947024,
+      "step": 1850
+    },
+    {
+      "epoch": 6.156405990016639,
+      "loss": 0.5439035892486572,
+      "loss_ce": 0.0003855030226986855,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 115947024,
+      "step": 1850
+    },
+    {
+      "epoch": 6.159733777038269,
+      "grad_norm": 12.673142433166504,
+      "learning_rate": 5e-06,
+      "loss": 0.591,
+      "num_input_tokens_seen": 116008540,
+      "step": 1851
+    },
+    {
+      "epoch": 6.159733777038269,
+      "loss": 0.5002042651176453,
+      "loss_ce": 0.0006009893259033561,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 116008540,
+      "step": 1851
+    },
+    {
+      "epoch": 6.1630615640599,
+      "grad_norm": 12.410579681396484,
+      "learning_rate": 5e-06,
+      "loss": 0.5276,
+      "num_input_tokens_seen": 116069856,
+      "step": 1852
+    },
+    {
+      "epoch": 6.1630615640599,
+      "loss": 0.6053777933120728,
+      "loss_ce": 0.00015313336916733533,
+      "loss_iou": 0.203125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 116069856,
+      "step": 1852
+    },
+    {
+      "epoch": 6.166389351081531,
+      "grad_norm": 10.24417781829834,
+      "learning_rate": 5e-06,
+      "loss": 0.6968,
+      "num_input_tokens_seen": 116132820,
+      "step": 1853
+    },
+    {
+      "epoch": 6.166389351081531,
+      "loss": 0.45372575521469116,
+      "loss_ce": 0.0004481864161789417,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 116132820,
+      "step": 1853
+    },
+    {
+      "epoch": 6.169717138103161,
+      "grad_norm": 19.29709243774414,
+      "learning_rate": 5e-06,
+      "loss": 0.8242,
+      "num_input_tokens_seen": 116197008,
+      "step": 1854
+    },
+    {
+      "epoch": 6.169717138103161,
+      "loss": 0.7781339287757874,
+      "loss_ce": 0.0007902136421762407,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 116197008,
+      "step": 1854
+    },
+    {
+      "epoch": 6.173044925124792,
+      "grad_norm": 16.979761123657227,
+      "learning_rate": 5e-06,
+      "loss": 0.5969,
+      "num_input_tokens_seen": 116260052,
+      "step": 1855
+    },
+    {
+      "epoch": 6.173044925124792,
+      "loss": 0.5408188104629517,
+      "loss_ce": 4.733385139843449e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 116260052,
+      "step": 1855
+    },
+    {
+      "epoch": 6.1763727121464225,
+      "grad_norm": 25.549842834472656,
+      "learning_rate": 5e-06,
+      "loss": 0.8045,
+      "num_input_tokens_seen": 116322916,
+      "step": 1856
+    },
+    {
+      "epoch": 6.1763727121464225,
+      "loss": 0.9726593494415283,
+      "loss_ce": 0.0002472798223607242,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.97265625,
+      "num_input_tokens_seen": 116322916,
+      "step": 1856
+    },
+    {
+      "epoch": 6.179700499168053,
+      "grad_norm": 11.069473266601562,
+      "learning_rate": 5e-06,
+      "loss": 0.6182,
+      "num_input_tokens_seen": 116386300,
+      "step": 1857
+    },
+    {
+      "epoch": 6.179700499168053,
+      "loss": 0.8058730363845825,
+      "loss_ce": 0.0003310354077257216,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.046875,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 116386300,
+      "step": 1857
+    },
+    {
+      "epoch": 6.183028286189684,
+      "grad_norm": 11.940938949584961,
+      "learning_rate": 5e-06,
+      "loss": 0.6844,
+      "num_input_tokens_seen": 116448856,
+      "step": 1858
+    },
+    {
+      "epoch": 6.183028286189684,
+      "loss": 0.8730907440185547,
+      "loss_ce": 0.0008984009618870914,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 116448856,
+      "step": 1858
+    },
+    {
+      "epoch": 6.186356073211314,
+      "grad_norm": 26.180950164794922,
+      "learning_rate": 5e-06,
+      "loss": 0.6785,
+      "num_input_tokens_seen": 116513172,
+      "step": 1859
+    },
+    {
+      "epoch": 6.186356073211314,
+      "loss": 0.6941823959350586,
+      "loss_ce": 0.0003347777819726616,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 116513172,
+      "step": 1859
+    },
+    {
+      "epoch": 6.189683860232945,
+      "grad_norm": 38.279239654541016,
+      "learning_rate": 5e-06,
+      "loss": 0.5853,
+      "num_input_tokens_seen": 116575568,
+      "step": 1860
+    },
+    {
+      "epoch": 6.189683860232945,
+      "loss": 0.6565850973129272,
+      "loss_ce": 0.0003350722254253924,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 116575568,
+      "step": 1860
+    },
+    {
+      "epoch": 6.1930116472545755,
+      "grad_norm": 23.869171142578125,
+      "learning_rate": 5e-06,
+      "loss": 0.5414,
+      "num_input_tokens_seen": 116638880,
+      "step": 1861
+    },
+    {
+      "epoch": 6.1930116472545755,
+      "loss": 0.48156753182411194,
+      "loss_ce": 0.0002443046832922846,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.48046875,
+      "num_input_tokens_seen": 116638880,
+      "step": 1861
+    },
+    {
+      "epoch": 6.196339434276206,
+      "grad_norm": 36.01377487182617,
+      "learning_rate": 5e-06,
+      "loss": 0.6001,
+      "num_input_tokens_seen": 116702460,
+      "step": 1862
+    },
+    {
+      "epoch": 6.196339434276206,
+      "loss": 0.5464353561401367,
+      "loss_ce": 0.0007810404058545828,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 116702460,
+      "step": 1862
+    },
+    {
+      "epoch": 6.199667221297837,
+      "grad_norm": 21.87510108947754,
+      "learning_rate": 5e-06,
+      "loss": 0.6532,
+      "num_input_tokens_seen": 116765876,
+      "step": 1863
+    },
+    {
+      "epoch": 6.199667221297837,
+      "loss": 0.814652144908905,
+      "loss_ce": 0.00019903229258488864,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 116765876,
+      "step": 1863
+    },
+    {
+      "epoch": 6.202995008319467,
+      "grad_norm": 7.908597946166992,
+      "learning_rate": 5e-06,
+      "loss": 0.5938,
+      "num_input_tokens_seen": 116828836,
+      "step": 1864
+    },
+    {
+      "epoch": 6.202995008319467,
+      "loss": 0.7178022265434265,
+      "loss_ce": 0.000639147125184536,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 116828836,
+      "step": 1864
+    },
+    {
+      "epoch": 6.206322795341098,
+      "grad_norm": 11.291393280029297,
+      "learning_rate": 5e-06,
+      "loss": 0.5858,
+      "num_input_tokens_seen": 116890788,
+      "step": 1865
+    },
+    {
+      "epoch": 6.206322795341098,
+      "loss": 0.7302819490432739,
+      "loss_ce": 0.0004235214146319777,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 116890788,
+      "step": 1865
+    },
+    {
+      "epoch": 6.209650582362729,
+      "grad_norm": 27.98455238342285,
+      "learning_rate": 5e-06,
+      "loss": 0.6189,
+      "num_input_tokens_seen": 116953012,
+      "step": 1866
+    },
+    {
+      "epoch": 6.209650582362729,
+      "loss": 0.7605868577957153,
+      "loss_ce": 8.884212729753926e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.76171875,
+      "num_input_tokens_seen": 116953012,
+      "step": 1866
+    },
+    {
+      "epoch": 6.212978369384359,
+      "grad_norm": 37.32789611816406,
+      "learning_rate": 5e-06,
+      "loss": 0.6325,
+      "num_input_tokens_seen": 117016728,
+      "step": 1867
+    },
+    {
+      "epoch": 6.212978369384359,
+      "loss": 0.5575933456420898,
+      "loss_ce": 0.0004644446889869869,
+      "loss_iou": 0.1875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 117016728,
+      "step": 1867
+    },
+    {
+      "epoch": 6.21630615640599,
+      "grad_norm": 8.77429485321045,
+      "learning_rate": 5e-06,
+      "loss": 0.7978,
+      "num_input_tokens_seen": 117079776,
+      "step": 1868
+    },
+    {
+      "epoch": 6.21630615640599,
+      "loss": 0.5884289741516113,
+      "loss_ce": 0.0005383165553212166,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 117079776,
+      "step": 1868
+    },
+    {
+      "epoch": 6.21963394342762,
+      "grad_norm": 13.149292945861816,
+      "learning_rate": 5e-06,
+      "loss": 0.4875,
+      "num_input_tokens_seen": 117142040,
+      "step": 1869
+    },
+    {
+      "epoch": 6.21963394342762,
+      "loss": 0.41048645973205566,
+      "loss_ce": 2.5050567273865454e-05,
+      "loss_iou": 0.140625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.41015625,
+      "num_input_tokens_seen": 117142040,
+      "step": 1869
+    },
+    {
+      "epoch": 6.222961730449251,
+      "grad_norm": 22.748687744140625,
+      "learning_rate": 5e-06,
+      "loss": 0.6843,
+      "num_input_tokens_seen": 117204720,
+      "step": 1870
+    },
+    {
+      "epoch": 6.222961730449251,
+      "loss": 0.6726725101470947,
+      "loss_ce": 6.513913831440732e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 117204720,
+      "step": 1870
+    },
+    {
+      "epoch": 6.226289517470882,
+      "grad_norm": 25.107208251953125,
+      "learning_rate": 5e-06,
+      "loss": 0.5835,
+      "num_input_tokens_seen": 117268404,
+      "step": 1871
+    },
+    {
+      "epoch": 6.226289517470882,
+      "loss": 0.5766459703445435,
+      "loss_ce": 0.0013285500463098288,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.57421875,
+      "num_input_tokens_seen": 117268404,
+      "step": 1871
+    },
+    {
+      "epoch": 6.229617304492512,
+      "grad_norm": 15.994500160217285,
+      "learning_rate": 5e-06,
+      "loss": 0.6522,
+      "num_input_tokens_seen": 117331168,
+      "step": 1872
+    },
+    {
+      "epoch": 6.229617304492512,
+      "loss": 0.7275434732437134,
+      "loss_ce": 4.387545232020784e-06,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 117331168,
+      "step": 1872
+    },
+    {
+      "epoch": 6.232945091514143,
+      "grad_norm": 14.865208625793457,
+      "learning_rate": 5e-06,
+      "loss": 0.4913,
+      "num_input_tokens_seen": 117394412,
+      "step": 1873
+    },
+    {
+      "epoch": 6.232945091514143,
+      "loss": 0.4301111400127411,
+      "loss_ce": 5.7414017646806315e-05,
+      "loss_iou": 0.11962890625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 117394412,
+      "step": 1873
+    },
+    {
+      "epoch": 6.2362728785357735,
+      "grad_norm": 9.97636890411377,
+      "learning_rate": 5e-06,
+      "loss": 0.6502,
+      "num_input_tokens_seen": 117457176,
+      "step": 1874
+    },
+    {
+      "epoch": 6.2362728785357735,
+      "loss": 0.6446551084518433,
+      "loss_ce": 0.0006731529720127583,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 117457176,
+      "step": 1874
+    },
+    {
+      "epoch": 6.239600665557404,
+      "grad_norm": 8.297738075256348,
+      "learning_rate": 5e-06,
+      "loss": 0.5623,
+      "num_input_tokens_seen": 117518696,
+      "step": 1875
+    },
+    {
+      "epoch": 6.239600665557404,
+      "loss": 0.5319926142692566,
+      "loss_ce": 0.00019328358757775277,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 117518696,
+      "step": 1875
+    },
+    {
+      "epoch": 6.242928452579035,
+      "grad_norm": 7.489507675170898,
+      "learning_rate": 5e-06,
+      "loss": 0.5349,
+      "num_input_tokens_seen": 117581024,
+      "step": 1876
+    },
+    {
+      "epoch": 6.242928452579035,
+      "loss": 0.401066392660141,
+      "loss_ce": 4.387100489111617e-06,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 117581024,
+      "step": 1876
+    },
+    {
+      "epoch": 6.246256239600665,
+      "grad_norm": 19.296171188354492,
+      "learning_rate": 5e-06,
+      "loss": 0.6984,
+      "num_input_tokens_seen": 117646404,
+      "step": 1877
+    },
+    {
+      "epoch": 6.246256239600665,
+      "loss": 0.7436515092849731,
+      "loss_ce": 0.0007315932889468968,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 117646404,
+      "step": 1877
+    },
+    {
+      "epoch": 6.249584026622296,
+      "grad_norm": 29.02984046936035,
+      "learning_rate": 5e-06,
+      "loss": 0.6827,
+      "num_input_tokens_seen": 117709528,
+      "step": 1878
+    },
+    {
+      "epoch": 6.249584026622296,
+      "loss": 0.7689893245697021,
+      "loss_ce": 7.337753686442738e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 117709528,
+      "step": 1878
+    },
+    {
+      "epoch": 6.252911813643927,
+      "grad_norm": 45.192562103271484,
+      "learning_rate": 5e-06,
+      "loss": 0.6448,
+      "num_input_tokens_seen": 117772044,
+      "step": 1879
+    },
+    {
+      "epoch": 6.252911813643927,
+      "loss": 0.6723182201385498,
+      "loss_ce": 1.5984420315362513e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 117772044,
+      "step": 1879
+    },
+    {
+      "epoch": 6.256239600665557,
+      "grad_norm": 29.018625259399414,
+      "learning_rate": 5e-06,
+      "loss": 0.5516,
+      "num_input_tokens_seen": 117834036,
+      "step": 1880
+    },
+    {
+      "epoch": 6.256239600665557,
+      "loss": 0.5316672921180725,
+      "loss_ce": 0.0001731569936964661,
+      "loss_iou": 0.171875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 117834036,
+      "step": 1880
+    },
+    {
+      "epoch": 6.259567387687188,
+      "grad_norm": 23.841171264648438,
+      "learning_rate": 5e-06,
+      "loss": 0.7798,
+      "num_input_tokens_seen": 117895044,
+      "step": 1881
+    },
+    {
+      "epoch": 6.259567387687188,
+      "loss": 0.7895559072494507,
+      "loss_ce": 5.09536675963318e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.7890625,
+      "num_input_tokens_seen": 117895044,
+      "step": 1881
+    },
+    {
+      "epoch": 6.262895174708818,
+      "grad_norm": 28.532699584960938,
+      "learning_rate": 5e-06,
+      "loss": 0.5149,
+      "num_input_tokens_seen": 117956884,
+      "step": 1882
+    },
+    {
+      "epoch": 6.262895174708818,
+      "loss": 0.6902905702590942,
+      "loss_ce": 0.0007154002669267356,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.69140625,
+      "num_input_tokens_seen": 117956884,
+      "step": 1882
+    },
+    {
+      "epoch": 6.266222961730449,
+      "grad_norm": 17.87752342224121,
+      "learning_rate": 5e-06,
+      "loss": 0.7271,
+      "num_input_tokens_seen": 118019364,
+      "step": 1883
+    },
+    {
+      "epoch": 6.266222961730449,
+      "loss": 0.7441762089729309,
+      "loss_ce": 3.5563556593842804e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 118019364,
+      "step": 1883
+    },
+    {
+      "epoch": 6.26955074875208,
+      "grad_norm": 25.544570922851562,
+      "learning_rate": 5e-06,
+      "loss": 0.6711,
+      "num_input_tokens_seen": 118082380,
+      "step": 1884
+    },
+    {
+      "epoch": 6.26955074875208,
+      "loss": 0.7013899087905884,
+      "loss_ce": 0.0003401026770006865,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.69921875,
+      "num_input_tokens_seen": 118082380,
+      "step": 1884
+    },
+    {
+      "epoch": 6.27287853577371,
+      "grad_norm": 21.333023071289062,
+      "learning_rate": 5e-06,
+      "loss": 0.6432,
+      "num_input_tokens_seen": 118145304,
+      "step": 1885
+    },
+    {
+      "epoch": 6.27287853577371,
+      "loss": 0.3440302908420563,
+      "loss_ce": 0.00015822664136067033,
+      "loss_iou": 0.12109375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 118145304,
+      "step": 1885
+    },
+    {
+      "epoch": 6.276206322795341,
+      "grad_norm": 11.410968780517578,
+      "learning_rate": 5e-06,
+      "loss": 0.8493,
+      "num_input_tokens_seen": 118205404,
+      "step": 1886
+    },
+    {
+      "epoch": 6.276206322795341,
+      "loss": 0.6452671885490417,
+      "loss_ce": 3.5297755403007613e-06,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 118205404,
+      "step": 1886
+    },
+    {
+      "epoch": 6.2795341098169715,
+      "grad_norm": 15.16508674621582,
+      "learning_rate": 5e-06,
+      "loss": 0.746,
+      "num_input_tokens_seen": 118268188,
+      "step": 1887
+    },
+    {
+      "epoch": 6.2795341098169715,
+      "loss": 0.8573170900344849,
+      "loss_ce": 0.00013937031326349825,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 118268188,
+      "step": 1887
+    },
+    {
+      "epoch": 6.282861896838602,
+      "grad_norm": 19.193706512451172,
+      "learning_rate": 5e-06,
+      "loss": 0.5648,
+      "num_input_tokens_seen": 118331420,
+      "step": 1888
+    },
+    {
+      "epoch": 6.282861896838602,
+      "loss": 0.5550625324249268,
+      "loss_ce": 0.00037507241358980536,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.5546875,
+      "num_input_tokens_seen": 118331420,
+      "step": 1888
+    },
+    {
+      "epoch": 6.286189683860233,
+      "grad_norm": 10.598845481872559,
+      "learning_rate": 5e-06,
+      "loss": 0.4831,
+      "num_input_tokens_seen": 118394204,
+      "step": 1889
+    },
+    {
+      "epoch": 6.286189683860233,
+      "loss": 0.45692452788352966,
+      "loss_ce": 0.00047309871297329664,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.45703125,
+      "num_input_tokens_seen": 118394204,
+      "step": 1889
+    },
+    {
+      "epoch": 6.289517470881863,
+      "grad_norm": 18.339231491088867,
+      "learning_rate": 5e-06,
+      "loss": 0.3393,
+      "num_input_tokens_seen": 118456836,
+      "step": 1890
+    },
+    {
+      "epoch": 6.289517470881863,
+      "loss": 0.35476088523864746,
+      "loss_ce": 2.4531174858566374e-05,
+      "loss_iou": 0.125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 118456836,
+      "step": 1890
+    },
+    {
+      "epoch": 6.292845257903494,
+      "grad_norm": 42.98812484741211,
+      "learning_rate": 5e-06,
+      "loss": 0.9685,
+      "num_input_tokens_seen": 118520496,
+      "step": 1891
+    },
+    {
+      "epoch": 6.292845257903494,
+      "loss": 0.9763355255126953,
+      "loss_ce": 0.0005359815550036728,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 118520496,
+      "step": 1891
+    },
+    {
+      "epoch": 6.2961730449251245,
+      "grad_norm": 7.72786283493042,
+      "learning_rate": 5e-06,
+      "loss": 0.5755,
+      "num_input_tokens_seen": 118583420,
+      "step": 1892
+    },
+    {
+      "epoch": 6.2961730449251245,
+      "loss": 0.615013599395752,
+      "loss_ce": 2.342194420634769e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 118583420,
+      "step": 1892
+    },
+    {
+      "epoch": 6.299500831946755,
+      "grad_norm": 7.7686448097229,
+      "learning_rate": 5e-06,
+      "loss": 0.5277,
+      "num_input_tokens_seen": 118644200,
+      "step": 1893
+    },
+    {
+      "epoch": 6.299500831946755,
+      "loss": 0.6303321123123169,
+      "loss_ce": 8.307768439408392e-05,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 118644200,
+      "step": 1893
+    },
+    {
+      "epoch": 6.302828618968386,
+      "grad_norm": 14.28267765045166,
+      "learning_rate": 5e-06,
+      "loss": 0.6133,
+      "num_input_tokens_seen": 118706948,
+      "step": 1894
+    },
+    {
+      "epoch": 6.302828618968386,
+      "loss": 0.6571691036224365,
+      "loss_ce": 0.00036982051096856594,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 118706948,
+      "step": 1894
+    },
+    {
+      "epoch": 6.306156405990016,
+      "grad_norm": 14.473723411560059,
+      "learning_rate": 5e-06,
+      "loss": 0.7662,
+      "num_input_tokens_seen": 118770088,
+      "step": 1895
+    },
+    {
+      "epoch": 6.306156405990016,
+      "loss": 0.6408983469009399,
+      "loss_ce": 2.9183007427491248e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 118770088,
+      "step": 1895
+    },
+    {
+      "epoch": 6.309484193011647,
+      "grad_norm": 11.187668800354004,
+      "learning_rate": 5e-06,
+      "loss": 0.8763,
+      "num_input_tokens_seen": 118833268,
+      "step": 1896
+    },
+    {
+      "epoch": 6.309484193011647,
+      "loss": 0.8932377099990845,
+      "loss_ce": 0.00029332979465834796,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.89453125,
+      "num_input_tokens_seen": 118833268,
+      "step": 1896
+    },
+    {
+      "epoch": 6.312811980033278,
+      "grad_norm": 20.483861923217773,
+      "learning_rate": 5e-06,
+      "loss": 0.62,
+      "num_input_tokens_seen": 118895864,
+      "step": 1897
+    },
+    {
+      "epoch": 6.312811980033278,
+      "loss": 0.6993443965911865,
+      "loss_ce": 0.0007360352901741862,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.69921875,
+      "num_input_tokens_seen": 118895864,
+      "step": 1897
+    },
+    {
+      "epoch": 6.316139767054908,
+      "grad_norm": 8.918088912963867,
+      "learning_rate": 5e-06,
+      "loss": 0.5872,
+      "num_input_tokens_seen": 118958404,
+      "step": 1898
+    },
+    {
+      "epoch": 6.316139767054908,
+      "loss": 0.6274863481521606,
+      "loss_ce": 0.0011435841443017125,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 118958404,
+      "step": 1898
+    },
+    {
+      "epoch": 6.319467554076539,
+      "grad_norm": 13.76313591003418,
+      "learning_rate": 5e-06,
+      "loss": 0.7277,
+      "num_input_tokens_seen": 119021880,
+      "step": 1899
+    },
+    {
+      "epoch": 6.319467554076539,
+      "loss": 0.6522097587585449,
+      "loss_ce": 0.0009646408725529909,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 119021880,
+      "step": 1899
+    },
+    {
+      "epoch": 6.322795341098169,
+      "grad_norm": 8.077730178833008,
+      "learning_rate": 5e-06,
+      "loss": 0.5365,
+      "num_input_tokens_seen": 119085200,
+      "step": 1900
+    },
+    {
+      "epoch": 6.322795341098169,
+      "loss": 0.3900187015533447,
+      "loss_ce": 4.0382278712058906e-06,
+      "loss_iou": 0.11328125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 119085200,
+      "step": 1900
+    },
+    {
+      "epoch": 6.3261231281198,
+      "grad_norm": 20.717100143432617,
+      "learning_rate": 5e-06,
+      "loss": 0.662,
+      "num_input_tokens_seen": 119149340,
+      "step": 1901
+    },
+    {
+      "epoch": 6.3261231281198,
+      "loss": 0.5592089891433716,
+      "loss_ce": 0.0011035435600206256,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 119149340,
+      "step": 1901
+    },
+    {
+      "epoch": 6.329450915141431,
+      "grad_norm": 11.843432426452637,
+      "learning_rate": 5e-06,
+      "loss": 0.5203,
+      "num_input_tokens_seen": 119209956,
+      "step": 1902
+    },
+    {
+      "epoch": 6.329450915141431,
+      "loss": 0.6044921875,
+      "loss_ce": 0.0002441465330775827,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 119209956,
+      "step": 1902
+    },
+    {
+      "epoch": 6.332778702163061,
+      "grad_norm": 11.223464012145996,
+      "learning_rate": 5e-06,
+      "loss": 0.6212,
+      "num_input_tokens_seen": 119269532,
+      "step": 1903
+    },
+    {
+      "epoch": 6.332778702163061,
+      "loss": 0.6058506369590759,
+      "loss_ce": 1.5660873032175004e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 119269532,
+      "step": 1903
+    },
+    {
+      "epoch": 6.336106489184692,
+      "grad_norm": 27.761043548583984,
+      "learning_rate": 5e-06,
+      "loss": 0.8632,
+      "num_input_tokens_seen": 119333156,
+      "step": 1904
+    },
+    {
+      "epoch": 6.336106489184692,
+      "loss": 0.6953175067901611,
+      "loss_ce": 5.025212885811925e-06,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 119333156,
+      "step": 1904
+    },
+    {
+      "epoch": 6.3394342762063225,
+      "grad_norm": 22.749303817749023,
+      "learning_rate": 5e-06,
+      "loss": 0.5647,
+      "num_input_tokens_seen": 119394532,
+      "step": 1905
+    },
+    {
+      "epoch": 6.3394342762063225,
+      "loss": 0.5662988424301147,
+      "loss_ce": 0.001113282167352736,
+      "loss_iou": 0.203125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 119394532,
+      "step": 1905
+    },
+    {
+      "epoch": 6.342762063227953,
+      "grad_norm": 11.654192924499512,
+      "learning_rate": 5e-06,
+      "loss": 0.6636,
+      "num_input_tokens_seen": 119457736,
+      "step": 1906
+    },
+    {
+      "epoch": 6.342762063227953,
+      "loss": 0.8046139478683472,
+      "loss_ce": 0.0006588406395167112,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 119457736,
+      "step": 1906
+    },
+    {
+      "epoch": 6.346089850249584,
+      "grad_norm": 25.141996383666992,
+      "learning_rate": 5e-06,
+      "loss": 0.6719,
+      "num_input_tokens_seen": 119521436,
+      "step": 1907
+    },
+    {
+      "epoch": 6.346089850249584,
+      "loss": 0.4867073893547058,
+      "loss_ce": 1.305484965996584e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.486328125,
+      "num_input_tokens_seen": 119521436,
+      "step": 1907
+    },
+    {
+      "epoch": 6.349417637271214,
+      "grad_norm": 8.515308380126953,
+      "learning_rate": 5e-06,
+      "loss": 0.5758,
+      "num_input_tokens_seen": 119583924,
+      "step": 1908
+    },
+    {
+      "epoch": 6.349417637271214,
+      "loss": 0.5766646862030029,
+      "loss_ce": 4.5053284338791855e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 119583924,
+      "step": 1908
+    },
+    {
+      "epoch": 6.352745424292845,
+      "grad_norm": 25.861650466918945,
+      "learning_rate": 5e-06,
+      "loss": 0.5609,
+      "num_input_tokens_seen": 119644324,
+      "step": 1909
+    },
+    {
+      "epoch": 6.352745424292845,
+      "loss": 0.5296666026115417,
+      "loss_ce": 3.480890427454142e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 119644324,
+      "step": 1909
+    },
+    {
+      "epoch": 6.356073211314476,
+      "grad_norm": 17.014137268066406,
+      "learning_rate": 5e-06,
+      "loss": 0.4881,
+      "num_input_tokens_seen": 119707332,
+      "step": 1910
+    },
+    {
+      "epoch": 6.356073211314476,
+      "loss": 0.32322195172309875,
+      "loss_ce": 0.00010183690028497949,
+      "loss_iou": 0.1259765625,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 119707332,
+      "step": 1910
+    },
+    {
+      "epoch": 6.359400998336106,
+      "grad_norm": 42.42847442626953,
+      "learning_rate": 5e-06,
+      "loss": 0.6808,
+      "num_input_tokens_seen": 119770608,
+      "step": 1911
+    },
+    {
+      "epoch": 6.359400998336106,
+      "loss": 0.4963526129722595,
+      "loss_ce": 1.4706164620292839e-05,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.49609375,
+      "num_input_tokens_seen": 119770608,
+      "step": 1911
+    },
+    {
+      "epoch": 6.362728785357737,
+      "grad_norm": 90.03009033203125,
+      "learning_rate": 5e-06,
+      "loss": 0.6118,
+      "num_input_tokens_seen": 119832564,
+      "step": 1912
+    },
+    {
+      "epoch": 6.362728785357737,
+      "loss": 0.5688121914863586,
+      "loss_ce": 1.0294821549905464e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 119832564,
+      "step": 1912
+    },
+    {
+      "epoch": 6.366056572379367,
+      "grad_norm": 21.129972457885742,
+      "learning_rate": 5e-06,
+      "loss": 0.7614,
+      "num_input_tokens_seen": 119895428,
+      "step": 1913
+    },
+    {
+      "epoch": 6.366056572379367,
+      "loss": 0.6494866609573364,
+      "loss_ce": 0.0001946888369275257,
+      "loss_iou": 0.203125,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 119895428,
+      "step": 1913
+    },
+    {
+      "epoch": 6.369384359400998,
+      "grad_norm": 23.620243072509766,
+      "learning_rate": 5e-06,
+      "loss": 0.8151,
+      "num_input_tokens_seen": 119958696,
+      "step": 1914
+    },
+    {
+      "epoch": 6.369384359400998,
+      "loss": 0.8362605571746826,
+      "loss_ce": 7.890413689892739e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 119958696,
+      "step": 1914
+    },
+    {
+      "epoch": 6.372712146422629,
+      "grad_norm": 16.94724464416504,
+      "learning_rate": 5e-06,
+      "loss": 0.552,
+      "num_input_tokens_seen": 120021096,
+      "step": 1915
+    },
+    {
+      "epoch": 6.372712146422629,
+      "loss": 0.6032967567443848,
+      "loss_ce": 0.0002694588038139045,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 120021096,
+      "step": 1915
+    },
+    {
+      "epoch": 6.376039933444259,
+      "grad_norm": 15.850129127502441,
+      "learning_rate": 5e-06,
+      "loss": 0.5145,
+      "num_input_tokens_seen": 120081864,
+      "step": 1916
+    },
+    {
+      "epoch": 6.376039933444259,
+      "loss": 0.4399433135986328,
+      "loss_ce": 1.9184694792784285e-06,
+      "loss_iou": 0.140625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.439453125,
+      "num_input_tokens_seen": 120081864,
+      "step": 1916
+    },
+    {
+      "epoch": 6.37936772046589,
+      "grad_norm": 10.974051475524902,
+      "learning_rate": 5e-06,
+      "loss": 0.7344,
+      "num_input_tokens_seen": 120145864,
+      "step": 1917
+    },
+    {
+      "epoch": 6.37936772046589,
+      "loss": 0.6459264755249023,
+      "loss_ce": 0.0019140413496643305,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 120145864,
+      "step": 1917
+    },
+    {
+      "epoch": 6.3826955074875205,
+      "grad_norm": 22.09915542602539,
+      "learning_rate": 5e-06,
+      "loss": 0.5428,
+      "num_input_tokens_seen": 120208220,
+      "step": 1918
+    },
+    {
+      "epoch": 6.3826955074875205,
+      "loss": 0.6734851598739624,
+      "loss_ce": 0.0007557276985608041,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 120208220,
+      "step": 1918
+    },
+    {
+      "epoch": 6.386023294509151,
+      "grad_norm": 17.634904861450195,
+      "learning_rate": 5e-06,
+      "loss": 0.7517,
+      "num_input_tokens_seen": 120269832,
+      "step": 1919
+    },
+    {
+      "epoch": 6.386023294509151,
+      "loss": 0.6596292853355408,
+      "loss_ce": 8.335949678439647e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 120269832,
+      "step": 1919
+    },
+    {
+      "epoch": 6.389351081530782,
+      "grad_norm": 16.73068618774414,
+      "learning_rate": 5e-06,
+      "loss": 0.4661,
+      "num_input_tokens_seen": 120329824,
+      "step": 1920
+    },
+    {
+      "epoch": 6.389351081530782,
+      "loss": 0.5087909698486328,
+      "loss_ce": 1.963873046406661e-06,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.5078125,
+      "num_input_tokens_seen": 120329824,
+      "step": 1920
+    },
+    {
+      "epoch": 6.392678868552412,
+      "grad_norm": 26.34819984436035,
+      "learning_rate": 5e-06,
+      "loss": 0.6769,
+      "num_input_tokens_seen": 120392212,
+      "step": 1921
+    },
+    {
+      "epoch": 6.392678868552412,
+      "loss": 0.42969781160354614,
+      "loss_ce": 0.000803784467279911,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 120392212,
+      "step": 1921
+    },
+    {
+      "epoch": 6.396006655574043,
+      "grad_norm": 6.768531799316406,
+      "learning_rate": 5e-06,
+      "loss": 0.6683,
+      "num_input_tokens_seen": 120454584,
+      "step": 1922
+    },
+    {
+      "epoch": 6.396006655574043,
+      "loss": 0.9003890752792358,
+      "loss_ce": 0.00024259740894194692,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 120454584,
+      "step": 1922
+    },
+    {
+      "epoch": 6.3993344425956735,
+      "grad_norm": 8.30113697052002,
+      "learning_rate": 5e-06,
+      "loss": 0.6405,
+      "num_input_tokens_seen": 120514044,
+      "step": 1923
+    },
+    {
+      "epoch": 6.3993344425956735,
+      "loss": 0.7188661098480225,
+      "loss_ce": 0.00011609155626501888,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 120514044,
+      "step": 1923
+    },
+    {
+      "epoch": 6.402662229617304,
+      "grad_norm": 15.520536422729492,
+      "learning_rate": 5e-06,
+      "loss": 0.7244,
+      "num_input_tokens_seen": 120576424,
+      "step": 1924
+    },
+    {
+      "epoch": 6.402662229617304,
+      "loss": 0.5446481108665466,
+      "loss_ce": 0.00033657567109912634,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 120576424,
+      "step": 1924
+    },
+    {
+      "epoch": 6.405990016638935,
+      "grad_norm": 12.706613540649414,
+      "learning_rate": 5e-06,
+      "loss": 0.5809,
+      "num_input_tokens_seen": 120638852,
+      "step": 1925
+    },
+    {
+      "epoch": 6.405990016638935,
+      "loss": 0.6418596506118774,
+      "loss_ce": 0.0006242538802325726,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 120638852,
+      "step": 1925
+    },
+    {
+      "epoch": 6.409317803660565,
+      "grad_norm": 12.72490119934082,
+      "learning_rate": 5e-06,
+      "loss": 0.5454,
+      "num_input_tokens_seen": 120701328,
+      "step": 1926
+    },
+    {
+      "epoch": 6.409317803660565,
+      "loss": 0.5837253332138062,
+      "loss_ce": 0.0001071855440386571,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 120701328,
+      "step": 1926
+    },
+    {
+      "epoch": 6.412645590682196,
+      "grad_norm": 15.018410682678223,
+      "learning_rate": 5e-06,
+      "loss": 0.5549,
+      "num_input_tokens_seen": 120763848,
+      "step": 1927
+    },
+    {
+      "epoch": 6.412645590682196,
+      "loss": 0.5490936040878296,
+      "loss_ce": 0.0002655147691257298,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 120763848,
+      "step": 1927
+    },
+    {
+      "epoch": 6.415973377703827,
+      "grad_norm": 14.566852569580078,
+      "learning_rate": 5e-06,
+      "loss": 0.8448,
+      "num_input_tokens_seen": 120826228,
+      "step": 1928
+    },
+    {
+      "epoch": 6.415973377703827,
+      "loss": 1.2861429452896118,
+      "loss_ce": 1.0124242180609144e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.07861328125,
+      "loss_xval": 1.2890625,
+      "num_input_tokens_seen": 120826228,
+      "step": 1928
+    },
+    {
+      "epoch": 6.419301164725457,
+      "grad_norm": 11.734906196594238,
+      "learning_rate": 5e-06,
+      "loss": 0.584,
+      "num_input_tokens_seen": 120887768,
+      "step": 1929
+    },
+    {
+      "epoch": 6.419301164725457,
+      "loss": 0.5992169976234436,
+      "loss_ce": 0.0011334889568388462,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 120887768,
+      "step": 1929
+    },
+    {
+      "epoch": 6.422628951747088,
+      "grad_norm": 15.214763641357422,
+      "learning_rate": 5e-06,
+      "loss": 0.6457,
+      "num_input_tokens_seen": 120950624,
+      "step": 1930
+    },
+    {
+      "epoch": 6.422628951747088,
+      "loss": 0.8110338449478149,
+      "loss_ce": 0.00018182062194682658,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.8125,
+      "num_input_tokens_seen": 120950624,
+      "step": 1930
+    },
+    {
+      "epoch": 6.425956738768718,
+      "grad_norm": 28.60011863708496,
+      "learning_rate": 5e-06,
+      "loss": 0.6793,
+      "num_input_tokens_seen": 121013536,
+      "step": 1931
+    },
+    {
+      "epoch": 6.425956738768718,
+      "loss": 0.49274346232414246,
+      "loss_ce": 0.0004949223366566002,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 121013536,
+      "step": 1931
+    },
+    {
+      "epoch": 6.429284525790349,
+      "grad_norm": 12.213953971862793,
+      "learning_rate": 5e-06,
+      "loss": 0.7479,
+      "num_input_tokens_seen": 121075756,
+      "step": 1932
+    },
+    {
+      "epoch": 6.429284525790349,
+      "loss": 0.773937463760376,
+      "loss_ce": 1.1653934961941559e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 121075756,
+      "step": 1932
+    },
+    {
+      "epoch": 6.43261231281198,
+      "grad_norm": 15.571845054626465,
+      "learning_rate": 5e-06,
+      "loss": 0.6698,
+      "num_input_tokens_seen": 121139652,
+      "step": 1933
+    },
+    {
+      "epoch": 6.43261231281198,
+      "loss": 0.6814624071121216,
+      "loss_ce": 0.0027514593675732613,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 121139652,
+      "step": 1933
+    },
+    {
+      "epoch": 6.43594009983361,
+      "grad_norm": 13.807389259338379,
+      "learning_rate": 5e-06,
+      "loss": 0.9253,
+      "num_input_tokens_seen": 121203024,
+      "step": 1934
+    },
+    {
+      "epoch": 6.43594009983361,
+      "loss": 1.345597505569458,
+      "loss_ce": 0.0008709065150469542,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.08251953125,
+      "loss_xval": 1.34375,
+      "num_input_tokens_seen": 121203024,
+      "step": 1934
+    },
+    {
+      "epoch": 6.439267886855241,
+      "grad_norm": 12.07575511932373,
+      "learning_rate": 5e-06,
+      "loss": 0.5334,
+      "num_input_tokens_seen": 121266420,
+      "step": 1935
+    },
+    {
+      "epoch": 6.439267886855241,
+      "loss": 0.5801987648010254,
+      "loss_ce": 0.00012065586633980274,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 121266420,
+      "step": 1935
+    },
+    {
+      "epoch": 6.4425956738768715,
+      "grad_norm": 12.738320350646973,
+      "learning_rate": 5e-06,
+      "loss": 0.7129,
+      "num_input_tokens_seen": 121330352,
+      "step": 1936
+    },
+    {
+      "epoch": 6.4425956738768715,
+      "loss": 0.656618595123291,
+      "loss_ce": 0.00012446939945220947,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 121330352,
+      "step": 1936
+    },
+    {
+      "epoch": 6.445923460898502,
+      "grad_norm": 14.938615798950195,
+      "learning_rate": 5e-06,
+      "loss": 0.6309,
+      "num_input_tokens_seen": 121394128,
+      "step": 1937
+    },
+    {
+      "epoch": 6.445923460898502,
+      "loss": 0.8611018657684326,
+      "loss_ce": 0.0016048324760049582,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 121394128,
+      "step": 1937
+    },
+    {
+      "epoch": 6.449251247920133,
+      "grad_norm": 21.565048217773438,
+      "learning_rate": 5e-06,
+      "loss": 0.5475,
+      "num_input_tokens_seen": 121457512,
+      "step": 1938
+    },
+    {
+      "epoch": 6.449251247920133,
+      "loss": 0.39604151248931885,
+      "loss_ce": 4.542524766293354e-05,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.396484375,
+      "num_input_tokens_seen": 121457512,
+      "step": 1938
+    },
+    {
+      "epoch": 6.452579034941763,
+      "grad_norm": 35.06698989868164,
+      "learning_rate": 5e-06,
+      "loss": 0.8538,
+      "num_input_tokens_seen": 121520800,
+      "step": 1939
+    },
+    {
+      "epoch": 6.452579034941763,
+      "loss": 0.8477818965911865,
+      "loss_ce": 3.5593729990068823e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.84765625,
+      "num_input_tokens_seen": 121520800,
+      "step": 1939
+    },
+    {
+      "epoch": 6.455906821963394,
+      "grad_norm": 45.28481674194336,
+      "learning_rate": 5e-06,
+      "loss": 0.97,
+      "num_input_tokens_seen": 121585016,
+      "step": 1940
+    },
+    {
+      "epoch": 6.455906821963394,
+      "loss": 0.9006417989730835,
+      "loss_ce": 7.0053438321338035e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.90234375,
+      "num_input_tokens_seen": 121585016,
+      "step": 1940
+    },
+    {
+      "epoch": 6.4592346089850246,
+      "grad_norm": 39.1356086730957,
+      "learning_rate": 5e-06,
+      "loss": 0.6982,
+      "num_input_tokens_seen": 121647252,
+      "step": 1941
+    },
+    {
+      "epoch": 6.4592346089850246,
+      "loss": 0.6142480969429016,
+      "loss_ce": 0.0001123529946198687,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 121647252,
+      "step": 1941
+    },
+    {
+      "epoch": 6.462562396006655,
+      "grad_norm": 37.34075164794922,
+      "learning_rate": 5e-06,
+      "loss": 0.6763,
+      "num_input_tokens_seen": 121710464,
+      "step": 1942
+    },
+    {
+      "epoch": 6.462562396006655,
+      "loss": 0.7514750957489014,
+      "loss_ce": 1.022259584715357e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 121710464,
+      "step": 1942
+    },
+    {
+      "epoch": 6.465890183028286,
+      "grad_norm": 28.683731079101562,
+      "learning_rate": 5e-06,
+      "loss": 0.5492,
+      "num_input_tokens_seen": 121771684,
+      "step": 1943
+    },
+    {
+      "epoch": 6.465890183028286,
+      "loss": 0.34131908416748047,
+      "loss_ce": 0.0007429145043715835,
+      "loss_iou": 0.080078125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 121771684,
+      "step": 1943
+    },
+    {
+      "epoch": 6.469217970049916,
+      "grad_norm": 16.91063690185547,
+      "learning_rate": 5e-06,
+      "loss": 0.6058,
+      "num_input_tokens_seen": 121836040,
+      "step": 1944
+    },
+    {
+      "epoch": 6.469217970049916,
+      "loss": 0.6735157370567322,
+      "loss_ce": 0.0002979549753945321,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 121836040,
+      "step": 1944
+    },
+    {
+      "epoch": 6.472545757071547,
+      "grad_norm": 20.4154109954834,
+      "learning_rate": 5e-06,
+      "loss": 0.6285,
+      "num_input_tokens_seen": 121895316,
+      "step": 1945
+    },
+    {
+      "epoch": 6.472545757071547,
+      "loss": 0.8378450274467468,
+      "loss_ce": 7.650956104043871e-05,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 121895316,
+      "step": 1945
+    },
+    {
+      "epoch": 6.475873544093178,
+      "grad_norm": 42.6540412902832,
+      "learning_rate": 5e-06,
+      "loss": 0.7617,
+      "num_input_tokens_seen": 121958684,
+      "step": 1946
+    },
+    {
+      "epoch": 6.475873544093178,
+      "loss": 1.069746732711792,
+      "loss_ce": 0.0015093846013769507,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.05859375,
+      "loss_xval": 1.0703125,
+      "num_input_tokens_seen": 121958684,
+      "step": 1946
+    },
+    {
+      "epoch": 6.479201331114808,
+      "grad_norm": 15.946682929992676,
+      "learning_rate": 5e-06,
+      "loss": 0.7402,
+      "num_input_tokens_seen": 122020744,
+      "step": 1947
+    },
+    {
+      "epoch": 6.479201331114808,
+      "loss": 0.6518175601959229,
+      "loss_ce": 0.00026731760590337217,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 122020744,
+      "step": 1947
+    },
+    {
+      "epoch": 6.482529118136439,
+      "grad_norm": 10.553583145141602,
+      "learning_rate": 5e-06,
+      "loss": 0.5386,
+      "num_input_tokens_seen": 122083412,
+      "step": 1948
+    },
+    {
+      "epoch": 6.482529118136439,
+      "loss": 0.7277462482452393,
+      "loss_ce": 0.00020720537577290088,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 122083412,
+      "step": 1948
+    },
+    {
+      "epoch": 6.4858569051580695,
+      "grad_norm": 11.314140319824219,
+      "learning_rate": 5e-06,
+      "loss": 0.7428,
+      "num_input_tokens_seen": 122145288,
+      "step": 1949
+    },
+    {
+      "epoch": 6.4858569051580695,
+      "loss": 0.7091948986053467,
+      "loss_ce": 8.838271605782211e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 122145288,
+      "step": 1949
+    },
+    {
+      "epoch": 6.4891846921797,
+      "grad_norm": 67.86624145507812,
+      "learning_rate": 5e-06,
+      "loss": 0.7234,
+      "num_input_tokens_seen": 122208008,
+      "step": 1950
+    },
+    {
+      "epoch": 6.4891846921797,
+      "loss": 0.6662644147872925,
+      "loss_ce": 4.646051820600405e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 122208008,
+      "step": 1950
+    },
+    {
+      "epoch": 6.492512479201331,
+      "grad_norm": 15.304253578186035,
+      "learning_rate": 5e-06,
+      "loss": 0.5958,
+      "num_input_tokens_seen": 122268396,
+      "step": 1951
+    },
+    {
+      "epoch": 6.492512479201331,
+      "loss": 0.680047869682312,
+      "loss_ce": 0.00011620179429883137,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.6796875,
+      "num_input_tokens_seen": 122268396,
+      "step": 1951
+    },
+    {
+      "epoch": 6.495840266222961,
+      "grad_norm": 13.426730155944824,
+      "learning_rate": 5e-06,
+      "loss": 0.6756,
+      "num_input_tokens_seen": 122331456,
+      "step": 1952
+    },
+    {
+      "epoch": 6.495840266222961,
+      "loss": 0.5398004055023193,
+      "loss_ce": 5.481046173372306e-06,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 122331456,
+      "step": 1952
+    },
+    {
+      "epoch": 6.499168053244592,
+      "grad_norm": 6.796942710876465,
+      "learning_rate": 5e-06,
+      "loss": 0.5479,
+      "num_input_tokens_seen": 122395172,
+      "step": 1953
+    },
+    {
+      "epoch": 6.499168053244592,
+      "loss": 0.40055206418037415,
+      "loss_ce": 0.0006497133290395141,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 122395172,
+      "step": 1953
+    },
+    {
+      "epoch": 6.5024958402662225,
+      "grad_norm": 8.822606086730957,
+      "learning_rate": 5e-06,
+      "loss": 0.6339,
+      "num_input_tokens_seen": 122456692,
+      "step": 1954
+    },
+    {
+      "epoch": 6.5024958402662225,
+      "loss": 0.6271539330482483,
+      "loss_ce": 0.0003228959976695478,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 122456692,
+      "step": 1954
+    },
+    {
+      "epoch": 6.505823627287853,
+      "grad_norm": 14.753844261169434,
+      "learning_rate": 5e-06,
+      "loss": 0.7122,
+      "num_input_tokens_seen": 122517856,
+      "step": 1955
+    },
+    {
+      "epoch": 6.505823627287853,
+      "loss": 0.7032543420791626,
+      "loss_ce": 0.00012933007383253425,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.080078125,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 122517856,
+      "step": 1955
+    },
+    {
+      "epoch": 6.509151414309484,
+      "grad_norm": 15.91337776184082,
+      "learning_rate": 5e-06,
+      "loss": 0.5987,
+      "num_input_tokens_seen": 122581004,
+      "step": 1956
+    },
+    {
+      "epoch": 6.509151414309484,
+      "loss": 0.6297885179519653,
+      "loss_ce": 0.00014987353642936796,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 122581004,
+      "step": 1956
+    },
+    {
+      "epoch": 6.512479201331114,
+      "grad_norm": 9.528816223144531,
+      "learning_rate": 5e-06,
+      "loss": 0.6616,
+      "num_input_tokens_seen": 122644628,
+      "step": 1957
+    },
+    {
+      "epoch": 6.512479201331114,
+      "loss": 0.6756832003593445,
+      "loss_ce": 0.00039023166755214334,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 122644628,
+      "step": 1957
+    },
+    {
+      "epoch": 6.515806988352745,
+      "grad_norm": 27.174596786499023,
+      "learning_rate": 5e-06,
+      "loss": 0.6126,
+      "num_input_tokens_seen": 122708080,
+      "step": 1958
+    },
+    {
+      "epoch": 6.515806988352745,
+      "loss": 0.6474884152412415,
+      "loss_ce": 0.00027161132311448455,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 122708080,
+      "step": 1958
+    },
+    {
+      "epoch": 6.519134775374376,
+      "grad_norm": 19.91985321044922,
+      "learning_rate": 5e-06,
+      "loss": 0.5484,
+      "num_input_tokens_seen": 122770044,
+      "step": 1959
+    },
+    {
+      "epoch": 6.519134775374376,
+      "loss": 0.693800687789917,
+      "loss_ce": 1.4089351680013351e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 122770044,
+      "step": 1959
+    },
+    {
+      "epoch": 6.522462562396006,
+      "grad_norm": 8.022729873657227,
+      "learning_rate": 5e-06,
+      "loss": 0.576,
+      "num_input_tokens_seen": 122833492,
+      "step": 1960
+    },
+    {
+      "epoch": 6.522462562396006,
+      "loss": 0.7717412114143372,
+      "loss_ce": 0.00013476284220814705,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 122833492,
+      "step": 1960
+    },
+    {
+      "epoch": 6.525790349417637,
+      "grad_norm": 18.420011520385742,
+      "learning_rate": 5e-06,
+      "loss": 0.5904,
+      "num_input_tokens_seen": 122896032,
+      "step": 1961
+    },
+    {
+      "epoch": 6.525790349417637,
+      "loss": 0.7451998591423035,
+      "loss_ce": 2.1647940229740925e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.74609375,
+      "num_input_tokens_seen": 122896032,
+      "step": 1961
+    },
+    {
+      "epoch": 6.529118136439267,
+      "grad_norm": 14.324840545654297,
+      "learning_rate": 5e-06,
+      "loss": 0.6027,
+      "num_input_tokens_seen": 122957832,
+      "step": 1962
+    },
+    {
+      "epoch": 6.529118136439267,
+      "loss": 0.6533394455909729,
+      "loss_ce": 0.0006295054336078465,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 122957832,
+      "step": 1962
+    },
+    {
+      "epoch": 6.532445923460898,
+      "grad_norm": 11.380367279052734,
+      "learning_rate": 5e-06,
+      "loss": 0.6411,
+      "num_input_tokens_seen": 123019992,
+      "step": 1963
+    },
+    {
+      "epoch": 6.532445923460898,
+      "loss": 0.6608841419219971,
+      "loss_ce": 0.00048379399231635034,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 123019992,
+      "step": 1963
+    },
+    {
+      "epoch": 6.535773710482529,
+      "grad_norm": 21.338104248046875,
+      "learning_rate": 5e-06,
+      "loss": 0.6764,
+      "num_input_tokens_seen": 123083924,
+      "step": 1964
+    },
+    {
+      "epoch": 6.535773710482529,
+      "loss": 0.6724074482917786,
+      "loss_ce": 4.4186064769746736e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 123083924,
+      "step": 1964
+    },
+    {
+      "epoch": 6.539101497504159,
+      "grad_norm": 22.636884689331055,
+      "learning_rate": 5e-06,
+      "loss": 0.6027,
+      "num_input_tokens_seen": 123144872,
+      "step": 1965
+    },
+    {
+      "epoch": 6.539101497504159,
+      "loss": 0.6550688147544861,
+      "loss_ce": 0.0002073537471005693,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 123144872,
+      "step": 1965
+    },
+    {
+      "epoch": 6.54242928452579,
+      "grad_norm": 56.49250411987305,
+      "learning_rate": 5e-06,
+      "loss": 0.5544,
+      "num_input_tokens_seen": 123206232,
+      "step": 1966
+    },
+    {
+      "epoch": 6.54242928452579,
+      "loss": 0.5780704617500305,
+      "loss_ce": 6.4996402215911075e-06,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 123206232,
+      "step": 1966
+    },
+    {
+      "epoch": 6.5457570715474205,
+      "grad_norm": 10.766104698181152,
+      "learning_rate": 5e-06,
+      "loss": 0.4923,
+      "num_input_tokens_seen": 123268996,
+      "step": 1967
+    },
+    {
+      "epoch": 6.5457570715474205,
+      "loss": 0.5199794769287109,
+      "loss_ce": 0.00032618455588817596,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 123268996,
+      "step": 1967
+    },
+    {
+      "epoch": 6.549084858569051,
+      "grad_norm": 11.645246505737305,
+      "learning_rate": 5e-06,
+      "loss": 0.653,
+      "num_input_tokens_seen": 123334080,
+      "step": 1968
+    },
+    {
+      "epoch": 6.549084858569051,
+      "loss": 0.7981342673301697,
+      "loss_ce": 0.0012592482380568981,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 123334080,
+      "step": 1968
+    },
+    {
+      "epoch": 6.552412645590682,
+      "grad_norm": 8.64057731628418,
+      "learning_rate": 5e-06,
+      "loss": 0.3998,
+      "num_input_tokens_seen": 123396060,
+      "step": 1969
+    },
+    {
+      "epoch": 6.552412645590682,
+      "loss": 0.4468684196472168,
+      "loss_ce": 9.108871017815545e-05,
+      "loss_iou": 0.1064453125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.447265625,
+      "num_input_tokens_seen": 123396060,
+      "step": 1969
+    },
+    {
+      "epoch": 6.555740432612312,
+      "grad_norm": 24.30341339111328,
+      "learning_rate": 5e-06,
+      "loss": 0.7164,
+      "num_input_tokens_seen": 123460000,
+      "step": 1970
+    },
+    {
+      "epoch": 6.555740432612312,
+      "loss": 0.6704313158988953,
+      "loss_ce": 0.0004484155506361276,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 123460000,
+      "step": 1970
+    },
+    {
+      "epoch": 6.559068219633943,
+      "grad_norm": 39.814449310302734,
+      "learning_rate": 5e-06,
+      "loss": 0.9853,
+      "num_input_tokens_seen": 123523824,
+      "step": 1971
+    },
+    {
+      "epoch": 6.559068219633943,
+      "loss": 1.0057671070098877,
+      "loss_ce": 0.0018607999663800001,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.054443359375,
+      "loss_xval": 1.0,
+      "num_input_tokens_seen": 123523824,
+      "step": 1971
+    },
+    {
+      "epoch": 6.5623960066555735,
+      "grad_norm": 27.571672439575195,
+      "learning_rate": 5e-06,
+      "loss": 0.7632,
+      "num_input_tokens_seen": 123587972,
+      "step": 1972
+    },
+    {
+      "epoch": 6.5623960066555735,
+      "loss": 0.6259521245956421,
+      "loss_ce": 3.6633864510804415e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 123587972,
+      "step": 1972
+    },
+    {
+      "epoch": 6.565723793677205,
+      "grad_norm": 34.187774658203125,
+      "learning_rate": 5e-06,
+      "loss": 0.7046,
+      "num_input_tokens_seen": 123650356,
+      "step": 1973
+    },
+    {
+      "epoch": 6.565723793677205,
+      "loss": 0.6025407314300537,
+      "loss_ce": 1.6855144622240914e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 123650356,
+      "step": 1973
+    },
+    {
+      "epoch": 6.569051580698836,
+      "grad_norm": 20.649272918701172,
+      "learning_rate": 5e-06,
+      "loss": 0.6901,
+      "num_input_tokens_seen": 123713788,
+      "step": 1974
+    },
+    {
+      "epoch": 6.569051580698836,
+      "loss": 0.7274590730667114,
+      "loss_ce": 0.0004083071544300765,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.7265625,
+      "num_input_tokens_seen": 123713788,
+      "step": 1974
+    },
+    {
+      "epoch": 6.572379367720466,
+      "grad_norm": 13.874034881591797,
+      "learning_rate": 5e-06,
+      "loss": 0.6211,
+      "num_input_tokens_seen": 123777400,
+      "step": 1975
+    },
+    {
+      "epoch": 6.572379367720466,
+      "loss": 0.5335060358047485,
+      "loss_ce": 0.000791211670730263,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 123777400,
+      "step": 1975
+    },
+    {
+      "epoch": 6.575707154742097,
+      "grad_norm": 9.80647087097168,
+      "learning_rate": 5e-06,
+      "loss": 0.6299,
+      "num_input_tokens_seen": 123840460,
+      "step": 1976
+    },
+    {
+      "epoch": 6.575707154742097,
+      "loss": 0.4524020254611969,
+      "loss_ce": 9.460385626880452e-06,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 123840460,
+      "step": 1976
+    },
+    {
+      "epoch": 6.5790349417637275,
+      "grad_norm": 29.99645233154297,
+      "learning_rate": 5e-06,
+      "loss": 0.6651,
+      "num_input_tokens_seen": 123901060,
+      "step": 1977
+    },
+    {
+      "epoch": 6.5790349417637275,
+      "loss": 0.8397188186645508,
+      "loss_ce": 0.0008516378584317863,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.83984375,
+      "num_input_tokens_seen": 123901060,
+      "step": 1977
+    },
+    {
+      "epoch": 6.582362728785358,
+      "grad_norm": 48.624412536621094,
+      "learning_rate": 5e-06,
+      "loss": 0.5813,
+      "num_input_tokens_seen": 123962368,
+      "step": 1978
+    },
+    {
+      "epoch": 6.582362728785358,
+      "loss": 0.6665137410163879,
+      "loss_ce": 7.087117410264909e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.66796875,
+      "num_input_tokens_seen": 123962368,
+      "step": 1978
+    },
+    {
+      "epoch": 6.585690515806989,
+      "grad_norm": 15.941285133361816,
+      "learning_rate": 5e-06,
+      "loss": 0.4458,
+      "num_input_tokens_seen": 124024356,
+      "step": 1979
+    },
+    {
+      "epoch": 6.585690515806989,
+      "loss": 0.5257315635681152,
+      "loss_ce": 0.0004630337643902749,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 124024356,
+      "step": 1979
+    },
+    {
+      "epoch": 6.589018302828619,
+      "grad_norm": 11.377344131469727,
+      "learning_rate": 5e-06,
+      "loss": 0.7924,
+      "num_input_tokens_seen": 124087092,
+      "step": 1980
+    },
+    {
+      "epoch": 6.589018302828619,
+      "loss": 0.7506057024002075,
+      "loss_ce": 0.0005446606664918363,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.75,
+      "num_input_tokens_seen": 124087092,
+      "step": 1980
+    },
+    {
+      "epoch": 6.59234608985025,
+      "grad_norm": 22.64354133605957,
+      "learning_rate": 5e-06,
+      "loss": 0.6216,
+      "num_input_tokens_seen": 124149004,
+      "step": 1981
+    },
+    {
+      "epoch": 6.59234608985025,
+      "loss": 0.33752793073654175,
+      "loss_ce": 3.5264533835288603e-06,
+      "loss_iou": 0.11279296875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 124149004,
+      "step": 1981
+    },
+    {
+      "epoch": 6.595673876871881,
+      "grad_norm": 13.62070083618164,
+      "learning_rate": 5e-06,
+      "loss": 0.624,
+      "num_input_tokens_seen": 124212744,
+      "step": 1982
+    },
+    {
+      "epoch": 6.595673876871881,
+      "loss": 0.7718583345413208,
+      "loss_ce": 0.0008622486493550241,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 124212744,
+      "step": 1982
+    },
+    {
+      "epoch": 6.599001663893511,
+      "grad_norm": 32.08122253417969,
+      "learning_rate": 5e-06,
+      "loss": 0.7776,
+      "num_input_tokens_seen": 124275488,
+      "step": 1983
+    },
+    {
+      "epoch": 6.599001663893511,
+      "loss": 0.7189164757728577,
+      "loss_ce": 0.00028856488643214107,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 124275488,
+      "step": 1983
+    },
+    {
+      "epoch": 6.602329450915142,
+      "grad_norm": 23.14647102355957,
+      "learning_rate": 5e-06,
+      "loss": 0.5029,
+      "num_input_tokens_seen": 124336988,
+      "step": 1984
+    },
+    {
+      "epoch": 6.602329450915142,
+      "loss": 0.5258368253707886,
+      "loss_ce": 0.00020208263595122844,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 124336988,
+      "step": 1984
+    },
+    {
+      "epoch": 6.605657237936772,
+      "grad_norm": 8.717888832092285,
+      "learning_rate": 5e-06,
+      "loss": 0.6094,
+      "num_input_tokens_seen": 124398312,
+      "step": 1985
+    },
+    {
+      "epoch": 6.605657237936772,
+      "loss": 0.942054033279419,
+      "loss_ce": 0.0006478212890215218,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 124398312,
+      "step": 1985
+    },
+    {
+      "epoch": 6.608985024958403,
+      "grad_norm": 11.982193946838379,
+      "learning_rate": 5e-06,
+      "loss": 0.4813,
+      "num_input_tokens_seen": 124461176,
+      "step": 1986
+    },
+    {
+      "epoch": 6.608985024958403,
+      "loss": 0.5817021727561951,
+      "loss_ce": 0.0013798931613564491,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 124461176,
+      "step": 1986
+    },
+    {
+      "epoch": 6.612312811980034,
+      "grad_norm": 40.03938674926758,
+      "learning_rate": 5e-06,
+      "loss": 0.5984,
+      "num_input_tokens_seen": 124524312,
+      "step": 1987
+    },
+    {
+      "epoch": 6.612312811980034,
+      "loss": 0.7929661273956299,
+      "loss_ce": 0.00024154715356417,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.79296875,
+      "num_input_tokens_seen": 124524312,
+      "step": 1987
+    },
+    {
+      "epoch": 6.615640599001664,
+      "grad_norm": 20.670000076293945,
+      "learning_rate": 5e-06,
+      "loss": 0.7356,
+      "num_input_tokens_seen": 124586984,
+      "step": 1988
+    },
+    {
+      "epoch": 6.615640599001664,
+      "loss": 0.534072995185852,
+      "loss_ce": 0.000625723332632333,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 124586984,
+      "step": 1988
+    },
+    {
+      "epoch": 6.618968386023295,
+      "grad_norm": 13.344552993774414,
+      "learning_rate": 5e-06,
+      "loss": 0.7439,
+      "num_input_tokens_seen": 124648496,
+      "step": 1989
+    },
+    {
+      "epoch": 6.618968386023295,
+      "loss": 0.8475092053413391,
+      "loss_ce": 0.0009515842539258301,
+      "loss_iou": 0.28125,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.84765625,
+      "num_input_tokens_seen": 124648496,
+      "step": 1989
+    },
+    {
+      "epoch": 6.6222961730449255,
+      "grad_norm": 17.5792179107666,
+      "learning_rate": 5e-06,
+      "loss": 0.5477,
+      "num_input_tokens_seen": 124711420,
+      "step": 1990
+    },
+    {
+      "epoch": 6.6222961730449255,
+      "loss": 0.6316379904747009,
+      "loss_ce": 4.6201777877286077e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 124711420,
+      "step": 1990
+    },
+    {
+      "epoch": 6.625623960066556,
+      "grad_norm": 11.981757164001465,
+      "learning_rate": 5e-06,
+      "loss": 0.5592,
+      "num_input_tokens_seen": 124774256,
+      "step": 1991
+    },
+    {
+      "epoch": 6.625623960066556,
+      "loss": 0.4685099422931671,
+      "loss_ce": 4.079750851815334e-06,
+      "loss_iou": 0.115234375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.46875,
+      "num_input_tokens_seen": 124774256,
+      "step": 1991
+    },
+    {
+      "epoch": 6.628951747088187,
+      "grad_norm": 19.191776275634766,
+      "learning_rate": 5e-06,
+      "loss": 0.751,
+      "num_input_tokens_seen": 124837888,
+      "step": 1992
+    },
+    {
+      "epoch": 6.628951747088187,
+      "loss": 0.8945164084434509,
+      "loss_ce": 0.0010838116286322474,
+      "loss_iou": 0.34375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.89453125,
+      "num_input_tokens_seen": 124837888,
+      "step": 1992
+    },
+    {
+      "epoch": 6.632279534109817,
+      "grad_norm": 15.3238525390625,
+      "learning_rate": 5e-06,
+      "loss": 0.7391,
+      "num_input_tokens_seen": 124902180,
+      "step": 1993
+    },
+    {
+      "epoch": 6.632279534109817,
+      "loss": 0.8709877729415894,
+      "loss_ce": 0.0006264817784540355,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 124902180,
+      "step": 1993
+    },
+    {
+      "epoch": 6.635607321131448,
+      "grad_norm": 26.2684383392334,
+      "learning_rate": 5e-06,
+      "loss": 0.6892,
+      "num_input_tokens_seen": 124964652,
+      "step": 1994
+    },
+    {
+      "epoch": 6.635607321131448,
+      "loss": 0.5876778364181519,
+      "loss_ce": 3.134192229481414e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 124964652,
+      "step": 1994
+    },
+    {
+      "epoch": 6.6389351081530785,
+      "grad_norm": 15.246932029724121,
+      "learning_rate": 5e-06,
+      "loss": 0.5931,
+      "num_input_tokens_seen": 125027128,
+      "step": 1995
+    },
+    {
+      "epoch": 6.6389351081530785,
+      "loss": 0.5009132623672485,
+      "loss_ce": 0.0004249872581567615,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 125027128,
+      "step": 1995
+    },
+    {
+      "epoch": 6.642262895174709,
+      "grad_norm": 16.900373458862305,
+      "learning_rate": 5e-06,
+      "loss": 0.5838,
+      "num_input_tokens_seen": 125090272,
+      "step": 1996
+    },
+    {
+      "epoch": 6.642262895174709,
+      "loss": 0.6842046976089478,
+      "loss_ce": 0.0001226711319759488,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.68359375,
+      "num_input_tokens_seen": 125090272,
+      "step": 1996
+    },
+    {
+      "epoch": 6.64559068219634,
+      "grad_norm": 11.600629806518555,
+      "learning_rate": 5e-06,
+      "loss": 0.5266,
+      "num_input_tokens_seen": 125152640,
+      "step": 1997
+    },
+    {
+      "epoch": 6.64559068219634,
+      "loss": 0.5774567127227783,
+      "loss_ce": 3.054131411772687e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 125152640,
+      "step": 1997
+    },
+    {
+      "epoch": 6.64891846921797,
+      "grad_norm": 20.360536575317383,
+      "learning_rate": 5e-06,
+      "loss": 0.5939,
+      "num_input_tokens_seen": 125213584,
+      "step": 1998
+    },
+    {
+      "epoch": 6.64891846921797,
+      "loss": 0.7232832908630371,
+      "loss_ce": 0.0008712068083696067,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 125213584,
+      "step": 1998
+    },
+    {
+      "epoch": 6.652246256239601,
+      "grad_norm": 8.929659843444824,
+      "learning_rate": 5e-06,
+      "loss": 0.5807,
+      "num_input_tokens_seen": 125276516,
+      "step": 1999
+    },
+    {
+      "epoch": 6.652246256239601,
+      "loss": 0.4460247755050659,
+      "loss_ce": 0.00046813851804472506,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.4453125,
+      "num_input_tokens_seen": 125276516,
+      "step": 1999
+    },
+    {
+      "epoch": 6.655574043261232,
+      "grad_norm": 14.47555923461914,
+      "learning_rate": 5e-06,
+      "loss": 0.6939,
+      "num_input_tokens_seen": 125338692,
+      "step": 2000
+    },
+    {
+      "epoch": 6.655574043261232,
+      "eval_seeclick_CIoU": 0.1068805642426014,
+      "eval_seeclick_GIoU": 0.11891250684857368,
+      "eval_seeclick_IoU": 0.20469707250595093,
+      "eval_seeclick_MAE_all": 0.16567593067884445,
+      "eval_seeclick_MAE_h": 0.05109903775155544,
+      "eval_seeclick_MAE_w": 0.12995515763759613,
+      "eval_seeclick_MAE_x_boxes": 0.22886644303798676,
+      "eval_seeclick_MAE_y_boxes": 0.13595783710479736,
+      "eval_seeclick_NUM_probability": 0.9999660849571228,
+      "eval_seeclick_inside_bbox": 0.22500000149011612,
+      "eval_seeclick_loss": 2.713005304336548,
+      "eval_seeclick_loss_ce": 0.11975685507059097,
+      "eval_seeclick_loss_iou": 0.88720703125,
+      "eval_seeclick_loss_num": 0.1656036376953125,
+      "eval_seeclick_loss_xval": 2.6025390625,
+      "eval_seeclick_runtime": 69.2988,
+      "eval_seeclick_samples_per_second": 0.678,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 125338692,
+      "step": 2000
+    },
+    {
+      "epoch": 6.655574043261232,
+      "eval_icons_CIoU": 0.03829660825431347,
+      "eval_icons_GIoU": 0.1595398709177971,
+      "eval_icons_IoU": 0.19770432263612747,
+      "eval_icons_MAE_all": 0.16016975790262222,
+      "eval_icons_MAE_h": 0.0871335044503212,
+      "eval_icons_MAE_w": 0.15490344911813736,
+      "eval_icons_MAE_x_boxes": 0.17506500333547592,
+      "eval_icons_MAE_y_boxes": 0.0526757575571537,
+      "eval_icons_NUM_probability": 0.9999957382678986,
+      "eval_icons_inside_bbox": 0.3420138955116272,
+      "eval_icons_loss": 2.462501287460327,
+      "eval_icons_loss_ce": 1.5740541243758344e-06,
+      "eval_icons_loss_iou": 0.8251953125,
+      "eval_icons_loss_num": 0.1562976837158203,
+      "eval_icons_loss_xval": 2.430908203125,
+      "eval_icons_runtime": 71.6265,
+      "eval_icons_samples_per_second": 0.698,
+      "eval_icons_steps_per_second": 0.028,
+      "num_input_tokens_seen": 125338692,
+      "step": 2000
+    },
+    {
+      "epoch": 6.655574043261232,
+      "eval_screenspot_CIoU": 0.0693756896071136,
+      "eval_screenspot_GIoU": 0.13485626379648843,
+      "eval_screenspot_IoU": 0.21177987257639566,
+      "eval_screenspot_MAE_all": 0.1937429408232371,
+      "eval_screenspot_MAE_h": 0.06855036566654842,
+      "eval_screenspot_MAE_w": 0.1509904464085897,
+      "eval_screenspot_MAE_x_boxes": 0.2597081462542216,
+      "eval_screenspot_MAE_y_boxes": 0.1477653905749321,
+      "eval_screenspot_NUM_probability": 0.9999565482139587,
+      "eval_screenspot_inside_bbox": 0.37583333253860474,
+      "eval_screenspot_loss": 2.735307455062866,
+      "eval_screenspot_loss_ce": 5.104218810932556e-05,
+      "eval_screenspot_loss_iou": 0.8837890625,
+      "eval_screenspot_loss_num": 0.20152791341145834,
+      "eval_screenspot_loss_xval": 2.7766927083333335,
+      "eval_screenspot_runtime": 131.5463,
+      "eval_screenspot_samples_per_second": 0.677,
+      "eval_screenspot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 125338692,
+      "step": 2000
+    },
+    {
+      "epoch": 6.655574043261232,
+      "eval_compot_CIoU": 0.005251538008451462,
+      "eval_compot_GIoU": 0.10733498632907867,
+      "eval_compot_IoU": 0.17231258749961853,
+      "eval_compot_MAE_all": 0.19772262126207352,
+      "eval_compot_MAE_h": 0.05648845434188843,
+      "eval_compot_MAE_w": 0.24372312426567078,
+      "eval_compot_MAE_x_boxes": 0.20571433007717133,
+      "eval_compot_MAE_y_boxes": 0.10716105252504349,
+      "eval_compot_NUM_probability": 0.9999851286411285,
+      "eval_compot_inside_bbox": 0.3229166716337204,
+      "eval_compot_loss": 2.8182108402252197,
+      "eval_compot_loss_ce": 0.0029962222324684262,
+      "eval_compot_loss_iou": 0.91943359375,
+      "eval_compot_loss_num": 0.20189666748046875,
+      "eval_compot_loss_xval": 2.84716796875,
+      "eval_compot_runtime": 78.1032,
+      "eval_compot_samples_per_second": 0.64,
+      "eval_compot_steps_per_second": 0.026,
+      "num_input_tokens_seen": 125338692,
+      "step": 2000
+    },
+    {
+      "epoch": 6.655574043261232,
+      "eval_custom_ui_MAE_all": 0.07614851370453835,
+      "eval_custom_ui_MAE_x": 0.07330591231584549,
+      "eval_custom_ui_MAE_y": 0.0789911113679409,
+      "eval_custom_ui_NUM_probability": 0.9999920725822449,
+      "eval_custom_ui_loss": 0.3671107590198517,
+      "eval_custom_ui_loss_ce": 1.62217548904664e-06,
+      "eval_custom_ui_loss_num": 0.0698394775390625,
+      "eval_custom_ui_loss_xval": 0.34930419921875,
+      "eval_custom_ui_runtime": 54.0842,
+      "eval_custom_ui_samples_per_second": 0.924,
+      "eval_custom_ui_steps_per_second": 0.037,
+      "num_input_tokens_seen": 125338692,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 15000,
+  "num_input_tokens_seen": 125338692,
+  "num_train_epochs": 50,
+  "save_steps": 250,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.830993252785848e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}