diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9428 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 200,
+  "global_step": 1334,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0014992503748125937,
+      "grad_norm": 13.26502565972832,
+      "learning_rate": 9.999986134743949e-06,
+      "loss": 0.272,
+      "step": 1
+    },
+    {
+      "epoch": 0.0029985007496251873,
+      "grad_norm": 6.415792055142533,
+      "learning_rate": 9.99994453905269e-06,
+      "loss": 0.1857,
+      "step": 2
+    },
+    {
+      "epoch": 0.004497751124437781,
+      "grad_norm": 11.10978411864584,
+      "learning_rate": 9.999875213156919e-06,
+      "loss": 0.1699,
+      "step": 3
+    },
+    {
+      "epoch": 0.005997001499250375,
+      "grad_norm": 8.382081527954217,
+      "learning_rate": 9.999778157441126e-06,
+      "loss": 0.1367,
+      "step": 4
+    },
+    {
+      "epoch": 0.0074962518740629685,
+      "grad_norm": 5.5060982533100145,
+      "learning_rate": 9.99965337244359e-06,
+      "loss": 0.1497,
+      "step": 5
+    },
+    {
+      "epoch": 0.008995502248875561,
+      "grad_norm": 1.9437354048876796,
+      "learning_rate": 9.999500858856382e-06,
+      "loss": 0.1344,
+      "step": 6
+    },
+    {
+      "epoch": 0.010494752623688156,
+      "grad_norm": 3.1786028677810156,
+      "learning_rate": 9.999320617525356e-06,
+      "loss": 0.132,
+      "step": 7
+    },
+    {
+      "epoch": 0.01199400299850075,
+      "grad_norm": 1.6892173798813785,
+      "learning_rate": 9.999112649450154e-06,
+      "loss": 0.1215,
+      "step": 8
+    },
+    {
+      "epoch": 0.013493253373313344,
+      "grad_norm": 1.9819513149201022,
+      "learning_rate": 9.998876955784183e-06,
+      "loss": 0.0951,
+      "step": 9
+    },
+    {
+      "epoch": 0.014992503748125937,
+      "grad_norm": 2.298734499476876,
+      "learning_rate": 9.998613537834625e-06,
+      "loss": 0.1322,
+      "step": 10
+    },
+    {
+      "epoch": 0.01649175412293853,
+      "grad_norm": 1.694815130379713,
+      "learning_rate": 9.998322397062426e-06,
+      "loss": 0.1232,
+      "step": 11
+    },
+    {
+      "epoch": 0.017991004497751123,
+      "grad_norm": 1.3025934412272222,
+      "learning_rate": 9.99800353508228e-06,
+      "loss": 0.0971,
+      "step": 12
+    },
+    {
+      "epoch": 0.019490254872563718,
+      "grad_norm": 1.1700665880144996,
+      "learning_rate": 9.997656953662627e-06,
+      "loss": 0.0968,
+      "step": 13
+    },
+    {
+      "epoch": 0.020989505247376312,
+      "grad_norm": 1.8989369548737265,
+      "learning_rate": 9.997282654725645e-06,
+      "loss": 0.1239,
+      "step": 14
+    },
+    {
+      "epoch": 0.022488755622188907,
+      "grad_norm": 1.2375505688394786,
+      "learning_rate": 9.996880640347234e-06,
+      "loss": 0.0918,
+      "step": 15
+    },
+    {
+      "epoch": 0.0239880059970015,
+      "grad_norm": 1.2597616659301192,
+      "learning_rate": 9.99645091275701e-06,
+      "loss": 0.108,
+      "step": 16
+    },
+    {
+      "epoch": 0.025487256371814093,
+      "grad_norm": 8.017561344718406,
+      "learning_rate": 9.99599347433828e-06,
+      "loss": 0.1147,
+      "step": 17
+    },
+    {
+      "epoch": 0.026986506746626688,
+      "grad_norm": 1.7375340904830014,
+      "learning_rate": 9.99550832762805e-06,
+      "loss": 0.1356,
+      "step": 18
+    },
+    {
+      "epoch": 0.02848575712143928,
+      "grad_norm": 1.6319531131752782,
+      "learning_rate": 9.99499547531699e-06,
+      "loss": 0.1096,
+      "step": 19
+    },
+    {
+      "epoch": 0.029985007496251874,
+      "grad_norm": 1.505947233132876,
+      "learning_rate": 9.994454920249433e-06,
+      "loss": 0.1183,
+      "step": 20
+    },
+    {
+      "epoch": 0.031484257871064465,
+      "grad_norm": 1.1815974878939706,
+      "learning_rate": 9.993886665423348e-06,
+      "loss": 0.0944,
+      "step": 21
+    },
+    {
+      "epoch": 0.03298350824587706,
+      "grad_norm": 1.2999613016832219,
+      "learning_rate": 9.993290713990343e-06,
+      "loss": 0.1168,
+      "step": 22
+    },
+    {
+      "epoch": 0.034482758620689655,
+      "grad_norm": 1.1221514161840906,
+      "learning_rate": 9.99266706925562e-06,
+      "loss": 0.09,
+      "step": 23
+    },
+    {
+      "epoch": 0.035982008995502246,
+      "grad_norm": 1.7478360492082057,
+      "learning_rate": 9.992015734677979e-06,
+      "loss": 0.1206,
+      "step": 24
+    },
+    {
+      "epoch": 0.037481259370314844,
+      "grad_norm": 1.633172666008338,
+      "learning_rate": 9.991336713869785e-06,
+      "loss": 0.1063,
+      "step": 25
+    },
+    {
+      "epoch": 0.038980509745127435,
+      "grad_norm": 2.173462374358993,
+      "learning_rate": 9.99063001059696e-06,
+      "loss": 0.0878,
+      "step": 26
+    },
+    {
+      "epoch": 0.04047976011994003,
+      "grad_norm": 1.5232555771504384,
+      "learning_rate": 9.989895628778952e-06,
+      "loss": 0.1222,
+      "step": 27
+    },
+    {
+      "epoch": 0.041979010494752625,
+      "grad_norm": 1.231627475332525,
+      "learning_rate": 9.989133572488716e-06,
+      "loss": 0.0924,
+      "step": 28
+    },
+    {
+      "epoch": 0.043478260869565216,
+      "grad_norm": 1.5212678247247353,
+      "learning_rate": 9.988343845952697e-06,
+      "loss": 0.0975,
+      "step": 29
+    },
+    {
+      "epoch": 0.044977511244377814,
+      "grad_norm": 1.3904420397403086,
+      "learning_rate": 9.987526453550798e-06,
+      "loss": 0.0961,
+      "step": 30
+    },
+    {
+      "epoch": 0.046476761619190406,
+      "grad_norm": 1.4436589827404271,
+      "learning_rate": 9.98668139981636e-06,
+      "loss": 0.0956,
+      "step": 31
+    },
+    {
+      "epoch": 0.047976011994003,
+      "grad_norm": 1.650806118586378,
+      "learning_rate": 9.98580868943614e-06,
+      "loss": 0.1297,
+      "step": 32
+    },
+    {
+      "epoch": 0.049475262368815595,
+      "grad_norm": 0.9831438579696592,
+      "learning_rate": 9.984908327250278e-06,
+      "loss": 0.0823,
+      "step": 33
+    },
+    {
+      "epoch": 0.050974512743628186,
+      "grad_norm": 1.2506283780995278,
+      "learning_rate": 9.983980318252274e-06,
+      "loss": 0.081,
+      "step": 34
+    },
+    {
+      "epoch": 0.05247376311844078,
+      "grad_norm": 1.180710843211676,
+      "learning_rate": 9.983024667588961e-06,
+      "loss": 0.101,
+      "step": 35
+    },
+    {
+      "epoch": 0.053973013493253376,
+      "grad_norm": 1.6205414424965816,
+      "learning_rate": 9.982041380560476e-06,
+      "loss": 0.121,
+      "step": 36
+    },
+    {
+      "epoch": 0.05547226386806597,
+      "grad_norm": 1.4304796705358827,
+      "learning_rate": 9.98103046262023e-06,
+      "loss": 0.0986,
+      "step": 37
+    },
+    {
+      "epoch": 0.05697151424287856,
+      "grad_norm": 1.7261135450730076,
+      "learning_rate": 9.979991919374877e-06,
+      "loss": 0.1258,
+      "step": 38
+    },
+    {
+      "epoch": 0.05847076461769116,
+      "grad_norm": 1.216153229881134,
+      "learning_rate": 9.978925756584284e-06,
+      "loss": 0.1188,
+      "step": 39
+    },
+    {
+      "epoch": 0.05997001499250375,
+      "grad_norm": 1.4590418123205482,
+      "learning_rate": 9.9778319801615e-06,
+      "loss": 0.0872,
+      "step": 40
+    },
+    {
+      "epoch": 0.06146926536731634,
+      "grad_norm": 1.1730880138494157,
+      "learning_rate": 9.976710596172721e-06,
+      "loss": 0.0973,
+      "step": 41
+    },
+    {
+      "epoch": 0.06296851574212893,
+      "grad_norm": 1.5018162261660404,
+      "learning_rate": 9.975561610837254e-06,
+      "loss": 0.1151,
+      "step": 42
+    },
+    {
+      "epoch": 0.06446776611694154,
+      "grad_norm": 1.2176551032212017,
+      "learning_rate": 9.974385030527496e-06,
+      "loss": 0.0905,
+      "step": 43
+    },
+    {
+      "epoch": 0.06596701649175413,
+      "grad_norm": 1.3489830828915375,
+      "learning_rate": 9.973180861768874e-06,
+      "loss": 0.1081,
+      "step": 44
+    },
+    {
+      "epoch": 0.06746626686656672,
+      "grad_norm": 1.808347010432726,
+      "learning_rate": 9.971949111239838e-06,
+      "loss": 0.1148,
+      "step": 45
+    },
+    {
+      "epoch": 0.06896551724137931,
+      "grad_norm": 1.0288816559164147,
+      "learning_rate": 9.970689785771798e-06,
+      "loss": 0.0866,
+      "step": 46
+    },
+    {
+      "epoch": 0.0704647676161919,
+      "grad_norm": 1.0035100925977691,
+      "learning_rate": 9.969402892349105e-06,
+      "loss": 0.0721,
+      "step": 47
+    },
+    {
+      "epoch": 0.07196401799100449,
+      "grad_norm": 1.509817116280432,
+      "learning_rate": 9.968088438109002e-06,
+      "loss": 0.0795,
+      "step": 48
+    },
+    {
+      "epoch": 0.0734632683658171,
+      "grad_norm": 1.6752067001901925,
+      "learning_rate": 9.966746430341584e-06,
+      "loss": 0.0798,
+      "step": 49
+    },
+    {
+      "epoch": 0.07496251874062969,
+      "grad_norm": 5.63881375075983,
+      "learning_rate": 9.965376876489765e-06,
+      "loss": 0.1107,
+      "step": 50
+    },
+    {
+      "epoch": 0.07646176911544228,
+      "grad_norm": 3.1746071864052907,
+      "learning_rate": 9.963979784149232e-06,
+      "loss": 0.0917,
+      "step": 51
+    },
+    {
+      "epoch": 0.07796101949025487,
+      "grad_norm": 1.5070184852010142,
+      "learning_rate": 9.962555161068401e-06,
+      "loss": 0.0951,
+      "step": 52
+    },
+    {
+      "epoch": 0.07946026986506746,
+      "grad_norm": 1.1390904732886866,
+      "learning_rate": 9.961103015148376e-06,
+      "loss": 0.0948,
+      "step": 53
+    },
+    {
+      "epoch": 0.08095952023988005,
+      "grad_norm": 1.6227802283234323,
+      "learning_rate": 9.95962335444291e-06,
+      "loss": 0.0951,
+      "step": 54
+    },
+    {
+      "epoch": 0.08245877061469266,
+      "grad_norm": 1.12806435900969,
+      "learning_rate": 9.958116187158351e-06,
+      "loss": 0.0697,
+      "step": 55
+    },
+    {
+      "epoch": 0.08395802098950525,
+      "grad_norm": 1.3088220663550938,
+      "learning_rate": 9.956581521653604e-06,
+      "loss": 0.1219,
+      "step": 56
+    },
+    {
+      "epoch": 0.08545727136431784,
+      "grad_norm": 1.352351115916329,
+      "learning_rate": 9.955019366440082e-06,
+      "loss": 0.1019,
+      "step": 57
+    },
+    {
+      "epoch": 0.08695652173913043,
+      "grad_norm": 1.3956454676752919,
+      "learning_rate": 9.953429730181653e-06,
+      "loss": 0.0996,
+      "step": 58
+    },
+    {
+      "epoch": 0.08845577211394302,
+      "grad_norm": 0.9581560374558995,
+      "learning_rate": 9.95181262169461e-06,
+      "loss": 0.0869,
+      "step": 59
+    },
+    {
+      "epoch": 0.08995502248875563,
+      "grad_norm": 1.2541215303996898,
+      "learning_rate": 9.950168049947597e-06,
+      "loss": 0.0954,
+      "step": 60
+    },
+    {
+      "epoch": 0.09145427286356822,
+      "grad_norm": 5.803674366888778,
+      "learning_rate": 9.948496024061577e-06,
+      "loss": 0.0638,
+      "step": 61
+    },
+    {
+      "epoch": 0.09295352323838081,
+      "grad_norm": 1.144167919229693,
+      "learning_rate": 9.94679655330978e-06,
+      "loss": 0.1004,
+      "step": 62
+    },
+    {
+      "epoch": 0.0944527736131934,
+      "grad_norm": 1.4435443077654662,
+      "learning_rate": 9.945069647117645e-06,
+      "loss": 0.0928,
+      "step": 63
+    },
+    {
+      "epoch": 0.095952023988006,
+      "grad_norm": 1.0615094054663314,
+      "learning_rate": 9.943315315062766e-06,
+      "loss": 0.0922,
+      "step": 64
+    },
+    {
+      "epoch": 0.09745127436281859,
+      "grad_norm": 0.7356123022148706,
+      "learning_rate": 9.941533566874852e-06,
+      "loss": 0.0577,
+      "step": 65
+    },
+    {
+      "epoch": 0.09895052473763119,
+      "grad_norm": 1.3222133180093871,
+      "learning_rate": 9.939724412435661e-06,
+      "loss": 0.1021,
+      "step": 66
+    },
+    {
+      "epoch": 0.10044977511244378,
+      "grad_norm": 0.7820287705224593,
+      "learning_rate": 9.937887861778947e-06,
+      "loss": 0.0775,
+      "step": 67
+    },
+    {
+      "epoch": 0.10194902548725637,
+      "grad_norm": 1.394345647600291,
+      "learning_rate": 9.93602392509041e-06,
+      "loss": 0.0936,
+      "step": 68
+    },
+    {
+      "epoch": 0.10344827586206896,
+      "grad_norm": 1.4095954264322328,
+      "learning_rate": 9.934132612707631e-06,
+      "loss": 0.0789,
+      "step": 69
+    },
+    {
+      "epoch": 0.10494752623688156,
+      "grad_norm": 1.3025269242221205,
+      "learning_rate": 9.932213935120025e-06,
+      "loss": 0.0717,
+      "step": 70
+    },
+    {
+      "epoch": 0.10644677661169415,
+      "grad_norm": 0.6689191754554515,
+      "learning_rate": 9.930267902968774e-06,
+      "loss": 0.0613,
+      "step": 71
+    },
+    {
+      "epoch": 0.10794602698650675,
+      "grad_norm": 1.2000854107874113,
+      "learning_rate": 9.928294527046771e-06,
+      "loss": 0.1065,
+      "step": 72
+    },
+    {
+      "epoch": 0.10944527736131934,
+      "grad_norm": 1.3553103202961787,
+      "learning_rate": 9.92629381829856e-06,
+      "loss": 0.1065,
+      "step": 73
+    },
+    {
+      "epoch": 0.11094452773613193,
+      "grad_norm": 1.5380978893273833,
+      "learning_rate": 9.924265787820279e-06,
+      "loss": 0.1253,
+      "step": 74
+    },
+    {
+      "epoch": 0.11244377811094453,
+      "grad_norm": 1.2141635918772498,
+      "learning_rate": 9.92221044685959e-06,
+      "loss": 0.0986,
+      "step": 75
+    },
+    {
+      "epoch": 0.11394302848575712,
+      "grad_norm": 1.0175206943789166,
+      "learning_rate": 9.920127806815627e-06,
+      "loss": 0.0784,
+      "step": 76
+    },
+    {
+      "epoch": 0.11544227886056972,
+      "grad_norm": 0.9031372872633094,
+      "learning_rate": 9.918017879238922e-06,
+      "loss": 0.0833,
+      "step": 77
+    },
+    {
+      "epoch": 0.11694152923538231,
+      "grad_norm": 0.6734659366360752,
+      "learning_rate": 9.915880675831352e-06,
+      "loss": 0.0601,
+      "step": 78
+    },
+    {
+      "epoch": 0.1184407796101949,
+      "grad_norm": 1.110855562166447,
+      "learning_rate": 9.913716208446067e-06,
+      "loss": 0.0874,
+      "step": 79
+    },
+    {
+      "epoch": 0.1199400299850075,
+      "grad_norm": 1.0679202747350662,
+      "learning_rate": 9.91152448908742e-06,
+      "loss": 0.082,
+      "step": 80
+    },
+    {
+      "epoch": 0.12143928035982009,
+      "grad_norm": 0.9503217776466537,
+      "learning_rate": 9.909305529910917e-06,
+      "loss": 0.094,
+      "step": 81
+    },
+    {
+      "epoch": 0.12293853073463268,
+      "grad_norm": 1.1016844829202168,
+      "learning_rate": 9.907059343223129e-06,
+      "loss": 0.0912,
+      "step": 82
+    },
+    {
+      "epoch": 0.12443778110944528,
+      "grad_norm": 0.7126097332627687,
+      "learning_rate": 9.904785941481638e-06,
+      "loss": 0.0643,
+      "step": 83
+    },
+    {
+      "epoch": 0.12593703148425786,
+      "grad_norm": 0.624235737024501,
+      "learning_rate": 9.902485337294965e-06,
+      "loss": 0.0552,
+      "step": 84
+    },
+    {
+      "epoch": 0.12743628185907047,
+      "grad_norm": 0.8185402812240381,
+      "learning_rate": 9.900157543422493e-06,
+      "loss": 0.0744,
+      "step": 85
+    },
+    {
+      "epoch": 0.12893553223388307,
+      "grad_norm": 0.9213556640180655,
+      "learning_rate": 9.897802572774407e-06,
+      "loss": 0.0764,
+      "step": 86
+    },
+    {
+      "epoch": 0.13043478260869565,
+      "grad_norm": 0.8230687421136603,
+      "learning_rate": 9.895420438411616e-06,
+      "loss": 0.0704,
+      "step": 87
+    },
+    {
+      "epoch": 0.13193403298350825,
+      "grad_norm": 0.6821964376266765,
+      "learning_rate": 9.893011153545679e-06,
+      "loss": 0.0536,
+      "step": 88
+    },
+    {
+      "epoch": 0.13343328335832083,
+      "grad_norm": 1.001077071898709,
+      "learning_rate": 9.89057473153874e-06,
+      "loss": 0.0996,
+      "step": 89
+    },
+    {
+      "epoch": 0.13493253373313344,
+      "grad_norm": 0.9061349569266882,
+      "learning_rate": 9.888111185903442e-06,
+      "loss": 0.0717,
+      "step": 90
+    },
+    {
+      "epoch": 0.136431784107946,
+      "grad_norm": 0.9179497788520988,
+      "learning_rate": 9.885620530302865e-06,
+      "loss": 0.0846,
+      "step": 91
+    },
+    {
+      "epoch": 0.13793103448275862,
+      "grad_norm": 0.9573932222213358,
+      "learning_rate": 9.883102778550434e-06,
+      "loss": 0.088,
+      "step": 92
+    },
+    {
+      "epoch": 0.13943028485757122,
+      "grad_norm": 0.9063846833450139,
+      "learning_rate": 9.880557944609863e-06,
+      "loss": 0.079,
+      "step": 93
+    },
+    {
+      "epoch": 0.1409295352323838,
+      "grad_norm": 0.9987951130182463,
+      "learning_rate": 9.877986042595062e-06,
+      "loss": 0.0951,
+      "step": 94
+    },
+    {
+      "epoch": 0.1424287856071964,
+      "grad_norm": 1.1596730383061364,
+      "learning_rate": 9.87538708677006e-06,
+      "loss": 0.1094,
+      "step": 95
+    },
+    {
+      "epoch": 0.14392803598200898,
+      "grad_norm": 0.9564088046714287,
+      "learning_rate": 9.872761091548933e-06,
+      "loss": 0.0859,
+      "step": 96
+    },
+    {
+      "epoch": 0.1454272863568216,
+      "grad_norm": 1.2271391837227092,
+      "learning_rate": 9.870108071495721e-06,
+      "loss": 0.1035,
+      "step": 97
+    },
+    {
+      "epoch": 0.1469265367316342,
+      "grad_norm": 0.7869222097048603,
+      "learning_rate": 9.867428041324345e-06,
+      "loss": 0.0775,
+      "step": 98
+    },
+    {
+      "epoch": 0.14842578710644677,
+      "grad_norm": 0.850742004876203,
+      "learning_rate": 9.864721015898524e-06,
+      "loss": 0.0737,
+      "step": 99
+    },
+    {
+      "epoch": 0.14992503748125938,
+      "grad_norm": 1.0939502854790581,
+      "learning_rate": 9.861987010231701e-06,
+      "loss": 0.0991,
+      "step": 100
+    },
+    {
+      "epoch": 0.15142428785607195,
+      "grad_norm": 0.9848423955002873,
+      "learning_rate": 9.85922603948695e-06,
+      "loss": 0.0986,
+      "step": 101
+    },
+    {
+      "epoch": 0.15292353823088456,
+      "grad_norm": 0.9073892887118093,
+      "learning_rate": 9.856438118976899e-06,
+      "loss": 0.0848,
+      "step": 102
+    },
+    {
+      "epoch": 0.15442278860569716,
+      "grad_norm": 0.8511612396824224,
+      "learning_rate": 9.853623264163638e-06,
+      "loss": 0.0791,
+      "step": 103
+    },
+    {
+      "epoch": 0.15592203898050974,
+      "grad_norm": 0.8313478728471725,
+      "learning_rate": 9.850781490658643e-06,
+      "loss": 0.08,
+      "step": 104
+    },
+    {
+      "epoch": 0.15742128935532235,
+      "grad_norm": 0.914407500472306,
+      "learning_rate": 9.84791281422268e-06,
+      "loss": 0.09,
+      "step": 105
+    },
+    {
+      "epoch": 0.15892053973013492,
+      "grad_norm": 1.021241241999968,
+      "learning_rate": 9.845017250765721e-06,
+      "loss": 0.1124,
+      "step": 106
+    },
+    {
+      "epoch": 0.16041979010494753,
+      "grad_norm": 0.8114001034369749,
+      "learning_rate": 9.84209481634686e-06,
+      "loss": 0.0819,
+      "step": 107
+    },
+    {
+      "epoch": 0.1619190404797601,
+      "grad_norm": 1.1307151304487306,
+      "learning_rate": 9.839145527174216e-06,
+      "loss": 0.1024,
+      "step": 108
+    },
+    {
+      "epoch": 0.1634182908545727,
+      "grad_norm": 0.9101968984887977,
+      "learning_rate": 9.836169399604846e-06,
+      "loss": 0.0968,
+      "step": 109
+    },
+    {
+      "epoch": 0.16491754122938532,
+      "grad_norm": 0.8368486665148361,
+      "learning_rate": 9.833166450144665e-06,
+      "loss": 0.076,
+      "step": 110
+    },
+    {
+      "epoch": 0.1664167916041979,
+      "grad_norm": 0.9239885673615119,
+      "learning_rate": 9.830136695448334e-06,
+      "loss": 0.0873,
+      "step": 111
+    },
+    {
+      "epoch": 0.1679160419790105,
+      "grad_norm": 0.8668303510899839,
+      "learning_rate": 9.827080152319182e-06,
+      "loss": 0.0892,
+      "step": 112
+    },
+    {
+      "epoch": 0.16941529235382308,
+      "grad_norm": 1.161425645543233,
+      "learning_rate": 9.823996837709114e-06,
+      "loss": 0.0809,
+      "step": 113
+    },
+    {
+      "epoch": 0.17091454272863568,
+      "grad_norm": 0.9061182193424888,
+      "learning_rate": 9.820886768718503e-06,
+      "loss": 0.0969,
+      "step": 114
+    },
+    {
+      "epoch": 0.1724137931034483,
+      "grad_norm": 0.7412602744876524,
+      "learning_rate": 9.817749962596115e-06,
+      "loss": 0.0762,
+      "step": 115
+    },
+    {
+      "epoch": 0.17391304347826086,
+      "grad_norm": 1.3814891479513285,
+      "learning_rate": 9.814586436738998e-06,
+      "loss": 0.1659,
+      "step": 116
+    },
+    {
+      "epoch": 0.17541229385307347,
+      "grad_norm": 1.0612419425970216,
+      "learning_rate": 9.811396208692387e-06,
+      "loss": 0.0982,
+      "step": 117
+    },
+    {
+      "epoch": 0.17691154422788605,
+      "grad_norm": 1.2003038110538045,
+      "learning_rate": 9.808179296149616e-06,
+      "loss": 0.1194,
+      "step": 118
+    },
+    {
+      "epoch": 0.17841079460269865,
+      "grad_norm": 2.373943460574113,
+      "learning_rate": 9.804935716952011e-06,
+      "loss": 0.1066,
+      "step": 119
+    },
+    {
+      "epoch": 0.17991004497751126,
+      "grad_norm": 6.498688519039836,
+      "learning_rate": 9.801665489088795e-06,
+      "loss": 0.0882,
+      "step": 120
+    },
+    {
+      "epoch": 0.18140929535232383,
+      "grad_norm": 1.3023140240978248,
+      "learning_rate": 9.798368630696984e-06,
+      "loss": 0.105,
+      "step": 121
+    },
+    {
+      "epoch": 0.18290854572713644,
+      "grad_norm": 2.9396793364384406,
+      "learning_rate": 9.795045160061295e-06,
+      "loss": 0.1106,
+      "step": 122
+    },
+    {
+      "epoch": 0.18440779610194902,
+      "grad_norm": 2.0533773689086705,
+      "learning_rate": 9.791695095614036e-06,
+      "loss": 0.1042,
+      "step": 123
+    },
+    {
+      "epoch": 0.18590704647676162,
+      "grad_norm": 1.0069953729103869,
+      "learning_rate": 9.788318455935008e-06,
+      "loss": 0.075,
+      "step": 124
+    },
+    {
+      "epoch": 0.1874062968515742,
+      "grad_norm": 2.908459656497082,
+      "learning_rate": 9.7849152597514e-06,
+      "loss": 0.0666,
+      "step": 125
+    },
+    {
+      "epoch": 0.1889055472263868,
+      "grad_norm": 0.9669301483233409,
+      "learning_rate": 9.781485525937683e-06,
+      "loss": 0.0767,
+      "step": 126
+    },
+    {
+      "epoch": 0.1904047976011994,
+      "grad_norm": 0.9607909785372043,
+      "learning_rate": 9.778029273515519e-06,
+      "loss": 0.0772,
+      "step": 127
+    },
+    {
+      "epoch": 0.191904047976012,
+      "grad_norm": 2.288927363054112,
+      "learning_rate": 9.774546521653633e-06,
+      "loss": 0.1095,
+      "step": 128
+    },
+    {
+      "epoch": 0.1934032983508246,
+      "grad_norm": 1.2935263286068437,
+      "learning_rate": 9.771037289667726e-06,
+      "loss": 0.0789,
+      "step": 129
+    },
+    {
+      "epoch": 0.19490254872563717,
+      "grad_norm": 1.0249179408124138,
+      "learning_rate": 9.767501597020357e-06,
+      "loss": 0.0892,
+      "step": 130
+    },
+    {
+      "epoch": 0.19640179910044978,
+      "grad_norm": 0.9414782668773556,
+      "learning_rate": 9.76393946332084e-06,
+      "loss": 0.0868,
+      "step": 131
+    },
+    {
+      "epoch": 0.19790104947526238,
+      "grad_norm": 1.2343630490899091,
+      "learning_rate": 9.760350908325131e-06,
+      "loss": 0.1045,
+      "step": 132
+    },
+    {
+      "epoch": 0.19940029985007496,
+      "grad_norm": 0.8685100503446493,
+      "learning_rate": 9.756735951935725e-06,
+      "loss": 0.0967,
+      "step": 133
+    },
+    {
+      "epoch": 0.20089955022488756,
+      "grad_norm": 0.892612679207988,
+      "learning_rate": 9.753094614201542e-06,
+      "loss": 0.0728,
+      "step": 134
+    },
+    {
+      "epoch": 0.20239880059970014,
+      "grad_norm": 0.9334471016773784,
+      "learning_rate": 9.749426915317812e-06,
+      "loss": 0.086,
+      "step": 135
+    },
+    {
+      "epoch": 0.20389805097451275,
+      "grad_norm": 0.8414973972648457,
+      "learning_rate": 9.74573287562597e-06,
+      "loss": 0.0724,
+      "step": 136
+    },
+    {
+      "epoch": 0.20539730134932535,
+      "grad_norm": 0.94045602703502,
+      "learning_rate": 9.742012515613536e-06,
+      "loss": 0.0854,
+      "step": 137
+    },
+    {
+      "epoch": 0.20689655172413793,
+      "grad_norm": 0.864129730855834,
+      "learning_rate": 9.738265855914014e-06,
+      "loss": 0.0881,
+      "step": 138
+    },
+    {
+      "epoch": 0.20839580209895053,
+      "grad_norm": 0.9358420656149644,
+      "learning_rate": 9.734492917306754e-06,
+      "loss": 0.0886,
+      "step": 139
+    },
+    {
+      "epoch": 0.2098950524737631,
+      "grad_norm": 1.0141316085601595,
+      "learning_rate": 9.730693720716866e-06,
+      "loss": 0.1102,
+      "step": 140
+    },
+    {
+      "epoch": 0.21139430284857572,
+      "grad_norm": 0.8765552930169166,
+      "learning_rate": 9.72686828721508e-06,
+      "loss": 0.0838,
+      "step": 141
+    },
+    {
+      "epoch": 0.2128935532233883,
+      "grad_norm": 1.266315129173229,
+      "learning_rate": 9.723016638017644e-06,
+      "loss": 0.126,
+      "step": 142
+    },
+    {
+      "epoch": 0.2143928035982009,
+      "grad_norm": 0.9595423977029551,
+      "learning_rate": 9.719138794486198e-06,
+      "loss": 0.0839,
+      "step": 143
+    },
+    {
+      "epoch": 0.2158920539730135,
+      "grad_norm": 0.7296390061812317,
+      "learning_rate": 9.715234778127658e-06,
+      "loss": 0.0641,
+      "step": 144
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "grad_norm": 0.9495998107332504,
+      "learning_rate": 9.711304610594104e-06,
+      "loss": 0.0938,
+      "step": 145
+    },
+    {
+      "epoch": 0.21889055472263869,
+      "grad_norm": 1.1495919326296664,
+      "learning_rate": 9.70734831368264e-06,
+      "loss": 0.1175,
+      "step": 146
+    },
+    {
+      "epoch": 0.22038980509745126,
+      "grad_norm": 0.8185981763501038,
+      "learning_rate": 9.7033659093353e-06,
+      "loss": 0.0797,
+      "step": 147
+    },
+    {
+      "epoch": 0.22188905547226387,
+      "grad_norm": 1.0111925422801782,
+      "learning_rate": 9.699357419638904e-06,
+      "loss": 0.0948,
+      "step": 148
+    },
+    {
+      "epoch": 0.22338830584707647,
+      "grad_norm": 0.9729023112424593,
+      "learning_rate": 9.695322866824948e-06,
+      "loss": 0.0933,
+      "step": 149
+    },
+    {
+      "epoch": 0.22488755622188905,
+      "grad_norm": 0.9647403057621722,
+      "learning_rate": 9.691262273269472e-06,
+      "loss": 0.0915,
+      "step": 150
+    },
+    {
+      "epoch": 0.22638680659670166,
+      "grad_norm": 0.6865818012372632,
+      "learning_rate": 9.687175661492944e-06,
+      "loss": 0.0633,
+      "step": 151
+    },
+    {
+      "epoch": 0.22788605697151423,
+      "grad_norm": 0.7474197578330629,
+      "learning_rate": 9.683063054160136e-06,
+      "loss": 0.0597,
+      "step": 152
+    },
+    {
+      "epoch": 0.22938530734632684,
+      "grad_norm": 0.9151889181902987,
+      "learning_rate": 9.678924474079986e-06,
+      "loss": 0.0839,
+      "step": 153
+    },
+    {
+      "epoch": 0.23088455772113944,
+      "grad_norm": 0.7833819151357534,
+      "learning_rate": 9.67475994420548e-06,
+      "loss": 0.0633,
+      "step": 154
+    },
+    {
+      "epoch": 0.23238380809595202,
+      "grad_norm": 0.9259039350133348,
+      "learning_rate": 9.670569487633534e-06,
+      "loss": 0.0975,
+      "step": 155
+    },
+    {
+      "epoch": 0.23388305847076463,
+      "grad_norm": 1.061095783872958,
+      "learning_rate": 9.666353127604845e-06,
+      "loss": 0.1041,
+      "step": 156
+    },
+    {
+      "epoch": 0.2353823088455772,
+      "grad_norm": 0.7822417238058937,
+      "learning_rate": 9.66211088750378e-06,
+      "loss": 0.0715,
+      "step": 157
+    },
+    {
+      "epoch": 0.2368815592203898,
+      "grad_norm": 0.9190306969327839,
+      "learning_rate": 9.657842790858235e-06,
+      "loss": 0.0817,
+      "step": 158
+    },
+    {
+      "epoch": 0.2383808095952024,
+      "grad_norm": 1.8453962739395198,
+      "learning_rate": 9.65354886133951e-06,
+      "loss": 0.1533,
+      "step": 159
+    },
+    {
+      "epoch": 0.239880059970015,
+      "grad_norm": 0.7791441547414752,
+      "learning_rate": 9.64922912276218e-06,
+      "loss": 0.0701,
+      "step": 160
+    },
+    {
+      "epoch": 0.2413793103448276,
+      "grad_norm": 1.0457730941410868,
+      "learning_rate": 9.644883599083959e-06,
+      "loss": 0.083,
+      "step": 161
+    },
+    {
+      "epoch": 0.24287856071964017,
+      "grad_norm": 1.3353647295950095,
+      "learning_rate": 9.640512314405563e-06,
+      "loss": 0.0938,
+      "step": 162
+    },
+    {
+      "epoch": 0.24437781109445278,
+      "grad_norm": 0.7773466475084545,
+      "learning_rate": 9.636115292970587e-06,
+      "loss": 0.0719,
+      "step": 163
+    },
+    {
+      "epoch": 0.24587706146926536,
+      "grad_norm": 0.7527814689224679,
+      "learning_rate": 9.63169255916536e-06,
+      "loss": 0.0712,
+      "step": 164
+    },
+    {
+      "epoch": 0.24737631184407796,
+      "grad_norm": 0.9509688342120358,
+      "learning_rate": 9.627244137518821e-06,
+      "loss": 0.0748,
+      "step": 165
+    },
+    {
+      "epoch": 0.24887556221889057,
+      "grad_norm": 0.9555058792835063,
+      "learning_rate": 9.622770052702366e-06,
+      "loss": 0.0963,
+      "step": 166
+    },
+    {
+      "epoch": 0.25037481259370314,
+      "grad_norm": 0.8133602429116359,
+      "learning_rate": 9.618270329529734e-06,
+      "loss": 0.0765,
+      "step": 167
+    },
+    {
+      "epoch": 0.2518740629685157,
+      "grad_norm": 0.6461682346146611,
+      "learning_rate": 9.613744992956844e-06,
+      "loss": 0.0677,
+      "step": 168
+    },
+    {
+      "epoch": 0.25337331334332835,
+      "grad_norm": 0.7037787335713744,
+      "learning_rate": 9.609194068081682e-06,
+      "loss": 0.0686,
+      "step": 169
+    },
+    {
+      "epoch": 0.25487256371814093,
+      "grad_norm": 0.8617534185102779,
+      "learning_rate": 9.60461758014414e-06,
+      "loss": 0.0778,
+      "step": 170
+    },
+    {
+      "epoch": 0.2563718140929535,
+      "grad_norm": 0.9177885898974814,
+      "learning_rate": 9.60001555452589e-06,
+      "loss": 0.0811,
+      "step": 171
+    },
+    {
+      "epoch": 0.25787106446776614,
+      "grad_norm": 0.9918298108377994,
+      "learning_rate": 9.595388016750236e-06,
+      "loss": 0.0873,
+      "step": 172
+    },
+    {
+      "epoch": 0.2593703148425787,
+      "grad_norm": 0.8701576619033993,
+      "learning_rate": 9.590734992481978e-06,
+      "loss": 0.0846,
+      "step": 173
+    },
+    {
+      "epoch": 0.2608695652173913,
+      "grad_norm": 0.9690850104939416,
+      "learning_rate": 9.586056507527266e-06,
+      "loss": 0.0852,
+      "step": 174
+    },
+    {
+      "epoch": 0.2623688155922039,
+      "grad_norm": 1.0446491604695451,
+      "learning_rate": 9.581352587833455e-06,
+      "loss": 0.0883,
+      "step": 175
+    },
+    {
+      "epoch": 0.2638680659670165,
+      "grad_norm": 1.2908149857728721,
+      "learning_rate": 9.576623259488966e-06,
+      "loss": 0.1262,
+      "step": 176
+    },
+    {
+      "epoch": 0.2653673163418291,
+      "grad_norm": 1.136810559856467,
+      "learning_rate": 9.571868548723137e-06,
+      "loss": 0.1285,
+      "step": 177
+    },
+    {
+      "epoch": 0.26686656671664166,
+      "grad_norm": 0.8656495549678687,
+      "learning_rate": 9.567088481906084e-06,
+      "loss": 0.0766,
+      "step": 178
+    },
+    {
+      "epoch": 0.2683658170914543,
+      "grad_norm": 0.7855762784987734,
+      "learning_rate": 9.562283085548546e-06,
+      "loss": 0.0754,
+      "step": 179
+    },
+    {
+      "epoch": 0.2698650674662669,
+      "grad_norm": 1.7632928995174546,
+      "learning_rate": 9.55745238630174e-06,
+      "loss": 0.1212,
+      "step": 180
+    },
+    {
+      "epoch": 0.27136431784107945,
+      "grad_norm": 0.9567395216094423,
+      "learning_rate": 9.552596410957224e-06,
+      "loss": 0.1108,
+      "step": 181
+    },
+    {
+      "epoch": 0.272863568215892,
+      "grad_norm": 0.8643765672478392,
+      "learning_rate": 9.547715186446732e-06,
+      "loss": 0.0802,
+      "step": 182
+    },
+    {
+      "epoch": 0.27436281859070466,
+      "grad_norm": 0.8713675732750616,
+      "learning_rate": 9.542808739842034e-06,
+      "loss": 0.0972,
+      "step": 183
+    },
+    {
+      "epoch": 0.27586206896551724,
+      "grad_norm": 2.140737300013267,
+      "learning_rate": 9.537877098354787e-06,
+      "loss": 0.0903,
+      "step": 184
+    },
+    {
+      "epoch": 0.2773613193403298,
+      "grad_norm": 0.8600779905716623,
+      "learning_rate": 9.532920289336378e-06,
+      "loss": 0.0855,
+      "step": 185
+    },
+    {
+      "epoch": 0.27886056971514245,
+      "grad_norm": 0.8079437055894714,
+      "learning_rate": 9.52793834027778e-06,
+      "loss": 0.0827,
+      "step": 186
+    },
+    {
+      "epoch": 0.280359820089955,
+      "grad_norm": 0.7898108293405106,
+      "learning_rate": 9.522931278809393e-06,
+      "loss": 0.0798,
+      "step": 187
+    },
+    {
+      "epoch": 0.2818590704647676,
+      "grad_norm": 0.8239694952483635,
+      "learning_rate": 9.517899132700889e-06,
+      "loss": 0.084,
+      "step": 188
+    },
+    {
+      "epoch": 0.28335832083958024,
+      "grad_norm": 0.8291374920212224,
+      "learning_rate": 9.512841929861069e-06,
+      "loss": 0.078,
+      "step": 189
+    },
+    {
+      "epoch": 0.2848575712143928,
+      "grad_norm": 0.8626076437339099,
+      "learning_rate": 9.507759698337698e-06,
+      "loss": 0.0845,
+      "step": 190
+    },
+    {
+      "epoch": 0.2863568215892054,
+      "grad_norm": 5.669277288990632,
+      "learning_rate": 9.50265246631735e-06,
+      "loss": 0.0873,
+      "step": 191
+    },
+    {
+      "epoch": 0.28785607196401797,
+      "grad_norm": 2.763258381603397,
+      "learning_rate": 9.49752026212526e-06,
+      "loss": 0.0844,
+      "step": 192
+    },
+    {
+      "epoch": 0.2893553223388306,
+      "grad_norm": 1.1132625950904278,
+      "learning_rate": 9.492363114225156e-06,
+      "loss": 0.1105,
+      "step": 193
+    },
+    {
+      "epoch": 0.2908545727136432,
+      "grad_norm": 0.6389451414822085,
+      "learning_rate": 9.487181051219107e-06,
+      "loss": 0.066,
+      "step": 194
+    },
+    {
+      "epoch": 0.29235382308845576,
+      "grad_norm": 0.8163160315098639,
+      "learning_rate": 9.481974101847371e-06,
+      "loss": 0.0753,
+      "step": 195
+    },
+    {
+      "epoch": 0.2938530734632684,
+      "grad_norm": 0.7465398067344403,
+      "learning_rate": 9.476742294988214e-06,
+      "loss": 0.0623,
+      "step": 196
+    },
+    {
+      "epoch": 0.29535232383808097,
+      "grad_norm": 0.9821729687781496,
+      "learning_rate": 9.471485659657782e-06,
+      "loss": 0.09,
+      "step": 197
+    },
+    {
+      "epoch": 0.29685157421289354,
+      "grad_norm": 0.9293593233681421,
+      "learning_rate": 9.466204225009905e-06,
+      "loss": 0.0813,
+      "step": 198
+    },
+    {
+      "epoch": 0.2983508245877061,
+      "grad_norm": 1.0042461369200688,
+      "learning_rate": 9.460898020335964e-06,
+      "loss": 0.0919,
+      "step": 199
+    },
+    {
+      "epoch": 0.29985007496251875,
+      "grad_norm": 0.7069059976685538,
+      "learning_rate": 9.455567075064715e-06,
+      "loss": 0.0585,
+      "step": 200
+    },
+    {
+      "epoch": 0.29985007496251875,
+      "eval_loss": 0.09797929227352142,
+      "eval_runtime": 8.1181,
+      "eval_samples_per_second": 6.652,
+      "eval_steps_per_second": 1.725,
+      "step": 200
+    },
+    {
+      "epoch": 0.30134932533733133,
+      "grad_norm": 0.8346950311468646,
+      "learning_rate": 9.450211418762123e-06,
+      "loss": 0.0635,
+      "step": 201
+    },
+    {
+      "epoch": 0.3028485757121439,
+      "grad_norm": 1.0157402015328638,
+      "learning_rate": 9.444831081131209e-06,
+      "loss": 0.0804,
+      "step": 202
+    },
+    {
+      "epoch": 0.30434782608695654,
+      "grad_norm": 1.3853833932960662,
+      "learning_rate": 9.439426092011877e-06,
+      "loss": 0.0872,
+      "step": 203
+    },
+    {
+      "epoch": 0.3058470764617691,
+      "grad_norm": 0.7960343930121441,
+      "learning_rate": 9.433996481380747e-06,
+      "loss": 0.0621,
+      "step": 204
+    },
+    {
+      "epoch": 0.3073463268365817,
+      "grad_norm": 1.1929012323116646,
+      "learning_rate": 9.428542279351e-06,
+      "loss": 0.1018,
+      "step": 205
+    },
+    {
+      "epoch": 0.30884557721139433,
+      "grad_norm": 0.9800134071848488,
+      "learning_rate": 9.423063516172195e-06,
+      "loss": 0.1005,
+      "step": 206
+    },
+    {
+      "epoch": 0.3103448275862069,
+      "grad_norm": 0.8817635577015389,
+      "learning_rate": 9.417560222230115e-06,
+      "loss": 0.0888,
+      "step": 207
+    },
+    {
+      "epoch": 0.3118440779610195,
+      "grad_norm": 0.9957694048785869,
+      "learning_rate": 9.412032428046594e-06,
+      "loss": 0.092,
+      "step": 208
+    },
+    {
+      "epoch": 0.31334332833583206,
+      "grad_norm": 0.9468088589076104,
+      "learning_rate": 9.40648016427934e-06,
+      "loss": 0.0944,
+      "step": 209
+    },
+    {
+      "epoch": 0.3148425787106447,
+      "grad_norm": 1.0196913558289538,
+      "learning_rate": 9.400903461721783e-06,
+      "loss": 0.0803,
+      "step": 210
+    },
+    {
+      "epoch": 0.31634182908545727,
+      "grad_norm": 1.202642172017592,
+      "learning_rate": 9.395302351302881e-06,
+      "loss": 0.0793,
+      "step": 211
+    },
+    {
+      "epoch": 0.31784107946026985,
+      "grad_norm": 1.1228588084803688,
+      "learning_rate": 9.38967686408697e-06,
+      "loss": 0.0992,
+      "step": 212
+    },
+    {
+      "epoch": 0.3193403298350825,
+      "grad_norm": 1.346604009190712,
+      "learning_rate": 9.384027031273575e-06,
+      "loss": 0.1046,
+      "step": 213
+    },
+    {
+      "epoch": 0.32083958020989506,
+      "grad_norm": 0.9304384068036033,
+      "learning_rate": 9.37835288419725e-06,
+      "loss": 0.0878,
+      "step": 214
+    },
+    {
+      "epoch": 0.32233883058470764,
+      "grad_norm": 1.084680461810212,
+      "learning_rate": 9.372654454327394e-06,
+      "loss": 0.1036,
+      "step": 215
+    },
+    {
+      "epoch": 0.3238380809595202,
+      "grad_norm": 0.806322210148517,
+      "learning_rate": 9.366931773268083e-06,
+      "loss": 0.0867,
+      "step": 216
+    },
+    {
+      "epoch": 0.32533733133433285,
+      "grad_norm": 1.1617064906370642,
+      "learning_rate": 9.361184872757894e-06,
+      "loss": 0.098,
+      "step": 217
+    },
+    {
+      "epoch": 0.3268365817091454,
+      "grad_norm": 0.8772869468427767,
+      "learning_rate": 9.355413784669722e-06,
+      "loss": 0.075,
+      "step": 218
+    },
+    {
+      "epoch": 0.328335832083958,
+      "grad_norm": 0.9601260549574697,
+      "learning_rate": 9.349618541010616e-06,
+      "loss": 0.0824,
+      "step": 219
+    },
+    {
+      "epoch": 0.32983508245877063,
+      "grad_norm": 1.0879196945348022,
+      "learning_rate": 9.343799173921591e-06,
+      "loss": 0.1084,
+      "step": 220
+    },
+    {
+      "epoch": 0.3313343328335832,
+      "grad_norm": 0.732439182228758,
+      "learning_rate": 9.337955715677452e-06,
+      "loss": 0.0749,
+      "step": 221
+    },
+    {
+      "epoch": 0.3328335832083958,
+      "grad_norm": 1.0488372838396354,
+      "learning_rate": 9.332088198686618e-06,
+      "loss": 0.111,
+      "step": 222
+    },
+    {
+      "epoch": 0.3343328335832084,
+      "grad_norm": 0.7375725305396468,
+      "learning_rate": 9.326196655490935e-06,
+      "loss": 0.0736,
+      "step": 223
+    },
+    {
+      "epoch": 0.335832083958021,
+      "grad_norm": 1.1045496717870615,
+      "learning_rate": 9.32028111876551e-06,
+      "loss": 0.0952,
+      "step": 224
+    },
+    {
+      "epoch": 0.3373313343328336,
+      "grad_norm": 0.7986957753943419,
+      "learning_rate": 9.314341621318512e-06,
+      "loss": 0.0947,
+      "step": 225
+    },
+    {
+      "epoch": 0.33883058470764615,
+      "grad_norm": 1.0329194552745675,
+      "learning_rate": 9.308378196091006e-06,
+      "loss": 0.1171,
+      "step": 226
+    },
+    {
+      "epoch": 0.3403298350824588,
+      "grad_norm": 0.9137019226709812,
+      "learning_rate": 9.302390876156756e-06,
+      "loss": 0.0846,
+      "step": 227
+    },
+    {
+      "epoch": 0.34182908545727136,
+      "grad_norm": 0.8755300723211273,
+      "learning_rate": 9.296379694722051e-06,
+      "loss": 0.0975,
+      "step": 228
+    },
+    {
+      "epoch": 0.34332833583208394,
+      "grad_norm": 0.8478781565172734,
+      "learning_rate": 9.29034468512552e-06,
+      "loss": 0.085,
+      "step": 229
+    },
+    {
+      "epoch": 0.3448275862068966,
+      "grad_norm": 0.794690537152289,
+      "learning_rate": 9.284285880837947e-06,
+      "loss": 0.079,
+      "step": 230
+    },
+    {
+      "epoch": 0.34632683658170915,
+      "grad_norm": 0.7216787540842711,
+      "learning_rate": 9.278203315462078e-06,
+      "loss": 0.0719,
+      "step": 231
+    },
+    {
+      "epoch": 0.34782608695652173,
+      "grad_norm": 0.6830544019411962,
+      "learning_rate": 9.272097022732444e-06,
+      "loss": 0.072,
+      "step": 232
+    },
+    {
+      "epoch": 0.3493253373313343,
+      "grad_norm": 0.8145584532504779,
+      "learning_rate": 9.26596703651517e-06,
+      "loss": 0.0727,
+      "step": 233
+    },
+    {
+      "epoch": 0.35082458770614694,
+      "grad_norm": 0.9622223942879006,
+      "learning_rate": 9.259813390807788e-06,
+      "loss": 0.1043,
+      "step": 234
+    },
+    {
+      "epoch": 0.3523238380809595,
+      "grad_norm": 0.9429220093692995,
+      "learning_rate": 9.253636119739046e-06,
+      "loss": 0.0873,
+      "step": 235
+    },
+    {
+      "epoch": 0.3538230884557721,
+      "grad_norm": 1.0005274096878416,
+      "learning_rate": 9.247435257568724e-06,
+      "loss": 0.0887,
+      "step": 236
+    },
+    {
+      "epoch": 0.3553223388305847,
+      "grad_norm": 0.9752893546664142,
+      "learning_rate": 9.241210838687438e-06,
+      "loss": 0.1003,
+      "step": 237
+    },
+    {
+      "epoch": 0.3568215892053973,
+      "grad_norm": 0.8641037708082397,
+      "learning_rate": 9.23496289761645e-06,
+      "loss": 0.082,
+      "step": 238
+    },
+    {
+      "epoch": 0.3583208395802099,
+      "grad_norm": 1.2588894522104943,
+      "learning_rate": 9.228691469007487e-06,
+      "loss": 0.1143,
+      "step": 239
+    },
+    {
+      "epoch": 0.3598200899550225,
+      "grad_norm": 0.7879952798047372,
+      "learning_rate": 9.222396587642528e-06,
+      "loss": 0.073,
+      "step": 240
+    },
+    {
+      "epoch": 0.3613193403298351,
+      "grad_norm": 0.8671188799482393,
+      "learning_rate": 9.216078288433632e-06,
+      "loss": 0.0851,
+      "step": 241
+    },
+    {
+      "epoch": 0.36281859070464767,
+      "grad_norm": 1.0181685095399846,
+      "learning_rate": 9.209736606422736e-06,
+      "loss": 0.0901,
+      "step": 242
+    },
+    {
+      "epoch": 0.36431784107946025,
+      "grad_norm": 0.9721649016411203,
+      "learning_rate": 9.203371576781457e-06,
+      "loss": 0.0891,
+      "step": 243
+    },
+    {
+      "epoch": 0.3658170914542729,
+      "grad_norm": 0.8909168453870353,
+      "learning_rate": 9.1969832348109e-06,
+      "loss": 0.0694,
+      "step": 244
+    },
+    {
+      "epoch": 0.36731634182908546,
+      "grad_norm": 0.8077545822691103,
+      "learning_rate": 9.190571615941462e-06,
+      "loss": 0.0814,
+      "step": 245
+    },
+    {
+      "epoch": 0.36881559220389803,
+      "grad_norm": 0.8859702642206748,
+      "learning_rate": 9.18413675573264e-06,
+      "loss": 0.0758,
+      "step": 246
+    },
+    {
+      "epoch": 0.37031484257871067,
+      "grad_norm": 1.2492238388703283,
+      "learning_rate": 9.177678689872831e-06,
+      "loss": 0.1145,
+      "step": 247
+    },
+    {
+      "epoch": 0.37181409295352325,
+      "grad_norm": 0.625577066948349,
+      "learning_rate": 9.171197454179124e-06,
+      "loss": 0.0609,
+      "step": 248
+    },
+    {
+      "epoch": 0.3733133433283358,
+      "grad_norm": 1.2587726927222609,
+      "learning_rate": 9.16469308459712e-06,
+      "loss": 0.087,
+      "step": 249
+    },
+    {
+      "epoch": 0.3748125937031484,
+      "grad_norm": 0.8055549709456843,
+      "learning_rate": 9.158165617200717e-06,
+      "loss": 0.0827,
+      "step": 250
+    },
+    {
+      "epoch": 0.37631184407796103,
+      "grad_norm": 1.0713779153704084,
+      "learning_rate": 9.151615088191918e-06,
+      "loss": 0.083,
+      "step": 251
+    },
+    {
+      "epoch": 0.3778110944527736,
+      "grad_norm": 0.6782148860509931,
+      "learning_rate": 9.14504153390063e-06,
+      "loss": 0.0764,
+      "step": 252
+    },
+    {
+      "epoch": 0.3793103448275862,
+      "grad_norm": 1.00735049872519,
+      "learning_rate": 9.138444990784455e-06,
+      "loss": 0.1093,
+      "step": 253
+    },
+    {
+      "epoch": 0.3808095952023988,
+      "grad_norm": 0.7782935718193602,
+      "learning_rate": 9.131825495428496e-06,
+      "loss": 0.0688,
+      "step": 254
+    },
+    {
+      "epoch": 0.3823088455772114,
+      "grad_norm": 0.964496113803125,
+      "learning_rate": 9.125183084545158e-06,
+      "loss": 0.1057,
+      "step": 255
+    },
+    {
+      "epoch": 0.383808095952024,
+      "grad_norm": 0.9285629961384954,
+      "learning_rate": 9.118517794973925e-06,
+      "loss": 0.0976,
+      "step": 256
+    },
+    {
+      "epoch": 0.3853073463268366,
+      "grad_norm": 0.847043741072453,
+      "learning_rate": 9.111829663681182e-06,
+      "loss": 0.0894,
+      "step": 257
+    },
+    {
+      "epoch": 0.3868065967016492,
+      "grad_norm": 1.0102214813260426,
+      "learning_rate": 9.105118727759984e-06,
+      "loss": 0.0928,
+      "step": 258
+    },
+    {
+      "epoch": 0.38830584707646176,
+      "grad_norm": 1.0197348253682816,
+      "learning_rate": 9.098385024429875e-06,
+      "loss": 0.1041,
+      "step": 259
+    },
+    {
+      "epoch": 0.38980509745127434,
+      "grad_norm": 0.6988348185172609,
+      "learning_rate": 9.09162859103666e-06,
+      "loss": 0.0672,
+      "step": 260
+    },
+    {
+      "epoch": 0.391304347826087,
+      "grad_norm": 0.9973444040171555,
+      "learning_rate": 9.08484946505221e-06,
+      "loss": 0.0946,
+      "step": 261
+    },
+    {
+      "epoch": 0.39280359820089955,
+      "grad_norm": 0.9398315039019636,
+      "learning_rate": 9.078047684074254e-06,
+      "loss": 0.0924,
+      "step": 262
+    },
+    {
+      "epoch": 0.39430284857571213,
+      "grad_norm": 0.8760589156476465,
+      "learning_rate": 9.071223285826166e-06,
+      "loss": 0.0961,
+      "step": 263
+    },
+    {
+      "epoch": 0.39580209895052476,
+      "grad_norm": 0.8106760962171756,
+      "learning_rate": 9.064376308156754e-06,
+      "loss": 0.0859,
+      "step": 264
+    },
+    {
+      "epoch": 0.39730134932533734,
+      "grad_norm": 0.8314884835881416,
+      "learning_rate": 9.057506789040063e-06,
+      "loss": 0.0896,
+      "step": 265
+    },
+    {
+      "epoch": 0.3988005997001499,
+      "grad_norm": 0.8681713788497047,
+      "learning_rate": 9.050614766575147e-06,
+      "loss": 0.0981,
+      "step": 266
+    },
+    {
+      "epoch": 0.4002998500749625,
+      "grad_norm": 0.7489698727669879,
+      "learning_rate": 9.043700278985867e-06,
+      "loss": 0.072,
+      "step": 267
+    },
+    {
+      "epoch": 0.4017991004497751,
+      "grad_norm": 0.8493299975286843,
+      "learning_rate": 9.03676336462068e-06,
+      "loss": 0.0991,
+      "step": 268
+    },
+    {
+      "epoch": 0.4032983508245877,
+      "grad_norm": 0.7684654660855086,
+      "learning_rate": 9.029804061952426e-06,
+      "loss": 0.0816,
+      "step": 269
+    },
+    {
+      "epoch": 0.4047976011994003,
+      "grad_norm": 0.6225703007910903,
+      "learning_rate": 9.022822409578106e-06,
+      "loss": 0.0688,
+      "step": 270
+    },
+    {
+      "epoch": 0.4062968515742129,
+      "grad_norm": 0.660657471850874,
+      "learning_rate": 9.015818446218683e-06,
+      "loss": 0.0611,
+      "step": 271
+    },
+    {
+      "epoch": 0.4077961019490255,
+      "grad_norm": 1.0106639304425744,
+      "learning_rate": 9.008792210718854e-06,
+      "loss": 0.1114,
+      "step": 272
+    },
+    {
+      "epoch": 0.40929535232383807,
+      "grad_norm": 1.1964871789844203,
+      "learning_rate": 9.00174374204684e-06,
+      "loss": 0.1285,
+      "step": 273
+    },
+    {
+      "epoch": 0.4107946026986507,
+      "grad_norm": 0.9184483129996951,
+      "learning_rate": 8.994673079294171e-06,
+      "loss": 0.0936,
+      "step": 274
+    },
+    {
+      "epoch": 0.4122938530734633,
+      "grad_norm": 0.8865359640070075,
+      "learning_rate": 8.987580261675466e-06,
+      "loss": 0.0772,
+      "step": 275
+    },
+    {
+      "epoch": 0.41379310344827586,
+      "grad_norm": 0.9567877522003722,
+      "learning_rate": 8.98046532852822e-06,
+      "loss": 0.0914,
+      "step": 276
+    },
+    {
+      "epoch": 0.41529235382308843,
+      "grad_norm": 0.9202144720052845,
+      "learning_rate": 8.973328319312577e-06,
+      "loss": 0.0702,
+      "step": 277
+    },
+    {
+      "epoch": 0.41679160419790107,
+      "grad_norm": 0.7228688851308325,
+      "learning_rate": 8.966169273611125e-06,
+      "loss": 0.0823,
+      "step": 278
+    },
+    {
+      "epoch": 0.41829085457271364,
+      "grad_norm": 0.9874286048983996,
+      "learning_rate": 8.958988231128665e-06,
+      "loss": 0.1023,
+      "step": 279
+    },
+    {
+      "epoch": 0.4197901049475262,
+      "grad_norm": 0.781230691127796,
+      "learning_rate": 8.95178523169199e-06,
+      "loss": 0.0864,
+      "step": 280
+    },
+    {
+      "epoch": 0.42128935532233885,
+      "grad_norm": 0.7460790166709771,
+      "learning_rate": 8.944560315249676e-06,
+      "loss": 0.0669,
+      "step": 281
+    },
+    {
+      "epoch": 0.42278860569715143,
+      "grad_norm": 0.8041000653617603,
+      "learning_rate": 8.937313521871846e-06,
+      "loss": 0.085,
+      "step": 282
+    },
+    {
+      "epoch": 0.424287856071964,
+      "grad_norm": 0.8828435905469374,
+      "learning_rate": 8.930044891749962e-06,
+      "loss": 0.09,
+      "step": 283
+    },
+    {
+      "epoch": 0.4257871064467766,
+      "grad_norm": 0.9671451262759431,
+      "learning_rate": 8.922754465196591e-06,
+      "loss": 0.1156,
+      "step": 284
+    },
+    {
+      "epoch": 0.4272863568215892,
+      "grad_norm": 0.6420944490715579,
+      "learning_rate": 8.915442282645183e-06,
+      "loss": 0.0692,
+      "step": 285
+    },
+    {
+      "epoch": 0.4287856071964018,
+      "grad_norm": 0.8241523811807334,
+      "learning_rate": 8.908108384649856e-06,
+      "loss": 0.086,
+      "step": 286
+    },
+    {
+      "epoch": 0.4302848575712144,
+      "grad_norm": 0.9719980575109428,
+      "learning_rate": 8.900752811885152e-06,
+      "loss": 0.1062,
+      "step": 287
+    },
+    {
+      "epoch": 0.431784107946027,
+      "grad_norm": 0.6903171116615955,
+      "learning_rate": 8.893375605145837e-06,
+      "loss": 0.0685,
+      "step": 288
+    },
+    {
+      "epoch": 0.4332833583208396,
+      "grad_norm": 0.8576286500559104,
+      "learning_rate": 8.885976805346651e-06,
+      "loss": 0.1014,
+      "step": 289
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 0.7901130062556169,
+      "learning_rate": 8.8785564535221e-06,
+      "loss": 0.0786,
+      "step": 290
+    },
+    {
+      "epoch": 0.4362818590704648,
+      "grad_norm": 0.7787175182755985,
+      "learning_rate": 8.871114590826211e-06,
+      "loss": 0.0818,
+      "step": 291
+    },
+    {
+      "epoch": 0.43778110944527737,
+      "grad_norm": 1.2958543457689293,
+      "learning_rate": 8.86365125853232e-06,
+      "loss": 0.1087,
+      "step": 292
+    },
+    {
+      "epoch": 0.43928035982008995,
+      "grad_norm": 0.8318895908350302,
+      "learning_rate": 8.85616649803283e-06,
+      "loss": 0.085,
+      "step": 293
+    },
+    {
+      "epoch": 0.4407796101949025,
+      "grad_norm": 0.6157474875141361,
+      "learning_rate": 8.84866035083899e-06,
+      "loss": 0.0693,
+      "step": 294
+    },
+    {
+      "epoch": 0.44227886056971516,
+      "grad_norm": 0.9675717561900218,
+      "learning_rate": 8.841132858580661e-06,
+      "loss": 0.1056,
+      "step": 295
+    },
+    {
+      "epoch": 0.44377811094452774,
+      "grad_norm": 0.7600222728626678,
+      "learning_rate": 8.833584063006088e-06,
+      "loss": 0.0857,
+      "step": 296
+    },
+    {
+      "epoch": 0.4452773613193403,
+      "grad_norm": 0.9436040108499784,
+      "learning_rate": 8.826014005981662e-06,
+      "loss": 0.0829,
+      "step": 297
+    },
+    {
+      "epoch": 0.44677661169415295,
+      "grad_norm": 0.9392696703861476,
+      "learning_rate": 8.818422729491693e-06,
+      "loss": 0.0862,
+      "step": 298
+    },
+    {
+      "epoch": 0.4482758620689655,
+      "grad_norm": 0.8857711982078132,
+      "learning_rate": 8.810810275638183e-06,
+      "loss": 0.0987,
+      "step": 299
+    },
+    {
+      "epoch": 0.4497751124437781,
+      "grad_norm": 0.7373907411753713,
+      "learning_rate": 8.803176686640577e-06,
+      "loss": 0.0858,
+      "step": 300
+    },
+    {
+      "epoch": 0.4512743628185907,
+      "grad_norm": 1.0144127404906087,
+      "learning_rate": 8.795522004835543e-06,
+      "loss": 0.1083,
+      "step": 301
+    },
+    {
+      "epoch": 0.4527736131934033,
+      "grad_norm": 0.8217782622294467,
+      "learning_rate": 8.787846272676728e-06,
+      "loss": 0.0747,
+      "step": 302
+    },
+    {
+      "epoch": 0.4542728635682159,
+      "grad_norm": 0.7052740489545154,
+      "learning_rate": 8.780149532734531e-06,
+      "loss": 0.0672,
+      "step": 303
+    },
+    {
+      "epoch": 0.45577211394302847,
+      "grad_norm": 0.8314799318186591,
+      "learning_rate": 8.772431827695862e-06,
+      "loss": 0.0892,
+      "step": 304
+    },
+    {
+      "epoch": 0.4572713643178411,
+      "grad_norm": 0.8909823084770626,
+      "learning_rate": 8.764693200363897e-06,
+      "loss": 0.088,
+      "step": 305
+    },
+    {
+      "epoch": 0.4587706146926537,
+      "grad_norm": 0.7124231669044098,
+      "learning_rate": 8.756933693657863e-06,
+      "loss": 0.0758,
+      "step": 306
+    },
+    {
+      "epoch": 0.46026986506746626,
+      "grad_norm": 0.816782274783127,
+      "learning_rate": 8.749153350612774e-06,
+      "loss": 0.0912,
+      "step": 307
+    },
+    {
+      "epoch": 0.4617691154422789,
+      "grad_norm": 0.8547908514920887,
+      "learning_rate": 8.74135221437921e-06,
+      "loss": 0.0911,
+      "step": 308
+    },
+    {
+      "epoch": 0.46326836581709147,
+      "grad_norm": 0.717301363007696,
+      "learning_rate": 8.733530328223076e-06,
+      "loss": 0.0729,
+      "step": 309
+    },
+    {
+      "epoch": 0.46476761619190404,
+      "grad_norm": 0.6778815816838354,
+      "learning_rate": 8.725687735525347e-06,
+      "loss": 0.07,
+      "step": 310
+    },
+    {
+      "epoch": 0.4662668665667166,
+      "grad_norm": 1.0275185104608529,
+      "learning_rate": 8.71782447978185e-06,
+      "loss": 0.0996,
+      "step": 311
+    },
+    {
+      "epoch": 0.46776611694152925,
+      "grad_norm": 0.918024975588489,
+      "learning_rate": 8.709940604603006e-06,
+      "loss": 0.1078,
+      "step": 312
+    },
+    {
+      "epoch": 0.46926536731634183,
+      "grad_norm": 0.868160463907923,
+      "learning_rate": 8.702036153713594e-06,
+      "loss": 0.0908,
+      "step": 313
+    },
+    {
+      "epoch": 0.4707646176911544,
+      "grad_norm": 0.9593524760273816,
+      "learning_rate": 8.694111170952508e-06,
+      "loss": 0.1052,
+      "step": 314
+    },
+    {
+      "epoch": 0.47226386806596704,
+      "grad_norm": 0.7007726212272783,
+      "learning_rate": 8.686165700272513e-06,
+      "loss": 0.0798,
+      "step": 315
+    },
+    {
+      "epoch": 0.4737631184407796,
+      "grad_norm": 0.9317453458475099,
+      "learning_rate": 8.678199785740003e-06,
+      "loss": 0.0979,
+      "step": 316
+    },
+    {
+      "epoch": 0.4752623688155922,
+      "grad_norm": 0.7428126038726307,
+      "learning_rate": 8.670213471534759e-06,
+      "loss": 0.0842,
+      "step": 317
+    },
+    {
+      "epoch": 0.4767616191904048,
+      "grad_norm": 0.8261459727268365,
+      "learning_rate": 8.662206801949694e-06,
+      "loss": 0.0872,
+      "step": 318
+    },
+    {
+      "epoch": 0.4782608695652174,
+      "grad_norm": 0.6986348791814172,
+      "learning_rate": 8.65417982139062e-06,
+      "loss": 0.0755,
+      "step": 319
+    },
+    {
+      "epoch": 0.47976011994003,
+      "grad_norm": 0.8084229229543242,
+      "learning_rate": 8.646132574375994e-06,
+      "loss": 0.0816,
+      "step": 320
+    },
+    {
+      "epoch": 0.48125937031484256,
+      "grad_norm": 1.0622250592663878,
+      "learning_rate": 8.638065105536669e-06,
+      "loss": 0.1196,
+      "step": 321
+    },
+    {
+      "epoch": 0.4827586206896552,
+      "grad_norm": 0.7994934894932665,
+      "learning_rate": 8.629977459615655e-06,
+      "loss": 0.0855,
+      "step": 322
+    },
+    {
+      "epoch": 0.48425787106446777,
+      "grad_norm": 0.8707560142592776,
+      "learning_rate": 8.621869681467865e-06,
+      "loss": 0.0935,
+      "step": 323
+    },
+    {
+      "epoch": 0.48575712143928035,
+      "grad_norm": 0.6820273941191735,
+      "learning_rate": 8.613741816059867e-06,
+      "loss": 0.0695,
+      "step": 324
+    },
+    {
+      "epoch": 0.487256371814093,
+      "grad_norm": 0.9981393070141752,
+      "learning_rate": 8.605593908469635e-06,
+      "loss": 0.0852,
+      "step": 325
+    },
+    {
+      "epoch": 0.48875562218890556,
+      "grad_norm": 0.7405553963761845,
+      "learning_rate": 8.597426003886295e-06,
+      "loss": 0.084,
+      "step": 326
+    },
+    {
+      "epoch": 0.49025487256371814,
+      "grad_norm": 0.8612381269818579,
+      "learning_rate": 8.58923814760989e-06,
+      "loss": 0.0908,
+      "step": 327
+    },
+    {
+      "epoch": 0.4917541229385307,
+      "grad_norm": 0.6779752361188751,
+      "learning_rate": 8.581030385051105e-06,
+      "loss": 0.0757,
+      "step": 328
+    },
+    {
+      "epoch": 0.49325337331334335,
+      "grad_norm": 0.7398209314300788,
+      "learning_rate": 8.572802761731031e-06,
+      "loss": 0.0807,
+      "step": 329
+    },
+    {
+      "epoch": 0.4947526236881559,
+      "grad_norm": 1.126713053031775,
+      "learning_rate": 8.564555323280913e-06,
+      "loss": 0.1289,
+      "step": 330
+    },
+    {
+      "epoch": 0.4962518740629685,
+      "grad_norm": 0.787748256949515,
+      "learning_rate": 8.556288115441887e-06,
+      "loss": 0.0758,
+      "step": 331
+    },
+    {
+      "epoch": 0.49775112443778113,
+      "grad_norm": 0.7880491754211475,
+      "learning_rate": 8.548001184064733e-06,
+      "loss": 0.0846,
+      "step": 332
+    },
+    {
+      "epoch": 0.4992503748125937,
+      "grad_norm": 0.7640121528017574,
+      "learning_rate": 8.539694575109626e-06,
+      "loss": 0.0651,
+      "step": 333
+    },
+    {
+      "epoch": 0.5007496251874063,
+      "grad_norm": 0.6975832774216282,
+      "learning_rate": 8.531368334645865e-06,
+      "loss": 0.0798,
+      "step": 334
+    },
+    {
+      "epoch": 0.5022488755622189,
+      "grad_norm": 0.7270594844347897,
+      "learning_rate": 8.523022508851634e-06,
+      "loss": 0.0938,
+      "step": 335
+    },
+    {
+      "epoch": 0.5037481259370314,
+      "grad_norm": 0.7145526926432068,
+      "learning_rate": 8.514657144013738e-06,
+      "loss": 0.0756,
+      "step": 336
+    },
+    {
+      "epoch": 0.5052473763118441,
+      "grad_norm": 0.7763902169661998,
+      "learning_rate": 8.506272286527346e-06,
+      "loss": 0.0732,
+      "step": 337
+    },
+    {
+      "epoch": 0.5067466266866567,
+      "grad_norm": 0.7732819903553882,
+      "learning_rate": 8.497867982895741e-06,
+      "loss": 0.0741,
+      "step": 338
+    },
+    {
+      "epoch": 0.5082458770614693,
+      "grad_norm": 0.713777574582036,
+      "learning_rate": 8.489444279730046e-06,
+      "loss": 0.0763,
+      "step": 339
+    },
+    {
+      "epoch": 0.5097451274362819,
+      "grad_norm": 0.7115555422419503,
+      "learning_rate": 8.481001223748986e-06,
+      "loss": 0.0552,
+      "step": 340
+    },
+    {
+      "epoch": 0.5112443778110944,
+      "grad_norm": 0.849718799495384,
+      "learning_rate": 8.47253886177861e-06,
+      "loss": 0.0993,
+      "step": 341
+    },
+    {
+      "epoch": 0.512743628185907,
+      "grad_norm": 0.95763332801373,
+      "learning_rate": 8.464057240752046e-06,
+      "loss": 0.0976,
+      "step": 342
+    },
+    {
+      "epoch": 0.5142428785607196,
+      "grad_norm": 0.7670102475357184,
+      "learning_rate": 8.455556407709235e-06,
+      "loss": 0.0826,
+      "step": 343
+    },
+    {
+      "epoch": 0.5157421289355323,
+      "grad_norm": 0.7945131864992991,
+      "learning_rate": 8.447036409796663e-06,
+      "loss": 0.0876,
+      "step": 344
+    },
+    {
+      "epoch": 0.5172413793103449,
+      "grad_norm": 0.7607254847313331,
+      "learning_rate": 8.438497294267117e-06,
+      "loss": 0.0835,
+      "step": 345
+    },
+    {
+      "epoch": 0.5187406296851574,
+      "grad_norm": 0.8672653262889671,
+      "learning_rate": 8.429939108479403e-06,
+      "loss": 0.0844,
+      "step": 346
+    },
+    {
+      "epoch": 0.52023988005997,
+      "grad_norm": 0.8585360495206352,
+      "learning_rate": 8.421361899898095e-06,
+      "loss": 0.0816,
+      "step": 347
+    },
+    {
+      "epoch": 0.5217391304347826,
+      "grad_norm": 0.7044316512437083,
+      "learning_rate": 8.412765716093273e-06,
+      "loss": 0.0862,
+      "step": 348
+    },
+    {
+      "epoch": 0.5232383808095952,
+      "grad_norm": 0.8839958259170885,
+      "learning_rate": 8.404150604740248e-06,
+      "loss": 0.11,
+      "step": 349
+    },
+    {
+      "epoch": 0.5247376311844077,
+      "grad_norm": 0.8033883540498372,
+      "learning_rate": 8.395516613619315e-06,
+      "loss": 0.0744,
+      "step": 350
+    },
+    {
+      "epoch": 0.5262368815592204,
+      "grad_norm": 0.553454206942328,
+      "learning_rate": 8.386863790615472e-06,
+      "loss": 0.0576,
+      "step": 351
+    },
+    {
+      "epoch": 0.527736131934033,
+      "grad_norm": 0.8438383009068628,
+      "learning_rate": 8.378192183718158e-06,
+      "loss": 0.0964,
+      "step": 352
+    },
+    {
+      "epoch": 0.5292353823088456,
+      "grad_norm": 0.9920499202315903,
+      "learning_rate": 8.369501841021e-06,
+      "loss": 0.1036,
+      "step": 353
+    },
+    {
+      "epoch": 0.5307346326836582,
+      "grad_norm": 0.6109948992615272,
+      "learning_rate": 8.360792810721522e-06,
+      "loss": 0.0742,
+      "step": 354
+    },
+    {
+      "epoch": 0.5322338830584707,
+      "grad_norm": 0.8749266072773508,
+      "learning_rate": 8.352065141120902e-06,
+      "loss": 0.1014,
+      "step": 355
+    },
+    {
+      "epoch": 0.5337331334332833,
+      "grad_norm": 0.9245262937343904,
+      "learning_rate": 8.343318880623688e-06,
+      "loss": 0.0936,
+      "step": 356
+    },
+    {
+      "epoch": 0.5352323838080959,
+      "grad_norm": 0.882396710440555,
+      "learning_rate": 8.334554077737535e-06,
+      "loss": 0.0967,
+      "step": 357
+    },
+    {
+      "epoch": 0.5367316341829086,
+      "grad_norm": 0.9030125671711826,
+      "learning_rate": 8.325770781072939e-06,
+      "loss": 0.077,
+      "step": 358
+    },
+    {
+      "epoch": 0.5382308845577212,
+      "grad_norm": 0.6284179594634367,
+      "learning_rate": 8.316969039342963e-06,
+      "loss": 0.0743,
+      "step": 359
+    },
+    {
+      "epoch": 0.5397301349325337,
+      "grad_norm": 0.8311773289262309,
+      "learning_rate": 8.30814890136297e-06,
+      "loss": 0.101,
+      "step": 360
+    },
+    {
+      "epoch": 0.5412293853073463,
+      "grad_norm": 0.916130110958747,
+      "learning_rate": 8.299310416050345e-06,
+      "loss": 0.1204,
+      "step": 361
+    },
+    {
+      "epoch": 0.5427286356821589,
+      "grad_norm": 0.7033764421642134,
+      "learning_rate": 8.290453632424236e-06,
+      "loss": 0.0865,
+      "step": 362
+    },
+    {
+      "epoch": 0.5442278860569715,
+      "grad_norm": 1.0571799403058666,
+      "learning_rate": 8.281578599605269e-06,
+      "loss": 0.1033,
+      "step": 363
+    },
+    {
+      "epoch": 0.545727136431784,
+      "grad_norm": 0.9865596980059378,
+      "learning_rate": 8.272685366815287e-06,
+      "loss": 0.1153,
+      "step": 364
+    },
+    {
+      "epoch": 0.5472263868065967,
+      "grad_norm": 0.842257938344622,
+      "learning_rate": 8.26377398337707e-06,
+      "loss": 0.0813,
+      "step": 365
+    },
+    {
+      "epoch": 0.5487256371814093,
+      "grad_norm": 0.7801266484383078,
+      "learning_rate": 8.254844498714063e-06,
+      "loss": 0.0885,
+      "step": 366
+    },
+    {
+      "epoch": 0.5502248875562219,
+      "grad_norm": 0.79916042427441,
+      "learning_rate": 8.2458969623501e-06,
+      "loss": 0.0981,
+      "step": 367
+    },
+    {
+      "epoch": 0.5517241379310345,
+      "grad_norm": 0.7402662883095835,
+      "learning_rate": 8.23693142390914e-06,
+      "loss": 0.0875,
+      "step": 368
+    },
+    {
+      "epoch": 0.553223388305847,
+      "grad_norm": 0.8459094044233658,
+      "learning_rate": 8.227947933114971e-06,
+      "loss": 0.098,
+      "step": 369
+    },
+    {
+      "epoch": 0.5547226386806596,
+      "grad_norm": 0.8547179044918303,
+      "learning_rate": 8.218946539790957e-06,
+      "loss": 0.095,
+      "step": 370
+    },
+    {
+      "epoch": 0.5562218890554723,
+      "grad_norm": 0.7292163049064962,
+      "learning_rate": 8.209927293859746e-06,
+      "loss": 0.0851,
+      "step": 371
+    },
+    {
+      "epoch": 0.5577211394302849,
+      "grad_norm": 0.7343860351758368,
+      "learning_rate": 8.200890245342999e-06,
+      "loss": 0.0855,
+      "step": 372
+    },
+    {
+      "epoch": 0.5592203898050975,
+      "grad_norm": 1.0512438193559215,
+      "learning_rate": 8.191835444361113e-06,
+      "loss": 0.1282,
+      "step": 373
+    },
+    {
+      "epoch": 0.56071964017991,
+      "grad_norm": 0.6866930433736815,
+      "learning_rate": 8.182762941132944e-06,
+      "loss": 0.0712,
+      "step": 374
+    },
+    {
+      "epoch": 0.5622188905547226,
+      "grad_norm": 0.693144949411856,
+      "learning_rate": 8.173672785975522e-06,
+      "loss": 0.0664,
+      "step": 375
+    },
+    {
+      "epoch": 0.5637181409295352,
+      "grad_norm": 0.8200547989526404,
+      "learning_rate": 8.16456502930378e-06,
+      "loss": 0.0973,
+      "step": 376
+    },
+    {
+      "epoch": 0.5652173913043478,
+      "grad_norm": 0.8160855991831476,
+      "learning_rate": 8.155439721630265e-06,
+      "loss": 0.1014,
+      "step": 377
+    },
+    {
+      "epoch": 0.5667166416791605,
+      "grad_norm": 0.6948114047042937,
+      "learning_rate": 8.146296913564872e-06,
+      "loss": 0.0786,
+      "step": 378
+    },
+    {
+      "epoch": 0.568215892053973,
+      "grad_norm": 0.7453219285726429,
+      "learning_rate": 8.13713665581455e-06,
+      "loss": 0.0873,
+      "step": 379
+    },
+    {
+      "epoch": 0.5697151424287856,
+      "grad_norm": 0.6689061215559876,
+      "learning_rate": 8.127958999183027e-06,
+      "loss": 0.077,
+      "step": 380
+    },
+    {
+      "epoch": 0.5712143928035982,
+      "grad_norm": 0.7034705793903916,
+      "learning_rate": 8.118763994570528e-06,
+      "loss": 0.0666,
+      "step": 381
+    },
+    {
+      "epoch": 0.5727136431784108,
+      "grad_norm": 0.793373412732906,
+      "learning_rate": 8.109551692973487e-06,
+      "loss": 0.0896,
+      "step": 382
+    },
+    {
+      "epoch": 0.5742128935532234,
+      "grad_norm": 1.0515121788111412,
+      "learning_rate": 8.100322145484275e-06,
+      "loss": 0.1069,
+      "step": 383
+    },
+    {
+      "epoch": 0.5757121439280359,
+      "grad_norm": 0.8358541993809073,
+      "learning_rate": 8.091075403290905e-06,
+      "loss": 0.0934,
+      "step": 384
+    },
+    {
+      "epoch": 0.5772113943028486,
+      "grad_norm": 0.8659953433399471,
+      "learning_rate": 8.081811517676759e-06,
+      "loss": 0.0969,
+      "step": 385
+    },
+    {
+      "epoch": 0.5787106446776612,
+      "grad_norm": 0.7082320077656983,
+      "learning_rate": 8.072530540020294e-06,
+      "loss": 0.0791,
+      "step": 386
+    },
+    {
+      "epoch": 0.5802098950524738,
+      "grad_norm": 0.890957674375861,
+      "learning_rate": 8.063232521794762e-06,
+      "loss": 0.0982,
+      "step": 387
+    },
+    {
+      "epoch": 0.5817091454272864,
+      "grad_norm": 0.6041263951900956,
+      "learning_rate": 8.053917514567927e-06,
+      "loss": 0.0708,
+      "step": 388
+    },
+    {
+      "epoch": 0.5832083958020989,
+      "grad_norm": 0.7208263761126418,
+      "learning_rate": 8.04458557000177e-06,
+      "loss": 0.0758,
+      "step": 389
+    },
+    {
+      "epoch": 0.5847076461769115,
+      "grad_norm": 0.9369171110274785,
+      "learning_rate": 8.035236739852214e-06,
+      "loss": 0.1039,
+      "step": 390
+    },
+    {
+      "epoch": 0.5862068965517241,
+      "grad_norm": 0.6059040935664246,
+      "learning_rate": 8.025871075968828e-06,
+      "loss": 0.0583,
+      "step": 391
+    },
+    {
+      "epoch": 0.5877061469265368,
+      "grad_norm": 0.7334586797995069,
+      "learning_rate": 8.016488630294539e-06,
+      "loss": 0.0736,
+      "step": 392
+    },
+    {
+      "epoch": 0.5892053973013494,
+      "grad_norm": 0.8619212046706786,
+      "learning_rate": 8.007089454865358e-06,
+      "loss": 0.0819,
+      "step": 393
+    },
+    {
+      "epoch": 0.5907046476761619,
+      "grad_norm": 0.8590134886198357,
+      "learning_rate": 7.997673601810071e-06,
+      "loss": 0.0903,
+      "step": 394
+    },
+    {
+      "epoch": 0.5922038980509745,
+      "grad_norm": 0.6222252052444753,
+      "learning_rate": 7.988241123349965e-06,
+      "loss": 0.0683,
+      "step": 395
+    },
+    {
+      "epoch": 0.5937031484257871,
+      "grad_norm": 0.9345362748975566,
+      "learning_rate": 7.97879207179853e-06,
+      "loss": 0.1208,
+      "step": 396
+    },
+    {
+      "epoch": 0.5952023988005997,
+      "grad_norm": 0.7001732037572929,
+      "learning_rate": 7.969326499561173e-06,
+      "loss": 0.0841,
+      "step": 397
+    },
+    {
+      "epoch": 0.5967016491754122,
+      "grad_norm": 0.740202391154865,
+      "learning_rate": 7.95984445913493e-06,
+      "loss": 0.0809,
+      "step": 398
+    },
+    {
+      "epoch": 0.5982008995502249,
+      "grad_norm": 0.6709001947309992,
+      "learning_rate": 7.950346003108167e-06,
+      "loss": 0.075,
+      "step": 399
+    },
+    {
+      "epoch": 0.5997001499250375,
+      "grad_norm": 0.6699901878882981,
+      "learning_rate": 7.940831184160294e-06,
+      "loss": 0.0732,
+      "step": 400
+    },
+    {
+      "epoch": 0.5997001499250375,
+      "eval_loss": 0.09627355635166168,
+      "eval_runtime": 8.1045,
+      "eval_samples_per_second": 6.663,
+      "eval_steps_per_second": 1.727,
+      "step": 400
+    },
+    {
+      "epoch": 0.6011994002998501,
+      "grad_norm": 0.5470136779802762,
+      "learning_rate": 7.93130005506147e-06,
+      "loss": 0.0595,
+      "step": 401
+    },
+    {
+      "epoch": 0.6026986506746627,
+      "grad_norm": 0.7201758599744614,
+      "learning_rate": 7.921752668672316e-06,
+      "loss": 0.0947,
+      "step": 402
+    },
+    {
+      "epoch": 0.6041979010494752,
+      "grad_norm": 0.7429211888031387,
+      "learning_rate": 7.912189077943613e-06,
+      "loss": 0.0869,
+      "step": 403
+    },
+    {
+      "epoch": 0.6056971514242878,
+      "grad_norm": 0.8520710353490958,
+      "learning_rate": 7.902609335916015e-06,
+      "loss": 0.0789,
+      "step": 404
+    },
+    {
+      "epoch": 0.6071964017991005,
+      "grad_norm": 0.8340043144065545,
+      "learning_rate": 7.893013495719752e-06,
+      "loss": 0.0865,
+      "step": 405
+    },
+    {
+      "epoch": 0.6086956521739131,
+      "grad_norm": 0.8671572430752748,
+      "learning_rate": 7.883401610574338e-06,
+      "loss": 0.0929,
+      "step": 406
+    },
+    {
+      "epoch": 0.6101949025487257,
+      "grad_norm": 0.867761252148514,
+      "learning_rate": 7.873773733788268e-06,
+      "loss": 0.1013,
+      "step": 407
+    },
+    {
+      "epoch": 0.6116941529235382,
+      "grad_norm": 1.1340118926128642,
+      "learning_rate": 7.864129918758738e-06,
+      "loss": 0.1132,
+      "step": 408
+    },
+    {
+      "epoch": 0.6131934032983508,
+      "grad_norm": 0.7735660852223165,
+      "learning_rate": 7.854470218971333e-06,
+      "loss": 0.0899,
+      "step": 409
+    },
+    {
+      "epoch": 0.6146926536731634,
+      "grad_norm": 0.8641461625090249,
+      "learning_rate": 7.844794687999737e-06,
+      "loss": 0.0973,
+      "step": 410
+    },
+    {
+      "epoch": 0.616191904047976,
+      "grad_norm": 0.9223604320988085,
+      "learning_rate": 7.835103379505433e-06,
+      "loss": 0.1031,
+      "step": 411
+    },
+    {
+      "epoch": 0.6176911544227887,
+      "grad_norm": 0.6564052193473546,
+      "learning_rate": 7.825396347237413e-06,
+      "loss": 0.0589,
+      "step": 412
+    },
+    {
+      "epoch": 0.6191904047976012,
+      "grad_norm": 0.7758410933004322,
+      "learning_rate": 7.815673645031871e-06,
+      "loss": 0.0767,
+      "step": 413
+    },
+    {
+      "epoch": 0.6206896551724138,
+      "grad_norm": 0.7000826676230929,
+      "learning_rate": 7.805935326811913e-06,
+      "loss": 0.0697,
+      "step": 414
+    },
+    {
+      "epoch": 0.6221889055472264,
+      "grad_norm": 0.6202500859620549,
+      "learning_rate": 7.796181446587244e-06,
+      "loss": 0.06,
+      "step": 415
+    },
+    {
+      "epoch": 0.623688155922039,
+      "grad_norm": 0.9455528145415997,
+      "learning_rate": 7.786412058453886e-06,
+      "loss": 0.1086,
+      "step": 416
+    },
+    {
+      "epoch": 0.6251874062968515,
+      "grad_norm": 0.7633667907111196,
+      "learning_rate": 7.776627216593863e-06,
+      "loss": 0.0763,
+      "step": 417
+    },
+    {
+      "epoch": 0.6266866566716641,
+      "grad_norm": 0.5937272800567256,
+      "learning_rate": 7.766826975274916e-06,
+      "loss": 0.0759,
+      "step": 418
+    },
+    {
+      "epoch": 0.6281859070464768,
+      "grad_norm": 1.5979866516645662,
+      "learning_rate": 7.75701138885018e-06,
+      "loss": 0.1501,
+      "step": 419
+    },
+    {
+      "epoch": 0.6296851574212894,
+      "grad_norm": 0.9403199855807128,
+      "learning_rate": 7.747180511757908e-06,
+      "loss": 0.0924,
+      "step": 420
+    },
+    {
+      "epoch": 0.631184407796102,
+      "grad_norm": 0.6656064579805654,
+      "learning_rate": 7.737334398521149e-06,
+      "loss": 0.0747,
+      "step": 421
+    },
+    {
+      "epoch": 0.6326836581709145,
+      "grad_norm": 0.8955028022661206,
+      "learning_rate": 7.727473103747456e-06,
+      "loss": 0.1113,
+      "step": 422
+    },
+    {
+      "epoch": 0.6341829085457271,
+      "grad_norm": 0.7757018959756083,
+      "learning_rate": 7.717596682128578e-06,
+      "loss": 0.0795,
+      "step": 423
+    },
+    {
+      "epoch": 0.6356821589205397,
+      "grad_norm": 0.817576588272353,
+      "learning_rate": 7.707705188440165e-06,
+      "loss": 0.0894,
+      "step": 424
+    },
+    {
+      "epoch": 0.6371814092953523,
+      "grad_norm": 0.9432693394414009,
+      "learning_rate": 7.697798677541448e-06,
+      "loss": 0.1173,
+      "step": 425
+    },
+    {
+      "epoch": 0.638680659670165,
+      "grad_norm": 0.8938827851341684,
+      "learning_rate": 7.687877204374957e-06,
+      "loss": 0.0994,
+      "step": 426
+    },
+    {
+      "epoch": 0.6401799100449775,
+      "grad_norm": 0.7203224499911319,
+      "learning_rate": 7.677940823966196e-06,
+      "loss": 0.0785,
+      "step": 427
+    },
+    {
+      "epoch": 0.6416791604197901,
+      "grad_norm": 0.9591197714321589,
+      "learning_rate": 7.667989591423349e-06,
+      "loss": 0.0881,
+      "step": 428
+    },
+    {
+      "epoch": 0.6431784107946027,
+      "grad_norm": 0.6021675580029577,
+      "learning_rate": 7.658023561936966e-06,
+      "loss": 0.0683,
+      "step": 429
+    },
+    {
+      "epoch": 0.6446776611694153,
+      "grad_norm": 0.6977100790137684,
+      "learning_rate": 7.648042790779677e-06,
+      "loss": 0.0828,
+      "step": 430
+    },
+    {
+      "epoch": 0.6461769115442278,
+      "grad_norm": 0.7796243788420725,
+      "learning_rate": 7.638047333305853e-06,
+      "loss": 0.0835,
+      "step": 431
+    },
+    {
+      "epoch": 0.6476761619190404,
+      "grad_norm": 0.8394342072473704,
+      "learning_rate": 7.628037244951328e-06,
+      "loss": 0.0974,
+      "step": 432
+    },
+    {
+      "epoch": 0.6491754122938531,
+      "grad_norm": 0.7680913921855976,
+      "learning_rate": 7.618012581233076e-06,
+      "loss": 0.0792,
+      "step": 433
+    },
+    {
+      "epoch": 0.6506746626686657,
+      "grad_norm": 0.6361749050546337,
+      "learning_rate": 7.607973397748909e-06,
+      "loss": 0.0673,
+      "step": 434
+    },
+    {
+      "epoch": 0.6521739130434783,
+      "grad_norm": 0.8558591277710684,
+      "learning_rate": 7.597919750177168e-06,
+      "loss": 0.093,
+      "step": 435
+    },
+    {
+      "epoch": 0.6536731634182908,
+      "grad_norm": 0.8656304531926615,
+      "learning_rate": 7.587851694276412e-06,
+      "loss": 0.0968,
+      "step": 436
+    },
+    {
+      "epoch": 0.6551724137931034,
+      "grad_norm": 0.757437862552891,
+      "learning_rate": 7.57776928588511e-06,
+      "loss": 0.0682,
+      "step": 437
+    },
+    {
+      "epoch": 0.656671664167916,
+      "grad_norm": 0.6016983592897197,
+      "learning_rate": 7.56767258092133e-06,
+      "loss": 0.0606,
+      "step": 438
+    },
+    {
+      "epoch": 0.6581709145427287,
+      "grad_norm": 0.7562191443498486,
+      "learning_rate": 7.557561635382433e-06,
+      "loss": 0.081,
+      "step": 439
+    },
+    {
+      "epoch": 0.6596701649175413,
+      "grad_norm": 0.6004676790668859,
+      "learning_rate": 7.54743650534476e-06,
+      "loss": 0.0682,
+      "step": 440
+    },
+    {
+      "epoch": 0.6611694152923538,
+      "grad_norm": 0.9248812823105874,
+      "learning_rate": 7.537297246963316e-06,
+      "loss": 0.1031,
+      "step": 441
+    },
+    {
+      "epoch": 0.6626686656671664,
+      "grad_norm": 0.7939766575332708,
+      "learning_rate": 7.5271439164714695e-06,
+      "loss": 0.0877,
+      "step": 442
+    },
+    {
+      "epoch": 0.664167916041979,
+      "grad_norm": 0.6216181253514546,
+      "learning_rate": 7.5169765701806295e-06,
+      "loss": 0.0683,
+      "step": 443
+    },
+    {
+      "epoch": 0.6656671664167916,
+      "grad_norm": 0.541673198380648,
+      "learning_rate": 7.506795264479941e-06,
+      "loss": 0.0627,
+      "step": 444
+    },
+    {
+      "epoch": 0.6671664167916042,
+      "grad_norm": 0.7763898549928039,
+      "learning_rate": 7.4966000558359675e-06,
+      "loss": 0.0937,
+      "step": 445
+    },
+    {
+      "epoch": 0.6686656671664168,
+      "grad_norm": 0.9001332906389329,
+      "learning_rate": 7.486391000792379e-06,
+      "loss": 0.0912,
+      "step": 446
+    },
+    {
+      "epoch": 0.6701649175412294,
+      "grad_norm": 0.7812297877797904,
+      "learning_rate": 7.476168155969643e-06,
+      "loss": 0.0899,
+      "step": 447
+    },
+    {
+      "epoch": 0.671664167916042,
+      "grad_norm": 0.7851992836562671,
+      "learning_rate": 7.465931578064703e-06,
+      "loss": 0.085,
+      "step": 448
+    },
+    {
+      "epoch": 0.6731634182908546,
+      "grad_norm": 0.711644834670948,
+      "learning_rate": 7.455681323850669e-06,
+      "loss": 0.0687,
+      "step": 449
+    },
+    {
+      "epoch": 0.6746626686656672,
+      "grad_norm": 0.49458224936257544,
+      "learning_rate": 7.4454174501765e-06,
+      "loss": 0.0507,
+      "step": 450
+    },
+    {
+      "epoch": 0.6761619190404797,
+      "grad_norm": 0.7032827631974412,
+      "learning_rate": 7.4351400139666894e-06,
+      "loss": 0.0779,
+      "step": 451
+    },
+    {
+      "epoch": 0.6776611694152923,
+      "grad_norm": 0.672627597069967,
+      "learning_rate": 7.424849072220953e-06,
+      "loss": 0.0855,
+      "step": 452
+    },
+    {
+      "epoch": 0.679160419790105,
+      "grad_norm": 0.6964491969959253,
+      "learning_rate": 7.414544682013907e-06,
+      "loss": 0.0782,
+      "step": 453
+    },
+    {
+      "epoch": 0.6806596701649176,
+      "grad_norm": 0.6809063095473431,
+      "learning_rate": 7.404226900494753e-06,
+      "loss": 0.0647,
+      "step": 454
+    },
+    {
+      "epoch": 0.6821589205397302,
+      "grad_norm": 0.6430413578280697,
+      "learning_rate": 7.3938957848869684e-06,
+      "loss": 0.0719,
+      "step": 455
+    },
+    {
+      "epoch": 0.6836581709145427,
+      "grad_norm": 0.5645675814332473,
+      "learning_rate": 7.3835513924879755e-06,
+      "loss": 0.0528,
+      "step": 456
+    },
+    {
+      "epoch": 0.6851574212893553,
+      "grad_norm": 0.621222454181775,
+      "learning_rate": 7.373193780668835e-06,
+      "loss": 0.0638,
+      "step": 457
+    },
+    {
+      "epoch": 0.6866566716641679,
+      "grad_norm": 1.0453891621121392,
+      "learning_rate": 7.36282300687392e-06,
+      "loss": 0.1049,
+      "step": 458
+    },
+    {
+      "epoch": 0.6881559220389805,
+      "grad_norm": 0.8180436417351931,
+      "learning_rate": 7.35243912862061e-06,
+      "loss": 0.0856,
+      "step": 459
+    },
+    {
+      "epoch": 0.6896551724137931,
+      "grad_norm": 0.8756543614671445,
+      "learning_rate": 7.342042203498952e-06,
+      "loss": 0.0903,
+      "step": 460
+    },
+    {
+      "epoch": 0.6911544227886057,
+      "grad_norm": 0.949852411784083,
+      "learning_rate": 7.33163228917136e-06,
+      "loss": 0.0897,
+      "step": 461
+    },
+    {
+      "epoch": 0.6926536731634183,
+      "grad_norm": 0.7484073317060548,
+      "learning_rate": 7.321209443372284e-06,
+      "loss": 0.0885,
+      "step": 462
+    },
+    {
+      "epoch": 0.6941529235382309,
+      "grad_norm": 0.7655389155349908,
+      "learning_rate": 7.310773723907895e-06,
+      "loss": 0.0908,
+      "step": 463
+    },
+    {
+      "epoch": 0.6956521739130435,
+      "grad_norm": 1.0269719050985013,
+      "learning_rate": 7.300325188655762e-06,
+      "loss": 0.0995,
+      "step": 464
+    },
+    {
+      "epoch": 0.697151424287856,
+      "grad_norm": 0.7984890522025251,
+      "learning_rate": 7.289863895564531e-06,
+      "loss": 0.0939,
+      "step": 465
+    },
+    {
+      "epoch": 0.6986506746626686,
+      "grad_norm": 0.740579636384374,
+      "learning_rate": 7.279389902653606e-06,
+      "loss": 0.0848,
+      "step": 466
+    },
+    {
+      "epoch": 0.7001499250374813,
+      "grad_norm": 0.5994801560088966,
+      "learning_rate": 7.268903268012823e-06,
+      "loss": 0.0687,
+      "step": 467
+    },
+    {
+      "epoch": 0.7016491754122939,
+      "grad_norm": 0.6090171927611527,
+      "learning_rate": 7.258404049802135e-06,
+      "loss": 0.0662,
+      "step": 468
+    },
+    {
+      "epoch": 0.7031484257871065,
+      "grad_norm": 0.5262937966027181,
+      "learning_rate": 7.247892306251276e-06,
+      "loss": 0.0496,
+      "step": 469
+    },
+    {
+      "epoch": 0.704647676161919,
+      "grad_norm": 0.7159935848417084,
+      "learning_rate": 7.237368095659459e-06,
+      "loss": 0.0895,
+      "step": 470
+    },
+    {
+      "epoch": 0.7061469265367316,
+      "grad_norm": 0.802721251619299,
+      "learning_rate": 7.226831476395028e-06,
+      "loss": 0.0816,
+      "step": 471
+    },
+    {
+      "epoch": 0.7076461769115442,
+      "grad_norm": 0.801190321866234,
+      "learning_rate": 7.216282506895155e-06,
+      "loss": 0.0847,
+      "step": 472
+    },
+    {
+      "epoch": 0.7091454272863568,
+      "grad_norm": 0.7233072792835041,
+      "learning_rate": 7.2057212456655055e-06,
+      "loss": 0.0763,
+      "step": 473
+    },
+    {
+      "epoch": 0.7106446776611695,
+      "grad_norm": 0.7757156131228501,
+      "learning_rate": 7.195147751279915e-06,
+      "loss": 0.0922,
+      "step": 474
+    },
+    {
+      "epoch": 0.712143928035982,
+      "grad_norm": 0.9789850093924001,
+      "learning_rate": 7.184562082380069e-06,
+      "loss": 0.1208,
+      "step": 475
+    },
+    {
+      "epoch": 0.7136431784107946,
+      "grad_norm": 0.942355853277146,
+      "learning_rate": 7.173964297675168e-06,
+      "loss": 0.1051,
+      "step": 476
+    },
+    {
+      "epoch": 0.7151424287856072,
+      "grad_norm": 0.8569702955114691,
+      "learning_rate": 7.163354455941614e-06,
+      "loss": 0.093,
+      "step": 477
+    },
+    {
+      "epoch": 0.7166416791604198,
+      "grad_norm": 0.7625460476831917,
+      "learning_rate": 7.152732616022675e-06,
+      "loss": 0.0849,
+      "step": 478
+    },
+    {
+      "epoch": 0.7181409295352323,
+      "grad_norm": 0.8412431962050239,
+      "learning_rate": 7.142098836828162e-06,
+      "loss": 0.1001,
+      "step": 479
+    },
+    {
+      "epoch": 0.719640179910045,
+      "grad_norm": 0.9462780964971109,
+      "learning_rate": 7.131453177334103e-06,
+      "loss": 0.104,
+      "step": 480
+    },
+    {
+      "epoch": 0.7211394302848576,
+      "grad_norm": 0.65955944799275,
+      "learning_rate": 7.120795696582419e-06,
+      "loss": 0.0758,
+      "step": 481
+    },
+    {
+      "epoch": 0.7226386806596702,
+      "grad_norm": 0.7917483604378374,
+      "learning_rate": 7.1101264536805885e-06,
+      "loss": 0.0873,
+      "step": 482
+    },
+    {
+      "epoch": 0.7241379310344828,
+      "grad_norm": 0.6379284935146922,
+      "learning_rate": 7.099445507801324e-06,
+      "loss": 0.0714,
+      "step": 483
+    },
+    {
+      "epoch": 0.7256371814092953,
+      "grad_norm": 0.7005871748433659,
+      "learning_rate": 7.088752918182247e-06,
+      "loss": 0.0751,
+      "step": 484
+    },
+    {
+      "epoch": 0.7271364317841079,
+      "grad_norm": 0.8963560723631747,
+      "learning_rate": 7.078048744125553e-06,
+      "loss": 0.1075,
+      "step": 485
+    },
+    {
+      "epoch": 0.7286356821589205,
+      "grad_norm": 0.7812434075387101,
+      "learning_rate": 7.067333044997689e-06,
+      "loss": 0.0975,
+      "step": 486
+    },
+    {
+      "epoch": 0.7301349325337332,
+      "grad_norm": 0.9808191582318301,
+      "learning_rate": 7.0566058802290196e-06,
+      "loss": 0.1239,
+      "step": 487
+    },
+    {
+      "epoch": 0.7316341829085458,
+      "grad_norm": 0.6620883828753326,
+      "learning_rate": 7.045867309313499e-06,
+      "loss": 0.0741,
+      "step": 488
+    },
+    {
+      "epoch": 0.7331334332833583,
+      "grad_norm": 0.7349443495382586,
+      "learning_rate": 7.035117391808341e-06,
+      "loss": 0.0785,
+      "step": 489
+    },
+    {
+      "epoch": 0.7346326836581709,
+      "grad_norm": 0.7651682592945389,
+      "learning_rate": 7.024356187333692e-06,
+      "loss": 0.0906,
+      "step": 490
+    },
+    {
+      "epoch": 0.7361319340329835,
+      "grad_norm": 1.106398234961281,
+      "learning_rate": 7.01358375557229e-06,
+      "loss": 0.1282,
+      "step": 491
+    },
+    {
+      "epoch": 0.7376311844077961,
+      "grad_norm": 0.6480162972450583,
+      "learning_rate": 7.0028001562691475e-06,
+      "loss": 0.0785,
+      "step": 492
+    },
+    {
+      "epoch": 0.7391304347826086,
+      "grad_norm": 0.7633681826230366,
+      "learning_rate": 6.9920054492312086e-06,
+      "loss": 0.0959,
+      "step": 493
+    },
+    {
+      "epoch": 0.7406296851574213,
+      "grad_norm": 0.9147636790542241,
+      "learning_rate": 6.981199694327024e-06,
+      "loss": 0.0898,
+      "step": 494
+    },
+    {
+      "epoch": 0.7421289355322339,
+      "grad_norm": 0.7011794134324573,
+      "learning_rate": 6.97038295148642e-06,
+      "loss": 0.0855,
+      "step": 495
+    },
+    {
+      "epoch": 0.7436281859070465,
+      "grad_norm": 0.5439929370351676,
+      "learning_rate": 6.959555280700162e-06,
+      "loss": 0.0616,
+      "step": 496
+    },
+    {
+      "epoch": 0.7451274362818591,
+      "grad_norm": 0.8537614159042325,
+      "learning_rate": 6.948716742019616e-06,
+      "loss": 0.0871,
+      "step": 497
+    },
+    {
+      "epoch": 0.7466266866566716,
+      "grad_norm": 0.9367355000694778,
+      "learning_rate": 6.937867395556428e-06,
+      "loss": 0.0992,
+      "step": 498
+    },
+    {
+      "epoch": 0.7481259370314842,
+      "grad_norm": 0.5657939784744264,
+      "learning_rate": 6.927007301482187e-06,
+      "loss": 0.0558,
+      "step": 499
+    },
+    {
+      "epoch": 0.7496251874062968,
+      "grad_norm": 0.9142636042211669,
+      "learning_rate": 6.916136520028087e-06,
+      "loss": 0.1097,
+      "step": 500
+    },
+    {
+      "epoch": 0.7511244377811095,
+      "grad_norm": 0.6864954073855778,
+      "learning_rate": 6.905255111484592e-06,
+      "loss": 0.0732,
+      "step": 501
+    },
+    {
+      "epoch": 0.7526236881559221,
+      "grad_norm": 0.8878137152002674,
+      "learning_rate": 6.894363136201114e-06,
+      "loss": 0.1036,
+      "step": 502
+    },
+    {
+      "epoch": 0.7541229385307346,
+      "grad_norm": 0.7509870298728826,
+      "learning_rate": 6.88346065458566e-06,
+      "loss": 0.0969,
+      "step": 503
+    },
+    {
+      "epoch": 0.7556221889055472,
+      "grad_norm": 0.7100210714996809,
+      "learning_rate": 6.8725477271045085e-06,
+      "loss": 0.0822,
+      "step": 504
+    },
+    {
+      "epoch": 0.7571214392803598,
+      "grad_norm": 0.7749852946202924,
+      "learning_rate": 6.861624414281875e-06,
+      "loss": 0.0752,
+      "step": 505
+    },
+    {
+      "epoch": 0.7586206896551724,
+      "grad_norm": 0.7378340647477563,
+      "learning_rate": 6.850690776699574e-06,
+      "loss": 0.08,
+      "step": 506
+    },
+    {
+      "epoch": 0.760119940029985,
+      "grad_norm": 0.9798024053395626,
+      "learning_rate": 6.8397468749966735e-06,
+      "loss": 0.0753,
+      "step": 507
+    },
+    {
+      "epoch": 0.7616191904047976,
+      "grad_norm": 0.7186235146458679,
+      "learning_rate": 6.8287927698691745e-06,
+      "loss": 0.0805,
+      "step": 508
+    },
+    {
+      "epoch": 0.7631184407796102,
+      "grad_norm": 0.6686816735715482,
+      "learning_rate": 6.8178285220696686e-06,
+      "loss": 0.0798,
+      "step": 509
+    },
+    {
+      "epoch": 0.7646176911544228,
+      "grad_norm": 0.5921934443510786,
+      "learning_rate": 6.806854192406995e-06,
+      "loss": 0.0622,
+      "step": 510
+    },
+    {
+      "epoch": 0.7661169415292354,
+      "grad_norm": 0.7997892920198899,
+      "learning_rate": 6.795869841745912e-06,
+      "loss": 0.0855,
+      "step": 511
+    },
+    {
+      "epoch": 0.767616191904048,
+      "grad_norm": 0.6345294051027847,
+      "learning_rate": 6.784875531006751e-06,
+      "loss": 0.0642,
+      "step": 512
+    },
+    {
+      "epoch": 0.7691154422788605,
+      "grad_norm": 1.0223767126023742,
+      "learning_rate": 6.7738713211650885e-06,
+      "loss": 0.1061,
+      "step": 513
+    },
+    {
+      "epoch": 0.7706146926536732,
+      "grad_norm": 0.6346624712102827,
+      "learning_rate": 6.762857273251396e-06,
+      "loss": 0.0737,
+      "step": 514
+    },
+    {
+      "epoch": 0.7721139430284858,
+      "grad_norm": 0.8086529297214492,
+      "learning_rate": 6.751833448350713e-06,
+      "loss": 0.0806,
+      "step": 515
+    },
+    {
+      "epoch": 0.7736131934032984,
+      "grad_norm": 0.8037824498659087,
+      "learning_rate": 6.740799907602302e-06,
+      "loss": 0.087,
+      "step": 516
+    },
+    {
+      "epoch": 0.775112443778111,
+      "grad_norm": 0.6670210319552843,
+      "learning_rate": 6.729756712199309e-06,
+      "loss": 0.0741,
+      "step": 517
+    },
+    {
+      "epoch": 0.7766116941529235,
+      "grad_norm": 0.6351739458142562,
+      "learning_rate": 6.718703923388427e-06,
+      "loss": 0.0709,
+      "step": 518
+    },
+    {
+      "epoch": 0.7781109445277361,
+      "grad_norm": 0.6629224541538651,
+      "learning_rate": 6.707641602469554e-06,
+      "loss": 0.0717,
+      "step": 519
+    },
+    {
+      "epoch": 0.7796101949025487,
+      "grad_norm": 0.8300231523472757,
+      "learning_rate": 6.696569810795455e-06,
+      "loss": 0.091,
+      "step": 520
+    },
+    {
+      "epoch": 0.7811094452773614,
+      "grad_norm": 0.6185476465028584,
+      "learning_rate": 6.685488609771422e-06,
+      "loss": 0.063,
+      "step": 521
+    },
+    {
+      "epoch": 0.782608695652174,
+      "grad_norm": 0.7323462463745086,
+      "learning_rate": 6.674398060854931e-06,
+      "loss": 0.0729,
+      "step": 522
+    },
+    {
+      "epoch": 0.7841079460269865,
+      "grad_norm": 0.7318656436961414,
+      "learning_rate": 6.6632982255553004e-06,
+      "loss": 0.0856,
+      "step": 523
+    },
+    {
+      "epoch": 0.7856071964017991,
+      "grad_norm": 1.0213777797237156,
+      "learning_rate": 6.652189165433356e-06,
+      "loss": 0.0919,
+      "step": 524
+    },
+    {
+      "epoch": 0.7871064467766117,
+      "grad_norm": 0.6659457212149912,
+      "learning_rate": 6.64107094210108e-06,
+      "loss": 0.074,
+      "step": 525
+    },
+    {
+      "epoch": 0.7886056971514243,
+      "grad_norm": 0.6981380821595006,
+      "learning_rate": 6.62994361722128e-06,
+      "loss": 0.0728,
+      "step": 526
+    },
+    {
+      "epoch": 0.7901049475262368,
+      "grad_norm": 0.7107882641503626,
+      "learning_rate": 6.618807252507238e-06,
+      "loss": 0.0761,
+      "step": 527
+    },
+    {
+      "epoch": 0.7916041979010495,
+      "grad_norm": 1.0324581520146072,
+      "learning_rate": 6.6076619097223735e-06,
+      "loss": 0.1217,
+      "step": 528
+    },
+    {
+      "epoch": 0.7931034482758621,
+      "grad_norm": 1.1094383063424573,
+      "learning_rate": 6.5965076506799e-06,
+      "loss": 0.1219,
+      "step": 529
+    },
+    {
+      "epoch": 0.7946026986506747,
+      "grad_norm": 0.839519507935226,
+      "learning_rate": 6.5853445372424805e-06,
+      "loss": 0.0957,
+      "step": 530
+    },
+    {
+      "epoch": 0.7961019490254873,
+      "grad_norm": 0.6568192692149819,
+      "learning_rate": 6.574172631321885e-06,
+      "loss": 0.0709,
+      "step": 531
+    },
+    {
+      "epoch": 0.7976011994002998,
+      "grad_norm": 0.799260561849818,
+      "learning_rate": 6.562991994878649e-06,
+      "loss": 0.0826,
+      "step": 532
+    },
+    {
+      "epoch": 0.7991004497751124,
+      "grad_norm": 0.7702607490358493,
+      "learning_rate": 6.551802689921726e-06,
+      "loss": 0.0903,
+      "step": 533
+    },
+    {
+      "epoch": 0.800599700149925,
+      "grad_norm": 0.8443658202571149,
+      "learning_rate": 6.5406047785081485e-06,
+      "loss": 0.0954,
+      "step": 534
+    },
+    {
+      "epoch": 0.8020989505247377,
+      "grad_norm": 0.6267123790888645,
+      "learning_rate": 6.529398322742677e-06,
+      "loss": 0.066,
+      "step": 535
+    },
+    {
+      "epoch": 0.8035982008995503,
+      "grad_norm": 0.5811866462250391,
+      "learning_rate": 6.518183384777468e-06,
+      "loss": 0.0613,
+      "step": 536
+    },
+    {
+      "epoch": 0.8050974512743628,
+      "grad_norm": 1.06847844444903,
+      "learning_rate": 6.506960026811712e-06,
+      "loss": 0.1068,
+      "step": 537
+    },
+    {
+      "epoch": 0.8065967016491754,
+      "grad_norm": 0.8348885875479208,
+      "learning_rate": 6.495728311091303e-06,
+      "loss": 0.1124,
+      "step": 538
+    },
+    {
+      "epoch": 0.808095952023988,
+      "grad_norm": 0.7492538167176246,
+      "learning_rate": 6.484488299908487e-06,
+      "loss": 0.0793,
+      "step": 539
+    },
+    {
+      "epoch": 0.8095952023988006,
+      "grad_norm": 0.7712306329996408,
+      "learning_rate": 6.473240055601517e-06,
+      "loss": 0.0795,
+      "step": 540
+    },
+    {
+      "epoch": 0.8110944527736131,
+      "grad_norm": 0.726341024671012,
+      "learning_rate": 6.46198364055431e-06,
+      "loss": 0.0888,
+      "step": 541
+    },
+    {
+      "epoch": 0.8125937031484258,
+      "grad_norm": 1.1109314677888769,
+      "learning_rate": 6.450719117196094e-06,
+      "loss": 0.1025,
+      "step": 542
+    },
+    {
+      "epoch": 0.8140929535232384,
+      "grad_norm": 0.9078055032047169,
+      "learning_rate": 6.439446548001069e-06,
+      "loss": 0.0867,
+      "step": 543
+    },
+    {
+      "epoch": 0.815592203898051,
+      "grad_norm": 0.888156054189124,
+      "learning_rate": 6.4281659954880605e-06,
+      "loss": 0.0929,
+      "step": 544
+    },
+    {
+      "epoch": 0.8170914542728636,
+      "grad_norm": 0.7408716056283109,
+      "learning_rate": 6.416877522220167e-06,
+      "loss": 0.0697,
+      "step": 545
+    },
+    {
+      "epoch": 0.8185907046476761,
+      "grad_norm": 0.8109065209002204,
+      "learning_rate": 6.405581190804418e-06,
+      "loss": 0.0728,
+      "step": 546
+    },
+    {
+      "epoch": 0.8200899550224887,
+      "grad_norm": 0.564030750722701,
+      "learning_rate": 6.394277063891422e-06,
+      "loss": 0.0613,
+      "step": 547
+    },
+    {
+      "epoch": 0.8215892053973014,
+      "grad_norm": 0.863046090229823,
+      "learning_rate": 6.382965204175027e-06,
+      "loss": 0.1167,
+      "step": 548
+    },
+    {
+      "epoch": 0.823088455772114,
+      "grad_norm": 0.5812649535238091,
+      "learning_rate": 6.371645674391967e-06,
+      "loss": 0.0634,
+      "step": 549
+    },
+    {
+      "epoch": 0.8245877061469266,
+      "grad_norm": 0.8799656793943876,
+      "learning_rate": 6.3603185373215105e-06,
+      "loss": 0.0884,
+      "step": 550
+    },
+    {
+      "epoch": 0.8260869565217391,
+      "grad_norm": 0.8583338866157422,
+      "learning_rate": 6.348983855785122e-06,
+      "loss": 0.1149,
+      "step": 551
+    },
+    {
+      "epoch": 0.8275862068965517,
+      "grad_norm": 0.8804636715263047,
+      "learning_rate": 6.337641692646106e-06,
+      "loss": 0.0948,
+      "step": 552
+    },
+    {
+      "epoch": 0.8290854572713643,
+      "grad_norm": 0.5809382274557015,
+      "learning_rate": 6.326292110809258e-06,
+      "loss": 0.0674,
+      "step": 553
+    },
+    {
+      "epoch": 0.8305847076461769,
+      "grad_norm": 0.5536153244513229,
+      "learning_rate": 6.314935173220524e-06,
+      "loss": 0.0647,
+      "step": 554
+    },
+    {
+      "epoch": 0.8320839580209896,
+      "grad_norm": 0.9618444433996294,
+      "learning_rate": 6.303570942866643e-06,
+      "loss": 0.1239,
+      "step": 555
+    },
+    {
+      "epoch": 0.8335832083958021,
+      "grad_norm": 0.670482772933402,
+      "learning_rate": 6.2921994827748e-06,
+      "loss": 0.0746,
+      "step": 556
+    },
+    {
+      "epoch": 0.8350824587706147,
+      "grad_norm": 0.7836633227952995,
+      "learning_rate": 6.280820856012277e-06,
+      "loss": 0.0694,
+      "step": 557
+    },
+    {
+      "epoch": 0.8365817091454273,
+      "grad_norm": 0.58278624676119,
+      "learning_rate": 6.269435125686105e-06,
+      "loss": 0.0632,
+      "step": 558
+    },
+    {
+      "epoch": 0.8380809595202399,
+      "grad_norm": 0.8126109848960543,
+      "learning_rate": 6.258042354942708e-06,
+      "loss": 0.0865,
+      "step": 559
+    },
+    {
+      "epoch": 0.8395802098950524,
+      "grad_norm": 0.7419025048939819,
+      "learning_rate": 6.2466426069675626e-06,
+      "loss": 0.0858,
+      "step": 560
+    },
+    {
+      "epoch": 0.841079460269865,
+      "grad_norm": 0.6312659449297664,
+      "learning_rate": 6.235235944984835e-06,
+      "loss": 0.0746,
+      "step": 561
+    },
+    {
+      "epoch": 0.8425787106446777,
+      "grad_norm": 0.666958134751936,
+      "learning_rate": 6.223822432257043e-06,
+      "loss": 0.0632,
+      "step": 562
+    },
+    {
+      "epoch": 0.8440779610194903,
+      "grad_norm": 0.8762918271039484,
+      "learning_rate": 6.212402132084697e-06,
+      "loss": 0.1017,
+      "step": 563
+    },
+    {
+      "epoch": 0.8455772113943029,
+      "grad_norm": 0.7839953321354922,
+      "learning_rate": 6.200975107805951e-06,
+      "loss": 0.0739,
+      "step": 564
+    },
+    {
+      "epoch": 0.8470764617691154,
+      "grad_norm": 0.7485105389653173,
+      "learning_rate": 6.189541422796254e-06,
+      "loss": 0.0804,
+      "step": 565
+    },
+    {
+      "epoch": 0.848575712143928,
+      "grad_norm": 0.6744963232341645,
+      "learning_rate": 6.1781011404679905e-06,
+      "loss": 0.083,
+      "step": 566
+    },
+    {
+      "epoch": 0.8500749625187406,
+      "grad_norm": 0.6563190372626161,
+      "learning_rate": 6.16665432427014e-06,
+      "loss": 0.0652,
+      "step": 567
+    },
+    {
+      "epoch": 0.8515742128935532,
+      "grad_norm": 0.8381827444905121,
+      "learning_rate": 6.155201037687917e-06,
+      "loss": 0.108,
+      "step": 568
+    },
+    {
+      "epoch": 0.8530734632683659,
+      "grad_norm": 0.694845276727635,
+      "learning_rate": 6.1437413442424236e-06,
+      "loss": 0.074,
+      "step": 569
+    },
+    {
+      "epoch": 0.8545727136431784,
+      "grad_norm": 0.6584176625660612,
+      "learning_rate": 6.132275307490291e-06,
+      "loss": 0.0688,
+      "step": 570
+    },
+    {
+      "epoch": 0.856071964017991,
+      "grad_norm": 0.6950564035482574,
+      "learning_rate": 6.120802991023334e-06,
+      "loss": 0.0638,
+      "step": 571
+    },
+    {
+      "epoch": 0.8575712143928036,
+      "grad_norm": 0.7560759592155073,
+      "learning_rate": 6.109324458468198e-06,
+      "loss": 0.0939,
+      "step": 572
+    },
+    {
+      "epoch": 0.8590704647676162,
+      "grad_norm": 0.7198533185055043,
+      "learning_rate": 6.097839773485995e-06,
+      "loss": 0.0702,
+      "step": 573
+    },
+    {
+      "epoch": 0.8605697151424287,
+      "grad_norm": 0.8818127292711702,
+      "learning_rate": 6.086348999771967e-06,
+      "loss": 0.096,
+      "step": 574
+    },
+    {
+      "epoch": 0.8620689655172413,
+      "grad_norm": 0.6549762547827207,
+      "learning_rate": 6.074852201055121e-06,
+      "loss": 0.0726,
+      "step": 575
+    },
+    {
+      "epoch": 0.863568215892054,
+      "grad_norm": 0.7425748915189844,
+      "learning_rate": 6.063349441097881e-06,
+      "loss": 0.0828,
+      "step": 576
+    },
+    {
+      "epoch": 0.8650674662668666,
+      "grad_norm": 0.7363298137416725,
+      "learning_rate": 6.051840783695731e-06,
+      "loss": 0.0773,
+      "step": 577
+    },
+    {
+      "epoch": 0.8665667166416792,
+      "grad_norm": 0.8053103146175553,
+      "learning_rate": 6.040326292676865e-06,
+      "loss": 0.0896,
+      "step": 578
+    },
+    {
+      "epoch": 0.8680659670164917,
+      "grad_norm": 1.0064327327485305,
+      "learning_rate": 6.028806031901829e-06,
+      "loss": 0.1043,
+      "step": 579
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 0.7022270979019841,
+      "learning_rate": 6.0172800652631706e-06,
+      "loss": 0.0871,
+      "step": 580
+    },
+    {
+      "epoch": 0.8710644677661169,
+      "grad_norm": 0.7179866246500819,
+      "learning_rate": 6.005748456685077e-06,
+      "loss": 0.0702,
+      "step": 581
+    },
+    {
+      "epoch": 0.8725637181409296,
+      "grad_norm": 0.8233256498187941,
+      "learning_rate": 5.994211270123034e-06,
+      "loss": 0.0961,
+      "step": 582
+    },
+    {
+      "epoch": 0.8740629685157422,
+      "grad_norm": 0.8334695960673352,
+      "learning_rate": 5.9826685695634575e-06,
+      "loss": 0.0988,
+      "step": 583
+    },
+    {
+      "epoch": 0.8755622188905547,
+      "grad_norm": 0.7245217443690574,
+      "learning_rate": 5.971120419023349e-06,
+      "loss": 0.0683,
+      "step": 584
+    },
+    {
+      "epoch": 0.8770614692653673,
+      "grad_norm": 0.6096318018271903,
+      "learning_rate": 5.959566882549936e-06,
+      "loss": 0.0827,
+      "step": 585
+    },
+    {
+      "epoch": 0.8785607196401799,
+      "grad_norm": 0.6757886976397195,
+      "learning_rate": 5.948008024220311e-06,
+      "loss": 0.0746,
+      "step": 586
+    },
+    {
+      "epoch": 0.8800599700149925,
+      "grad_norm": 0.6358045441807924,
+      "learning_rate": 5.936443908141088e-06,
+      "loss": 0.0688,
+      "step": 587
+    },
+    {
+      "epoch": 0.881559220389805,
+      "grad_norm": 0.7934275737016375,
+      "learning_rate": 5.924874598448038e-06,
+      "loss": 0.0842,
+      "step": 588
+    },
+    {
+      "epoch": 0.8830584707646177,
+      "grad_norm": 0.5966005008973805,
+      "learning_rate": 5.913300159305741e-06,
+      "loss": 0.0741,
+      "step": 589
+    },
+    {
+      "epoch": 0.8845577211394303,
+      "grad_norm": 0.785363496175796,
+      "learning_rate": 5.901720654907217e-06,
+      "loss": 0.0712,
+      "step": 590
+    },
+    {
+      "epoch": 0.8860569715142429,
+      "grad_norm": 0.7269388260326227,
+      "learning_rate": 5.8901361494735874e-06,
+      "loss": 0.0793,
+      "step": 591
+    },
+    {
+      "epoch": 0.8875562218890555,
+      "grad_norm": 0.7977720299623906,
+      "learning_rate": 5.878546707253704e-06,
+      "loss": 0.0797,
+      "step": 592
+    },
+    {
+      "epoch": 0.889055472263868,
+      "grad_norm": 0.786599111108691,
+      "learning_rate": 5.8669523925238e-06,
+      "loss": 0.0882,
+      "step": 593
+    },
+    {
+      "epoch": 0.8905547226386806,
+      "grad_norm": 0.6486721761805913,
+      "learning_rate": 5.855353269587134e-06,
+      "loss": 0.0712,
+      "step": 594
+    },
+    {
+      "epoch": 0.8920539730134932,
+      "grad_norm": 0.6895666611207213,
+      "learning_rate": 5.843749402773629e-06,
+      "loss": 0.0623,
+      "step": 595
+    },
+    {
+      "epoch": 0.8935532233883059,
+      "grad_norm": 0.6530659413676487,
+      "learning_rate": 5.8321408564395165e-06,
+      "loss": 0.0687,
+      "step": 596
+    },
+    {
+      "epoch": 0.8950524737631185,
+      "grad_norm": 0.6681449675140796,
+      "learning_rate": 5.820527694966988e-06,
+      "loss": 0.0717,
+      "step": 597
+    },
+    {
+      "epoch": 0.896551724137931,
+      "grad_norm": 0.5459922105010183,
+      "learning_rate": 5.808909982763825e-06,
+      "loss": 0.0517,
+      "step": 598
+    },
+    {
+      "epoch": 0.8980509745127436,
+      "grad_norm": 0.6590483783728908,
+      "learning_rate": 5.797287784263047e-06,
+      "loss": 0.0881,
+      "step": 599
+    },
+    {
+      "epoch": 0.8995502248875562,
+      "grad_norm": 0.7988646454636542,
+      "learning_rate": 5.785661163922558e-06,
+      "loss": 0.0844,
+      "step": 600
+    },
+    {
+      "epoch": 0.8995502248875562,
+      "eval_loss": 0.09048319607973099,
+      "eval_runtime": 8.1057,
+      "eval_samples_per_second": 6.662,
+      "eval_steps_per_second": 1.727,
+      "step": 600
+    },
+    {
+      "epoch": 0.9010494752623688,
+      "grad_norm": 0.7855451198577023,
+      "learning_rate": 5.774030186224786e-06,
+      "loss": 0.0757,
+      "step": 601
+    },
+    {
+      "epoch": 0.9025487256371814,
+      "grad_norm": 0.5466333268835563,
+      "learning_rate": 5.762394915676325e-06,
+      "loss": 0.0603,
+      "step": 602
+    },
+    {
+      "epoch": 0.904047976011994,
+      "grad_norm": 0.8889359528679897,
+      "learning_rate": 5.750755416807575e-06,
+      "loss": 0.1035,
+      "step": 603
+    },
+    {
+      "epoch": 0.9055472263868066,
+      "grad_norm": 0.7175343211158108,
+      "learning_rate": 5.7391117541723914e-06,
+      "loss": 0.0733,
+      "step": 604
+    },
+    {
+      "epoch": 0.9070464767616192,
+      "grad_norm": 0.7661058124177177,
+      "learning_rate": 5.727463992347719e-06,
+      "loss": 0.0904,
+      "step": 605
+    },
+    {
+      "epoch": 0.9085457271364318,
+      "grad_norm": 0.6799463689982046,
+      "learning_rate": 5.715812195933238e-06,
+      "loss": 0.0697,
+      "step": 606
+    },
+    {
+      "epoch": 0.9100449775112444,
+      "grad_norm": 0.873781710584012,
+      "learning_rate": 5.704156429551004e-06,
+      "loss": 0.0754,
+      "step": 607
+    },
+    {
+      "epoch": 0.9115442278860569,
+      "grad_norm": 0.7660186076937884,
+      "learning_rate": 5.692496757845092e-06,
+      "loss": 0.08,
+      "step": 608
+    },
+    {
+      "epoch": 0.9130434782608695,
+      "grad_norm": 0.7561630536431578,
+      "learning_rate": 5.680833245481234e-06,
+      "loss": 0.0819,
+      "step": 609
+    },
+    {
+      "epoch": 0.9145427286356822,
+      "grad_norm": 1.0436431234241734,
+      "learning_rate": 5.6691659571464655e-06,
+      "loss": 0.119,
+      "step": 610
+    },
+    {
+      "epoch": 0.9160419790104948,
+      "grad_norm": 0.5701148152271438,
+      "learning_rate": 5.657494957548761e-06,
+      "loss": 0.0688,
+      "step": 611
+    },
+    {
+      "epoch": 0.9175412293853074,
+      "grad_norm": 0.7262221376192396,
+      "learning_rate": 5.645820311416681e-06,
+      "loss": 0.0608,
+      "step": 612
+    },
+    {
+      "epoch": 0.9190404797601199,
+      "grad_norm": 0.8135839181332992,
+      "learning_rate": 5.63414208349901e-06,
+      "loss": 0.0901,
+      "step": 613
+    },
+    {
+      "epoch": 0.9205397301349325,
+      "grad_norm": 0.7318936586698559,
+      "learning_rate": 5.622460338564393e-06,
+      "loss": 0.0791,
+      "step": 614
+    },
+    {
+      "epoch": 0.9220389805097451,
+      "grad_norm": 0.614959103261924,
+      "learning_rate": 5.610775141400986e-06,
+      "loss": 0.0615,
+      "step": 615
+    },
+    {
+      "epoch": 0.9235382308845578,
+      "grad_norm": 0.7357849238068866,
+      "learning_rate": 5.599086556816089e-06,
+      "loss": 0.0809,
+      "step": 616
+    },
+    {
+      "epoch": 0.9250374812593704,
+      "grad_norm": 0.7900517082682416,
+      "learning_rate": 5.587394649635789e-06,
+      "loss": 0.0884,
+      "step": 617
+    },
+    {
+      "epoch": 0.9265367316341829,
+      "grad_norm": 0.756030164230796,
+      "learning_rate": 5.575699484704599e-06,
+      "loss": 0.0749,
+      "step": 618
+    },
+    {
+      "epoch": 0.9280359820089955,
+      "grad_norm": 0.8428781982180006,
+      "learning_rate": 5.564001126885106e-06,
+      "loss": 0.0887,
+      "step": 619
+    },
+    {
+      "epoch": 0.9295352323838081,
+      "grad_norm": 0.8781280021209953,
+      "learning_rate": 5.552299641057596e-06,
+      "loss": 0.0932,
+      "step": 620
+    },
+    {
+      "epoch": 0.9310344827586207,
+      "grad_norm": 0.7983205376631062,
+      "learning_rate": 5.540595092119709e-06,
+      "loss": 0.09,
+      "step": 621
+    },
+    {
+      "epoch": 0.9325337331334332,
+      "grad_norm": 0.7475405884109432,
+      "learning_rate": 5.5288875449860745e-06,
+      "loss": 0.0785,
+      "step": 622
+    },
+    {
+      "epoch": 0.9340329835082459,
+      "grad_norm": 0.6652240242208931,
+      "learning_rate": 5.517177064587945e-06,
+      "loss": 0.0773,
+      "step": 623
+    },
+    {
+      "epoch": 0.9355322338830585,
+      "grad_norm": 0.5724534948955192,
+      "learning_rate": 5.505463715872846e-06,
+      "loss": 0.0678,
+      "step": 624
+    },
+    {
+      "epoch": 0.9370314842578711,
+      "grad_norm": 0.7210936186040761,
+      "learning_rate": 5.493747563804211e-06,
+      "loss": 0.0736,
+      "step": 625
+    },
+    {
+      "epoch": 0.9385307346326837,
+      "grad_norm": 0.5640294739403683,
+      "learning_rate": 5.482028673361015e-06,
+      "loss": 0.0557,
+      "step": 626
+    },
+    {
+      "epoch": 0.9400299850074962,
+      "grad_norm": 0.744802819563649,
+      "learning_rate": 5.470307109537427e-06,
+      "loss": 0.0784,
+      "step": 627
+    },
+    {
+      "epoch": 0.9415292353823088,
+      "grad_norm": 0.7824196105494375,
+      "learning_rate": 5.45858293734244e-06,
+      "loss": 0.0802,
+      "step": 628
+    },
+    {
+      "epoch": 0.9430284857571214,
+      "grad_norm": 1.0555503748280866,
+      "learning_rate": 5.446856221799515e-06,
+      "loss": 0.0992,
+      "step": 629
+    },
+    {
+      "epoch": 0.9445277361319341,
+      "grad_norm": 0.7693330537692353,
+      "learning_rate": 5.435127027946215e-06,
+      "loss": 0.092,
+      "step": 630
+    },
+    {
+      "epoch": 0.9460269865067467,
+      "grad_norm": 0.8873031638532757,
+      "learning_rate": 5.423395420833853e-06,
+      "loss": 0.0938,
+      "step": 631
+    },
+    {
+      "epoch": 0.9475262368815592,
+      "grad_norm": 0.7629921974979079,
+      "learning_rate": 5.411661465527123e-06,
+      "loss": 0.0758,
+      "step": 632
+    },
+    {
+      "epoch": 0.9490254872563718,
+      "grad_norm": 0.7087812016939877,
+      "learning_rate": 5.39992522710374e-06,
+      "loss": 0.0778,
+      "step": 633
+    },
+    {
+      "epoch": 0.9505247376311844,
+      "grad_norm": 0.625790703152573,
+      "learning_rate": 5.38818677065409e-06,
+      "loss": 0.0545,
+      "step": 634
+    },
+    {
+      "epoch": 0.952023988005997,
+      "grad_norm": 0.6491141933326614,
+      "learning_rate": 5.376446161280851e-06,
+      "loss": 0.068,
+      "step": 635
+    },
+    {
+      "epoch": 0.9535232383808095,
+      "grad_norm": 0.7846313564100897,
+      "learning_rate": 5.364703464098645e-06,
+      "loss": 0.0902,
+      "step": 636
+    },
+    {
+      "epoch": 0.9550224887556222,
+      "grad_norm": 0.7169752016542214,
+      "learning_rate": 5.352958744233673e-06,
+      "loss": 0.0739,
+      "step": 637
+    },
+    {
+      "epoch": 0.9565217391304348,
+      "grad_norm": 0.6871519689427388,
+      "learning_rate": 5.341212066823356e-06,
+      "loss": 0.0743,
+      "step": 638
+    },
+    {
+      "epoch": 0.9580209895052474,
+      "grad_norm": 0.6697034374459067,
+      "learning_rate": 5.329463497015969e-06,
+      "loss": 0.0761,
+      "step": 639
+    },
+    {
+      "epoch": 0.95952023988006,
+      "grad_norm": 0.7032885840774589,
+      "learning_rate": 5.317713099970283e-06,
+      "loss": 0.0734,
+      "step": 640
+    },
+    {
+      "epoch": 0.9610194902548725,
+      "grad_norm": 0.7228962909646095,
+      "learning_rate": 5.305960940855205e-06,
+      "loss": 0.0905,
+      "step": 641
+    },
+    {
+      "epoch": 0.9625187406296851,
+      "grad_norm": 0.8327756806202219,
+      "learning_rate": 5.294207084849412e-06,
+      "loss": 0.0762,
+      "step": 642
+    },
+    {
+      "epoch": 0.9640179910044977,
+      "grad_norm": 0.5684794007820115,
+      "learning_rate": 5.282451597140994e-06,
+      "loss": 0.0602,
+      "step": 643
+    },
+    {
+      "epoch": 0.9655172413793104,
+      "grad_norm": 0.9426713930453604,
+      "learning_rate": 5.270694542927089e-06,
+      "loss": 0.1,
+      "step": 644
+    },
+    {
+      "epoch": 0.967016491754123,
+      "grad_norm": 1.1142477081832662,
+      "learning_rate": 5.258935987413524e-06,
+      "loss": 0.1051,
+      "step": 645
+    },
+    {
+      "epoch": 0.9685157421289355,
+      "grad_norm": 0.6672889046515589,
+      "learning_rate": 5.247175995814452e-06,
+      "loss": 0.068,
+      "step": 646
+    },
+    {
+      "epoch": 0.9700149925037481,
+      "grad_norm": 0.8998941155728255,
+      "learning_rate": 5.235414633351992e-06,
+      "loss": 0.0769,
+      "step": 647
+    },
+    {
+      "epoch": 0.9715142428785607,
+      "grad_norm": 0.8037197119234065,
+      "learning_rate": 5.223651965255864e-06,
+      "loss": 0.1032,
+      "step": 648
+    },
+    {
+      "epoch": 0.9730134932533733,
+      "grad_norm": 0.6182927867003201,
+      "learning_rate": 5.211888056763029e-06,
+      "loss": 0.0568,
+      "step": 649
+    },
+    {
+      "epoch": 0.974512743628186,
+      "grad_norm": 0.8084233917507633,
+      "learning_rate": 5.20012297311733e-06,
+      "loss": 0.0833,
+      "step": 650
+    },
+    {
+      "epoch": 0.9760119940029985,
+      "grad_norm": 0.6041714276348784,
+      "learning_rate": 5.188356779569125e-06,
+      "loss": 0.0682,
+      "step": 651
+    },
+    {
+      "epoch": 0.9775112443778111,
+      "grad_norm": 0.789140223987544,
+      "learning_rate": 5.176589541374929e-06,
+      "loss": 0.0751,
+      "step": 652
+    },
+    {
+      "epoch": 0.9790104947526237,
+      "grad_norm": 0.8642974840768147,
+      "learning_rate": 5.164821323797051e-06,
+      "loss": 0.0848,
+      "step": 653
+    },
+    {
+      "epoch": 0.9805097451274363,
+      "grad_norm": 0.7847363496303763,
+      "learning_rate": 5.1530521921032305e-06,
+      "loss": 0.0716,
+      "step": 654
+    },
+    {
+      "epoch": 0.9820089955022488,
+      "grad_norm": 0.6976514963327977,
+      "learning_rate": 5.141282211566276e-06,
+      "loss": 0.0842,
+      "step": 655
+    },
+    {
+      "epoch": 0.9835082458770614,
+      "grad_norm": 0.8213938393361362,
+      "learning_rate": 5.129511447463705e-06,
+      "loss": 0.0773,
+      "step": 656
+    },
+    {
+      "epoch": 0.9850074962518741,
+      "grad_norm": 0.9392354299859436,
+      "learning_rate": 5.117739965077382e-06,
+      "loss": 0.0953,
+      "step": 657
+    },
+    {
+      "epoch": 0.9865067466266867,
+      "grad_norm": 0.5356346743064295,
+      "learning_rate": 5.105967829693155e-06,
+      "loss": 0.0545,
+      "step": 658
+    },
+    {
+      "epoch": 0.9880059970014993,
+      "grad_norm": 0.7786311187076186,
+      "learning_rate": 5.0941951066004906e-06,
+      "loss": 0.0701,
+      "step": 659
+    },
+    {
+      "epoch": 0.9895052473763118,
+      "grad_norm": 0.8120448796375336,
+      "learning_rate": 5.082421861092116e-06,
+      "loss": 0.0823,
+      "step": 660
+    },
+    {
+      "epoch": 0.9910044977511244,
+      "grad_norm": 0.6917251114265367,
+      "learning_rate": 5.0706481584636605e-06,
+      "loss": 0.0811,
+      "step": 661
+    },
+    {
+      "epoch": 0.992503748125937,
+      "grad_norm": 0.7225585502772094,
+      "learning_rate": 5.0588740640132805e-06,
+      "loss": 0.0722,
+      "step": 662
+    },
+    {
+      "epoch": 0.9940029985007496,
+      "grad_norm": 0.6585549798696143,
+      "learning_rate": 5.047099643041312e-06,
+      "loss": 0.0673,
+      "step": 663
+    },
+    {
+      "epoch": 0.9955022488755623,
+      "grad_norm": 0.6749201342466775,
+      "learning_rate": 5.0353249608499e-06,
+      "loss": 0.0749,
+      "step": 664
+    },
+    {
+      "epoch": 0.9970014992503748,
+      "grad_norm": 0.9439178539262554,
+      "learning_rate": 5.023550082742637e-06,
+      "loss": 0.1103,
+      "step": 665
+    },
+    {
+      "epoch": 0.9985007496251874,
+      "grad_norm": 0.996289761613092,
+      "learning_rate": 5.011775074024202e-06,
+      "loss": 0.0892,
+      "step": 666
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5633389618403231,
+      "learning_rate": 5e-06,
+      "loss": 0.0476,
+      "step": 667
+    },
+    {
+      "epoch": 1.0014992503748126,
+      "grad_norm": 0.3809980823439671,
+      "learning_rate": 4.988224925975799e-06,
+      "loss": 0.0286,
+      "step": 668
+    },
+    {
+      "epoch": 1.0029985007496252,
+      "grad_norm": 0.49091959647249894,
+      "learning_rate": 4.976449917257365e-06,
+      "loss": 0.0298,
+      "step": 669
+    },
+    {
+      "epoch": 1.0044977511244377,
+      "grad_norm": 0.5187531791558389,
+      "learning_rate": 4.964675039150102e-06,
+      "loss": 0.0325,
+      "step": 670
+    },
+    {
+      "epoch": 1.0059970014992503,
+      "grad_norm": 0.45354105354685575,
+      "learning_rate": 4.952900356958689e-06,
+      "loss": 0.0343,
+      "step": 671
+    },
+    {
+      "epoch": 1.0074962518740629,
+      "grad_norm": 0.4806651128122064,
+      "learning_rate": 4.941125935986721e-06,
+      "loss": 0.0268,
+      "step": 672
+    },
+    {
+      "epoch": 1.0089955022488755,
+      "grad_norm": 0.4811976447619468,
+      "learning_rate": 4.929351841536342e-06,
+      "loss": 0.0323,
+      "step": 673
+    },
+    {
+      "epoch": 1.0104947526236883,
+      "grad_norm": 0.38106810211851155,
+      "learning_rate": 4.917578138907884e-06,
+      "loss": 0.0272,
+      "step": 674
+    },
+    {
+      "epoch": 1.0119940029985008,
+      "grad_norm": 0.4476102215823303,
+      "learning_rate": 4.90580489339951e-06,
+      "loss": 0.0264,
+      "step": 675
+    },
+    {
+      "epoch": 1.0134932533733134,
+      "grad_norm": 0.41357274081893003,
+      "learning_rate": 4.894032170306846e-06,
+      "loss": 0.0239,
+      "step": 676
+    },
+    {
+      "epoch": 1.014992503748126,
+      "grad_norm": 0.41395068501608717,
+      "learning_rate": 4.882260034922618e-06,
+      "loss": 0.0252,
+      "step": 677
+    },
+    {
+      "epoch": 1.0164917541229386,
+      "grad_norm": 0.28383736168464613,
+      "learning_rate": 4.870488552536296e-06,
+      "loss": 0.0181,
+      "step": 678
+    },
+    {
+      "epoch": 1.0179910044977512,
+      "grad_norm": 0.6878396493620234,
+      "learning_rate": 4.858717788433725e-06,
+      "loss": 0.0425,
+      "step": 679
+    },
+    {
+      "epoch": 1.0194902548725637,
+      "grad_norm": 0.46520730599746907,
+      "learning_rate": 4.846947807896771e-06,
+      "loss": 0.0344,
+      "step": 680
+    },
+    {
+      "epoch": 1.0209895052473763,
+      "grad_norm": 0.7801047189684277,
+      "learning_rate": 4.83517867620295e-06,
+      "loss": 0.0362,
+      "step": 681
+    },
+    {
+      "epoch": 1.0224887556221889,
+      "grad_norm": 0.29158374597557807,
+      "learning_rate": 4.823410458625072e-06,
+      "loss": 0.019,
+      "step": 682
+    },
+    {
+      "epoch": 1.0239880059970015,
+      "grad_norm": 0.4630290226691293,
+      "learning_rate": 4.811643220430877e-06,
+      "loss": 0.0289,
+      "step": 683
+    },
+    {
+      "epoch": 1.025487256371814,
+      "grad_norm": 0.544310175215777,
+      "learning_rate": 4.7998770268826726e-06,
+      "loss": 0.0303,
+      "step": 684
+    },
+    {
+      "epoch": 1.0269865067466266,
+      "grad_norm": 0.35362684967390356,
+      "learning_rate": 4.788111943236973e-06,
+      "loss": 0.0189,
+      "step": 685
+    },
+    {
+      "epoch": 1.0284857571214392,
+      "grad_norm": 0.6120844827093161,
+      "learning_rate": 4.7763480347441395e-06,
+      "loss": 0.035,
+      "step": 686
+    },
+    {
+      "epoch": 1.0299850074962518,
+      "grad_norm": 0.3632047830080879,
+      "learning_rate": 4.7645853666480104e-06,
+      "loss": 0.0205,
+      "step": 687
+    },
+    {
+      "epoch": 1.0314842578710646,
+      "grad_norm": 0.4131459089699973,
+      "learning_rate": 4.752824004185548e-06,
+      "loss": 0.0204,
+      "step": 688
+    },
+    {
+      "epoch": 1.0329835082458771,
+      "grad_norm": 0.4566281737953875,
+      "learning_rate": 4.7410640125864785e-06,
+      "loss": 0.0217,
+      "step": 689
+    },
+    {
+      "epoch": 1.0344827586206897,
+      "grad_norm": 0.5541356245832727,
+      "learning_rate": 4.729305457072913e-06,
+      "loss": 0.0289,
+      "step": 690
+    },
+    {
+      "epoch": 1.0359820089955023,
+      "grad_norm": 0.3494052633534004,
+      "learning_rate": 4.717548402859008e-06,
+      "loss": 0.0185,
+      "step": 691
+    },
+    {
+      "epoch": 1.0374812593703149,
+      "grad_norm": 0.5591197740374505,
+      "learning_rate": 4.7057929151505895e-06,
+      "loss": 0.0277,
+      "step": 692
+    },
+    {
+      "epoch": 1.0389805097451275,
+      "grad_norm": 0.4235854780296674,
+      "learning_rate": 4.694039059144797e-06,
+      "loss": 0.0178,
+      "step": 693
+    },
+    {
+      "epoch": 1.04047976011994,
+      "grad_norm": 0.4688111558851222,
+      "learning_rate": 4.6822869000297185e-06,
+      "loss": 0.0213,
+      "step": 694
+    },
+    {
+      "epoch": 1.0419790104947526,
+      "grad_norm": 0.6713782912874572,
+      "learning_rate": 4.670536502984033e-06,
+      "loss": 0.0346,
+      "step": 695
+    },
+    {
+      "epoch": 1.0434782608695652,
+      "grad_norm": 0.3896710552965651,
+      "learning_rate": 4.6587879331766465e-06,
+      "loss": 0.0153,
+      "step": 696
+    },
+    {
+      "epoch": 1.0449775112443778,
+      "grad_norm": 0.3822310907728224,
+      "learning_rate": 4.647041255766329e-06,
+      "loss": 0.0183,
+      "step": 697
+    },
+    {
+      "epoch": 1.0464767616191903,
+      "grad_norm": 0.42896967868015895,
+      "learning_rate": 4.6352965359013576e-06,
+      "loss": 0.0169,
+      "step": 698
+    },
+    {
+      "epoch": 1.047976011994003,
+      "grad_norm": 0.4967892385935207,
+      "learning_rate": 4.623553838719151e-06,
+      "loss": 0.0286,
+      "step": 699
+    },
+    {
+      "epoch": 1.0494752623688155,
+      "grad_norm": 0.48627102835023317,
+      "learning_rate": 4.611813229345911e-06,
+      "loss": 0.0221,
+      "step": 700
+    },
+    {
+      "epoch": 1.050974512743628,
+      "grad_norm": 0.272752884391699,
+      "learning_rate": 4.6000747728962606e-06,
+      "loss": 0.0098,
+      "step": 701
+    },
+    {
+      "epoch": 1.0524737631184409,
+      "grad_norm": 0.46493224035889,
+      "learning_rate": 4.588338534472878e-06,
+      "loss": 0.0167,
+      "step": 702
+    },
+    {
+      "epoch": 1.0539730134932535,
+      "grad_norm": 0.7208580112573136,
+      "learning_rate": 4.576604579166147e-06,
+      "loss": 0.0375,
+      "step": 703
+    },
+    {
+      "epoch": 1.055472263868066,
+      "grad_norm": 0.568651903219971,
+      "learning_rate": 4.564872972053786e-06,
+      "loss": 0.0206,
+      "step": 704
+    },
+    {
+      "epoch": 1.0569715142428786,
+      "grad_norm": 0.703486589898687,
+      "learning_rate": 4.553143778200486e-06,
+      "loss": 0.0387,
+      "step": 705
+    },
+    {
+      "epoch": 1.0584707646176912,
+      "grad_norm": 0.6318457170171868,
+      "learning_rate": 4.541417062657561e-06,
+      "loss": 0.032,
+      "step": 706
+    },
+    {
+      "epoch": 1.0599700149925038,
+      "grad_norm": 0.5876711844364544,
+      "learning_rate": 4.529692890462574e-06,
+      "loss": 0.0195,
+      "step": 707
+    },
+    {
+      "epoch": 1.0614692653673163,
+      "grad_norm": 0.47802205244078005,
+      "learning_rate": 4.5179713266389866e-06,
+      "loss": 0.0199,
+      "step": 708
+    },
+    {
+      "epoch": 1.062968515742129,
+      "grad_norm": 0.4933615275078171,
+      "learning_rate": 4.50625243619579e-06,
+      "loss": 0.0154,
+      "step": 709
+    },
+    {
+      "epoch": 1.0644677661169415,
+      "grad_norm": 0.521175513714302,
+      "learning_rate": 4.494536284127155e-06,
+      "loss": 0.0191,
+      "step": 710
+    },
+    {
+      "epoch": 1.065967016491754,
+      "grad_norm": 0.4852237140304914,
+      "learning_rate": 4.4828229354120565e-06,
+      "loss": 0.015,
+      "step": 711
+    },
+    {
+      "epoch": 1.0674662668665666,
+      "grad_norm": 0.7010605708539257,
+      "learning_rate": 4.471112455013928e-06,
+      "loss": 0.0294,
+      "step": 712
+    },
+    {
+      "epoch": 1.0689655172413792,
+      "grad_norm": 0.5272862883890127,
+      "learning_rate": 4.459404907880293e-06,
+      "loss": 0.0194,
+      "step": 713
+    },
+    {
+      "epoch": 1.0704647676161918,
+      "grad_norm": 0.6464878542718074,
+      "learning_rate": 4.447700358942407e-06,
+      "loss": 0.0266,
+      "step": 714
+    },
+    {
+      "epoch": 1.0719640179910046,
+      "grad_norm": 0.5137953425218037,
+      "learning_rate": 4.435998873114895e-06,
+      "loss": 0.0174,
+      "step": 715
+    },
+    {
+      "epoch": 1.0734632683658172,
+      "grad_norm": 0.7267234338314322,
+      "learning_rate": 4.424300515295401e-06,
+      "loss": 0.025,
+      "step": 716
+    },
+    {
+      "epoch": 1.0749625187406298,
+      "grad_norm": 0.5660408165406456,
+      "learning_rate": 4.412605350364213e-06,
+      "loss": 0.0157,
+      "step": 717
+    },
+    {
+      "epoch": 1.0764617691154423,
+      "grad_norm": 0.4823656253788781,
+      "learning_rate": 4.400913443183913e-06,
+      "loss": 0.011,
+      "step": 718
+    },
+    {
+      "epoch": 1.077961019490255,
+      "grad_norm": 0.7867880052617067,
+      "learning_rate": 4.389224858599015e-06,
+      "loss": 0.0238,
+      "step": 719
+    },
+    {
+      "epoch": 1.0794602698650675,
+      "grad_norm": 0.5395480144927652,
+      "learning_rate": 4.377539661435608e-06,
+      "loss": 0.023,
+      "step": 720
+    },
+    {
+      "epoch": 1.08095952023988,
+      "grad_norm": 0.5759906043019002,
+      "learning_rate": 4.365857916500991e-06,
+      "loss": 0.0228,
+      "step": 721
+    },
+    {
+      "epoch": 1.0824587706146926,
+      "grad_norm": 0.5486020437905004,
+      "learning_rate": 4.35417968858332e-06,
+      "loss": 0.0279,
+      "step": 722
+    },
+    {
+      "epoch": 1.0839580209895052,
+      "grad_norm": 0.4552033958253014,
+      "learning_rate": 4.3425050424512405e-06,
+      "loss": 0.0167,
+      "step": 723
+    },
+    {
+      "epoch": 1.0854572713643178,
+      "grad_norm": 0.7506222854082746,
+      "learning_rate": 4.330834042853537e-06,
+      "loss": 0.0289,
+      "step": 724
+    },
+    {
+      "epoch": 1.0869565217391304,
+      "grad_norm": 1.1788209696711796,
+      "learning_rate": 4.319166754518768e-06,
+      "loss": 0.0483,
+      "step": 725
+    },
+    {
+      "epoch": 1.088455772113943,
+      "grad_norm": 0.48451300617333715,
+      "learning_rate": 4.30750324215491e-06,
+      "loss": 0.0163,
+      "step": 726
+    },
+    {
+      "epoch": 1.0899550224887555,
+      "grad_norm": 0.5783085136043674,
+      "learning_rate": 4.295843570448998e-06,
+      "loss": 0.0236,
+      "step": 727
+    },
+    {
+      "epoch": 1.0914542728635683,
+      "grad_norm": 0.6151248856545246,
+      "learning_rate": 4.284187804066764e-06,
+      "loss": 0.0267,
+      "step": 728
+    },
+    {
+      "epoch": 1.092953523238381,
+      "grad_norm": 0.6016276774635799,
+      "learning_rate": 4.272536007652281e-06,
+      "loss": 0.0259,
+      "step": 729
+    },
+    {
+      "epoch": 1.0944527736131935,
+      "grad_norm": 0.5173538742047529,
+      "learning_rate": 4.260888245827608e-06,
+      "loss": 0.0206,
+      "step": 730
+    },
+    {
+      "epoch": 1.095952023988006,
+      "grad_norm": 0.6064865939150763,
+      "learning_rate": 4.249244583192425e-06,
+      "loss": 0.0205,
+      "step": 731
+    },
+    {
+      "epoch": 1.0974512743628186,
+      "grad_norm": 0.6326015319518571,
+      "learning_rate": 4.237605084323676e-06,
+      "loss": 0.0193,
+      "step": 732
+    },
+    {
+      "epoch": 1.0989505247376312,
+      "grad_norm": 0.5777260412516618,
+      "learning_rate": 4.225969813775215e-06,
+      "loss": 0.0282,
+      "step": 733
+    },
+    {
+      "epoch": 1.1004497751124438,
+      "grad_norm": 0.47143856513091925,
+      "learning_rate": 4.214338836077444e-06,
+      "loss": 0.0249,
+      "step": 734
+    },
+    {
+      "epoch": 1.1019490254872564,
+      "grad_norm": 0.5093551655170319,
+      "learning_rate": 4.202712215736955e-06,
+      "loss": 0.0207,
+      "step": 735
+    },
+    {
+      "epoch": 1.103448275862069,
+      "grad_norm": 0.44893934975136457,
+      "learning_rate": 4.191090017236177e-06,
+      "loss": 0.0201,
+      "step": 736
+    },
+    {
+      "epoch": 1.1049475262368815,
+      "grad_norm": 0.4490424692478759,
+      "learning_rate": 4.1794723050330125e-06,
+      "loss": 0.0209,
+      "step": 737
+    },
+    {
+      "epoch": 1.106446776611694,
+      "grad_norm": 0.5052940778578281,
+      "learning_rate": 4.167859143560484e-06,
+      "loss": 0.0251,
+      "step": 738
+    },
+    {
+      "epoch": 1.1079460269865067,
+      "grad_norm": 0.5408832646587565,
+      "learning_rate": 4.1562505972263735e-06,
+      "loss": 0.0273,
+      "step": 739
+    },
+    {
+      "epoch": 1.1094452773613193,
+      "grad_norm": 0.6483993385769797,
+      "learning_rate": 4.144646730412868e-06,
+      "loss": 0.0299,
+      "step": 740
+    },
+    {
+      "epoch": 1.1109445277361318,
+      "grad_norm": 0.5161392056894014,
+      "learning_rate": 4.133047607476202e-06,
+      "loss": 0.0234,
+      "step": 741
+    },
+    {
+      "epoch": 1.1124437781109444,
+      "grad_norm": 0.5066851903708308,
+      "learning_rate": 4.121453292746297e-06,
+      "loss": 0.0245,
+      "step": 742
+    },
+    {
+      "epoch": 1.1139430284857572,
+      "grad_norm": 0.4726931175740864,
+      "learning_rate": 4.109863850526413e-06,
+      "loss": 0.0276,
+      "step": 743
+    },
+    {
+      "epoch": 1.1154422788605698,
+      "grad_norm": 0.41912410565450925,
+      "learning_rate": 4.098279345092783e-06,
+      "loss": 0.0183,
+      "step": 744
+    },
+    {
+      "epoch": 1.1169415292353824,
+      "grad_norm": 0.4247530101628744,
+      "learning_rate": 4.086699840694262e-06,
+      "loss": 0.0164,
+      "step": 745
+    },
+    {
+      "epoch": 1.118440779610195,
+      "grad_norm": 0.46193448179248514,
+      "learning_rate": 4.075125401551963e-06,
+      "loss": 0.0213,
+      "step": 746
+    },
+    {
+      "epoch": 1.1199400299850075,
+      "grad_norm": 0.4844773028321289,
+      "learning_rate": 4.063556091858914e-06,
+      "loss": 0.0195,
+      "step": 747
+    },
+    {
+      "epoch": 1.12143928035982,
+      "grad_norm": 0.5924040291604895,
+      "learning_rate": 4.051991975779691e-06,
+      "loss": 0.0299,
+      "step": 748
+    },
+    {
+      "epoch": 1.1229385307346327,
+      "grad_norm": 0.7348048146082546,
+      "learning_rate": 4.040433117450066e-06,
+      "loss": 0.0314,
+      "step": 749
+    },
+    {
+      "epoch": 1.1244377811094453,
+      "grad_norm": 0.5865108234375199,
+      "learning_rate": 4.0288795809766516e-06,
+      "loss": 0.0255,
+      "step": 750
+    },
+    {
+      "epoch": 1.1259370314842578,
+      "grad_norm": 0.6423480557578461,
+      "learning_rate": 4.017331430436543e-06,
+      "loss": 0.0354,
+      "step": 751
+    },
+    {
+      "epoch": 1.1274362818590704,
+      "grad_norm": 0.6539523373963867,
+      "learning_rate": 4.005788729876968e-06,
+      "loss": 0.0265,
+      "step": 752
+    },
+    {
+      "epoch": 1.128935532233883,
+      "grad_norm": 0.4905987000097847,
+      "learning_rate": 3.994251543314925e-06,
+      "loss": 0.0198,
+      "step": 753
+    },
+    {
+      "epoch": 1.1304347826086956,
+      "grad_norm": 0.4693791216996954,
+      "learning_rate": 3.982719934736832e-06,
+      "loss": 0.0196,
+      "step": 754
+    },
+    {
+      "epoch": 1.1319340329835081,
+      "grad_norm": 0.779435375100094,
+      "learning_rate": 3.971193968098172e-06,
+      "loss": 0.0351,
+      "step": 755
+    },
+    {
+      "epoch": 1.133433283358321,
+      "grad_norm": 0.5711127762282515,
+      "learning_rate": 3.959673707323135e-06,
+      "loss": 0.0307,
+      "step": 756
+    },
+    {
+      "epoch": 1.1349325337331335,
+      "grad_norm": 0.5860100845265914,
+      "learning_rate": 3.948159216304269e-06,
+      "loss": 0.03,
+      "step": 757
+    },
+    {
+      "epoch": 1.136431784107946,
+      "grad_norm": 0.44243204908376227,
+      "learning_rate": 3.93665055890212e-06,
+      "loss": 0.0156,
+      "step": 758
+    },
+    {
+      "epoch": 1.1379310344827587,
+      "grad_norm": 0.6414017669228939,
+      "learning_rate": 3.92514779894488e-06,
+      "loss": 0.0338,
+      "step": 759
+    },
+    {
+      "epoch": 1.1394302848575713,
+      "grad_norm": 0.6827079548014926,
+      "learning_rate": 3.9136510002280344e-06,
+      "loss": 0.025,
+      "step": 760
+    },
+    {
+      "epoch": 1.1409295352323838,
+      "grad_norm": 0.3537937042148946,
+      "learning_rate": 3.902160226514007e-06,
+      "loss": 0.0122,
+      "step": 761
+    },
+    {
+      "epoch": 1.1424287856071964,
+      "grad_norm": 0.5150647403816233,
+      "learning_rate": 3.8906755415318045e-06,
+      "loss": 0.0265,
+      "step": 762
+    },
+    {
+      "epoch": 1.143928035982009,
+      "grad_norm": 0.4248302967786211,
+      "learning_rate": 3.8791970089766665e-06,
+      "loss": 0.0199,
+      "step": 763
+    },
+    {
+      "epoch": 1.1454272863568216,
+      "grad_norm": 0.5500572004309948,
+      "learning_rate": 3.86772469250971e-06,
+      "loss": 0.0291,
+      "step": 764
+    },
+    {
+      "epoch": 1.1469265367316341,
+      "grad_norm": 0.4973536012123906,
+      "learning_rate": 3.856258655757578e-06,
+      "loss": 0.019,
+      "step": 765
+    },
+    {
+      "epoch": 1.1484257871064467,
+      "grad_norm": 0.5650663032957277,
+      "learning_rate": 3.844798962312085e-06,
+      "loss": 0.0276,
+      "step": 766
+    },
+    {
+      "epoch": 1.1499250374812593,
+      "grad_norm": 0.3994716671260626,
+      "learning_rate": 3.833345675729863e-06,
+      "loss": 0.0176,
+      "step": 767
+    },
+    {
+      "epoch": 1.1514242878560719,
+      "grad_norm": 0.5362001440843644,
+      "learning_rate": 3.821898859532013e-06,
+      "loss": 0.0233,
+      "step": 768
+    },
+    {
+      "epoch": 1.1529235382308847,
+      "grad_norm": 0.6375362235810208,
+      "learning_rate": 3.8104585772037493e-06,
+      "loss": 0.0166,
+      "step": 769
+    },
+    {
+      "epoch": 1.1544227886056972,
+      "grad_norm": 0.756855667852829,
+      "learning_rate": 3.7990248921940485e-06,
+      "loss": 0.0335,
+      "step": 770
+    },
+    {
+      "epoch": 1.1559220389805098,
+      "grad_norm": 0.7686117357571411,
+      "learning_rate": 3.787597867915303e-06,
+      "loss": 0.0301,
+      "step": 771
+    },
+    {
+      "epoch": 1.1574212893553224,
+      "grad_norm": 0.6101202569369653,
+      "learning_rate": 3.7761775677429567e-06,
+      "loss": 0.0221,
+      "step": 772
+    },
+    {
+      "epoch": 1.158920539730135,
+      "grad_norm": 0.5043189066105882,
+      "learning_rate": 3.7647640550151666e-06,
+      "loss": 0.0196,
+      "step": 773
+    },
+    {
+      "epoch": 1.1604197901049476,
+      "grad_norm": 0.48182507477926056,
+      "learning_rate": 3.7533573930324395e-06,
+      "loss": 0.018,
+      "step": 774
+    },
+    {
+      "epoch": 1.1619190404797601,
+      "grad_norm": 0.511150067153958,
+      "learning_rate": 3.7419576450572924e-06,
+      "loss": 0.0157,
+      "step": 775
+    },
+    {
+      "epoch": 1.1634182908545727,
+      "grad_norm": 0.5807912860158657,
+      "learning_rate": 3.7305648743138966e-06,
+      "loss": 0.0244,
+      "step": 776
+    },
+    {
+      "epoch": 1.1649175412293853,
+      "grad_norm": 0.5605633190251349,
+      "learning_rate": 3.7191791439877236e-06,
+      "loss": 0.0227,
+      "step": 777
+    },
+    {
+      "epoch": 1.1664167916041979,
+      "grad_norm": 0.3947259447983112,
+      "learning_rate": 3.7078005172252015e-06,
+      "loss": 0.0129,
+      "step": 778
+    },
+    {
+      "epoch": 1.1679160419790104,
+      "grad_norm": 0.45291547445711866,
+      "learning_rate": 3.6964290571333583e-06,
+      "loss": 0.019,
+      "step": 779
+    },
+    {
+      "epoch": 1.169415292353823,
+      "grad_norm": 0.545695737314298,
+      "learning_rate": 3.6850648267794776e-06,
+      "loss": 0.0179,
+      "step": 780
+    },
+    {
+      "epoch": 1.1709145427286356,
+      "grad_norm": 0.46123366052813847,
+      "learning_rate": 3.673707889190744e-06,
+      "loss": 0.0182,
+      "step": 781
+    },
+    {
+      "epoch": 1.1724137931034484,
+      "grad_norm": 0.5129814051767793,
+      "learning_rate": 3.662358307353897e-06,
+      "loss": 0.0164,
+      "step": 782
+    },
+    {
+      "epoch": 1.1739130434782608,
+      "grad_norm": 0.5839578290685485,
+      "learning_rate": 3.6510161442148783e-06,
+      "loss": 0.0227,
+      "step": 783
+    },
+    {
+      "epoch": 1.1754122938530736,
+      "grad_norm": 0.4596524778342608,
+      "learning_rate": 3.63968146267849e-06,
+      "loss": 0.0176,
+      "step": 784
+    },
+    {
+      "epoch": 1.1769115442278861,
+      "grad_norm": 0.5723746187720152,
+      "learning_rate": 3.6283543256080334e-06,
+      "loss": 0.0214,
+      "step": 785
+    },
+    {
+      "epoch": 1.1784107946026987,
+      "grad_norm": 0.5052446370445101,
+      "learning_rate": 3.6170347958249728e-06,
+      "loss": 0.0202,
+      "step": 786
+    },
+    {
+      "epoch": 1.1799100449775113,
+      "grad_norm": 0.6034064106167325,
+      "learning_rate": 3.605722936108579e-06,
+      "loss": 0.0201,
+      "step": 787
+    },
+    {
+      "epoch": 1.1814092953523239,
+      "grad_norm": 0.5960041235879205,
+      "learning_rate": 3.5944188091955843e-06,
+      "loss": 0.0278,
+      "step": 788
+    },
+    {
+      "epoch": 1.1829085457271364,
+      "grad_norm": 0.4044248711460727,
+      "learning_rate": 3.5831224777798346e-06,
+      "loss": 0.0158,
+      "step": 789
+    },
+    {
+      "epoch": 1.184407796101949,
+      "grad_norm": 0.3552214358162709,
+      "learning_rate": 3.5718340045119416e-06,
+      "loss": 0.0124,
+      "step": 790
+    },
+    {
+      "epoch": 1.1859070464767616,
+      "grad_norm": 0.5039788542394908,
+      "learning_rate": 3.5605534519989327e-06,
+      "loss": 0.0206,
+      "step": 791
+    },
+    {
+      "epoch": 1.1874062968515742,
+      "grad_norm": 0.487129035849391,
+      "learning_rate": 3.5492808828039083e-06,
+      "loss": 0.0146,
+      "step": 792
+    },
+    {
+      "epoch": 1.1889055472263867,
+      "grad_norm": 0.4917076445685134,
+      "learning_rate": 3.538016359445692e-06,
+      "loss": 0.0207,
+      "step": 793
+    },
+    {
+      "epoch": 1.1904047976011993,
+      "grad_norm": 0.6097610028128922,
+      "learning_rate": 3.5267599443984848e-06,
+      "loss": 0.0258,
+      "step": 794
+    },
+    {
+      "epoch": 1.191904047976012,
+      "grad_norm": 0.656383280577153,
+      "learning_rate": 3.5155117000915153e-06,
+      "loss": 0.0315,
+      "step": 795
+    },
+    {
+      "epoch": 1.1934032983508245,
+      "grad_norm": 0.6064853518886382,
+      "learning_rate": 3.5042716889086998e-06,
+      "loss": 0.0219,
+      "step": 796
+    },
+    {
+      "epoch": 1.1949025487256373,
+      "grad_norm": 0.6661248524695417,
+      "learning_rate": 3.493039973188289e-06,
+      "loss": 0.0291,
+      "step": 797
+    },
+    {
+      "epoch": 1.1964017991004499,
+      "grad_norm": 0.4889550005413012,
+      "learning_rate": 3.481816615222533e-06,
+      "loss": 0.0139,
+      "step": 798
+    },
+    {
+      "epoch": 1.1979010494752624,
+      "grad_norm": 0.44643118683980454,
+      "learning_rate": 3.470601677257323e-06,
+      "loss": 0.0156,
+      "step": 799
+    },
+    {
+      "epoch": 1.199400299850075,
+      "grad_norm": 0.5207360828651232,
+      "learning_rate": 3.459395221491853e-06,
+      "loss": 0.0165,
+      "step": 800
+    },
+    {
+      "epoch": 1.199400299850075,
+      "eval_loss": 0.1017409935593605,
+      "eval_runtime": 8.1098,
+      "eval_samples_per_second": 6.659,
+      "eval_steps_per_second": 1.726,
+      "step": 800
+    },
+    {
+      "epoch": 1.2008995502248876,
+      "grad_norm": 0.6189065843801586,
+      "learning_rate": 3.4481973100782756e-06,
+      "loss": 0.025,
+      "step": 801
+    },
+    {
+      "epoch": 1.2023988005997002,
+      "grad_norm": 0.7764345409592021,
+      "learning_rate": 3.4370080051213527e-06,
+      "loss": 0.037,
+      "step": 802
+    },
+    {
+      "epoch": 1.2038980509745127,
+      "grad_norm": 0.4271615967116821,
+      "learning_rate": 3.4258273686781156e-06,
+      "loss": 0.0177,
+      "step": 803
+    },
+    {
+      "epoch": 1.2053973013493253,
+      "grad_norm": 0.7051681371299896,
+      "learning_rate": 3.4146554627575207e-06,
+      "loss": 0.0244,
+      "step": 804
+    },
+    {
+      "epoch": 1.206896551724138,
+      "grad_norm": 0.5876839735148423,
+      "learning_rate": 3.403492349320101e-06,
+      "loss": 0.0242,
+      "step": 805
+    },
+    {
+      "epoch": 1.2083958020989505,
+      "grad_norm": 0.2981882834969981,
+      "learning_rate": 3.392338090277628e-06,
+      "loss": 0.0096,
+      "step": 806
+    },
+    {
+      "epoch": 1.209895052473763,
+      "grad_norm": 0.4860127054955284,
+      "learning_rate": 3.3811927474927644e-06,
+      "loss": 0.0217,
+      "step": 807
+    },
+    {
+      "epoch": 1.2113943028485756,
+      "grad_norm": 0.6431446205865087,
+      "learning_rate": 3.3700563827787224e-06,
+      "loss": 0.0251,
+      "step": 808
+    },
+    {
+      "epoch": 1.2128935532233882,
+      "grad_norm": 0.5165066424249244,
+      "learning_rate": 3.358929057898922e-06,
+      "loss": 0.0167,
+      "step": 809
+    },
+    {
+      "epoch": 1.214392803598201,
+      "grad_norm": 0.619327844693917,
+      "learning_rate": 3.3478108345666456e-06,
+      "loss": 0.0269,
+      "step": 810
+    },
+    {
+      "epoch": 1.2158920539730136,
+      "grad_norm": 0.5145216222382173,
+      "learning_rate": 3.3367017744446995e-06,
+      "loss": 0.0246,
+      "step": 811
+    },
+    {
+      "epoch": 1.2173913043478262,
+      "grad_norm": 0.7565087088235075,
+      "learning_rate": 3.3256019391450696e-06,
+      "loss": 0.0358,
+      "step": 812
+    },
+    {
+      "epoch": 1.2188905547226387,
+      "grad_norm": 0.4509811677574059,
+      "learning_rate": 3.314511390228578e-06,
+      "loss": 0.0167,
+      "step": 813
+    },
+    {
+      "epoch": 1.2203898050974513,
+      "grad_norm": 0.4955287260584956,
+      "learning_rate": 3.303430189204545e-06,
+      "loss": 0.0216,
+      "step": 814
+    },
+    {
+      "epoch": 1.221889055472264,
+      "grad_norm": 0.80009861439639,
+      "learning_rate": 3.2923583975304474e-06,
+      "loss": 0.0369,
+      "step": 815
+    },
+    {
+      "epoch": 1.2233883058470765,
+      "grad_norm": 0.4906122598962174,
+      "learning_rate": 3.2812960766115747e-06,
+      "loss": 0.018,
+      "step": 816
+    },
+    {
+      "epoch": 1.224887556221889,
+      "grad_norm": 0.6754573557996651,
+      "learning_rate": 3.270243287800693e-06,
+      "loss": 0.033,
+      "step": 817
+    },
+    {
+      "epoch": 1.2263868065967016,
+      "grad_norm": 0.43988888662125986,
+      "learning_rate": 3.2592000923976997e-06,
+      "loss": 0.0149,
+      "step": 818
+    },
+    {
+      "epoch": 1.2278860569715142,
+      "grad_norm": 0.5297095822921374,
+      "learning_rate": 3.2481665516492876e-06,
+      "loss": 0.0258,
+      "step": 819
+    },
+    {
+      "epoch": 1.2293853073463268,
+      "grad_norm": 0.46193018431719046,
+      "learning_rate": 3.2371427267486044e-06,
+      "loss": 0.0211,
+      "step": 820
+    },
+    {
+      "epoch": 1.2308845577211394,
+      "grad_norm": 0.4950714239310738,
+      "learning_rate": 3.2261286788349127e-06,
+      "loss": 0.0194,
+      "step": 821
+    },
+    {
+      "epoch": 1.232383808095952,
+      "grad_norm": 0.5631596108276786,
+      "learning_rate": 3.2151244689932505e-06,
+      "loss": 0.0264,
+      "step": 822
+    },
+    {
+      "epoch": 1.2338830584707647,
+      "grad_norm": 0.5896237455780339,
+      "learning_rate": 3.2041301582540903e-06,
+      "loss": 0.0213,
+      "step": 823
+    },
+    {
+      "epoch": 1.235382308845577,
+      "grad_norm": 0.5195137733424808,
+      "learning_rate": 3.1931458075930046e-06,
+      "loss": 0.0224,
+      "step": 824
+    },
+    {
+      "epoch": 1.23688155922039,
+      "grad_norm": 0.3965664740361069,
+      "learning_rate": 3.182171477930332e-06,
+      "loss": 0.0142,
+      "step": 825
+    },
+    {
+      "epoch": 1.2383808095952025,
+      "grad_norm": 0.36388656837812033,
+      "learning_rate": 3.171207230130826e-06,
+      "loss": 0.0138,
+      "step": 826
+    },
+    {
+      "epoch": 1.239880059970015,
+      "grad_norm": 0.5467301774037692,
+      "learning_rate": 3.1602531250033286e-06,
+      "loss": 0.0273,
+      "step": 827
+    },
+    {
+      "epoch": 1.2413793103448276,
+      "grad_norm": 0.44984185750936967,
+      "learning_rate": 3.149309223300428e-06,
+      "loss": 0.0183,
+      "step": 828
+    },
+    {
+      "epoch": 1.2428785607196402,
+      "grad_norm": 0.5162843270202278,
+      "learning_rate": 3.1383755857181253e-06,
+      "loss": 0.0209,
+      "step": 829
+    },
+    {
+      "epoch": 1.2443778110944528,
+      "grad_norm": 0.32362652726984936,
+      "learning_rate": 3.1274522728954928e-06,
+      "loss": 0.012,
+      "step": 830
+    },
+    {
+      "epoch": 1.2458770614692654,
+      "grad_norm": 0.5794131996516068,
+      "learning_rate": 3.1165393454143423e-06,
+      "loss": 0.0257,
+      "step": 831
+    },
+    {
+      "epoch": 1.247376311844078,
+      "grad_norm": 0.6471108705645412,
+      "learning_rate": 3.1056368637988876e-06,
+      "loss": 0.0324,
+      "step": 832
+    },
+    {
+      "epoch": 1.2488755622188905,
+      "grad_norm": 0.5299106903222452,
+      "learning_rate": 3.0947448885154085e-06,
+      "loss": 0.0232,
+      "step": 833
+    },
+    {
+      "epoch": 1.250374812593703,
+      "grad_norm": 0.556767044994862,
+      "learning_rate": 3.0838634799719157e-06,
+      "loss": 0.028,
+      "step": 834
+    },
+    {
+      "epoch": 1.2518740629685157,
+      "grad_norm": 0.38574939926568,
+      "learning_rate": 3.072992698517815e-06,
+      "loss": 0.0128,
+      "step": 835
+    },
+    {
+      "epoch": 1.2533733133433285,
+      "grad_norm": 0.6107798929179674,
+      "learning_rate": 3.0621326044435738e-06,
+      "loss": 0.02,
+      "step": 836
+    },
+    {
+      "epoch": 1.2548725637181408,
+      "grad_norm": 0.44566068205299164,
+      "learning_rate": 3.0512832579803873e-06,
+      "loss": 0.0197,
+      "step": 837
+    },
+    {
+      "epoch": 1.2563718140929536,
+      "grad_norm": 0.3950354736894474,
+      "learning_rate": 3.0404447192998398e-06,
+      "loss": 0.0107,
+      "step": 838
+    },
+    {
+      "epoch": 1.2578710644677662,
+      "grad_norm": 0.557352619629793,
+      "learning_rate": 3.029617048513579e-06,
+      "loss": 0.0166,
+      "step": 839
+    },
+    {
+      "epoch": 1.2593703148425788,
+      "grad_norm": 0.40833684638157214,
+      "learning_rate": 3.0188003056729752e-06,
+      "loss": 0.0197,
+      "step": 840
+    },
+    {
+      "epoch": 1.2608695652173914,
+      "grad_norm": 0.4253439599135722,
+      "learning_rate": 3.007994550768793e-06,
+      "loss": 0.0141,
+      "step": 841
+    },
+    {
+      "epoch": 1.262368815592204,
+      "grad_norm": 0.554056142050489,
+      "learning_rate": 2.9971998437308546e-06,
+      "loss": 0.0227,
+      "step": 842
+    },
+    {
+      "epoch": 1.2638680659670165,
+      "grad_norm": 0.6212590608390086,
+      "learning_rate": 2.9864162444277118e-06,
+      "loss": 0.0259,
+      "step": 843
+    },
+    {
+      "epoch": 1.265367316341829,
+      "grad_norm": 0.6216037540485144,
+      "learning_rate": 2.97564381266631e-06,
+      "loss": 0.0184,
+      "step": 844
+    },
+    {
+      "epoch": 1.2668665667166417,
+      "grad_norm": 0.35377463610476834,
+      "learning_rate": 2.964882608191659e-06,
+      "loss": 0.0094,
+      "step": 845
+    },
+    {
+      "epoch": 1.2683658170914542,
+      "grad_norm": 0.4861643186184286,
+      "learning_rate": 2.954132690686502e-06,
+      "loss": 0.013,
+      "step": 846
+    },
+    {
+      "epoch": 1.2698650674662668,
+      "grad_norm": 0.7562176916260294,
+      "learning_rate": 2.9433941197709813e-06,
+      "loss": 0.0352,
+      "step": 847
+    },
+    {
+      "epoch": 1.2713643178410794,
+      "grad_norm": 0.5952488641689456,
+      "learning_rate": 2.9326669550023124e-06,
+      "loss": 0.0264,
+      "step": 848
+    },
+    {
+      "epoch": 1.272863568215892,
+      "grad_norm": 0.49574814725772337,
+      "learning_rate": 2.921951255874449e-06,
+      "loss": 0.0172,
+      "step": 849
+    },
+    {
+      "epoch": 1.2743628185907045,
+      "grad_norm": 0.5172818186815066,
+      "learning_rate": 2.9112470818177563e-06,
+      "loss": 0.0194,
+      "step": 850
+    },
+    {
+      "epoch": 1.2758620689655173,
+      "grad_norm": 0.5782722376094496,
+      "learning_rate": 2.9005544921986774e-06,
+      "loss": 0.0243,
+      "step": 851
+    },
+    {
+      "epoch": 1.2773613193403297,
+      "grad_norm": 0.48679615726757264,
+      "learning_rate": 2.8898735463194128e-06,
+      "loss": 0.0192,
+      "step": 852
+    },
+    {
+      "epoch": 1.2788605697151425,
+      "grad_norm": 0.7060881280481219,
+      "learning_rate": 2.8792043034175817e-06,
+      "loss": 0.0338,
+      "step": 853
+    },
+    {
+      "epoch": 1.280359820089955,
+      "grad_norm": 0.45552337855447605,
+      "learning_rate": 2.8685468226658974e-06,
+      "loss": 0.0173,
+      "step": 854
+    },
+    {
+      "epoch": 1.2818590704647677,
+      "grad_norm": 0.48847423393689643,
+      "learning_rate": 2.85790116317184e-06,
+      "loss": 0.017,
+      "step": 855
+    },
+    {
+      "epoch": 1.2833583208395802,
+      "grad_norm": 0.397339999699135,
+      "learning_rate": 2.8472673839773267e-06,
+      "loss": 0.0123,
+      "step": 856
+    },
+    {
+      "epoch": 1.2848575712143928,
+      "grad_norm": 0.5291722736171363,
+      "learning_rate": 2.8366455440583874e-06,
+      "loss": 0.0286,
+      "step": 857
+    },
+    {
+      "epoch": 1.2863568215892054,
+      "grad_norm": 0.6130804467995362,
+      "learning_rate": 2.8260357023248323e-06,
+      "loss": 0.0233,
+      "step": 858
+    },
+    {
+      "epoch": 1.287856071964018,
+      "grad_norm": 0.5098302126621519,
+      "learning_rate": 2.815437917619932e-06,
+      "loss": 0.021,
+      "step": 859
+    },
+    {
+      "epoch": 1.2893553223388305,
+      "grad_norm": 0.5522991431279791,
+      "learning_rate": 2.804852248720085e-06,
+      "loss": 0.0234,
+      "step": 860
+    },
+    {
+      "epoch": 1.2908545727136431,
+      "grad_norm": 0.448104097460281,
+      "learning_rate": 2.7942787543344957e-06,
+      "loss": 0.0149,
+      "step": 861
+    },
+    {
+      "epoch": 1.2923538230884557,
+      "grad_norm": 0.44600922315032776,
+      "learning_rate": 2.783717493104846e-06,
+      "loss": 0.0116,
+      "step": 862
+    },
+    {
+      "epoch": 1.2938530734632683,
+      "grad_norm": 0.6124497741240319,
+      "learning_rate": 2.7731685236049745e-06,
+      "loss": 0.0327,
+      "step": 863
+    },
+    {
+      "epoch": 1.295352323838081,
+      "grad_norm": 0.6249431312124162,
+      "learning_rate": 2.762631904340546e-06,
+      "loss": 0.0253,
+      "step": 864
+    },
+    {
+      "epoch": 1.2968515742128934,
+      "grad_norm": 0.7438101459616736,
+      "learning_rate": 2.7521076937487248e-06,
+      "loss": 0.0361,
+      "step": 865
+    },
+    {
+      "epoch": 1.2983508245877062,
+      "grad_norm": 0.32818613313950656,
+      "learning_rate": 2.7415959501978674e-06,
+      "loss": 0.0089,
+      "step": 866
+    },
+    {
+      "epoch": 1.2998500749625188,
+      "grad_norm": 0.5172844769978538,
+      "learning_rate": 2.731096731987177e-06,
+      "loss": 0.0193,
+      "step": 867
+    },
+    {
+      "epoch": 1.3013493253373314,
+      "grad_norm": 0.5455559481876842,
+      "learning_rate": 2.7206100973463958e-06,
+      "loss": 0.018,
+      "step": 868
+    },
+    {
+      "epoch": 1.302848575712144,
+      "grad_norm": 0.6731888358494801,
+      "learning_rate": 2.71013610443547e-06,
+      "loss": 0.0342,
+      "step": 869
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 0.6308825997186805,
+      "learning_rate": 2.6996748113442397e-06,
+      "loss": 0.0215,
+      "step": 870
+    },
+    {
+      "epoch": 1.3058470764617691,
+      "grad_norm": 0.43009958383863106,
+      "learning_rate": 2.689226276092107e-06,
+      "loss": 0.0192,
+      "step": 871
+    },
+    {
+      "epoch": 1.3073463268365817,
+      "grad_norm": 0.5117005377270278,
+      "learning_rate": 2.6787905566277185e-06,
+      "loss": 0.0252,
+      "step": 872
+    },
+    {
+      "epoch": 1.3088455772113943,
+      "grad_norm": 0.4391719953510318,
+      "learning_rate": 2.6683677108286423e-06,
+      "loss": 0.0178,
+      "step": 873
+    },
+    {
+      "epoch": 1.3103448275862069,
+      "grad_norm": 0.6499239994734337,
+      "learning_rate": 2.65795779650105e-06,
+      "loss": 0.0319,
+      "step": 874
+    },
+    {
+      "epoch": 1.3118440779610194,
+      "grad_norm": 0.4815362170071885,
+      "learning_rate": 2.6475608713793923e-06,
+      "loss": 0.0188,
+      "step": 875
+    },
+    {
+      "epoch": 1.313343328335832,
+      "grad_norm": 0.45742772613023264,
+      "learning_rate": 2.6371769931260806e-06,
+      "loss": 0.0161,
+      "step": 876
+    },
+    {
+      "epoch": 1.3148425787106448,
+      "grad_norm": 0.4751316879095394,
+      "learning_rate": 2.6268062193311672e-06,
+      "loss": 0.0153,
+      "step": 877
+    },
+    {
+      "epoch": 1.3163418290854572,
+      "grad_norm": 0.5769786082137504,
+      "learning_rate": 2.6164486075120245e-06,
+      "loss": 0.0232,
+      "step": 878
+    },
+    {
+      "epoch": 1.31784107946027,
+      "grad_norm": 0.5971000519801224,
+      "learning_rate": 2.606104215113033e-06,
+      "loss": 0.0239,
+      "step": 879
+    },
+    {
+      "epoch": 1.3193403298350825,
+      "grad_norm": 0.47506609753255247,
+      "learning_rate": 2.5957730995052477e-06,
+      "loss": 0.0187,
+      "step": 880
+    },
+    {
+      "epoch": 1.3208395802098951,
+      "grad_norm": 0.550641804245094,
+      "learning_rate": 2.585455317986095e-06,
+      "loss": 0.0264,
+      "step": 881
+    },
+    {
+      "epoch": 1.3223388305847077,
+      "grad_norm": 0.6104892354040901,
+      "learning_rate": 2.5751509277790487e-06,
+      "loss": 0.0264,
+      "step": 882
+    },
+    {
+      "epoch": 1.3238380809595203,
+      "grad_norm": 0.45225897631183587,
+      "learning_rate": 2.5648599860333122e-06,
+      "loss": 0.0173,
+      "step": 883
+    },
+    {
+      "epoch": 1.3253373313343328,
+      "grad_norm": 0.4241833066892494,
+      "learning_rate": 2.554582549823502e-06,
+      "loss": 0.013,
+      "step": 884
+    },
+    {
+      "epoch": 1.3268365817091454,
+      "grad_norm": 0.6539705620097195,
+      "learning_rate": 2.5443186761493327e-06,
+      "loss": 0.0258,
+      "step": 885
+    },
+    {
+      "epoch": 1.328335832083958,
+      "grad_norm": 0.4727867303827312,
+      "learning_rate": 2.5340684219352977e-06,
+      "loss": 0.0124,
+      "step": 886
+    },
+    {
+      "epoch": 1.3298350824587706,
+      "grad_norm": 0.8105130474313901,
+      "learning_rate": 2.523831844030358e-06,
+      "loss": 0.0373,
+      "step": 887
+    },
+    {
+      "epoch": 1.3313343328335832,
+      "grad_norm": 0.4303328666217054,
+      "learning_rate": 2.513608999207622e-06,
+      "loss": 0.0157,
+      "step": 888
+    },
+    {
+      "epoch": 1.3328335832083957,
+      "grad_norm": 0.6175537689994001,
+      "learning_rate": 2.503399944164035e-06,
+      "loss": 0.0295,
+      "step": 889
+    },
+    {
+      "epoch": 1.3343328335832085,
+      "grad_norm": 0.5397713363484046,
+      "learning_rate": 2.4932047355200613e-06,
+      "loss": 0.0199,
+      "step": 890
+    },
+    {
+      "epoch": 1.3358320839580209,
+      "grad_norm": 0.6053336871272007,
+      "learning_rate": 2.483023429819372e-06,
+      "loss": 0.0286,
+      "step": 891
+    },
+    {
+      "epoch": 1.3373313343328337,
+      "grad_norm": 0.595382720575653,
+      "learning_rate": 2.472856083528531e-06,
+      "loss": 0.0284,
+      "step": 892
+    },
+    {
+      "epoch": 1.338830584707646,
+      "grad_norm": 0.5330999700790942,
+      "learning_rate": 2.4627027530366836e-06,
+      "loss": 0.022,
+      "step": 893
+    },
+    {
+      "epoch": 1.3403298350824588,
+      "grad_norm": 0.42346390246555854,
+      "learning_rate": 2.4525634946552405e-06,
+      "loss": 0.0217,
+      "step": 894
+    },
+    {
+      "epoch": 1.3418290854572714,
+      "grad_norm": 0.680446704239564,
+      "learning_rate": 2.442438364617567e-06,
+      "loss": 0.0328,
+      "step": 895
+    },
+    {
+      "epoch": 1.343328335832084,
+      "grad_norm": 0.5769714723645571,
+      "learning_rate": 2.4323274190786703e-06,
+      "loss": 0.0238,
+      "step": 896
+    },
+    {
+      "epoch": 1.3448275862068966,
+      "grad_norm": 0.5436208483593954,
+      "learning_rate": 2.422230714114891e-06,
+      "loss": 0.0225,
+      "step": 897
+    },
+    {
+      "epoch": 1.3463268365817092,
+      "grad_norm": 0.49624844997123313,
+      "learning_rate": 2.4121483057235884e-06,
+      "loss": 0.0256,
+      "step": 898
+    },
+    {
+      "epoch": 1.3478260869565217,
+      "grad_norm": 0.664372324503363,
+      "learning_rate": 2.4020802498228333e-06,
+      "loss": 0.0284,
+      "step": 899
+    },
+    {
+      "epoch": 1.3493253373313343,
+      "grad_norm": 0.37071267381445966,
+      "learning_rate": 2.392026602251093e-06,
+      "loss": 0.0107,
+      "step": 900
+    },
+    {
+      "epoch": 1.3508245877061469,
+      "grad_norm": 0.6696393973839908,
+      "learning_rate": 2.3819874187669266e-06,
+      "loss": 0.0269,
+      "step": 901
+    },
+    {
+      "epoch": 1.3523238380809595,
+      "grad_norm": 0.4705532077601151,
+      "learning_rate": 2.371962755048675e-06,
+      "loss": 0.0196,
+      "step": 902
+    },
+    {
+      "epoch": 1.353823088455772,
+      "grad_norm": 0.5144332227911089,
+      "learning_rate": 2.36195266669415e-06,
+      "loss": 0.0278,
+      "step": 903
+    },
+    {
+      "epoch": 1.3553223388305846,
+      "grad_norm": 0.3745079641162486,
+      "learning_rate": 2.351957209220326e-06,
+      "loss": 0.0126,
+      "step": 904
+    },
+    {
+      "epoch": 1.3568215892053974,
+      "grad_norm": 0.3932790974847323,
+      "learning_rate": 2.341976438063035e-06,
+      "loss": 0.0119,
+      "step": 905
+    },
+    {
+      "epoch": 1.3583208395802098,
+      "grad_norm": 0.41463171825130424,
+      "learning_rate": 2.332010408576653e-06,
+      "loss": 0.0154,
+      "step": 906
+    },
+    {
+      "epoch": 1.3598200899550226,
+      "grad_norm": 0.5309141810072499,
+      "learning_rate": 2.3220591760338046e-06,
+      "loss": 0.0232,
+      "step": 907
+    },
+    {
+      "epoch": 1.3613193403298351,
+      "grad_norm": 0.46698918395691136,
+      "learning_rate": 2.3121227956250435e-06,
+      "loss": 0.0203,
+      "step": 908
+    },
+    {
+      "epoch": 1.3628185907046477,
+      "grad_norm": 0.6217345145362444,
+      "learning_rate": 2.302201322458552e-06,
+      "loss": 0.0281,
+      "step": 909
+    },
+    {
+      "epoch": 1.3643178410794603,
+      "grad_norm": 0.31573701953113587,
+      "learning_rate": 2.292294811559837e-06,
+      "loss": 0.0097,
+      "step": 910
+    },
+    {
+      "epoch": 1.3658170914542729,
+      "grad_norm": 0.6123164077355115,
+      "learning_rate": 2.282403317871422e-06,
+      "loss": 0.0253,
+      "step": 911
+    },
+    {
+      "epoch": 1.3673163418290855,
+      "grad_norm": 0.5336505635862737,
+      "learning_rate": 2.2725268962525454e-06,
+      "loss": 0.0235,
+      "step": 912
+    },
+    {
+      "epoch": 1.368815592203898,
+      "grad_norm": 0.626061844369073,
+      "learning_rate": 2.262665601478852e-06,
+      "loss": 0.0329,
+      "step": 913
+    },
+    {
+      "epoch": 1.3703148425787106,
+      "grad_norm": 0.34781584830296586,
+      "learning_rate": 2.252819488242093e-06,
+      "loss": 0.0126,
+      "step": 914
+    },
+    {
+      "epoch": 1.3718140929535232,
+      "grad_norm": 0.5272987258866434,
+      "learning_rate": 2.24298861114982e-06,
+      "loss": 0.0193,
+      "step": 915
+    },
+    {
+      "epoch": 1.3733133433283358,
+      "grad_norm": 0.4108323344157199,
+      "learning_rate": 2.2331730247250857e-06,
+      "loss": 0.0109,
+      "step": 916
+    },
+    {
+      "epoch": 1.3748125937031483,
+      "grad_norm": 0.649249207793475,
+      "learning_rate": 2.223372783406137e-06,
+      "loss": 0.0313,
+      "step": 917
+    },
+    {
+      "epoch": 1.3763118440779611,
+      "grad_norm": 0.584976217948012,
+      "learning_rate": 2.2135879415461152e-06,
+      "loss": 0.0185,
+      "step": 918
+    },
+    {
+      "epoch": 1.3778110944527735,
+      "grad_norm": 0.8268126729828987,
+      "learning_rate": 2.203818553412757e-06,
+      "loss": 0.0375,
+      "step": 919
+    },
+    {
+      "epoch": 1.3793103448275863,
+      "grad_norm": 0.43587470121652855,
+      "learning_rate": 2.1940646731880887e-06,
+      "loss": 0.0125,
+      "step": 920
+    },
+    {
+      "epoch": 1.3808095952023989,
+      "grad_norm": 0.3276917322161156,
+      "learning_rate": 2.1843263549681287e-06,
+      "loss": 0.0085,
+      "step": 921
+    },
+    {
+      "epoch": 1.3823088455772115,
+      "grad_norm": 0.47936598294328125,
+      "learning_rate": 2.174603652762588e-06,
+      "loss": 0.0184,
+      "step": 922
+    },
+    {
+      "epoch": 1.383808095952024,
+      "grad_norm": 0.7021552862401628,
+      "learning_rate": 2.164896620494569e-06,
+      "loss": 0.031,
+      "step": 923
+    },
+    {
+      "epoch": 1.3853073463268366,
+      "grad_norm": 0.43039705633546455,
+      "learning_rate": 2.1552053120002655e-06,
+      "loss": 0.0167,
+      "step": 924
+    },
+    {
+      "epoch": 1.3868065967016492,
+      "grad_norm": 0.445834265123432,
+      "learning_rate": 2.145529781028668e-06,
+      "loss": 0.0188,
+      "step": 925
+    },
+    {
+      "epoch": 1.3883058470764618,
+      "grad_norm": 0.38241223056818285,
+      "learning_rate": 2.1358700812412625e-06,
+      "loss": 0.0113,
+      "step": 926
+    },
+    {
+      "epoch": 1.3898050974512743,
+      "grad_norm": 0.3580443931262484,
+      "learning_rate": 2.1262262662117327e-06,
+      "loss": 0.0129,
+      "step": 927
+    },
+    {
+      "epoch": 1.391304347826087,
+      "grad_norm": 0.5546776375896582,
+      "learning_rate": 2.1165983894256647e-06,
+      "loss": 0.0222,
+      "step": 928
+    },
+    {
+      "epoch": 1.3928035982008995,
+      "grad_norm": 0.49282105308902024,
+      "learning_rate": 2.1069865042802502e-06,
+      "loss": 0.0169,
+      "step": 929
+    },
+    {
+      "epoch": 1.394302848575712,
+      "grad_norm": 0.5971052430118794,
+      "learning_rate": 2.0973906640839867e-06,
+      "loss": 0.0205,
+      "step": 930
+    },
+    {
+      "epoch": 1.3958020989505249,
+      "grad_norm": 0.4043193195662047,
+      "learning_rate": 2.0878109220563884e-06,
+      "loss": 0.0132,
+      "step": 931
+    },
+    {
+      "epoch": 1.3973013493253372,
+      "grad_norm": 0.6428054194201187,
+      "learning_rate": 2.078247331327685e-06,
+      "loss": 0.0243,
+      "step": 932
+    },
+    {
+      "epoch": 1.39880059970015,
+      "grad_norm": 0.3571237727437659,
+      "learning_rate": 2.0686999449385286e-06,
+      "loss": 0.0103,
+      "step": 933
+    },
+    {
+      "epoch": 1.4002998500749624,
+      "grad_norm": 0.43641399153894267,
+      "learning_rate": 2.0591688158397054e-06,
+      "loss": 0.0164,
+      "step": 934
+    },
+    {
+      "epoch": 1.4017991004497752,
+      "grad_norm": 0.4891225897304742,
+      "learning_rate": 2.0496539968918342e-06,
+      "loss": 0.0177,
+      "step": 935
+    },
+    {
+      "epoch": 1.4032983508245878,
+      "grad_norm": 0.3823153361135867,
+      "learning_rate": 2.0401555408650714e-06,
+      "loss": 0.0136,
+      "step": 936
+    },
+    {
+      "epoch": 1.4047976011994003,
+      "grad_norm": 0.5393369787083407,
+      "learning_rate": 2.030673500438828e-06,
+      "loss": 0.0166,
+      "step": 937
+    },
+    {
+      "epoch": 1.406296851574213,
+      "grad_norm": 0.45145118871052936,
+      "learning_rate": 2.0212079282014725e-06,
+      "loss": 0.0162,
+      "step": 938
+    },
+    {
+      "epoch": 1.4077961019490255,
+      "grad_norm": 0.43207612678207774,
+      "learning_rate": 2.0117588766500375e-06,
+      "loss": 0.0147,
+      "step": 939
+    },
+    {
+      "epoch": 1.409295352323838,
+      "grad_norm": 0.585613173470247,
+      "learning_rate": 2.002326398189931e-06,
+      "loss": 0.025,
+      "step": 940
+    },
+    {
+      "epoch": 1.4107946026986506,
+      "grad_norm": 0.5021736337104517,
+      "learning_rate": 1.9929105451346436e-06,
+      "loss": 0.0119,
+      "step": 941
+    },
+    {
+      "epoch": 1.4122938530734632,
+      "grad_norm": 0.4631372860258595,
+      "learning_rate": 1.983511369705462e-06,
+      "loss": 0.0133,
+      "step": 942
+    },
+    {
+      "epoch": 1.4137931034482758,
+      "grad_norm": 0.47040389698509644,
+      "learning_rate": 1.9741289240311757e-06,
+      "loss": 0.0167,
+      "step": 943
+    },
+    {
+      "epoch": 1.4152923538230884,
+      "grad_norm": 0.5850461054894891,
+      "learning_rate": 1.9647632601477877e-06,
+      "loss": 0.0214,
+      "step": 944
+    },
+    {
+      "epoch": 1.416791604197901,
+      "grad_norm": 0.6349823616266396,
+      "learning_rate": 1.9554144299982314e-06,
+      "loss": 0.0285,
+      "step": 945
+    },
+    {
+      "epoch": 1.4182908545727138,
+      "grad_norm": 0.44951313626943834,
+      "learning_rate": 1.9460824854320755e-06,
+      "loss": 0.0156,
+      "step": 946
+    },
+    {
+      "epoch": 1.419790104947526,
+      "grad_norm": 0.4627259668612246,
+      "learning_rate": 1.9367674782052376e-06,
+      "loss": 0.0172,
+      "step": 947
+    },
+    {
+      "epoch": 1.421289355322339,
+      "grad_norm": 0.4135278246785498,
+      "learning_rate": 1.9274694599797067e-06,
+      "loss": 0.0133,
+      "step": 948
+    },
+    {
+      "epoch": 1.4227886056971515,
+      "grad_norm": 0.45051199098625627,
+      "learning_rate": 1.918188482323242e-06,
+      "loss": 0.0173,
+      "step": 949
+    },
+    {
+      "epoch": 1.424287856071964,
+      "grad_norm": 0.5350791150007976,
+      "learning_rate": 1.9089245967090952e-06,
+      "loss": 0.0298,
+      "step": 950
+    },
+    {
+      "epoch": 1.4257871064467766,
+      "grad_norm": 0.6676528877456653,
+      "learning_rate": 1.8996778545157263e-06,
+      "loss": 0.0206,
+      "step": 951
+    },
+    {
+      "epoch": 1.4272863568215892,
+      "grad_norm": 0.3951187255739535,
+      "learning_rate": 1.8904483070265133e-06,
+      "loss": 0.0144,
+      "step": 952
+    },
+    {
+      "epoch": 1.4287856071964018,
+      "grad_norm": 0.717819004359565,
+      "learning_rate": 1.8812360054294725e-06,
+      "loss": 0.0286,
+      "step": 953
+    },
+    {
+      "epoch": 1.4302848575712144,
+      "grad_norm": 0.4758339666266969,
+      "learning_rate": 1.8720410008169727e-06,
+      "loss": 0.0219,
+      "step": 954
+    },
+    {
+      "epoch": 1.431784107946027,
+      "grad_norm": 0.5566730026962932,
+      "learning_rate": 1.8628633441854515e-06,
+      "loss": 0.0198,
+      "step": 955
+    },
+    {
+      "epoch": 1.4332833583208395,
+      "grad_norm": 0.39320345343810625,
+      "learning_rate": 1.8537030864351303e-06,
+      "loss": 0.0159,
+      "step": 956
+    },
+    {
+      "epoch": 1.434782608695652,
+      "grad_norm": 0.6380647500645542,
+      "learning_rate": 1.8445602783697375e-06,
+      "loss": 0.0285,
+      "step": 957
+    },
+    {
+      "epoch": 1.4362818590704647,
+      "grad_norm": 0.5701399913343512,
+      "learning_rate": 1.8354349706962243e-06,
+      "loss": 0.0245,
+      "step": 958
+    },
+    {
+      "epoch": 1.4377811094452775,
+      "grad_norm": 0.3864943426497106,
+      "learning_rate": 1.8263272140244803e-06,
+      "loss": 0.0126,
+      "step": 959
+    },
+    {
+      "epoch": 1.4392803598200898,
+      "grad_norm": 0.432282483284804,
+      "learning_rate": 1.8172370588670563e-06,
+      "loss": 0.0145,
+      "step": 960
+    },
+    {
+      "epoch": 1.4407796101949026,
+      "grad_norm": 0.470033860030005,
+      "learning_rate": 1.8081645556388866e-06,
+      "loss": 0.0193,
+      "step": 961
+    },
+    {
+      "epoch": 1.4422788605697152,
+      "grad_norm": 0.39410035573761637,
+      "learning_rate": 1.7991097546570018e-06,
+      "loss": 0.0116,
+      "step": 962
+    },
+    {
+      "epoch": 1.4437781109445278,
+      "grad_norm": 0.7612418257887412,
+      "learning_rate": 1.7900727061402556e-06,
+      "loss": 0.0291,
+      "step": 963
+    },
+    {
+      "epoch": 1.4452773613193404,
+      "grad_norm": 0.7220388186746732,
+      "learning_rate": 1.7810534602090445e-06,
+      "loss": 0.02,
+      "step": 964
+    },
+    {
+      "epoch": 1.446776611694153,
+      "grad_norm": 0.6933150545041857,
+      "learning_rate": 1.77205206688503e-06,
+      "loss": 0.0284,
+      "step": 965
+    },
+    {
+      "epoch": 1.4482758620689655,
+      "grad_norm": 0.5689173288643403,
+      "learning_rate": 1.7630685760908623e-06,
+      "loss": 0.021,
+      "step": 966
+    },
+    {
+      "epoch": 1.449775112443778,
+      "grad_norm": 0.419874582750232,
+      "learning_rate": 1.7541030376499002e-06,
+      "loss": 0.0197,
+      "step": 967
+    },
+    {
+      "epoch": 1.4512743628185907,
+      "grad_norm": 0.49868122239143964,
+      "learning_rate": 1.745155501285939e-06,
+      "loss": 0.0249,
+      "step": 968
+    },
+    {
+      "epoch": 1.4527736131934033,
+      "grad_norm": 0.4990000130682441,
+      "learning_rate": 1.736226016622931e-06,
+      "loss": 0.0236,
+      "step": 969
+    },
+    {
+      "epoch": 1.4542728635682158,
+      "grad_norm": 0.5617429277123729,
+      "learning_rate": 1.727314633184714e-06,
+      "loss": 0.026,
+      "step": 970
+    },
+    {
+      "epoch": 1.4557721139430284,
+      "grad_norm": 0.6788498430631721,
+      "learning_rate": 1.718421400394732e-06,
+      "loss": 0.0331,
+      "step": 971
+    },
+    {
+      "epoch": 1.4572713643178412,
+      "grad_norm": 0.4582019183487326,
+      "learning_rate": 1.7095463675757656e-06,
+      "loss": 0.0141,
+      "step": 972
+    },
+    {
+      "epoch": 1.4587706146926536,
+      "grad_norm": 0.653593652362452,
+      "learning_rate": 1.7006895839496557e-06,
+      "loss": 0.0292,
+      "step": 973
+    },
+    {
+      "epoch": 1.4602698650674664,
+      "grad_norm": 0.4431860889187162,
+      "learning_rate": 1.6918510986370312e-06,
+      "loss": 0.014,
+      "step": 974
+    },
+    {
+      "epoch": 1.461769115442279,
+      "grad_norm": 0.7106746602505594,
+      "learning_rate": 1.6830309606570372e-06,
+      "loss": 0.0241,
+      "step": 975
+    },
+    {
+      "epoch": 1.4632683658170915,
+      "grad_norm": 0.37285958372758793,
+      "learning_rate": 1.674229218927062e-06,
+      "loss": 0.0158,
+      "step": 976
+    },
+    {
+      "epoch": 1.464767616191904,
+      "grad_norm": 0.758937528071681,
+      "learning_rate": 1.665445922262467e-06,
+      "loss": 0.029,
+      "step": 977
+    },
+    {
+      "epoch": 1.4662668665667167,
+      "grad_norm": 0.707455812914321,
+      "learning_rate": 1.6566811193763149e-06,
+      "loss": 0.0245,
+      "step": 978
+    },
+    {
+      "epoch": 1.4677661169415293,
+      "grad_norm": 0.5486194682763901,
+      "learning_rate": 1.6479348588791e-06,
+      "loss": 0.0216,
+      "step": 979
+    },
+    {
+      "epoch": 1.4692653673163418,
+      "grad_norm": 0.4669430843844355,
+      "learning_rate": 1.6392071892784789e-06,
+      "loss": 0.0169,
+      "step": 980
+    },
+    {
+      "epoch": 1.4707646176911544,
+      "grad_norm": 0.6276923753909117,
+      "learning_rate": 1.6304981589790015e-06,
+      "loss": 0.0211,
+      "step": 981
+    },
+    {
+      "epoch": 1.472263868065967,
+      "grad_norm": 0.6138410353810674,
+      "learning_rate": 1.6218078162818418e-06,
+      "loss": 0.0273,
+      "step": 982
+    },
+    {
+      "epoch": 1.4737631184407796,
+      "grad_norm": 0.6699080413141051,
+      "learning_rate": 1.6131362093845299e-06,
+      "loss": 0.0301,
+      "step": 983
+    },
+    {
+      "epoch": 1.4752623688155921,
+      "grad_norm": 0.49298297868204805,
+      "learning_rate": 1.6044833863806864e-06,
+      "loss": 0.0182,
+      "step": 984
+    },
+    {
+      "epoch": 1.4767616191904047,
+      "grad_norm": 0.49993661426633546,
+      "learning_rate": 1.5958493952597536e-06,
+      "loss": 0.0172,
+      "step": 985
+    },
+    {
+      "epoch": 1.4782608695652173,
+      "grad_norm": 0.3721194603227216,
+      "learning_rate": 1.5872342839067305e-06,
+      "loss": 0.0136,
+      "step": 986
+    },
+    {
+      "epoch": 1.47976011994003,
+      "grad_norm": 0.49354085151642174,
+      "learning_rate": 1.5786381001019052e-06,
+      "loss": 0.0202,
+      "step": 987
+    },
+    {
+      "epoch": 1.4812593703148424,
+      "grad_norm": 0.4182061566777323,
+      "learning_rate": 1.5700608915205978e-06,
+      "loss": 0.0168,
+      "step": 988
+    },
+    {
+      "epoch": 1.4827586206896552,
+      "grad_norm": 0.5421178065630526,
+      "learning_rate": 1.561502705732883e-06,
+      "loss": 0.0197,
+      "step": 989
+    },
+    {
+      "epoch": 1.4842578710644678,
+      "grad_norm": 0.5085497806258572,
+      "learning_rate": 1.5529635902033358e-06,
+      "loss": 0.0143,
+      "step": 990
+    },
+    {
+      "epoch": 1.4857571214392804,
+      "grad_norm": 0.8004774893792993,
+      "learning_rate": 1.5444435922907669e-06,
+      "loss": 0.0395,
+      "step": 991
+    },
+    {
+      "epoch": 1.487256371814093,
+      "grad_norm": 0.7316349010154469,
+      "learning_rate": 1.5359427592479553e-06,
+      "loss": 0.0326,
+      "step": 992
+    },
+    {
+      "epoch": 1.4887556221889056,
+      "grad_norm": 0.47447666404106514,
+      "learning_rate": 1.5274611382213922e-06,
+      "loss": 0.0166,
+      "step": 993
+    },
+    {
+      "epoch": 1.4902548725637181,
+      "grad_norm": 0.6917952596069579,
+      "learning_rate": 1.5189987762510167e-06,
+      "loss": 0.026,
+      "step": 994
+    },
+    {
+      "epoch": 1.4917541229385307,
+      "grad_norm": 0.4372722459379206,
+      "learning_rate": 1.510555720269955e-06,
+      "loss": 0.0172,
+      "step": 995
+    },
+    {
+      "epoch": 1.4932533733133433,
+      "grad_norm": 0.5848751817340585,
+      "learning_rate": 1.5021320171042608e-06,
+      "loss": 0.0129,
+      "step": 996
+    },
+    {
+      "epoch": 1.4947526236881559,
+      "grad_norm": 0.6719678517021768,
+      "learning_rate": 1.4937277134726542e-06,
+      "loss": 0.0286,
+      "step": 997
+    },
+    {
+      "epoch": 1.4962518740629684,
+      "grad_norm": 0.40820429189849405,
+      "learning_rate": 1.4853428559862637e-06,
+      "loss": 0.0142,
+      "step": 998
+    },
+    {
+      "epoch": 1.497751124437781,
+      "grad_norm": 0.5344984767173455,
+      "learning_rate": 1.4769774911483686e-06,
+      "loss": 0.0206,
+      "step": 999
+    },
+    {
+      "epoch": 1.4992503748125938,
+      "grad_norm": 0.500968599929968,
+      "learning_rate": 1.4686316653541377e-06,
+      "loss": 0.0192,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4992503748125938,
+      "eval_loss": 0.09554397314786911,
+      "eval_runtime": 8.1104,
+      "eval_samples_per_second": 6.658,
+      "eval_steps_per_second": 1.726,
+      "step": 1000
+    },
+    {
+      "epoch": 1.5007496251874062,
+      "grad_norm": 0.6898342895100004,
+      "learning_rate": 1.4603054248903752e-06,
+      "loss": 0.0307,
+      "step": 1001
+    },
+    {
+      "epoch": 1.502248875562219,
+      "grad_norm": 0.5359600980485157,
+      "learning_rate": 1.4519988159352665e-06,
+      "loss": 0.0175,
+      "step": 1002
+    },
+    {
+      "epoch": 1.5037481259370313,
+      "grad_norm": 0.7184525329881966,
+      "learning_rate": 1.4437118845581138e-06,
+      "loss": 0.0224,
+      "step": 1003
+    },
+    {
+      "epoch": 1.5052473763118441,
+      "grad_norm": 0.6084541262431876,
+      "learning_rate": 1.4354446767190873e-06,
+      "loss": 0.0251,
+      "step": 1004
+    },
+    {
+      "epoch": 1.5067466266866567,
+      "grad_norm": 0.7162085607096316,
+      "learning_rate": 1.4271972382689685e-06,
+      "loss": 0.0275,
+      "step": 1005
+    },
+    {
+      "epoch": 1.5082458770614693,
+      "grad_norm": 0.626852136806899,
+      "learning_rate": 1.4189696149488956e-06,
+      "loss": 0.0226,
+      "step": 1006
+    },
+    {
+      "epoch": 1.5097451274362819,
+      "grad_norm": 0.38086342553353764,
+      "learning_rate": 1.4107618523901101e-06,
+      "loss": 0.012,
+      "step": 1007
+    },
+    {
+      "epoch": 1.5112443778110944,
+      "grad_norm": 0.5404544549174051,
+      "learning_rate": 1.4025739961137043e-06,
+      "loss": 0.0217,
+      "step": 1008
+    },
+    {
+      "epoch": 1.512743628185907,
+      "grad_norm": 0.4283859360769096,
+      "learning_rate": 1.394406091530367e-06,
+      "loss": 0.0125,
+      "step": 1009
+    },
+    {
+      "epoch": 1.5142428785607196,
+      "grad_norm": 0.5172146535693602,
+      "learning_rate": 1.3862581839401346e-06,
+      "loss": 0.0172,
+      "step": 1010
+    },
+    {
+      "epoch": 1.5157421289355324,
+      "grad_norm": 0.5295538077274659,
+      "learning_rate": 1.3781303185321377e-06,
+      "loss": 0.0225,
+      "step": 1011
+    },
+    {
+      "epoch": 1.5172413793103448,
+      "grad_norm": 0.5271684762815159,
+      "learning_rate": 1.370022540384347e-06,
+      "loss": 0.024,
+      "step": 1012
+    },
+    {
+      "epoch": 1.5187406296851576,
+      "grad_norm": 0.45909053996765475,
+      "learning_rate": 1.3619348944633331e-06,
+      "loss": 0.0138,
+      "step": 1013
+    },
+    {
+      "epoch": 1.52023988005997,
+      "grad_norm": 0.44771434501448115,
+      "learning_rate": 1.3538674256240087e-06,
+      "loss": 0.0174,
+      "step": 1014
+    },
+    {
+      "epoch": 1.5217391304347827,
+      "grad_norm": 0.5217617684644401,
+      "learning_rate": 1.3458201786093795e-06,
+      "loss": 0.0165,
+      "step": 1015
+    },
+    {
+      "epoch": 1.523238380809595,
+      "grad_norm": 0.7053433640167447,
+      "learning_rate": 1.3377931980503055e-06,
+      "loss": 0.0252,
+      "step": 1016
+    },
+    {
+      "epoch": 1.5247376311844079,
+      "grad_norm": 0.4415119497218534,
+      "learning_rate": 1.3297865284652417e-06,
+      "loss": 0.0115,
+      "step": 1017
+    },
+    {
+      "epoch": 1.5262368815592204,
+      "grad_norm": 0.5985139765697872,
+      "learning_rate": 1.3218002142599973e-06,
+      "loss": 0.0234,
+      "step": 1018
+    },
+    {
+      "epoch": 1.527736131934033,
+      "grad_norm": 0.5749783303004483,
+      "learning_rate": 1.3138342997274883e-06,
+      "loss": 0.022,
+      "step": 1019
+    },
+    {
+      "epoch": 1.5292353823088456,
+      "grad_norm": 0.5720013225461984,
+      "learning_rate": 1.3058888290474937e-06,
+      "loss": 0.0243,
+      "step": 1020
+    },
+    {
+      "epoch": 1.5307346326836582,
+      "grad_norm": 0.7384737555274363,
+      "learning_rate": 1.2979638462864069e-06,
+      "loss": 0.029,
+      "step": 1021
+    },
+    {
+      "epoch": 1.5322338830584707,
+      "grad_norm": 0.7031868603722948,
+      "learning_rate": 1.2900593953969947e-06,
+      "loss": 0.0137,
+      "step": 1022
+    },
+    {
+      "epoch": 1.5337331334332833,
+      "grad_norm": 0.578041922711177,
+      "learning_rate": 1.2821755202181503e-06,
+      "loss": 0.0267,
+      "step": 1023
+    },
+    {
+      "epoch": 1.535232383808096,
+      "grad_norm": 0.43853562088656906,
+      "learning_rate": 1.2743122644746536e-06,
+      "loss": 0.016,
+      "step": 1024
+    },
+    {
+      "epoch": 1.5367316341829085,
+      "grad_norm": 0.7724098948311205,
+      "learning_rate": 1.266469671776926e-06,
+      "loss": 0.0293,
+      "step": 1025
+    },
+    {
+      "epoch": 1.5382308845577213,
+      "grad_norm": 0.49897112952082545,
+      "learning_rate": 1.2586477856207902e-06,
+      "loss": 0.0129,
+      "step": 1026
+    },
+    {
+      "epoch": 1.5397301349325336,
+      "grad_norm": 0.4978090534692373,
+      "learning_rate": 1.2508466493872273e-06,
+      "loss": 0.0193,
+      "step": 1027
+    },
+    {
+      "epoch": 1.5412293853073464,
+      "grad_norm": 0.690077401719104,
+      "learning_rate": 1.2430663063421388e-06,
+      "loss": 0.0306,
+      "step": 1028
+    },
+    {
+      "epoch": 1.5427286356821588,
+      "grad_norm": 0.5778145451169988,
+      "learning_rate": 1.2353067996361034e-06,
+      "loss": 0.0344,
+      "step": 1029
+    },
+    {
+      "epoch": 1.5442278860569716,
+      "grad_norm": 0.5198335236215647,
+      "learning_rate": 1.2275681723041406e-06,
+      "loss": 0.0226,
+      "step": 1030
+    },
+    {
+      "epoch": 1.545727136431784,
+      "grad_norm": 0.6790200980893414,
+      "learning_rate": 1.2198504672654694e-06,
+      "loss": 0.0274,
+      "step": 1031
+    },
+    {
+      "epoch": 1.5472263868065967,
+      "grad_norm": 0.3601763806879082,
+      "learning_rate": 1.212153727323273e-06,
+      "loss": 0.0111,
+      "step": 1032
+    },
+    {
+      "epoch": 1.5487256371814093,
+      "grad_norm": 0.4243614544051403,
+      "learning_rate": 1.2044779951644586e-06,
+      "loss": 0.0193,
+      "step": 1033
+    },
+    {
+      "epoch": 1.550224887556222,
+      "grad_norm": 0.5537388208954206,
+      "learning_rate": 1.1968233133594243e-06,
+      "loss": 0.0211,
+      "step": 1034
+    },
+    {
+      "epoch": 1.5517241379310345,
+      "grad_norm": 0.48759392409602653,
+      "learning_rate": 1.1891897243618184e-06,
+      "loss": 0.012,
+      "step": 1035
+    },
+    {
+      "epoch": 1.553223388305847,
+      "grad_norm": 0.4753142425531343,
+      "learning_rate": 1.1815772705083072e-06,
+      "loss": 0.0105,
+      "step": 1036
+    },
+    {
+      "epoch": 1.5547226386806596,
+      "grad_norm": 0.7652867680760843,
+      "learning_rate": 1.17398599401834e-06,
+      "loss": 0.0226,
+      "step": 1037
+    },
+    {
+      "epoch": 1.5562218890554722,
+      "grad_norm": 0.5367356285426131,
+      "learning_rate": 1.1664159369939137e-06,
+      "loss": 0.0239,
+      "step": 1038
+    },
+    {
+      "epoch": 1.557721139430285,
+      "grad_norm": 0.5468645155853205,
+      "learning_rate": 1.1588671414193397e-06,
+      "loss": 0.0204,
+      "step": 1039
+    },
+    {
+      "epoch": 1.5592203898050974,
+      "grad_norm": 0.40430616263724684,
+      "learning_rate": 1.1513396491610113e-06,
+      "loss": 0.0209,
+      "step": 1040
+    },
+    {
+      "epoch": 1.5607196401799102,
+      "grad_norm": 0.38621305703077424,
+      "learning_rate": 1.1438335019671715e-06,
+      "loss": 0.0145,
+      "step": 1041
+    },
+    {
+      "epoch": 1.5622188905547225,
+      "grad_norm": 0.5560396048782303,
+      "learning_rate": 1.1363487414676805e-06,
+      "loss": 0.0226,
+      "step": 1042
+    },
+    {
+      "epoch": 1.5637181409295353,
+      "grad_norm": 0.5244488243765995,
+      "learning_rate": 1.128885409173789e-06,
+      "loss": 0.0213,
+      "step": 1043
+    },
+    {
+      "epoch": 1.5652173913043477,
+      "grad_norm": 0.38084753614372174,
+      "learning_rate": 1.1214435464779006e-06,
+      "loss": 0.0113,
+      "step": 1044
+    },
+    {
+      "epoch": 1.5667166416791605,
+      "grad_norm": 0.4340665893815063,
+      "learning_rate": 1.1140231946533486e-06,
+      "loss": 0.015,
+      "step": 1045
+    },
+    {
+      "epoch": 1.568215892053973,
+      "grad_norm": 0.9160035276808725,
+      "learning_rate": 1.1066243948541638e-06,
+      "loss": 0.0182,
+      "step": 1046
+    },
+    {
+      "epoch": 1.5697151424287856,
+      "grad_norm": 0.3596492595363518,
+      "learning_rate": 1.0992471881148497e-06,
+      "loss": 0.011,
+      "step": 1047
+    },
+    {
+      "epoch": 1.5712143928035982,
+      "grad_norm": 0.580120171060122,
+      "learning_rate": 1.091891615350147e-06,
+      "loss": 0.0206,
+      "step": 1048
+    },
+    {
+      "epoch": 1.5727136431784108,
+      "grad_norm": 0.6395470678871679,
+      "learning_rate": 1.0845577173548172e-06,
+      "loss": 0.0329,
+      "step": 1049
+    },
+    {
+      "epoch": 1.5742128935532234,
+      "grad_norm": 0.3474836726260322,
+      "learning_rate": 1.07724553480341e-06,
+      "loss": 0.0129,
+      "step": 1050
+    },
+    {
+      "epoch": 1.575712143928036,
+      "grad_norm": 0.5927650241183463,
+      "learning_rate": 1.0699551082500387e-06,
+      "loss": 0.0228,
+      "step": 1051
+    },
+    {
+      "epoch": 1.5772113943028487,
+      "grad_norm": 0.6276619655177309,
+      "learning_rate": 1.0626864781281553e-06,
+      "loss": 0.0222,
+      "step": 1052
+    },
+    {
+      "epoch": 1.578710644677661,
+      "grad_norm": 0.4066551320653785,
+      "learning_rate": 1.0554396847503272e-06,
+      "loss": 0.0128,
+      "step": 1053
+    },
+    {
+      "epoch": 1.580209895052474,
+      "grad_norm": 0.4664697892428956,
+      "learning_rate": 1.0482147683080125e-06,
+      "loss": 0.0157,
+      "step": 1054
+    },
+    {
+      "epoch": 1.5817091454272862,
+      "grad_norm": 0.6135556316287063,
+      "learning_rate": 1.0410117688713366e-06,
+      "loss": 0.0257,
+      "step": 1055
+    },
+    {
+      "epoch": 1.583208395802099,
+      "grad_norm": 0.46913886461072907,
+      "learning_rate": 1.0338307263888748e-06,
+      "loss": 0.0119,
+      "step": 1056
+    },
+    {
+      "epoch": 1.5847076461769114,
+      "grad_norm": 0.5287026380753799,
+      "learning_rate": 1.0266716806874227e-06,
+      "loss": 0.0193,
+      "step": 1057
+    },
+    {
+      "epoch": 1.5862068965517242,
+      "grad_norm": 0.3964062042058069,
+      "learning_rate": 1.0195346714717813e-06,
+      "loss": 0.0135,
+      "step": 1058
+    },
+    {
+      "epoch": 1.5877061469265368,
+      "grad_norm": 0.6662722433360334,
+      "learning_rate": 1.0124197383245344e-06,
+      "loss": 0.0248,
+      "step": 1059
+    },
+    {
+      "epoch": 1.5892053973013494,
+      "grad_norm": 0.2700982020397101,
+      "learning_rate": 1.0053269207058298e-06,
+      "loss": 0.0087,
+      "step": 1060
+    },
+    {
+      "epoch": 1.590704647676162,
+      "grad_norm": 0.46070378010182783,
+      "learning_rate": 9.982562579531607e-07,
+      "loss": 0.0187,
+      "step": 1061
+    },
+    {
+      "epoch": 1.5922038980509745,
+      "grad_norm": 0.3414460769456052,
+      "learning_rate": 9.912077892811473e-07,
+      "loss": 0.0103,
+      "step": 1062
+    },
+    {
+      "epoch": 1.593703148425787,
+      "grad_norm": 0.5203201363999499,
+      "learning_rate": 9.841815537813177e-07,
+      "loss": 0.02,
+      "step": 1063
+    },
+    {
+      "epoch": 1.5952023988005997,
+      "grad_norm": 0.5505774044320573,
+      "learning_rate": 9.77177590421895e-07,
+      "loss": 0.0159,
+      "step": 1064
+    },
+    {
+      "epoch": 1.5967016491754122,
+      "grad_norm": 0.7078649767751176,
+      "learning_rate": 9.70195938047576e-07,
+      "loss": 0.0321,
+      "step": 1065
+    },
+    {
+      "epoch": 1.5982008995502248,
+      "grad_norm": 0.691862331346126,
+      "learning_rate": 9.63236635379321e-07,
+      "loss": 0.0234,
+      "step": 1066
+    },
+    {
+      "epoch": 1.5997001499250376,
+      "grad_norm": 0.7071407604233519,
+      "learning_rate": 9.562997210141355e-07,
+      "loss": 0.0243,
+      "step": 1067
+    },
+    {
+      "epoch": 1.60119940029985,
+      "grad_norm": 0.48384964688592375,
+      "learning_rate": 9.49385233424856e-07,
+      "loss": 0.0197,
+      "step": 1068
+    },
+    {
+      "epoch": 1.6026986506746628,
+      "grad_norm": 0.6201173341328386,
+      "learning_rate": 9.424932109599372e-07,
+      "loss": 0.0246,
+      "step": 1069
+    },
+    {
+      "epoch": 1.6041979010494751,
+      "grad_norm": 0.6249698919608127,
+      "learning_rate": 9.356236918432454e-07,
+      "loss": 0.0245,
+      "step": 1070
+    },
+    {
+      "epoch": 1.605697151424288,
+      "grad_norm": 0.6619316484941568,
+      "learning_rate": 9.287767141738352e-07,
+      "loss": 0.0312,
+      "step": 1071
+    },
+    {
+      "epoch": 1.6071964017991005,
+      "grad_norm": 0.426901589443811,
+      "learning_rate": 9.21952315925746e-07,
+      "loss": 0.017,
+      "step": 1072
+    },
+    {
+      "epoch": 1.608695652173913,
+      "grad_norm": 0.3476470582013292,
+      "learning_rate": 9.151505349477901e-07,
+      "loss": 0.0116,
+      "step": 1073
+    },
+    {
+      "epoch": 1.6101949025487257,
+      "grad_norm": 0.6911271705804719,
+      "learning_rate": 9.08371408963341e-07,
+      "loss": 0.0256,
+      "step": 1074
+    },
+    {
+      "epoch": 1.6116941529235382,
+      "grad_norm": 0.48973044276037775,
+      "learning_rate": 9.016149755701259e-07,
+      "loss": 0.018,
+      "step": 1075
+    },
+    {
+      "epoch": 1.6131934032983508,
+      "grad_norm": 0.4588433264717821,
+      "learning_rate": 8.948812722400157e-07,
+      "loss": 0.0201,
+      "step": 1076
+    },
+    {
+      "epoch": 1.6146926536731634,
+      "grad_norm": 0.7526940107492812,
+      "learning_rate": 8.881703363188199e-07,
+      "loss": 0.0332,
+      "step": 1077
+    },
+    {
+      "epoch": 1.616191904047976,
+      "grad_norm": 0.47831323515229335,
+      "learning_rate": 8.814822050260758e-07,
+      "loss": 0.0229,
+      "step": 1078
+    },
+    {
+      "epoch": 1.6176911544227885,
+      "grad_norm": 0.6058233457379996,
+      "learning_rate": 8.748169154548448e-07,
+      "loss": 0.0295,
+      "step": 1079
+    },
+    {
+      "epoch": 1.6191904047976013,
+      "grad_norm": 0.5363051821459227,
+      "learning_rate": 8.681745045715045e-07,
+      "loss": 0.0306,
+      "step": 1080
+    },
+    {
+      "epoch": 1.6206896551724137,
+      "grad_norm": 0.48406047823644177,
+      "learning_rate": 8.615550092155478e-07,
+      "loss": 0.0195,
+      "step": 1081
+    },
+    {
+      "epoch": 1.6221889055472265,
+      "grad_norm": 0.3828905223431728,
+      "learning_rate": 8.549584660993726e-07,
+      "loss": 0.0126,
+      "step": 1082
+    },
+    {
+      "epoch": 1.6236881559220389,
+      "grad_norm": 0.5501103408352331,
+      "learning_rate": 8.483849118080828e-07,
+      "loss": 0.0191,
+      "step": 1083
+    },
+    {
+      "epoch": 1.6251874062968517,
+      "grad_norm": 0.47984075012262645,
+      "learning_rate": 8.418343827992842e-07,
+      "loss": 0.0143,
+      "step": 1084
+    },
+    {
+      "epoch": 1.626686656671664,
+      "grad_norm": 0.7034770903419695,
+      "learning_rate": 8.353069154028814e-07,
+      "loss": 0.037,
+      "step": 1085
+    },
+    {
+      "epoch": 1.6281859070464768,
+      "grad_norm": 0.5099072409564612,
+      "learning_rate": 8.28802545820877e-07,
+      "loss": 0.0221,
+      "step": 1086
+    },
+    {
+      "epoch": 1.6296851574212894,
+      "grad_norm": 0.6871978210800703,
+      "learning_rate": 8.223213101271709e-07,
+      "loss": 0.0292,
+      "step": 1087
+    },
+    {
+      "epoch": 1.631184407796102,
+      "grad_norm": 0.49952829812291416,
+      "learning_rate": 8.158632442673603e-07,
+      "loss": 0.0196,
+      "step": 1088
+    },
+    {
+      "epoch": 1.6326836581709145,
+      "grad_norm": 0.5219355784406893,
+      "learning_rate": 8.094283840585398e-07,
+      "loss": 0.0314,
+      "step": 1089
+    },
+    {
+      "epoch": 1.6341829085457271,
+      "grad_norm": 0.510433555917818,
+      "learning_rate": 8.03016765189103e-07,
+      "loss": 0.0185,
+      "step": 1090
+    },
+    {
+      "epoch": 1.6356821589205397,
+      "grad_norm": 0.5582313205888276,
+      "learning_rate": 7.966284232185451e-07,
+      "loss": 0.0223,
+      "step": 1091
+    },
+    {
+      "epoch": 1.6371814092953523,
+      "grad_norm": 0.5618542977637488,
+      "learning_rate": 7.902633935772647e-07,
+      "loss": 0.0221,
+      "step": 1092
+    },
+    {
+      "epoch": 1.638680659670165,
+      "grad_norm": 0.6145182864595798,
+      "learning_rate": 7.839217115663683e-07,
+      "loss": 0.026,
+      "step": 1093
+    },
+    {
+      "epoch": 1.6401799100449774,
+      "grad_norm": 0.32591191417360615,
+      "learning_rate": 7.776034123574738e-07,
+      "loss": 0.0094,
+      "step": 1094
+    },
+    {
+      "epoch": 1.6416791604197902,
+      "grad_norm": 0.4128833210959961,
+      "learning_rate": 7.713085309925156e-07,
+      "loss": 0.0148,
+      "step": 1095
+    },
+    {
+      "epoch": 1.6431784107946026,
+      "grad_norm": 0.8003563940113999,
+      "learning_rate": 7.650371023835495e-07,
+      "loss": 0.0356,
+      "step": 1096
+    },
+    {
+      "epoch": 1.6446776611694154,
+      "grad_norm": 0.5743805289919706,
+      "learning_rate": 7.587891613125631e-07,
+      "loss": 0.0245,
+      "step": 1097
+    },
+    {
+      "epoch": 1.6461769115442277,
+      "grad_norm": 0.57739364314371,
+      "learning_rate": 7.525647424312766e-07,
+      "loss": 0.0253,
+      "step": 1098
+    },
+    {
+      "epoch": 1.6476761619190405,
+      "grad_norm": 0.5291389243287572,
+      "learning_rate": 7.46363880260954e-07,
+      "loss": 0.0203,
+      "step": 1099
+    },
+    {
+      "epoch": 1.6491754122938531,
+      "grad_norm": 0.597266377266116,
+      "learning_rate": 7.401866091922133e-07,
+      "loss": 0.0261,
+      "step": 1100
+    },
+    {
+      "epoch": 1.6506746626686657,
+      "grad_norm": 0.8054552384644502,
+      "learning_rate": 7.340329634848309e-07,
+      "loss": 0.0309,
+      "step": 1101
+    },
+    {
+      "epoch": 1.6521739130434783,
+      "grad_norm": 0.638166835148322,
+      "learning_rate": 7.279029772675572e-07,
+      "loss": 0.0203,
+      "step": 1102
+    },
+    {
+      "epoch": 1.6536731634182908,
+      "grad_norm": 0.43414960720118245,
+      "learning_rate": 7.217966845379243e-07,
+      "loss": 0.0138,
+      "step": 1103
+    },
+    {
+      "epoch": 1.6551724137931034,
+      "grad_norm": 0.5076613652797889,
+      "learning_rate": 7.157141191620548e-07,
+      "loss": 0.0159,
+      "step": 1104
+    },
+    {
+      "epoch": 1.656671664167916,
+      "grad_norm": 0.38271932454897406,
+      "learning_rate": 7.096553148744806e-07,
+      "loss": 0.0097,
+      "step": 1105
+    },
+    {
+      "epoch": 1.6581709145427288,
+      "grad_norm": 0.43399435436526007,
+      "learning_rate": 7.036203052779506e-07,
+      "loss": 0.0163,
+      "step": 1106
+    },
+    {
+      "epoch": 1.6596701649175412,
+      "grad_norm": 0.516120405565092,
+      "learning_rate": 6.97609123843247e-07,
+      "loss": 0.0175,
+      "step": 1107
+    },
+    {
+      "epoch": 1.661169415292354,
+      "grad_norm": 0.5753657581364159,
+      "learning_rate": 6.916218039089961e-07,
+      "loss": 0.0219,
+      "step": 1108
+    },
+    {
+      "epoch": 1.6626686656671663,
+      "grad_norm": 0.4538528972263157,
+      "learning_rate": 6.856583786814891e-07,
+      "loss": 0.0167,
+      "step": 1109
+    },
+    {
+      "epoch": 1.6641679160419791,
+      "grad_norm": 0.6243665449286745,
+      "learning_rate": 6.797188812344907e-07,
+      "loss": 0.0221,
+      "step": 1110
+    },
+    {
+      "epoch": 1.6656671664167915,
+      "grad_norm": 0.6325487154492438,
+      "learning_rate": 6.738033445090653e-07,
+      "loss": 0.0337,
+      "step": 1111
+    },
+    {
+      "epoch": 1.6671664167916043,
+      "grad_norm": 0.539380667716522,
+      "learning_rate": 6.67911801313384e-07,
+      "loss": 0.0208,
+      "step": 1112
+    },
+    {
+      "epoch": 1.6686656671664168,
+      "grad_norm": 0.6833057565354582,
+      "learning_rate": 6.620442843225483e-07,
+      "loss": 0.0234,
+      "step": 1113
+    },
+    {
+      "epoch": 1.6701649175412294,
+      "grad_norm": 0.48697418229917216,
+      "learning_rate": 6.562008260784092e-07,
+      "loss": 0.0128,
+      "step": 1114
+    },
+    {
+      "epoch": 1.671664167916042,
+      "grad_norm": 0.5302933492148422,
+      "learning_rate": 6.503814589893836e-07,
+      "loss": 0.0184,
+      "step": 1115
+    },
+    {
+      "epoch": 1.6731634182908546,
+      "grad_norm": 0.5270295896309886,
+      "learning_rate": 6.445862153302784e-07,
+      "loss": 0.0248,
+      "step": 1116
+    },
+    {
+      "epoch": 1.6746626686656672,
+      "grad_norm": 0.4819095073660573,
+      "learning_rate": 6.388151272421078e-07,
+      "loss": 0.0173,
+      "step": 1117
+    },
+    {
+      "epoch": 1.6761619190404797,
+      "grad_norm": 0.39766618429389194,
+      "learning_rate": 6.330682267319177e-07,
+      "loss": 0.0157,
+      "step": 1118
+    },
+    {
+      "epoch": 1.6776611694152923,
+      "grad_norm": 0.4132580057263917,
+      "learning_rate": 6.273455456726074e-07,
+      "loss": 0.0114,
+      "step": 1119
+    },
+    {
+      "epoch": 1.6791604197901049,
+      "grad_norm": 0.5507747714106712,
+      "learning_rate": 6.216471158027515e-07,
+      "loss": 0.0262,
+      "step": 1120
+    },
+    {
+      "epoch": 1.6806596701649177,
+      "grad_norm": 0.503435638594848,
+      "learning_rate": 6.159729687264254e-07,
+      "loss": 0.0203,
+      "step": 1121
+    },
+    {
+      "epoch": 1.68215892053973,
+      "grad_norm": 0.7379447834143609,
+      "learning_rate": 6.103231359130308e-07,
+      "loss": 0.0208,
+      "step": 1122
+    },
+    {
+      "epoch": 1.6836581709145428,
+      "grad_norm": 0.541048407672562,
+      "learning_rate": 6.046976486971201e-07,
+      "loss": 0.0237,
+      "step": 1123
+    },
+    {
+      "epoch": 1.6851574212893552,
+      "grad_norm": 0.5342705056793224,
+      "learning_rate": 5.990965382782177e-07,
+      "loss": 0.0184,
+      "step": 1124
+    },
+    {
+      "epoch": 1.686656671664168,
+      "grad_norm": 0.5041980643969215,
+      "learning_rate": 5.935198357206595e-07,
+      "loss": 0.0146,
+      "step": 1125
+    },
+    {
+      "epoch": 1.6881559220389803,
+      "grad_norm": 0.4328454213717956,
+      "learning_rate": 5.879675719534078e-07,
+      "loss": 0.0135,
+      "step": 1126
+    },
+    {
+      "epoch": 1.6896551724137931,
+      "grad_norm": 0.5211204073090064,
+      "learning_rate": 5.824397777698859e-07,
+      "loss": 0.0246,
+      "step": 1127
+    },
+    {
+      "epoch": 1.6911544227886057,
+      "grad_norm": 0.5796070433012782,
+      "learning_rate": 5.769364838278063e-07,
+      "loss": 0.022,
+      "step": 1128
+    },
+    {
+      "epoch": 1.6926536731634183,
+      "grad_norm": 0.548751601468487,
+      "learning_rate": 5.714577206490018e-07,
+      "loss": 0.0251,
+      "step": 1129
+    },
+    {
+      "epoch": 1.6941529235382309,
+      "grad_norm": 0.4301557036276666,
+      "learning_rate": 5.660035186192531e-07,
+      "loss": 0.0153,
+      "step": 1130
+    },
+    {
+      "epoch": 1.6956521739130435,
+      "grad_norm": 0.38599016187945795,
+      "learning_rate": 5.60573907988124e-07,
+      "loss": 0.0112,
+      "step": 1131
+    },
+    {
+      "epoch": 1.697151424287856,
+      "grad_norm": 0.5886311328260946,
+      "learning_rate": 5.551689188687909e-07,
+      "loss": 0.0228,
+      "step": 1132
+    },
+    {
+      "epoch": 1.6986506746626686,
+      "grad_norm": 0.4774446429614018,
+      "learning_rate": 5.497885812378772e-07,
+      "loss": 0.0152,
+      "step": 1133
+    },
+    {
+      "epoch": 1.7001499250374814,
+      "grad_norm": 0.5162843465303787,
+      "learning_rate": 5.444329249352859e-07,
+      "loss": 0.0189,
+      "step": 1134
+    },
+    {
+      "epoch": 1.7016491754122938,
+      "grad_norm": 0.580120083068201,
+      "learning_rate": 5.391019796640362e-07,
+      "loss": 0.0271,
+      "step": 1135
+    },
+    {
+      "epoch": 1.7031484257871066,
+      "grad_norm": 0.5320379110972915,
+      "learning_rate": 5.337957749900958e-07,
+      "loss": 0.0214,
+      "step": 1136
+    },
+    {
+      "epoch": 1.704647676161919,
+      "grad_norm": 0.29590992441691727,
+      "learning_rate": 5.285143403422188e-07,
+      "loss": 0.0077,
+      "step": 1137
+    },
+    {
+      "epoch": 1.7061469265367317,
+      "grad_norm": 0.42648866647948724,
+      "learning_rate": 5.23257705011786e-07,
+      "loss": 0.0162,
+      "step": 1138
+    },
+    {
+      "epoch": 1.707646176911544,
+      "grad_norm": 0.46411126666304164,
+      "learning_rate": 5.18025898152631e-07,
+      "loss": 0.0159,
+      "step": 1139
+    },
+    {
+      "epoch": 1.7091454272863569,
+      "grad_norm": 0.25362566462758873,
+      "learning_rate": 5.128189487808927e-07,
+      "loss": 0.0077,
+      "step": 1140
+    },
+    {
+      "epoch": 1.7106446776611695,
+      "grad_norm": 0.6941751735741677,
+      "learning_rate": 5.076368857748454e-07,
+      "loss": 0.0265,
+      "step": 1141
+    },
+    {
+      "epoch": 1.712143928035982,
+      "grad_norm": 0.6049107098813133,
+      "learning_rate": 5.024797378747414e-07,
+      "loss": 0.0221,
+      "step": 1142
+    },
+    {
+      "epoch": 1.7136431784107946,
+      "grad_norm": 0.5735445927551915,
+      "learning_rate": 4.973475336826506e-07,
+      "loss": 0.0225,
+      "step": 1143
+    },
+    {
+      "epoch": 1.7151424287856072,
+      "grad_norm": 0.5224472725067685,
+      "learning_rate": 4.922403016623034e-07,
+      "loss": 0.0186,
+      "step": 1144
+    },
+    {
+      "epoch": 1.7166416791604198,
+      "grad_norm": 0.6460084404514412,
+      "learning_rate": 4.871580701389316e-07,
+      "loss": 0.021,
+      "step": 1145
+    },
+    {
+      "epoch": 1.7181409295352323,
+      "grad_norm": 0.5492785849979425,
+      "learning_rate": 4.821008672991118e-07,
+      "loss": 0.0178,
+      "step": 1146
+    },
+    {
+      "epoch": 1.7196401799100451,
+      "grad_norm": 0.47542134793951046,
+      "learning_rate": 4.770687211906089e-07,
+      "loss": 0.0184,
+      "step": 1147
+    },
+    {
+      "epoch": 1.7211394302848575,
+      "grad_norm": 0.5087991486464657,
+      "learning_rate": 4.720616597222205e-07,
+      "loss": 0.0169,
+      "step": 1148
+    },
+    {
+      "epoch": 1.7226386806596703,
+      "grad_norm": 0.35298206463036275,
+      "learning_rate": 4.6707971066362324e-07,
+      "loss": 0.0113,
+      "step": 1149
+    },
+    {
+      "epoch": 1.7241379310344827,
+      "grad_norm": 0.4879196234317882,
+      "learning_rate": 4.6212290164521554e-07,
+      "loss": 0.0176,
+      "step": 1150
+    },
+    {
+      "epoch": 1.7256371814092955,
+      "grad_norm": 0.35808210437813254,
+      "learning_rate": 4.5719126015796757e-07,
+      "loss": 0.0103,
+      "step": 1151
+    },
+    {
+      "epoch": 1.7271364317841078,
+      "grad_norm": 0.4661112528297329,
+      "learning_rate": 4.522848135532698e-07,
+      "loss": 0.0148,
+      "step": 1152
+    },
+    {
+      "epoch": 1.7286356821589206,
+      "grad_norm": 0.6243656454529516,
+      "learning_rate": 4.474035890427769e-07,
+      "loss": 0.0277,
+      "step": 1153
+    },
+    {
+      "epoch": 1.7301349325337332,
+      "grad_norm": 0.5245233814409072,
+      "learning_rate": 4.4254761369825984e-07,
+      "loss": 0.0203,
+      "step": 1154
+    },
+    {
+      "epoch": 1.7316341829085458,
+      "grad_norm": 0.281723305858224,
+      "learning_rate": 4.377169144514554e-07,
+      "loss": 0.0077,
+      "step": 1155
+    },
+    {
+      "epoch": 1.7331334332833583,
+      "grad_norm": 0.5995017677645802,
+      "learning_rate": 4.329115180939164e-07,
+      "loss": 0.0264,
+      "step": 1156
+    },
+    {
+      "epoch": 1.734632683658171,
+      "grad_norm": 0.6422348669753307,
+      "learning_rate": 4.281314512768625e-07,
+      "loss": 0.0227,
+      "step": 1157
+    },
+    {
+      "epoch": 1.7361319340329835,
+      "grad_norm": 0.6655659854234758,
+      "learning_rate": 4.2337674051103504e-07,
+      "loss": 0.025,
+      "step": 1158
+    },
+    {
+      "epoch": 1.737631184407796,
+      "grad_norm": 0.6933499189903861,
+      "learning_rate": 4.186474121665468e-07,
+      "loss": 0.0297,
+      "step": 1159
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 0.38257146697952876,
+      "learning_rate": 4.139434924727359e-07,
+      "loss": 0.0152,
+      "step": 1160
+    },
+    {
+      "epoch": 1.7406296851574212,
+      "grad_norm": 0.43199610262667665,
+      "learning_rate": 4.092650075180232e-07,
+      "loss": 0.0194,
+      "step": 1161
+    },
+    {
+      "epoch": 1.742128935532234,
+      "grad_norm": 0.5340969239774258,
+      "learning_rate": 4.046119832497658e-07,
+      "loss": 0.0219,
+      "step": 1162
+    },
+    {
+      "epoch": 1.7436281859070464,
+      "grad_norm": 0.4159623918478173,
+      "learning_rate": 3.9998444547411255e-07,
+      "loss": 0.0126,
+      "step": 1163
+    },
+    {
+      "epoch": 1.7451274362818592,
+      "grad_norm": 0.48195531893791077,
+      "learning_rate": 3.9538241985586144e-07,
+      "loss": 0.019,
+      "step": 1164
+    },
+    {
+      "epoch": 1.7466266866566715,
+      "grad_norm": 0.608809771084013,
+      "learning_rate": 3.908059319183194e-07,
+      "loss": 0.0322,
+      "step": 1165
+    },
+    {
+      "epoch": 1.7481259370314843,
+      "grad_norm": 0.36507596757767385,
+      "learning_rate": 3.8625500704315645e-07,
+      "loss": 0.0093,
+      "step": 1166
+    },
+    {
+      "epoch": 1.7496251874062967,
+      "grad_norm": 0.62549369309446,
+      "learning_rate": 3.8172967047026834e-07,
+      "loss": 0.0196,
+      "step": 1167
+    },
+    {
+      "epoch": 1.7511244377811095,
+      "grad_norm": 0.48229366851435274,
+      "learning_rate": 3.7722994729763427e-07,
+      "loss": 0.0185,
+      "step": 1168
+    },
+    {
+      "epoch": 1.752623688155922,
+      "grad_norm": 0.591216575026225,
+      "learning_rate": 3.7275586248118114e-07,
+      "loss": 0.027,
+      "step": 1169
+    },
+    {
+      "epoch": 1.7541229385307346,
+      "grad_norm": 0.4288879435325373,
+      "learning_rate": 3.683074408346404e-07,
+      "loss": 0.0145,
+      "step": 1170
+    },
+    {
+      "epoch": 1.7556221889055472,
+      "grad_norm": 0.5919461604008727,
+      "learning_rate": 3.6388470702941436e-07,
+      "loss": 0.0304,
+      "step": 1171
+    },
+    {
+      "epoch": 1.7571214392803598,
+      "grad_norm": 0.8303121520049621,
+      "learning_rate": 3.594876855944385e-07,
+      "loss": 0.0301,
+      "step": 1172
+    },
+    {
+      "epoch": 1.7586206896551724,
+      "grad_norm": 0.47623707315018393,
+      "learning_rate": 3.5511640091604293e-07,
+      "loss": 0.0193,
+      "step": 1173
+    },
+    {
+      "epoch": 1.760119940029985,
+      "grad_norm": 0.35010189422768667,
+      "learning_rate": 3.50770877237821e-07,
+      "loss": 0.0113,
+      "step": 1174
+    },
+    {
+      "epoch": 1.7616191904047978,
+      "grad_norm": 0.4904394147131819,
+      "learning_rate": 3.4645113866049187e-07,
+      "loss": 0.0165,
+      "step": 1175
+    },
+    {
+      "epoch": 1.76311844077961,
+      "grad_norm": 0.42009301791998965,
+      "learning_rate": 3.42157209141768e-07,
+      "loss": 0.0166,
+      "step": 1176
+    },
+    {
+      "epoch": 1.764617691154423,
+      "grad_norm": 0.5125210150072099,
+      "learning_rate": 3.3788911249622194e-07,
+      "loss": 0.0147,
+      "step": 1177
+    },
+    {
+      "epoch": 1.7661169415292353,
+      "grad_norm": 0.5853653482821087,
+      "learning_rate": 3.336468723951558e-07,
+      "loss": 0.0179,
+      "step": 1178
+    },
+    {
+      "epoch": 1.767616191904048,
+      "grad_norm": 0.4786873782979341,
+      "learning_rate": 3.294305123664665e-07,
+      "loss": 0.0129,
+      "step": 1179
+    },
+    {
+      "epoch": 1.7691154422788604,
+      "grad_norm": 0.4582612750180858,
+      "learning_rate": 3.2524005579452014e-07,
+      "loss": 0.019,
+      "step": 1180
+    },
+    {
+      "epoch": 1.7706146926536732,
+      "grad_norm": 0.5080977377931006,
+      "learning_rate": 3.2107552592001657e-07,
+      "loss": 0.0143,
+      "step": 1181
+    },
+    {
+      "epoch": 1.7721139430284858,
+      "grad_norm": 0.4234412023779566,
+      "learning_rate": 3.169369458398652e-07,
+      "loss": 0.0118,
+      "step": 1182
+    },
+    {
+      "epoch": 1.7736131934032984,
+      "grad_norm": 0.6614632198056868,
+      "learning_rate": 3.128243385070562e-07,
+      "loss": 0.0291,
+      "step": 1183
+    },
+    {
+      "epoch": 1.775112443778111,
+      "grad_norm": 0.4334579569338365,
+      "learning_rate": 3.087377267305297e-07,
+      "loss": 0.0248,
+      "step": 1184
+    },
+    {
+      "epoch": 1.7766116941529235,
+      "grad_norm": 0.4424075940067335,
+      "learning_rate": 3.0467713317505363e-07,
+      "loss": 0.0137,
+      "step": 1185
+    },
+    {
+      "epoch": 1.778110944527736,
+      "grad_norm": 0.4199325292664545,
+      "learning_rate": 3.006425803610963e-07,
+      "loss": 0.0174,
+      "step": 1186
+    },
+    {
+      "epoch": 1.7796101949025487,
+      "grad_norm": 0.5044497942994316,
+      "learning_rate": 2.9663409066470025e-07,
+      "loss": 0.0147,
+      "step": 1187
+    },
+    {
+      "epoch": 1.7811094452773615,
+      "grad_norm": 0.4008691643677315,
+      "learning_rate": 2.9265168631736005e-07,
+      "loss": 0.013,
+      "step": 1188
+    },
+    {
+      "epoch": 1.7826086956521738,
+      "grad_norm": 0.5000241259230002,
+      "learning_rate": 2.88695389405898e-07,
+      "loss": 0.0149,
+      "step": 1189
+    },
+    {
+      "epoch": 1.7841079460269866,
+      "grad_norm": 0.5013930123673712,
+      "learning_rate": 2.8476522187234177e-07,
+      "loss": 0.0212,
+      "step": 1190
+    },
+    {
+      "epoch": 1.785607196401799,
+      "grad_norm": 0.6295283712216917,
+      "learning_rate": 2.808612055138038e-07,
+      "loss": 0.0237,
+      "step": 1191
+    },
+    {
+      "epoch": 1.7871064467766118,
+      "grad_norm": 0.41129608628961056,
+      "learning_rate": 2.76983361982357e-07,
+      "loss": 0.0153,
+      "step": 1192
+    },
+    {
+      "epoch": 1.7886056971514241,
+      "grad_norm": 0.5147146090399922,
+      "learning_rate": 2.731317127849209e-07,
+      "loss": 0.0206,
+      "step": 1193
+    },
+    {
+      "epoch": 1.790104947526237,
+      "grad_norm": 0.5483909719071896,
+      "learning_rate": 2.693062792831358e-07,
+      "loss": 0.0229,
+      "step": 1194
+    },
+    {
+      "epoch": 1.7916041979010495,
+      "grad_norm": 0.465177289033475,
+      "learning_rate": 2.655070826932471e-07,
+      "loss": 0.0139,
+      "step": 1195
+    },
+    {
+      "epoch": 1.793103448275862,
+      "grad_norm": 0.5787320225151523,
+      "learning_rate": 2.617341440859883e-07,
+      "loss": 0.022,
+      "step": 1196
+    },
+    {
+      "epoch": 1.7946026986506747,
+      "grad_norm": 0.3539693287409812,
+      "learning_rate": 2.5798748438646326e-07,
+      "loss": 0.0154,
+      "step": 1197
+    },
+    {
+      "epoch": 1.7961019490254873,
+      "grad_norm": 0.4178260778038222,
+      "learning_rate": 2.5426712437403134e-07,
+      "loss": 0.0175,
+      "step": 1198
+    },
+    {
+      "epoch": 1.7976011994002998,
+      "grad_norm": 0.30574592784773713,
+      "learning_rate": 2.5057308468218913e-07,
+      "loss": 0.0085,
+      "step": 1199
+    },
+    {
+      "epoch": 1.7991004497751124,
+      "grad_norm": 0.43789508175406455,
+      "learning_rate": 2.4690538579845933e-07,
+      "loss": 0.0161,
+      "step": 1200
+    },
+    {
+      "epoch": 1.7991004497751124,
+      "eval_loss": 0.09375476837158203,
+      "eval_runtime": 8.1107,
+      "eval_samples_per_second": 6.658,
+      "eval_steps_per_second": 1.726,
+      "step": 1200
+    },
+    {
+      "epoch": 1.800599700149925,
+      "grad_norm": 0.5823918161449702,
+      "learning_rate": 2.432640480642756e-07,
+      "loss": 0.0311,
+      "step": 1201
+    },
+    {
+      "epoch": 1.8020989505247376,
+      "grad_norm": 0.4156212086061445,
+      "learning_rate": 2.396490916748706e-07,
+      "loss": 0.0141,
+      "step": 1202
+    },
+    {
+      "epoch": 1.8035982008995504,
+      "grad_norm": 0.8719997255307311,
+      "learning_rate": 2.360605366791624e-07,
+      "loss": 0.0461,
+      "step": 1203
+    },
+    {
+      "epoch": 1.8050974512743627,
+      "grad_norm": 0.6439982877002179,
+      "learning_rate": 2.32498402979644e-07,
+      "loss": 0.0247,
+      "step": 1204
+    },
+    {
+      "epoch": 1.8065967016491755,
+      "grad_norm": 0.49587513574538505,
+      "learning_rate": 2.2896271033227392e-07,
+      "loss": 0.0197,
+      "step": 1205
+    },
+    {
+      "epoch": 1.8080959520239879,
+      "grad_norm": 0.5478395152669084,
+      "learning_rate": 2.2545347834636632e-07,
+      "loss": 0.034,
+      "step": 1206
+    },
+    {
+      "epoch": 1.8095952023988007,
+      "grad_norm": 0.41444956227593244,
+      "learning_rate": 2.219707264844806e-07,
+      "loss": 0.0175,
+      "step": 1207
+    },
+    {
+      "epoch": 1.811094452773613,
+      "grad_norm": 0.526008029314854,
+      "learning_rate": 2.1851447406231573e-07,
+      "loss": 0.0201,
+      "step": 1208
+    },
+    {
+      "epoch": 1.8125937031484258,
+      "grad_norm": 0.5851053731474306,
+      "learning_rate": 2.1508474024860171e-07,
+      "loss": 0.0218,
+      "step": 1209
+    },
+    {
+      "epoch": 1.8140929535232384,
+      "grad_norm": 0.4628719383752043,
+      "learning_rate": 2.1168154406499275e-07,
+      "loss": 0.0149,
+      "step": 1210
+    },
+    {
+      "epoch": 1.815592203898051,
+      "grad_norm": 0.5770985424064698,
+      "learning_rate": 2.0830490438596418e-07,
+      "loss": 0.0233,
+      "step": 1211
+    },
+    {
+      "epoch": 1.8170914542728636,
+      "grad_norm": 0.33022093774131595,
+      "learning_rate": 2.0495483993870578e-07,
+      "loss": 0.0131,
+      "step": 1212
+    },
+    {
+      "epoch": 1.8185907046476761,
+      "grad_norm": 0.3446939058642555,
+      "learning_rate": 2.0163136930301696e-07,
+      "loss": 0.0097,
+      "step": 1213
+    },
+    {
+      "epoch": 1.8200899550224887,
+      "grad_norm": 0.6026302725668491,
+      "learning_rate": 1.9833451091120727e-07,
+      "loss": 0.0235,
+      "step": 1214
+    },
+    {
+      "epoch": 1.8215892053973013,
+      "grad_norm": 0.7705976743349287,
+      "learning_rate": 1.9506428304799095e-07,
+      "loss": 0.0275,
+      "step": 1215
+    },
+    {
+      "epoch": 1.823088455772114,
+      "grad_norm": 0.4111972792890735,
+      "learning_rate": 1.9182070385038555e-07,
+      "loss": 0.0137,
+      "step": 1216
+    },
+    {
+      "epoch": 1.8245877061469264,
+      "grad_norm": 0.3151222470087879,
+      "learning_rate": 1.886037913076144e-07,
+      "loss": 0.0085,
+      "step": 1217
+    },
+    {
+      "epoch": 1.8260869565217392,
+      "grad_norm": 0.6262152782143581,
+      "learning_rate": 1.8541356326100436e-07,
+      "loss": 0.0267,
+      "step": 1218
+    },
+    {
+      "epoch": 1.8275862068965516,
+      "grad_norm": 0.5631490974568135,
+      "learning_rate": 1.8225003740388546e-07,
+      "loss": 0.0242,
+      "step": 1219
+    },
+    {
+      "epoch": 1.8290854572713644,
+      "grad_norm": 0.487945022650714,
+      "learning_rate": 1.791132312814975e-07,
+      "loss": 0.0157,
+      "step": 1220
+    },
+    {
+      "epoch": 1.8305847076461768,
+      "grad_norm": 0.41023271846996723,
+      "learning_rate": 1.760031622908881e-07,
+      "loss": 0.0153,
+      "step": 1221
+    },
+    {
+      "epoch": 1.8320839580209896,
+      "grad_norm": 0.4508922406656978,
+      "learning_rate": 1.729198476808186e-07,
+      "loss": 0.0176,
+      "step": 1222
+    },
+    {
+      "epoch": 1.8335832083958021,
+      "grad_norm": 0.488751573667202,
+      "learning_rate": 1.6986330455166733e-07,
+      "loss": 0.0182,
+      "step": 1223
+    },
+    {
+      "epoch": 1.8350824587706147,
+      "grad_norm": 0.5550698475033068,
+      "learning_rate": 1.6683354985533583e-07,
+      "loss": 0.0194,
+      "step": 1224
+    },
+    {
+      "epoch": 1.8365817091454273,
+      "grad_norm": 0.41114727714900323,
+      "learning_rate": 1.6383060039515343e-07,
+      "loss": 0.0132,
+      "step": 1225
+    },
+    {
+      "epoch": 1.8380809595202399,
+      "grad_norm": 0.46930421878386913,
+      "learning_rate": 1.6085447282578548e-07,
+      "loss": 0.0124,
+      "step": 1226
+    },
+    {
+      "epoch": 1.8395802098950524,
+      "grad_norm": 0.559735898801587,
+      "learning_rate": 1.579051836531409e-07,
+      "loss": 0.0198,
+      "step": 1227
+    },
+    {
+      "epoch": 1.841079460269865,
+      "grad_norm": 0.5680010127348258,
+      "learning_rate": 1.5498274923427925e-07,
+      "loss": 0.0251,
+      "step": 1228
+    },
+    {
+      "epoch": 1.8425787106446778,
+      "grad_norm": 0.5275128813335922,
+      "learning_rate": 1.5208718577732096e-07,
+      "loss": 0.0209,
+      "step": 1229
+    },
+    {
+      "epoch": 1.8440779610194902,
+      "grad_norm": 0.3341080863053713,
+      "learning_rate": 1.4921850934135785e-07,
+      "loss": 0.0103,
+      "step": 1230
+    },
+    {
+      "epoch": 1.845577211394303,
+      "grad_norm": 0.6027550964993424,
+      "learning_rate": 1.463767358363627e-07,
+      "loss": 0.026,
+      "step": 1231
+    },
+    {
+      "epoch": 1.8470764617691153,
+      "grad_norm": 0.48570803619615205,
+      "learning_rate": 1.4356188102310266e-07,
+      "loss": 0.0188,
+      "step": 1232
+    },
+    {
+      "epoch": 1.8485757121439281,
+      "grad_norm": 0.6356577573932962,
+      "learning_rate": 1.4077396051305093e-07,
+      "loss": 0.0307,
+      "step": 1233
+    },
+    {
+      "epoch": 1.8500749625187405,
+      "grad_norm": 0.5509718807764291,
+      "learning_rate": 1.3801298976830025e-07,
+      "loss": 0.024,
+      "step": 1234
+    },
+    {
+      "epoch": 1.8515742128935533,
+      "grad_norm": 0.40423235793071405,
+      "learning_rate": 1.3527898410147677e-07,
+      "loss": 0.0124,
+      "step": 1235
+    },
+    {
+      "epoch": 1.8530734632683659,
+      "grad_norm": 0.5679090326115342,
+      "learning_rate": 1.325719586756563e-07,
+      "loss": 0.0205,
+      "step": 1236
+    },
+    {
+      "epoch": 1.8545727136431784,
+      "grad_norm": 0.5693690902132705,
+      "learning_rate": 1.2989192850427933e-07,
+      "loss": 0.0275,
+      "step": 1237
+    },
+    {
+      "epoch": 1.856071964017991,
+      "grad_norm": 0.5873212686539716,
+      "learning_rate": 1.2723890845106723e-07,
+      "loss": 0.0296,
+      "step": 1238
+    },
+    {
+      "epoch": 1.8575712143928036,
+      "grad_norm": 0.46481336998840667,
+      "learning_rate": 1.2461291322994118e-07,
+      "loss": 0.0219,
+      "step": 1239
+    },
+    {
+      "epoch": 1.8590704647676162,
+      "grad_norm": 0.49583796979812417,
+      "learning_rate": 1.2201395740493948e-07,
+      "loss": 0.0226,
+      "step": 1240
+    },
+    {
+      "epoch": 1.8605697151424287,
+      "grad_norm": 0.6222080093103628,
+      "learning_rate": 1.1944205539013708e-07,
+      "loss": 0.0288,
+      "step": 1241
+    },
+    {
+      "epoch": 1.8620689655172413,
+      "grad_norm": 0.3471552456471302,
+      "learning_rate": 1.1689722144956672e-07,
+      "loss": 0.0137,
+      "step": 1242
+    },
+    {
+      "epoch": 1.863568215892054,
+      "grad_norm": 0.4035349983647765,
+      "learning_rate": 1.1437946969713731e-07,
+      "loss": 0.0193,
+      "step": 1243
+    },
+    {
+      "epoch": 1.8650674662668667,
+      "grad_norm": 0.5667306131662019,
+      "learning_rate": 1.1188881409655849e-07,
+      "loss": 0.0207,
+      "step": 1244
+    },
+    {
+      "epoch": 1.866566716641679,
+      "grad_norm": 0.38094119318719766,
+      "learning_rate": 1.0942526846126122e-07,
+      "loss": 0.0156,
+      "step": 1245
+    },
+    {
+      "epoch": 1.8680659670164919,
+      "grad_norm": 0.5434028141481656,
+      "learning_rate": 1.0698884645432117e-07,
+      "loss": 0.0151,
+      "step": 1246
+    },
+    {
+      "epoch": 1.8695652173913042,
+      "grad_norm": 0.4907173134907601,
+      "learning_rate": 1.0457956158838545e-07,
+      "loss": 0.0189,
+      "step": 1247
+    },
+    {
+      "epoch": 1.871064467766117,
+      "grad_norm": 0.4801096581098125,
+      "learning_rate": 1.0219742722559433e-07,
+      "loss": 0.0179,
+      "step": 1248
+    },
+    {
+      "epoch": 1.8725637181409296,
+      "grad_norm": 0.4701995294286294,
+      "learning_rate": 9.984245657750857e-08,
+      "loss": 0.0174,
+      "step": 1249
+    },
+    {
+      "epoch": 1.8740629685157422,
+      "grad_norm": 0.9910041419710222,
+      "learning_rate": 9.751466270503718e-08,
+      "loss": 0.0343,
+      "step": 1250
+    },
+    {
+      "epoch": 1.8755622188905547,
+      "grad_norm": 0.6185799611148916,
+      "learning_rate": 9.521405851836252e-08,
+      "loss": 0.0278,
+      "step": 1251
+    },
+    {
+      "epoch": 1.8770614692653673,
+      "grad_norm": 0.45154472102236554,
+      "learning_rate": 9.294065677687202e-08,
+      "loss": 0.0159,
+      "step": 1252
+    },
+    {
+      "epoch": 1.87856071964018,
+      "grad_norm": 0.5798728190358127,
+      "learning_rate": 9.069447008908383e-08,
+      "loss": 0.0239,
+      "step": 1253
+    },
+    {
+      "epoch": 1.8800599700149925,
+      "grad_norm": 0.4625272810378491,
+      "learning_rate": 8.847551091257956e-08,
+      "loss": 0.0164,
+      "step": 1254
+    },
+    {
+      "epoch": 1.881559220389805,
+      "grad_norm": 0.52030482492452,
+      "learning_rate": 8.62837915539344e-08,
+      "loss": 0.0249,
+      "step": 1255
+    },
+    {
+      "epoch": 1.8830584707646176,
+      "grad_norm": 0.4484408605469282,
+      "learning_rate": 8.411932416864832e-08,
+      "loss": 0.0159,
+      "step": 1256
+    },
+    {
+      "epoch": 1.8845577211394304,
+      "grad_norm": 0.42631724015214423,
+      "learning_rate": 8.198212076107881e-08,
+      "loss": 0.0203,
+      "step": 1257
+    },
+    {
+      "epoch": 1.8860569715142428,
+      "grad_norm": 0.3307501389835731,
+      "learning_rate": 7.987219318437489e-08,
+      "loss": 0.0101,
+      "step": 1258
+    },
+    {
+      "epoch": 1.8875562218890556,
+      "grad_norm": 0.6037727172959414,
+      "learning_rate": 7.778955314041103e-08,
+      "loss": 0.0268,
+      "step": 1259
+    },
+    {
+      "epoch": 1.889055472263868,
+      "grad_norm": 0.5393439839929652,
+      "learning_rate": 7.573421217972222e-08,
+      "loss": 0.0181,
+      "step": 1260
+    },
+    {
+      "epoch": 1.8905547226386807,
+      "grad_norm": 0.6550776404941613,
+      "learning_rate": 7.370618170144062e-08,
+      "loss": 0.0302,
+      "step": 1261
+    },
+    {
+      "epoch": 1.892053973013493,
+      "grad_norm": 0.5228883000301223,
+      "learning_rate": 7.170547295323016e-08,
+      "loss": 0.0286,
+      "step": 1262
+    },
+    {
+      "epoch": 1.893553223388306,
+      "grad_norm": 0.3566280620564002,
+      "learning_rate": 6.973209703122652e-08,
+      "loss": 0.0137,
+      "step": 1263
+    },
+    {
+      "epoch": 1.8950524737631185,
+      "grad_norm": 0.6042019477302487,
+      "learning_rate": 6.778606487997496e-08,
+      "loss": 0.0256,
+      "step": 1264
+    },
+    {
+      "epoch": 1.896551724137931,
+      "grad_norm": 0.5295295851315833,
+      "learning_rate": 6.58673872923693e-08,
+      "loss": 0.0205,
+      "step": 1265
+    },
+    {
+      "epoch": 1.8980509745127436,
+      "grad_norm": 0.5676980427887119,
+      "learning_rate": 6.397607490959134e-08,
+      "loss": 0.0151,
+      "step": 1266
+    },
+    {
+      "epoch": 1.8995502248875562,
+      "grad_norm": 0.5630645035796176,
+      "learning_rate": 6.211213822105378e-08,
+      "loss": 0.0238,
+      "step": 1267
+    },
+    {
+      "epoch": 1.9010494752623688,
+      "grad_norm": 0.6824690529597209,
+      "learning_rate": 6.027558756434015e-08,
+      "loss": 0.0257,
+      "step": 1268
+    },
+    {
+      "epoch": 1.9025487256371814,
+      "grad_norm": 0.4496618205563172,
+      "learning_rate": 5.846643312514888e-08,
+      "loss": 0.0158,
+      "step": 1269
+    },
+    {
+      "epoch": 1.9040479760119942,
+      "grad_norm": 0.5525013511048589,
+      "learning_rate": 5.668468493723489e-08,
+      "loss": 0.0168,
+      "step": 1270
+    },
+    {
+      "epoch": 1.9055472263868065,
+      "grad_norm": 0.4342510393251234,
+      "learning_rate": 5.4930352882357486e-08,
+      "loss": 0.0146,
+      "step": 1271
+    },
+    {
+      "epoch": 1.9070464767616193,
+      "grad_norm": 0.7521060060304364,
+      "learning_rate": 5.3203446690220374e-08,
+      "loss": 0.03,
+      "step": 1272
+    },
+    {
+      "epoch": 1.9085457271364317,
+      "grad_norm": 0.4187671259415702,
+      "learning_rate": 5.1503975938422824e-08,
+      "loss": 0.0175,
+      "step": 1273
+    },
+    {
+      "epoch": 1.9100449775112445,
+      "grad_norm": 0.43006513391134743,
+      "learning_rate": 4.983195005240415e-08,
+      "loss": 0.0161,
+      "step": 1274
+    },
+    {
+      "epoch": 1.9115442278860568,
+      "grad_norm": 0.47016154547458827,
+      "learning_rate": 4.8187378305390994e-08,
+      "loss": 0.0228,
+      "step": 1275
+    },
+    {
+      "epoch": 1.9130434782608696,
+      "grad_norm": 0.3897414998085116,
+      "learning_rate": 4.657026981834623e-08,
+      "loss": 0.0127,
+      "step": 1276
+    },
+    {
+      "epoch": 1.9145427286356822,
+      "grad_norm": 0.44874252607204496,
+      "learning_rate": 4.498063355991955e-08,
+      "loss": 0.019,
+      "step": 1277
+    },
+    {
+      "epoch": 1.9160419790104948,
+      "grad_norm": 0.5202521923551553,
+      "learning_rate": 4.341847834639645e-08,
+      "loss": 0.021,
+      "step": 1278
+    },
+    {
+      "epoch": 1.9175412293853074,
+      "grad_norm": 0.7389911916613162,
+      "learning_rate": 4.188381284164933e-08,
+      "loss": 0.0242,
+      "step": 1279
+    },
+    {
+      "epoch": 1.91904047976012,
+      "grad_norm": 0.48233836375391353,
+      "learning_rate": 4.0376645557090864e-08,
+      "loss": 0.018,
+      "step": 1280
+    },
+    {
+      "epoch": 1.9205397301349325,
+      "grad_norm": 0.8310399921718331,
+      "learning_rate": 3.889698485162463e-08,
+      "loss": 0.0378,
+      "step": 1281
+    },
+    {
+      "epoch": 1.922038980509745,
+      "grad_norm": 0.5671332493920915,
+      "learning_rate": 3.744483893160067e-08,
+      "loss": 0.0241,
+      "step": 1282
+    },
+    {
+      "epoch": 1.9235382308845579,
+      "grad_norm": 0.45610047707168994,
+      "learning_rate": 3.602021585076942e-08,
+      "loss": 0.0156,
+      "step": 1283
+    },
+    {
+      "epoch": 1.9250374812593702,
+      "grad_norm": 0.8281691639626697,
+      "learning_rate": 3.462312351023567e-08,
+      "loss": 0.0394,
+      "step": 1284
+    },
+    {
+      "epoch": 1.926536731634183,
+      "grad_norm": 0.6378798868078879,
+      "learning_rate": 3.325356965841686e-08,
+      "loss": 0.0149,
+      "step": 1285
+    },
+    {
+      "epoch": 1.9280359820089954,
+      "grad_norm": 0.37459436436258153,
+      "learning_rate": 3.191156189099931e-08,
+      "loss": 0.0126,
+      "step": 1286
+    },
+    {
+      "epoch": 1.9295352323838082,
+      "grad_norm": 0.4247084534527821,
+      "learning_rate": 3.0597107650894855e-08,
+      "loss": 0.0135,
+      "step": 1287
+    },
+    {
+      "epoch": 1.9310344827586206,
+      "grad_norm": 0.3205757342441444,
+      "learning_rate": 2.9310214228202016e-08,
+      "loss": 0.0079,
+      "step": 1288
+    },
+    {
+      "epoch": 1.9325337331334334,
+      "grad_norm": 0.539908383764795,
+      "learning_rate": 2.8050888760163265e-08,
+      "loss": 0.0206,
+      "step": 1289
+    },
+    {
+      "epoch": 1.934032983508246,
+      "grad_norm": 0.5859102960306886,
+      "learning_rate": 2.6819138231126695e-08,
+      "loss": 0.0278,
+      "step": 1290
+    },
+    {
+      "epoch": 1.9355322338830585,
+      "grad_norm": 0.4619953721662239,
+      "learning_rate": 2.5614969472506634e-08,
+      "loss": 0.0154,
+      "step": 1291
+    },
+    {
+      "epoch": 1.937031484257871,
+      "grad_norm": 0.4221258017425029,
+      "learning_rate": 2.4438389162746434e-08,
+      "loss": 0.0148,
+      "step": 1292
+    },
+    {
+      "epoch": 1.9385307346326837,
+      "grad_norm": 0.4713780929226789,
+      "learning_rate": 2.3289403827281287e-08,
+      "loss": 0.0159,
+      "step": 1293
+    },
+    {
+      "epoch": 1.9400299850074962,
+      "grad_norm": 0.5018361895934237,
+      "learning_rate": 2.2168019838501032e-08,
+      "loss": 0.0233,
+      "step": 1294
+    },
+    {
+      "epoch": 1.9415292353823088,
+      "grad_norm": 0.45230343893246255,
+      "learning_rate": 2.1074243415716288e-08,
+      "loss": 0.0181,
+      "step": 1295
+    },
+    {
+      "epoch": 1.9430284857571214,
+      "grad_norm": 0.48310911582138816,
+      "learning_rate": 2.0008080625124048e-08,
+      "loss": 0.018,
+      "step": 1296
+    },
+    {
+      "epoch": 1.944527736131934,
+      "grad_norm": 0.3787828920524053,
+      "learning_rate": 1.896953737977103e-08,
+      "loss": 0.0121,
+      "step": 1297
+    },
+    {
+      "epoch": 1.9460269865067468,
+      "grad_norm": 0.4562164632010712,
+      "learning_rate": 1.7958619439524817e-08,
+      "loss": 0.0163,
+      "step": 1298
+    },
+    {
+      "epoch": 1.9475262368815591,
+      "grad_norm": 0.49548888900892796,
+      "learning_rate": 1.6975332411040547e-08,
+      "loss": 0.0174,
+      "step": 1299
+    },
+    {
+      "epoch": 1.949025487256372,
+      "grad_norm": 0.5240321671778967,
+      "learning_rate": 1.601968174772761e-08,
+      "loss": 0.0168,
+      "step": 1300
+    },
+    {
+      "epoch": 1.9505247376311843,
+      "grad_norm": 0.5785677293885776,
+      "learning_rate": 1.5091672749723564e-08,
+      "loss": 0.0291,
+      "step": 1301
+    },
+    {
+      "epoch": 1.952023988005997,
+      "grad_norm": 0.5686256203304928,
+      "learning_rate": 1.4191310563860806e-08,
+      "loss": 0.0187,
+      "step": 1302
+    },
+    {
+      "epoch": 1.9535232383808094,
+      "grad_norm": 0.47752404513305463,
+      "learning_rate": 1.331860018363995e-08,
+      "loss": 0.0145,
+      "step": 1303
+    },
+    {
+      "epoch": 1.9550224887556222,
+      "grad_norm": 0.5251901851436875,
+      "learning_rate": 1.2473546449203178e-08,
+      "loss": 0.0152,
+      "step": 1304
+    },
+    {
+      "epoch": 1.9565217391304348,
+      "grad_norm": 0.41043784536271305,
+      "learning_rate": 1.1656154047303691e-08,
+      "loss": 0.0186,
+      "step": 1305
+    },
+    {
+      "epoch": 1.9580209895052474,
+      "grad_norm": 0.4083450622258475,
+      "learning_rate": 1.0866427511285194e-08,
+      "loss": 0.0123,
+      "step": 1306
+    },
+    {
+      "epoch": 1.95952023988006,
+      "grad_norm": 0.38770277702625333,
+      "learning_rate": 1.0104371221050236e-08,
+      "loss": 0.0126,
+      "step": 1307
+    },
+    {
+      "epoch": 1.9610194902548725,
+      "grad_norm": 0.6740211829853863,
+      "learning_rate": 9.369989403041347e-09,
+      "loss": 0.0295,
+      "step": 1308
+    },
+    {
+      "epoch": 1.9625187406296851,
+      "grad_norm": 0.45020048476551494,
+      "learning_rate": 8.663286130216608e-09,
+      "loss": 0.0149,
+      "step": 1309
+    },
+    {
+      "epoch": 1.9640179910044977,
+      "grad_norm": 0.6374767825522553,
+      "learning_rate": 7.984265322023011e-09,
+      "loss": 0.0189,
+      "step": 1310
+    },
+    {
+      "epoch": 1.9655172413793105,
+      "grad_norm": 0.5227629544487636,
+      "learning_rate": 7.332930744380906e-09,
+      "loss": 0.0301,
+      "step": 1311
+    },
+    {
+      "epoch": 1.9670164917541229,
+      "grad_norm": 0.5464776291743795,
+      "learning_rate": 6.709286009657368e-09,
+      "loss": 0.0197,
+      "step": 1312
+    },
+    {
+      "epoch": 1.9685157421289357,
+      "grad_norm": 0.6614254575464079,
+      "learning_rate": 6.1133345766511975e-09,
+      "loss": 0.025,
+      "step": 1313
+    },
+    {
+      "epoch": 1.970014992503748,
+      "grad_norm": 0.53627172259005,
+      "learning_rate": 5.5450797505690605e-09,
+      "loss": 0.0218,
+      "step": 1314
+    },
+    {
+      "epoch": 1.9715142428785608,
+      "grad_norm": 0.325913033828512,
+      "learning_rate": 5.004524683011048e-09,
+      "loss": 0.0092,
+      "step": 1315
+    },
+    {
+      "epoch": 1.9730134932533732,
+      "grad_norm": 0.4814264646824674,
+      "learning_rate": 4.491672371950695e-09,
+      "loss": 0.0192,
+      "step": 1316
+    },
+    {
+      "epoch": 1.974512743628186,
+      "grad_norm": 0.6215176135824605,
+      "learning_rate": 4.0065256617199954e-09,
+      "loss": 0.0225,
+      "step": 1317
+    },
+    {
+      "epoch": 1.9760119940029985,
+      "grad_norm": 0.6076251674892379,
+      "learning_rate": 3.5490872429910784e-09,
+      "loss": 0.0222,
+      "step": 1318
+    },
+    {
+      "epoch": 1.9775112443778111,
+      "grad_norm": 0.503029034321492,
+      "learning_rate": 3.119359652765108e-09,
+      "loss": 0.0186,
+      "step": 1319
+    },
+    {
+      "epoch": 1.9790104947526237,
+      "grad_norm": 0.68322896044437,
+      "learning_rate": 2.7173452743550767e-09,
+      "loss": 0.0277,
+      "step": 1320
+    },
+    {
+      "epoch": 1.9805097451274363,
+      "grad_norm": 0.6014987327411556,
+      "learning_rate": 2.343046337374144e-09,
+      "loss": 0.0196,
+      "step": 1321
+    },
+    {
+      "epoch": 1.9820089955022488,
+      "grad_norm": 0.5336859827851407,
+      "learning_rate": 1.9964649177223184e-09,
+      "loss": 0.0211,
+      "step": 1322
+    },
+    {
+      "epoch": 1.9835082458770614,
+      "grad_norm": 0.506766825967047,
+      "learning_rate": 1.6776029375759062e-09,
+      "loss": 0.0198,
+      "step": 1323
+    },
+    {
+      "epoch": 1.9850074962518742,
+      "grad_norm": 0.7304033237326711,
+      "learning_rate": 1.386462165375857e-09,
+      "loss": 0.0278,
+      "step": 1324
+    },
+    {
+      "epoch": 1.9865067466266866,
+      "grad_norm": 0.473486083985319,
+      "learning_rate": 1.1230442158188804e-09,
+      "loss": 0.0187,
+      "step": 1325
+    },
+    {
+      "epoch": 1.9880059970014994,
+      "grad_norm": 0.665948392484204,
+      "learning_rate": 8.873505498474544e-10,
+      "loss": 0.0259,
+      "step": 1326
+    },
+    {
+      "epoch": 1.9895052473763117,
+      "grad_norm": 0.7048482541340728,
+      "learning_rate": 6.793824746437194e-10,
+      "loss": 0.0266,
+      "step": 1327
+    },
+    {
+      "epoch": 1.9910044977511245,
+      "grad_norm": 0.46197518412598776,
+      "learning_rate": 4.991411436189308e-10,
+      "loss": 0.0143,
+      "step": 1328
+    },
+    {
+      "epoch": 1.992503748125937,
+      "grad_norm": 0.4986404688539546,
+      "learning_rate": 3.466275564101285e-10,
+      "loss": 0.0207,
+      "step": 1329
+    },
+    {
+      "epoch": 1.9940029985007497,
+      "grad_norm": 0.524222800952551,
+      "learning_rate": 2.2184255887403028e-10,
+      "loss": 0.0187,
+      "step": 1330
+    },
+    {
+      "epoch": 1.9955022488755623,
+      "grad_norm": 0.5020002557383357,
+      "learning_rate": 1.2478684308037115e-10,
+      "loss": 0.0185,
+      "step": 1331
+    },
+    {
+      "epoch": 1.9970014992503748,
+      "grad_norm": 0.5167378768133697,
+      "learning_rate": 5.5460947310237435e-11,
+      "loss": 0.0183,
+      "step": 1332
+    },
+    {
+      "epoch": 1.9985007496251874,
+      "grad_norm": 0.38751877067201607,
+      "learning_rate": 1.3865256052181252e-11,
+      "loss": 0.0126,
+      "step": 1333
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.25045675932878464,
+      "learning_rate": 0.0,
+      "loss": 0.0056,
+      "step": 1334
+    },
+    {
+      "epoch": 2.0,
+      "step": 1334,
+      "total_flos": 13753039749120.0,
+      "train_loss": 0.05433965472272713,
+      "train_runtime": 2996.5328,
+      "train_samples_per_second": 3.557,
+      "train_steps_per_second": 0.445
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1334,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 13753039749120.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}