diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5621 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8313847752663029,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 1229.812744140625,
+      "learning_rate": 5e-06,
+      "loss": 30.4769,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 309.7672119140625,
+      "learning_rate": 1e-05,
+      "loss": 27.4271,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 192.40821838378906,
+      "learning_rate": 1.5e-05,
+      "loss": 23.191,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 101.36874389648438,
+      "learning_rate": 2e-05,
+      "loss": 18.2568,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 56.78367614746094,
+      "learning_rate": 2.5e-05,
+      "loss": 13.6882,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 22.516868591308594,
+      "learning_rate": 3e-05,
+      "loss": 10.1169,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 13.551531791687012,
+      "learning_rate": 3.5e-05,
+      "loss": 8.5886,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 12.628911018371582,
+      "learning_rate": 4e-05,
+      "loss": 8.2086,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 13.996498107910156,
+      "learning_rate": 4.5e-05,
+      "loss": 8.1833,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.634190559387207,
+      "learning_rate": 5e-05,
+      "loss": 8.0995,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.44438362121582,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 8.0283,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.912370204925537,
+      "learning_rate": 6e-05,
+      "loss": 7.9718,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.106713771820068,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 7.9734,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.923928260803223,
+      "learning_rate": 7e-05,
+      "loss": 7.8768,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 17.63799285888672,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 7.7631,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 14.726093292236328,
+      "learning_rate": 8e-05,
+      "loss": 7.557,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 12.786348342895508,
+      "learning_rate": 8.5e-05,
+      "loss": 7.4308,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.956538677215576,
+      "learning_rate": 9e-05,
+      "loss": 7.2131,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.928043365478516,
+      "learning_rate": 9.5e-05,
+      "loss": 7.0953,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.65689468383789,
+      "learning_rate": 0.0001,
+      "loss": 6.9548,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.416757583618164,
+      "learning_rate": 9.999972205865686e-05,
+      "loss": 6.7371,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.33735179901123,
+      "learning_rate": 9.999888823771751e-05,
+      "loss": 6.6175,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.704575061798096,
+      "learning_rate": 9.999749854645204e-05,
+      "loss": 6.432,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 10.699024200439453,
+      "learning_rate": 9.99955530003106e-05,
+      "loss": 6.2739,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.885202407836914,
+      "learning_rate": 9.99930516209231e-05,
+      "loss": 6.1744,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 9.077549934387207,
+      "learning_rate": 9.998999443609897e-05,
+      "loss": 5.9334,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.783599853515625,
+      "learning_rate": 9.998638147982696e-05,
+      "loss": 5.9392,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.453628063201904,
+      "learning_rate": 9.998221279227467e-05,
+      "loss": 5.8259,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.611512660980225,
+      "learning_rate": 9.997748841978812e-05,
+      "loss": 5.7614,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.822922706604004,
+      "learning_rate": 9.997220841489122e-05,
+      "loss": 5.6118,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.933773994445801,
+      "learning_rate": 9.996637283628528e-05,
+      "loss": 5.522,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.299020767211914,
+      "learning_rate": 9.995998174884821e-05,
+      "loss": 5.4244,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.691274166107178,
+      "learning_rate": 9.995303522363394e-05,
+      "loss": 5.3919,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.277099132537842,
+      "learning_rate": 9.99455333378715e-05,
+      "loss": 5.3489,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7971572875976562,
+      "learning_rate": 9.993747617496428e-05,
+      "loss": 5.2339,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9199624061584473,
+      "learning_rate": 9.9928863824489e-05,
+      "loss": 5.1713,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2884421348571777,
+      "learning_rate": 9.99196963821948e-05,
+      "loss": 5.13,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.122908592224121,
+      "learning_rate": 9.990997395000217e-05,
+      "loss": 5.0172,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.893023729324341,
+      "learning_rate": 9.989969663600169e-05,
+      "loss": 5.0076,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7886602878570557,
+      "learning_rate": 9.9888864554453e-05,
+      "loss": 5.0649,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7208797931671143,
+      "learning_rate": 9.987747782578342e-05,
+      "loss": 4.8919,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6963469982147217,
+      "learning_rate": 9.986553657658668e-05,
+      "loss": 4.7996,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.400980472564697,
+      "learning_rate": 9.985304093962145e-05,
+      "loss": 4.9208,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.301896095275879,
+      "learning_rate": 9.983999105380988e-05,
+      "loss": 4.7683,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1751794815063477,
+      "learning_rate": 9.982638706423608e-05,
+      "loss": 4.8343,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.351551055908203,
+      "learning_rate": 9.98122291221445e-05,
+      "loss": 4.6515,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5814805030822754,
+      "learning_rate": 9.979751738493826e-05,
+      "loss": 4.671,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3766465187072754,
+      "learning_rate": 9.978225201617732e-05,
+      "loss": 4.641,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2348690032958984,
+      "learning_rate": 9.976643318557678e-05,
+      "loss": 4.6256,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7532126903533936,
+      "learning_rate": 9.975006106900495e-05,
+      "loss": 4.4895,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8409101963043213,
+      "learning_rate": 9.973313584848132e-05,
+      "loss": 4.6573,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0017435550689697,
+      "learning_rate": 9.971565771217464e-05,
+      "loss": 4.5779,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3714191913604736,
+      "learning_rate": 9.969762685440076e-05,
+      "loss": 4.5046,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.982652187347412,
+      "learning_rate": 9.967904347562054e-05,
+      "loss": 4.5502,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0139358043670654,
+      "learning_rate": 9.965990778243755e-05,
+      "loss": 4.4332,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.529115915298462,
+      "learning_rate": 9.964021998759577e-05,
+      "loss": 4.3909,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.1116204261779785,
+      "learning_rate": 9.961998030997733e-05,
+      "loss": 4.3632,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.561253309249878,
+      "learning_rate": 9.95991889745999e-05,
+      "loss": 4.4698,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0149197578430176,
+      "learning_rate": 9.957784621261441e-05,
+      "loss": 4.4313,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.329812049865723,
+      "learning_rate": 9.955595226130226e-05,
+      "loss": 4.344,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1647002696990967,
+      "learning_rate": 9.953350736407282e-05,
+      "loss": 4.2796,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.213648796081543,
+      "learning_rate": 9.951051177046069e-05,
+      "loss": 4.2897,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8537462949752808,
+      "learning_rate": 9.948696573612292e-05,
+      "loss": 4.2881,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3809573650360107,
+      "learning_rate": 9.946286952283618e-05,
+      "loss": 4.3036,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.212860107421875,
+      "learning_rate": 9.943822339849381e-05,
+      "loss": 4.2527,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1665050983428955,
+      "learning_rate": 9.941302763710288e-05,
+      "loss": 4.2345,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2548954486846924,
+      "learning_rate": 9.938728251878116e-05,
+      "loss": 4.2502,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.7102184295654297,
+      "learning_rate": 9.936098832975393e-05,
+      "loss": 4.3006,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.172853469848633,
+      "learning_rate": 9.933414536235091e-05,
+      "loss": 4.246,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.476499557495117,
+      "learning_rate": 9.93067539150029e-05,
+      "loss": 4.1389,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0237905979156494,
+      "learning_rate": 9.927881429223853e-05,
+      "loss": 4.2699,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0171585083007812,
+      "learning_rate": 9.925032680468085e-05,
+      "loss": 4.1228,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2988505363464355,
+      "learning_rate": 9.922129176904388e-05,
+      "loss": 4.1321,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.95521879196167,
+      "learning_rate": 9.919170950812911e-05,
+      "loss": 4.049,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.220893144607544,
+      "learning_rate": 9.916158035082184e-05,
+      "loss": 4.1948,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9344232082366943,
+      "learning_rate": 9.913090463208763e-05,
+      "loss": 4.0705,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3103578090667725,
+      "learning_rate": 9.90996826929685e-05,
+      "loss": 4.1431,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.692507743835449,
+      "learning_rate": 9.906791488057916e-05,
+      "loss": 4.016,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.183342218399048,
+      "learning_rate": 9.903560154810313e-05,
+      "loss": 4.1322,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.961439609527588,
+      "learning_rate": 9.900274305478887e-05,
+      "loss": 3.9821,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1895840167999268,
+      "learning_rate": 9.896933976594572e-05,
+      "loss": 4.126,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0895981788635254,
+      "learning_rate": 9.893539205293989e-05,
+      "loss": 4.0772,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.85659122467041,
+      "learning_rate": 9.890090029319028e-05,
+      "loss": 3.9736,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9629476070404053,
+      "learning_rate": 9.886586487016433e-05,
+      "loss": 4.0457,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3512821197509766,
+      "learning_rate": 9.883028617337378e-05,
+      "loss": 4.0104,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0070366859436035,
+      "learning_rate": 9.879416459837022e-05,
+      "loss": 4.0205,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.785737156867981,
+      "learning_rate": 9.875750054674082e-05,
+      "loss": 4.0264,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6705329418182373,
+      "learning_rate": 9.872029442610382e-05,
+      "loss": 3.9757,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.837396502494812,
+      "learning_rate": 9.8682546650104e-05,
+      "loss": 4.0169,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0840909481048584,
+      "learning_rate": 9.864425763840802e-05,
+      "loss": 3.98,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6531455516815186,
+      "learning_rate": 9.860542781669988e-05,
+      "loss": 4.0411,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8012709617614746,
+      "learning_rate": 9.85660576166761e-05,
+      "loss": 3.9331,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9777172803878784,
+      "learning_rate": 9.852614747604093e-05,
+      "loss": 3.8908,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.182304859161377,
+      "learning_rate": 9.848569783850145e-05,
+      "loss": 3.9088,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.920596122741699,
+      "learning_rate": 9.844470915376278e-05,
+      "loss": 3.9597,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.971817135810852,
+      "learning_rate": 9.840318187752292e-05,
+      "loss": 3.8223,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.046733856201172,
+      "learning_rate": 9.836111647146771e-05,
+      "loss": 3.9495,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4442347288131714,
+      "learning_rate": 9.831851340326577e-05,
+      "loss": 3.9561,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5948429107666016,
+      "learning_rate": 9.82753731465633e-05,
+      "loss": 3.9302,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8231884241104126,
+      "learning_rate": 9.823169618097871e-05,
+      "loss": 3.9291,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.004462480545044,
+      "learning_rate": 9.81874829920974e-05,
+      "loss": 3.8685,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7805441617965698,
+      "learning_rate": 9.814273407146623e-05,
+      "loss": 3.8173,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4531677961349487,
+      "learning_rate": 9.809744991658829e-05,
+      "loss": 3.8398,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6244157552719116,
+      "learning_rate": 9.805163103091708e-05,
+      "loss": 3.8935,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.486452341079712,
+      "learning_rate": 9.800527792385112e-05,
+      "loss": 3.8778,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.142691135406494,
+      "learning_rate": 9.79583911107282e-05,
+      "loss": 3.919,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5212677717208862,
+      "learning_rate": 9.791097111281968e-05,
+      "loss": 3.8523,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7378300428390503,
+      "learning_rate": 9.786301845732467e-05,
+      "loss": 3.835,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5763616561889648,
+      "learning_rate": 9.781453367736418e-05,
+      "loss": 3.8253,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.663087248802185,
+      "learning_rate": 9.776551731197524e-05,
+      "loss": 3.7616,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.545979619026184,
+      "learning_rate": 9.771596990610478e-05,
+      "loss": 3.8516,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.53286874294281,
+      "learning_rate": 9.766589201060372e-05,
+      "loss": 3.77,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.624846935272217,
+      "learning_rate": 9.761528418222077e-05,
+      "loss": 3.7981,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.213869571685791,
+      "learning_rate": 9.756414698359624e-05,
+      "loss": 3.7741,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.046447515487671,
+      "learning_rate": 9.75124809832558e-05,
+      "loss": 3.8351,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3796635866165161,
+      "learning_rate": 9.746028675560413e-05,
+      "loss": 3.7778,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6972594261169434,
+      "learning_rate": 9.740756488091861e-05,
+      "loss": 3.7298,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.337314486503601,
+      "learning_rate": 9.735431594534277e-05,
+      "loss": 3.7911,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4657835960388184,
+      "learning_rate": 9.730054054087983e-05,
+      "loss": 3.8175,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.644771933555603,
+      "learning_rate": 9.724623926538612e-05,
+      "loss": 3.7631,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5241328477859497,
+      "learning_rate": 9.719141272256443e-05,
+      "loss": 3.7058,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7510253190994263,
+      "learning_rate": 9.713606152195726e-05,
+      "loss": 3.8659,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.608384370803833,
+      "learning_rate": 9.708018627894011e-05,
+      "loss": 3.7078,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.44278085231781,
+      "learning_rate": 9.702378761471456e-05,
+      "loss": 3.7982,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9439903497695923,
+      "learning_rate": 9.696686615630146e-05,
+      "loss": 3.7216,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7295832633972168,
+      "learning_rate": 9.690942253653385e-05,
+      "loss": 3.7839,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4564378261566162,
+      "learning_rate": 9.685145739405002e-05,
+      "loss": 3.7383,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3756966590881348,
+      "learning_rate": 9.679297137328634e-05,
+      "loss": 3.6504,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4621644020080566,
+      "learning_rate": 9.673396512447013e-05,
+      "loss": 3.7754,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.62283194065094,
+      "learning_rate": 9.667443930361247e-05,
+      "loss": 3.6665,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5261329412460327,
+      "learning_rate": 9.661439457250076e-05,
+      "loss": 3.7083,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.560340166091919,
+      "learning_rate": 9.655383159869158e-05,
+      "loss": 3.744,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6525319814682007,
+      "learning_rate": 9.649275105550309e-05,
+      "loss": 3.7419,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4129501581192017,
+      "learning_rate": 9.643115362200762e-05,
+      "loss": 3.7203,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.337715983390808,
+      "learning_rate": 9.636903998302409e-05,
+      "loss": 3.6888,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7548418045043945,
+      "learning_rate": 9.630641082911045e-05,
+      "loss": 3.6591,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4376401901245117,
+      "learning_rate": 9.624326685655593e-05,
+      "loss": 3.5796,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5718088150024414,
+      "learning_rate": 9.617960876737337e-05,
+      "loss": 3.6026,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4157310724258423,
+      "learning_rate": 9.611543726929134e-05,
+      "loss": 3.7093,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5425848960876465,
+      "learning_rate": 9.605075307574635e-05,
+      "loss": 3.7303,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.287452220916748,
+      "learning_rate": 9.598555690587487e-05,
+      "loss": 3.6574,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.592053771018982,
+      "learning_rate": 9.591984948450532e-05,
+      "loss": 3.6116,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5864181518554688,
+      "learning_rate": 9.585363154215008e-05,
+      "loss": 3.6425,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4077820777893066,
+      "learning_rate": 9.578690381499728e-05,
+      "loss": 3.686,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3487296104431152,
+      "learning_rate": 9.571966704490271e-05,
+      "loss": 3.6217,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9488983154296875,
+      "learning_rate": 9.565192197938148e-05,
+      "loss": 3.5956,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4107372760772705,
+      "learning_rate": 9.558366937159977e-05,
+      "loss": 3.737,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4383083581924438,
+      "learning_rate": 9.551490998036646e-05,
+      "loss": 3.648,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3736857175827026,
+      "learning_rate": 9.544564457012463e-05,
+      "loss": 3.6411,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3991824388504028,
+      "learning_rate": 9.537587391094314e-05,
+      "loss": 3.6154,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4682576656341553,
+      "learning_rate": 9.5305598778508e-05,
+      "loss": 3.6828,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.429841160774231,
+      "learning_rate": 9.52348199541138e-05,
+      "loss": 3.5335,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5785666704177856,
+      "learning_rate": 9.516353822465504e-05,
+      "loss": 3.6533,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5011215209960938,
+      "learning_rate": 9.509175438261726e-05,
+      "loss": 3.659,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.679041862487793,
+      "learning_rate": 9.501946922606838e-05,
+      "loss": 3.579,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.58845853805542,
+      "learning_rate": 9.494668355864973e-05,
+      "loss": 3.5899,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2959877252578735,
+      "learning_rate": 9.487339818956716e-05,
+      "loss": 3.5654,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5493981838226318,
+      "learning_rate": 9.479961393358203e-05,
+      "loss": 3.6012,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5047404766082764,
+      "learning_rate": 9.472533161100215e-05,
+      "loss": 3.6367,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3275983333587646,
+      "learning_rate": 9.465055204767265e-05,
+      "loss": 3.6274,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3751220703125,
+      "learning_rate": 9.457527607496685e-05,
+      "loss": 3.5199,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3353458642959595,
+      "learning_rate": 9.44995045297769e-05,
+      "loss": 3.5321,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2492051124572754,
+      "learning_rate": 9.442323825450464e-05,
+      "loss": 3.6332,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0786397457122803,
+      "learning_rate": 9.43464780970521e-05,
+      "loss": 3.6151,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4719089269638062,
+      "learning_rate": 9.426922491081212e-05,
+      "loss": 3.4847,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3098955154418945,
+      "learning_rate": 9.419147955465888e-05,
+      "loss": 3.5445,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4527875185012817,
+      "learning_rate": 9.411324289293832e-05,
+      "loss": 3.5209,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2426012754440308,
+      "learning_rate": 9.403451579545859e-05,
+      "loss": 3.5557,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3350285291671753,
+      "learning_rate": 9.395529913748025e-05,
+      "loss": 3.5265,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.466402530670166,
+      "learning_rate": 9.387559379970672e-05,
+      "loss": 3.5044,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9047514200210571,
+      "learning_rate": 9.379540066827431e-05,
+      "loss": 3.4028,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1885713338851929,
+      "learning_rate": 9.371472063474248e-05,
+      "loss": 3.5112,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.418020486831665,
+      "learning_rate": 9.363355459608394e-05,
+      "loss": 3.5393,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5337977409362793,
+      "learning_rate": 9.355190345467457e-05,
+      "loss": 3.498,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5761481523513794,
+      "learning_rate": 9.346976811828352e-05,
+      "loss": 3.4533,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.394687533378601,
+      "learning_rate": 9.338714950006297e-05,
+      "loss": 3.6147,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4166648387908936,
+      "learning_rate": 9.330404851853817e-05,
+      "loss": 3.5151,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5528873205184937,
+      "learning_rate": 9.3220466097597e-05,
+      "loss": 3.5674,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3615038394927979,
+      "learning_rate": 9.313640316647991e-05,
+      "loss": 3.4514,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3137192726135254,
+      "learning_rate": 9.305186065976945e-05,
+      "loss": 3.5113,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2795889377593994,
+      "learning_rate": 9.296683951737993e-05,
+      "loss": 3.5497,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1420422792434692,
+      "learning_rate": 9.288134068454697e-05,
+      "loss": 3.5683,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9743645191192627,
+      "learning_rate": 9.2795365111817e-05,
+      "loss": 3.5278,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2144616842269897,
+      "learning_rate": 9.270891375503665e-05,
+      "loss": 3.5173,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5434770584106445,
+      "learning_rate": 9.262198757534218e-05,
+      "loss": 3.5055,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.326240062713623,
+      "learning_rate": 9.253458753914874e-05,
+      "loss": 3.5729,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3340810537338257,
+      "learning_rate": 9.244671461813969e-05,
+      "loss": 3.5003,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2375303506851196,
+      "learning_rate": 9.235836978925572e-05,
+      "loss": 3.5078,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.30227530002594,
+      "learning_rate": 9.226955403468406e-05,
+      "loss": 3.5024,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.679872751235962,
+      "learning_rate": 9.21802683418475e-05,
+      "loss": 3.5012,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2751423120498657,
+      "learning_rate": 9.209051370339347e-05,
+      "loss": 3.504,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7539143562316895,
+      "learning_rate": 9.200029111718295e-05,
+      "loss": 3.5367,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2914196252822876,
+      "learning_rate": 9.190960158627941e-05,
+      "loss": 3.4866,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.5011461973190308,
+      "learning_rate": 9.181844611893766e-05,
+      "loss": 3.5024,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.258388638496399,
+      "learning_rate": 9.172682572859261e-05,
+      "loss": 3.3675,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1322250366210938,
+      "learning_rate": 9.163474143384806e-05,
+      "loss": 3.4236,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1992788314819336,
+      "learning_rate": 9.154219425846528e-05,
+      "loss": 3.4403,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2604423761367798,
+      "learning_rate": 9.144918523135175e-05,
+      "loss": 3.4361,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2565964460372925,
+      "learning_rate": 9.13557153865496e-05,
+      "loss": 3.513,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2667522430419922,
+      "learning_rate": 9.12617857632242e-05,
+      "loss": 3.3979,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.21877920627594,
+      "learning_rate": 9.116739740565259e-05,
+      "loss": 3.5425,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.243408441543579,
+      "learning_rate": 9.107255136321184e-05,
+      "loss": 3.4138,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.416135549545288,
+      "learning_rate": 9.09772486903674e-05,
+      "loss": 3.4064,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.505253791809082,
+      "learning_rate": 9.08814904466614e-05,
+      "loss": 3.4608,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.468630313873291,
+      "learning_rate": 9.078527769670085e-05,
+      "loss": 3.4985,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.4280701875686646,
+      "learning_rate": 9.068861151014575e-05,
+      "loss": 3.4031,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1800689697265625,
+      "learning_rate": 9.05914929616973e-05,
+      "loss": 3.3947,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3605331182479858,
+      "learning_rate": 9.04939231310859e-05,
+      "loss": 3.4994,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2450255155563354,
+      "learning_rate": 9.039590310305914e-05,
+      "loss": 3.4788,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.280288815498352,
+      "learning_rate": 9.029743396736974e-05,
+      "loss": 3.3975,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3635969161987305,
+      "learning_rate": 9.019851681876348e-05,
+      "loss": 3.4155,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2809624671936035,
+      "learning_rate": 9.009915275696693e-05,
+      "loss": 3.4478,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5251933336257935,
+      "learning_rate": 8.999934288667534e-05,
+      "loss": 3.6159,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2694870233535767,
+      "learning_rate": 8.989908831754028e-05,
+      "loss": 3.4153,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2396315336227417,
+      "learning_rate": 8.979839016415735e-05,
+      "loss": 3.4796,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.390388011932373,
+      "learning_rate": 8.969724954605373e-05,
+      "loss": 3.4476,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3664170503616333,
+      "learning_rate": 8.959566758767581e-05,
+      "loss": 3.4889,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.537876844406128,
+      "learning_rate": 8.949364541837661e-05,
+      "loss": 3.4718,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.397546648979187,
+      "learning_rate": 8.939118417240329e-05,
+      "loss": 3.5307,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.2351093292236328,
+      "learning_rate": 8.92882849888845e-05,
+      "loss": 3.3648,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.4052352905273438,
+      "learning_rate": 8.918494901181773e-05,
+      "loss": 3.436,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.319915771484375,
+      "learning_rate": 8.908117739005659e-05,
+      "loss": 3.4592,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1869890689849854,
+      "learning_rate": 8.897697127729805e-05,
+      "loss": 3.4234,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.2945029735565186,
+      "learning_rate": 8.887233183206957e-05,
+      "loss": 3.3872,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.2278895378112793,
+      "learning_rate": 8.876726021771627e-05,
+      "loss": 3.4017,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1132441759109497,
+      "learning_rate": 8.866175760238798e-05,
+      "loss": 3.4569,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2101727724075317,
+      "learning_rate": 8.855582515902625e-05,
+      "loss": 3.46,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4088022708892822,
+      "learning_rate": 8.844946406535131e-05,
+      "loss": 3.329,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5124653577804565,
+      "learning_rate": 8.834267550384893e-05,
+      "loss": 3.407,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2741353511810303,
+      "learning_rate": 8.823546066175741e-05,
+      "loss": 3.4585,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.128075361251831,
+      "learning_rate": 8.81278207310542e-05,
+      "loss": 3.3034,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.306127667427063,
+      "learning_rate": 8.801975690844278e-05,
+      "loss": 3.4291,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1444599628448486,
+      "learning_rate": 8.791127039533934e-05,
+      "loss": 3.3713,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2846177816390991,
+      "learning_rate": 8.780236239785935e-05,
+      "loss": 3.3423,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2519681453704834,
+      "learning_rate": 8.76930341268042e-05,
+      "loss": 3.3863,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1739153861999512,
+      "learning_rate": 8.758328679764776e-05,
+      "loss": 3.3694,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2181482315063477,
+      "learning_rate": 8.747312163052284e-05,
+      "loss": 3.3413,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2021340131759644,
+      "learning_rate": 8.736253985020761e-05,
+      "loss": 3.382,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2046457529067993,
+      "learning_rate": 8.725154268611203e-05,
+      "loss": 3.3032,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1830154657363892,
+      "learning_rate": 8.714013137226411e-05,
+      "loss": 3.4315,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2441033124923706,
+      "learning_rate": 8.702830714729628e-05,
+      "loss": 3.4461,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1201214790344238,
+      "learning_rate": 8.691607125443153e-05,
+      "loss": 3.3202,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1856852769851685,
+      "learning_rate": 8.680342494146967e-05,
+      "loss": 3.3683,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.256367564201355,
+      "learning_rate": 8.66903694607734e-05,
+      "loss": 3.4501,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2275352478027344,
+      "learning_rate": 8.65769060692544e-05,
+      "loss": 3.3701,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.179276704788208,
+      "learning_rate": 8.646303602835936e-05,
+      "loss": 3.4582,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.1398378610610962,
+      "learning_rate": 8.634876060405597e-05,
+      "loss": 3.3763,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.2729588747024536,
+      "learning_rate": 8.623408106681884e-05,
+      "loss": 3.4178,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.4245160818099976,
+      "learning_rate": 8.611899869161535e-05,
+      "loss": 3.4572,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.11946439743042,
+      "learning_rate": 8.600351475789147e-05,
+      "loss": 3.3231,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.1230659484863281,
+      "learning_rate": 8.588763054955764e-05,
+      "loss": 3.3122,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.2719365358352661,
+      "learning_rate": 8.57713473549743e-05,
+      "loss": 3.3193,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.1483792066574097,
+      "learning_rate": 8.565466646693778e-05,
+      "loss": 3.2684,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.1922017335891724,
+      "learning_rate": 8.553758918266578e-05,
+      "loss": 3.4181,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.228605031967163,
+      "learning_rate": 8.5420116803783e-05,
+      "loss": 3.3389,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.23223876953125,
+      "learning_rate": 8.530225063630668e-05,
+      "loss": 3.4186,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.138245701789856,
+      "learning_rate": 8.518399199063205e-05,
+      "loss": 3.3469,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.2012712955474854,
+      "learning_rate": 8.50653421815178e-05,
+      "loss": 3.3803,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.1761367321014404,
+      "learning_rate": 8.494630252807138e-05,
+      "loss": 3.3373,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.1363654136657715,
+      "learning_rate": 8.482687435373449e-05,
+      "loss": 3.4441,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.247525930404663,
+      "learning_rate": 8.470705898626817e-05,
+      "loss": 3.348,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.1885273456573486,
+      "learning_rate": 8.458685775773822e-05,
+      "loss": 3.3302,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.1339763402938843,
+      "learning_rate": 8.446627200450025e-05,
+      "loss": 3.3123,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0527379512786865,
+      "learning_rate": 8.434530306718493e-05,
+      "loss": 3.3242,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.3958696126937866,
+      "learning_rate": 8.4223952290683e-05,
+      "loss": 3.3615,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.123361587524414,
+      "learning_rate": 8.41022210241304e-05,
+      "loss": 3.3276,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1145395040512085,
+      "learning_rate": 8.398011062089316e-05,
+      "loss": 3.2965,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1637969017028809,
+      "learning_rate": 8.385762243855249e-05,
+      "loss": 3.3045,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1752508878707886,
+      "learning_rate": 8.373475783888958e-05,
+      "loss": 3.2793,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1751797199249268,
+      "learning_rate": 8.36115181878705e-05,
+      "loss": 3.375,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.2501639127731323,
+      "learning_rate": 8.348790485563101e-05,
+      "loss": 3.3518,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4987150430679321,
+      "learning_rate": 8.336391921646134e-05,
+      "loss": 3.2781,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1865673065185547,
+      "learning_rate": 8.323956264879089e-05,
+      "loss": 3.2958,
+      "step": 1365
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1272923946380615,
+      "learning_rate": 8.311483653517294e-05,
+      "loss": 3.2893,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.13765287399292,
+      "learning_rate": 8.298974226226919e-05,
+      "loss": 3.2754,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.2192575931549072,
+      "learning_rate": 8.28642812208345e-05,
+      "loss": 3.3247,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.0930324792861938,
+      "learning_rate": 8.273845480570123e-05,
+      "loss": 3.3496,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.430282711982727,
+      "learning_rate": 8.26122644157639e-05,
+      "loss": 3.343,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1841108798980713,
+      "learning_rate": 8.248571145396362e-05,
+      "loss": 3.3798,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1953662633895874,
+      "learning_rate": 8.235879732727236e-05,
+      "loss": 3.3721,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1492798328399658,
+      "learning_rate": 8.223152344667745e-05,
+      "loss": 3.3421,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1533442735671997,
+      "learning_rate": 8.21038912271658e-05,
+      "loss": 3.236,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3999956846237183,
+      "learning_rate": 8.197590208770824e-05,
+      "loss": 3.2426,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1835312843322754,
+      "learning_rate": 8.184755745124371e-05,
+      "loss": 3.3129,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1276029348373413,
+      "learning_rate": 8.171885874466342e-05,
+      "loss": 3.2657,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.065759539604187,
+      "learning_rate": 8.158980739879507e-05,
+      "loss": 3.335,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.103005290031433,
+      "learning_rate": 8.146040484838677e-05,
+      "loss": 3.2582,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1258820295333862,
+      "learning_rate": 8.133065253209132e-05,
+      "loss": 3.3361,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1314821243286133,
+      "learning_rate": 8.120055189245e-05,
+      "loss": 3.3443,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3966840505599976,
+      "learning_rate": 8.10701043758767e-05,
+      "loss": 3.2131,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3545833826065063,
+      "learning_rate": 8.093931143264174e-05,
+      "loss": 3.279,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.2523497343063354,
+      "learning_rate": 8.080817451685576e-05,
+      "loss": 3.3573,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1441946029663086,
+      "learning_rate": 8.067669508645356e-05,
+      "loss": 3.2457,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1906113624572754,
+      "learning_rate": 8.054487460317797e-05,
+      "loss": 3.3027,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1724116802215576,
+      "learning_rate": 8.041271453256345e-05,
+      "loss": 3.3604,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.189816951751709,
+      "learning_rate": 8.02802163439199e-05,
+      "loss": 3.2289,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4529297351837158,
+      "learning_rate": 8.01473815103163e-05,
+      "loss": 3.2517,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.2230924367904663,
+      "learning_rate": 8.001421150856434e-05,
+      "loss": 3.2214,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1627131700515747,
+      "learning_rate": 7.988070781920197e-05,
+      "loss": 3.2423,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.193337321281433,
+      "learning_rate": 7.9746871926477e-05,
+      "loss": 3.2245,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.08041250705719,
+      "learning_rate": 7.961270531833052e-05,
+      "loss": 3.1718,
+      "step": 1505
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.207674264907837,
+      "learning_rate": 7.947820948638045e-05,
+      "loss": 3.3168,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1085306406021118,
+      "learning_rate": 7.934338592590486e-05,
+      "loss": 3.3181,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.2514351606369019,
+      "learning_rate": 7.92082361358254e-05,
+      "loss": 3.2197,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1564998626708984,
+      "learning_rate": 7.907276161869065e-05,
+      "loss": 3.1698,
+      "step": 1525
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.152014136314392,
+      "learning_rate": 7.893696388065936e-05,
+      "loss": 3.3014,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1435388326644897,
+      "learning_rate": 7.88008444314838e-05,
+      "loss": 3.2903,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.0654579401016235,
+      "learning_rate": 7.866440478449283e-05,
+      "loss": 3.248,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.0817036628723145,
+      "learning_rate": 7.852764645657522e-05,
+      "loss": 3.2552,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1185789108276367,
+      "learning_rate": 7.839057096816271e-05,
+      "loss": 3.3086,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.160009503364563,
+      "learning_rate": 7.82531798432131e-05,
+      "loss": 3.3689,
+      "step": 1555
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.2025865316390991,
+      "learning_rate": 7.811547460919333e-05,
+      "loss": 3.2443,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.1262131929397583,
+      "learning_rate": 7.797745679706254e-05,
+      "loss": 3.2118,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.1751744747161865,
+      "learning_rate": 7.783912794125496e-05,
+      "loss": 3.319,
+      "step": 1570
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2392690181732178,
+      "learning_rate": 7.770048957966291e-05,
+      "loss": 3.2533,
+      "step": 1575
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.1627576351165771,
+      "learning_rate": 7.756154325361967e-05,
+      "loss": 3.2821,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2147516012191772,
+      "learning_rate": 7.74222905078824e-05,
+      "loss": 3.3135,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.495335340499878,
+      "learning_rate": 7.728273289061489e-05,
+      "loss": 3.2368,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2422459125518799,
+      "learning_rate": 7.714287195337044e-05,
+      "loss": 3.1872,
+      "step": 1595
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.0939897298812866,
+      "learning_rate": 7.700270925107448e-05,
+      "loss": 3.2568,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2069770097732544,
+      "learning_rate": 7.686224634200742e-05,
+      "loss": 3.2425,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.1149344444274902,
+      "learning_rate": 7.672148478778722e-05,
+      "loss": 3.2489,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.1385297775268555,
+      "learning_rate": 7.658042615335212e-05,
+      "loss": 3.2202,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.2272182703018188,
+      "learning_rate": 7.643907200694318e-05,
+      "loss": 3.1268,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.0669175386428833,
+      "learning_rate": 7.629742392008684e-05,
+      "loss": 3.201,
+      "step": 1625
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.39219069480896,
+      "learning_rate": 7.615548346757749e-05,
+      "loss": 3.2481,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.3066028356552124,
+      "learning_rate": 7.60132522274599e-05,
+      "loss": 3.2454,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.1821011304855347,
+      "learning_rate": 7.587073178101178e-05,
+      "loss": 3.2107,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.1814302206039429,
+      "learning_rate": 7.572792371272609e-05,
+      "loss": 3.2949,
+      "step": 1645
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.2134374380111694,
+      "learning_rate": 7.55848296102935e-05,
+      "loss": 3.2682,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.2237021923065186,
+      "learning_rate": 7.544145106458465e-05,
+      "loss": 3.2396,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1517393589019775,
+      "learning_rate": 7.529778966963259e-05,
+      "loss": 3.1926,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.0898007154464722,
+      "learning_rate": 7.515384702261496e-05,
+      "loss": 3.1994,
+      "step": 1665
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1451698541641235,
+      "learning_rate": 7.500962472383627e-05,
+      "loss": 3.1632,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.080106258392334,
+      "learning_rate": 7.486512437671011e-05,
+      "loss": 3.1445,
+      "step": 1675
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1294046640396118,
+      "learning_rate": 7.472034758774128e-05,
+      "loss": 3.1905,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.0651402473449707,
+      "learning_rate": 7.457529596650797e-05,
+      "loss": 3.2374,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3194466829299927,
+      "learning_rate": 7.442997112564392e-05,
+      "loss": 3.2435,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1329237222671509,
+      "learning_rate": 7.428437468082037e-05,
+      "loss": 3.3095,
+      "step": 1695
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1752973794937134,
+      "learning_rate": 7.413850825072817e-05,
+      "loss": 3.1358,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.0169298648834229,
+      "learning_rate": 7.39923734570598e-05,
+      "loss": 3.2223,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0975102186203003,
+      "learning_rate": 7.384597192449126e-05,
+      "loss": 3.2105,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2045339345932007,
+      "learning_rate": 7.369930528066412e-05,
+      "loss": 3.2965,
+      "step": 1715
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0657718181610107,
+      "learning_rate": 7.355237515616732e-05,
+      "loss": 3.1994,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.1542158126831055,
+      "learning_rate": 7.340518318451914e-05,
+      "loss": 3.2854,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2519826889038086,
+      "learning_rate": 7.325773100214893e-05,
+      "loss": 3.2101,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.1538292169570923,
+      "learning_rate": 7.311002024837899e-05,
+      "loss": 3.1743,
+      "step": 1735
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.3539190292358398,
+      "learning_rate": 7.296205256540633e-05,
+      "loss": 3.3418,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2613801956176758,
+      "learning_rate": 7.281382959828443e-05,
+      "loss": 3.2483,
+      "step": 1745
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0935710668563843,
+      "learning_rate": 7.26653529949049e-05,
+      "loss": 3.2259,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0635665655136108,
+      "learning_rate": 7.25166244059792e-05,
+      "loss": 3.1394,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1002237796783447,
+      "learning_rate": 7.236764548502029e-05,
+      "loss": 3.2551,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.0740032196044922,
+      "learning_rate": 7.221841788832421e-05,
+      "loss": 3.2319,
+      "step": 1765
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1868687868118286,
+      "learning_rate": 7.206894327495173e-05,
+      "loss": 3.1983,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1227601766586304,
+      "learning_rate": 7.191922330670982e-05,
+      "loss": 3.2175,
+      "step": 1775
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1705244779586792,
+      "learning_rate": 7.176925964813326e-05,
+      "loss": 3.1498,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4268603324890137,
+      "learning_rate": 7.161905396646607e-05,
+      "loss": 3.2766,
+      "step": 1785
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1663274765014648,
+      "learning_rate": 7.146860793164299e-05,
+      "loss": 3.2363,
+      "step": 1790
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.143417477607727,
+      "learning_rate": 7.131792321627098e-05,
+      "loss": 3.2707,
+      "step": 1795
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.0310893058776855,
+      "learning_rate": 7.116700149561048e-05,
+      "loss": 3.1236,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.0529377460479736,
+      "learning_rate": 7.101584444755696e-05,
+      "loss": 3.219,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1156840324401855,
+      "learning_rate": 7.086445375262212e-05,
+      "loss": 3.1371,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1329755783081055,
+      "learning_rate": 7.071283109391528e-05,
+      "loss": 3.2273,
+      "step": 1815
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.060715913772583,
+      "learning_rate": 7.056097815712466e-05,
+      "loss": 3.1384,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9990625977516174,
+      "learning_rate": 7.040889663049862e-05,
+      "loss": 3.1813,
+      "step": 1825
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1197738647460938,
+      "learning_rate": 7.025658820482693e-05,
+      "loss": 3.1422,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.136229157447815,
+      "learning_rate": 7.010405457342192e-05,
+      "loss": 3.2805,
+      "step": 1835
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1726741790771484,
+      "learning_rate": 6.995129743209967e-05,
+      "loss": 3.1917,
+      "step": 1840
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1462650299072266,
+      "learning_rate": 6.97983184791612e-05,
+      "loss": 3.2554,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.136553406715393,
+      "learning_rate": 6.964511941537355e-05,
+      "loss": 3.2403,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0682226419448853,
+      "learning_rate": 6.949170194395083e-05,
+      "loss": 3.1274,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0860931873321533,
+      "learning_rate": 6.933806777053536e-05,
+      "loss": 3.2707,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0465028285980225,
+      "learning_rate": 6.918421860317872e-05,
+      "loss": 3.1675,
+      "step": 1865
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.100434422492981,
+      "learning_rate": 6.903015615232263e-05,
+      "loss": 3.2351,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0993447303771973,
+      "learning_rate": 6.887588213078012e-05,
+      "loss": 3.2404,
+      "step": 1875
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0658077001571655,
+      "learning_rate": 6.87213982537163e-05,
+      "loss": 3.2169,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0856738090515137,
+      "learning_rate": 6.856670623862943e-05,
+      "loss": 3.1248,
+      "step": 1885
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.1014505624771118,
+      "learning_rate": 6.841180780533179e-05,
+      "loss": 3.1161,
+      "step": 1890
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0805455446243286,
+      "learning_rate": 6.82567046759305e-05,
+      "loss": 3.1812,
+      "step": 1895
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.082909345626831,
+      "learning_rate": 6.810139857480844e-05,
+      "loss": 3.1427,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.1378077268600464,
+      "learning_rate": 6.794589122860509e-05,
+      "loss": 3.0901,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0580015182495117,
+      "learning_rate": 6.779018436619725e-05,
+      "loss": 3.2417,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.1437370777130127,
+      "learning_rate": 6.763427971867992e-05,
+      "loss": 3.1671,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0612616539001465,
+      "learning_rate": 6.747817901934699e-05,
+      "loss": 3.1191,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.3342070579528809,
+      "learning_rate": 6.732188400367197e-05,
+      "loss": 3.2476,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.2217915058135986,
+      "learning_rate": 6.716539640928871e-05,
+      "loss": 3.2108,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.1861838102340698,
+      "learning_rate": 6.70087179759721e-05,
+      "loss": 3.1584,
+      "step": 1935
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0744543075561523,
+      "learning_rate": 6.685185044561874e-05,
+      "loss": 3.2015,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.2175796031951904,
+      "learning_rate": 6.669479556222747e-05,
+      "loss": 3.1226,
+      "step": 1945
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.1044460535049438,
+      "learning_rate": 6.653755507188013e-05,
+      "loss": 3.1128,
+      "step": 1950
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.1110048294067383,
+      "learning_rate": 6.638013072272205e-05,
+      "loss": 3.1404,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.1088255643844604,
+      "learning_rate": 6.622252426494259e-05,
+      "loss": 3.1565,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0379773378372192,
+      "learning_rate": 6.606473745075581e-05,
+      "loss": 3.1129,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0543949604034424,
+      "learning_rate": 6.590677203438084e-05,
+      "loss": 3.2115,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0617347955703735,
+      "learning_rate": 6.574862977202252e-05,
+      "loss": 3.1027,
+      "step": 1975
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.2704519033432007,
+      "learning_rate": 6.559031242185174e-05,
+      "loss": 3.1921,
+      "step": 1980
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.120644450187683,
+      "learning_rate": 6.543182174398597e-05,
+      "loss": 3.212,
+      "step": 1985
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.186946988105774,
+      "learning_rate": 6.52731595004697e-05,
+      "loss": 3.164,
+      "step": 1990
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.1211094856262207,
+      "learning_rate": 6.51143274552548e-05,
+      "loss": 3.1306,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0689617395401,
+      "learning_rate": 6.495532737418098e-05,
+      "loss": 3.1182,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0167192220687866,
+      "learning_rate": 6.479616102495605e-05,
+      "loss": 3.2118,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0616776943206787,
+      "learning_rate": 6.463683017713638e-05,
+      "loss": 3.1227,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.060515284538269,
+      "learning_rate": 6.447733660210715e-05,
+      "loss": 3.1194,
+      "step": 2015
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0889480113983154,
+      "learning_rate": 6.431768207306272e-05,
+      "loss": 3.0782,
+      "step": 2020
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0803637504577637,
+      "learning_rate": 6.415786836498684e-05,
+      "loss": 3.1184,
+      "step": 2025
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.064009428024292,
+      "learning_rate": 6.399789725463298e-05,
+      "loss": 3.1652,
+      "step": 2030
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0829176902770996,
+      "learning_rate": 6.383777052050458e-05,
+      "loss": 3.1046,
+      "step": 2035
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0532939434051514,
+      "learning_rate": 6.367748994283518e-05,
+      "loss": 3.0792,
+      "step": 2040
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.196413516998291,
+      "learning_rate": 6.351705730356877e-05,
+      "loss": 3.1204,
+      "step": 2045
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.1028838157653809,
+      "learning_rate": 6.335647438633987e-05,
+      "loss": 3.1243,
+      "step": 2050
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.0548689365386963,
+      "learning_rate": 6.319574297645374e-05,
+      "loss": 3.1802,
+      "step": 2055
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.1255388259887695,
+      "learning_rate": 6.303486486086654e-05,
+      "loss": 3.1596,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.1859532594680786,
+      "learning_rate": 6.287384182816546e-05,
+      "loss": 3.1474,
+      "step": 2065
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.064068078994751,
+      "learning_rate": 6.271267566854883e-05,
+      "loss": 3.1942,
+      "step": 2070
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.0523204803466797,
+      "learning_rate": 6.255136817380618e-05,
+      "loss": 3.1562,
+      "step": 2075
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.1364482641220093,
+      "learning_rate": 6.23899211372984e-05,
+      "loss": 3.0299,
+      "step": 2080
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.1925618648529053,
+      "learning_rate": 6.222833635393772e-05,
+      "loss": 3.0583,
+      "step": 2085
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.0866798162460327,
+      "learning_rate": 6.206661562016782e-05,
+      "loss": 3.1356,
+      "step": 2090
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0916551351547241,
+      "learning_rate": 6.190476073394382e-05,
+      "loss": 3.045,
+      "step": 2095
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0586055517196655,
+      "learning_rate": 6.17427734947123e-05,
+      "loss": 3.1927,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0540417432785034,
+      "learning_rate": 6.158065570339127e-05,
+      "loss": 3.1094,
+      "step": 2105
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0485178232192993,
+      "learning_rate": 6.141840916235021e-05,
+      "loss": 3.1503,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.1303540468215942,
+      "learning_rate": 6.125603567539001e-05,
+      "loss": 3.1794,
+      "step": 2115
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0090653896331787,
+      "learning_rate": 6.109353704772284e-05,
+      "loss": 3.1685,
+      "step": 2120
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0770632028579712,
+      "learning_rate": 6.0930915085952164e-05,
+      "loss": 3.0851,
+      "step": 2125
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0763593912124634,
+      "learning_rate": 6.076817159805267e-05,
+      "loss": 3.0592,
+      "step": 2130
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.077744722366333,
+      "learning_rate": 6.06053083933501e-05,
+      "loss": 3.0575,
+      "step": 2135
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0479717254638672,
+      "learning_rate": 6.044232728250116e-05,
+      "loss": 3.2182,
+      "step": 2140
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.1060945987701416,
+      "learning_rate": 6.027923007747339e-05,
+      "loss": 3.1065,
+      "step": 2145
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.0798442363739014,
+      "learning_rate": 6.011601859152506e-05,
+      "loss": 2.9653,
+      "step": 2150
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.166008472442627,
+      "learning_rate": 5.995269463918495e-05,
+      "loss": 3.0984,
+      "step": 2155
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.1726555824279785,
+      "learning_rate": 5.97892600362322e-05,
+      "loss": 3.086,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.0481421947479248,
+      "learning_rate": 5.962571659967614e-05,
+      "loss": 3.1516,
+      "step": 2165
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.9961311221122742,
+      "learning_rate": 5.946206614773606e-05,
+      "loss": 3.1116,
+      "step": 2170
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.0177178382873535,
+      "learning_rate": 5.929831049982103e-05,
+      "loss": 3.1132,
+      "step": 2175
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.084097981452942,
+      "learning_rate": 5.9134451476509633e-05,
+      "loss": 3.1533,
+      "step": 2180
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.1300958395004272,
+      "learning_rate": 5.897049089952974e-05,
+      "loss": 3.069,
+      "step": 2185
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.1346057653427124,
+      "learning_rate": 5.880643059173826e-05,
+      "loss": 3.1014,
+      "step": 2190
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0111093521118164,
+      "learning_rate": 5.864227237710093e-05,
+      "loss": 3.1847,
+      "step": 2195
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.1515612602233887,
+      "learning_rate": 5.847801808067189e-05,
+      "loss": 3.1194,
+      "step": 2200
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0243968963623047,
+      "learning_rate": 5.831366952857357e-05,
+      "loss": 3.0739,
+      "step": 2205
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0244972705841064,
+      "learning_rate": 5.814922854797622e-05,
+      "loss": 3.0589,
+      "step": 2210
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.1009114980697632,
+      "learning_rate": 5.798469696707775e-05,
+      "loss": 3.0715,
+      "step": 2215
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0368943214416504,
+      "learning_rate": 5.782007661508331e-05,
+      "loss": 3.0485,
+      "step": 2220
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.1286373138427734,
+      "learning_rate": 5.765536932218495e-05,
+      "loss": 3.1076,
+      "step": 2225
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0644299983978271,
+      "learning_rate": 5.7490576919541315e-05,
+      "loss": 3.0415,
+      "step": 2230
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.1599712371826172,
+      "learning_rate": 5.732570123925729e-05,
+      "loss": 3.0608,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2678030729293823,
+      "learning_rate": 5.7160744114363593e-05,
+      "loss": 3.1866,
+      "step": 2240
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.0837265253067017,
+      "learning_rate": 5.699570737879641e-05,
+      "loss": 3.0818,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.1460355520248413,
+      "learning_rate": 5.683059286737702e-05,
+      "loss": 3.0558,
+      "step": 2250
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.1030579805374146,
+      "learning_rate": 5.666540241579139e-05,
+      "loss": 3.15,
+      "step": 2255
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2594730854034424,
+      "learning_rate": 5.6500137860569766e-05,
+      "loss": 3.1769,
+      "step": 2260
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2388299703598022,
+      "learning_rate": 5.633480103906624e-05,
+      "loss": 3.0187,
+      "step": 2265
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.030984878540039,
+      "learning_rate": 5.616939378943834e-05,
+      "loss": 3.0997,
+      "step": 2270
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.0577046871185303,
+      "learning_rate": 5.6003917950626595e-05,
+      "loss": 3.1403,
+      "step": 2275
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.1050175428390503,
+      "learning_rate": 5.583837536233407e-05,
+      "loss": 3.0793,
+      "step": 2280
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.063862681388855,
+      "learning_rate": 5.567276786500596e-05,
+      "loss": 3.0294,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.0375818014144897,
+      "learning_rate": 5.5507097299809054e-05,
+      "loss": 3.1218,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9979878664016724,
+      "learning_rate": 5.534136550861133e-05,
+      "loss": 3.1139,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.0724773406982422,
+      "learning_rate": 5.5175574333961465e-05,
+      "loss": 3.0563,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.01706063747406,
+      "learning_rate": 5.500972561906832e-05,
+      "loss": 3.1836,
+      "step": 2305
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.0690789222717285,
+      "learning_rate": 5.484382120778048e-05,
+      "loss": 3.1078,
+      "step": 2310
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.1447081565856934,
+      "learning_rate": 5.467786294456575e-05,
+      "loss": 3.0429,
+      "step": 2315
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.0642063617706299,
+      "learning_rate": 5.451185267449061e-05,
+      "loss": 3.0135,
+      "step": 2320
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.0696765184402466,
+      "learning_rate": 5.43457922431998e-05,
+      "loss": 3.1465,
+      "step": 2325
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.0368304252624512,
+      "learning_rate": 5.417968349689566e-05,
+      "loss": 2.9582,
+      "step": 2330
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.098090410232544,
+      "learning_rate": 5.401352828231772e-05,
+      "loss": 3.0758,
+      "step": 2335
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.0280003547668457,
+      "learning_rate": 5.384732844672211e-05,
+      "loss": 3.0935,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.0697113275527954,
+      "learning_rate": 5.368108583786107e-05,
+      "loss": 3.0966,
+      "step": 2345
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.1267234086990356,
+      "learning_rate": 5.3514802303962344e-05,
+      "loss": 3.0999,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.1041789054870605,
+      "learning_rate": 5.334847969370868e-05,
+      "loss": 3.0606,
+      "step": 2355
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.1038250923156738,
+      "learning_rate": 5.3182119856217284e-05,
+      "loss": 3.1977,
+      "step": 2360
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.0840222835540771,
+      "learning_rate": 5.3015724641019214e-05,
+      "loss": 3.0736,
+      "step": 2365
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.0754737854003906,
+      "learning_rate": 5.284929589803884e-05,
+      "loss": 3.115,
+      "step": 2370
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.0331034660339355,
+      "learning_rate": 5.2682835477573336e-05,
+      "loss": 3.0537,
+      "step": 2375
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.9996258616447449,
+      "learning_rate": 5.2516345230271965e-05,
+      "loss": 3.0924,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.054716944694519,
+      "learning_rate": 5.234982700711569e-05,
+      "loss": 3.0757,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0223090648651123,
+      "learning_rate": 5.218328265939643e-05,
+      "loss": 3.066,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.2190570831298828,
+      "learning_rate": 5.201671403869657e-05,
+      "loss": 3.0556,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1981064081192017,
+      "learning_rate": 5.1850122996868366e-05,
+      "loss": 3.087,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1506836414337158,
+      "learning_rate": 5.168351138601334e-05,
+      "loss": 3.0792,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1408671140670776,
+      "learning_rate": 5.1516881058461675e-05,
+      "loss": 3.151,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0261212587356567,
+      "learning_rate": 5.135023386675166e-05,
+      "loss": 3.0216,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0941778421401978,
+      "learning_rate": 5.118357166360906e-05,
+      "loss": 3.0108,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1151127815246582,
+      "learning_rate": 5.101689630192655e-05,
+      "loss": 3.13,
+      "step": 2425
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0719926357269287,
+      "learning_rate": 5.085020963474307e-05,
+      "loss": 2.9846,
+      "step": 2430
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.1672226190567017,
+      "learning_rate": 5.068351351522329e-05,
+      "loss": 3.1311,
+      "step": 2435
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.039842128753662,
+      "learning_rate": 5.0516809796636935e-05,
+      "loss": 3.1064,
+      "step": 2440
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.050147533416748,
+      "learning_rate": 5.035010033233821e-05,
+      "loss": 3.0082,
+      "step": 2445
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.9957916736602783,
+      "learning_rate": 5.018338697574523e-05,
+      "loss": 3.0923,
+      "step": 2450
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.055060625076294,
+      "learning_rate": 5.0016671580319354e-05,
+      "loss": 2.9972,
+      "step": 2455
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0098164081573486,
+      "learning_rate": 4.984995599954461e-05,
+      "loss": 3.1395,
+      "step": 2460
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0814324617385864,
+      "learning_rate": 4.968324208690712e-05,
+      "loss": 3.0554,
+      "step": 2465
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0801881551742554,
+      "learning_rate": 4.951653169587441e-05,
+      "loss": 3.0265,
+      "step": 2470
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0839911699295044,
+      "learning_rate": 4.93498266798749e-05,
+      "loss": 3.1256,
+      "step": 2475
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0653108358383179,
+      "learning_rate": 4.918312889227722e-05,
+      "loss": 3.1095,
+      "step": 2480
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0367735624313354,
+      "learning_rate": 4.901644018636966e-05,
+      "loss": 3.0546,
+      "step": 2485
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0559877157211304,
+      "learning_rate": 4.8849762415339526e-05,
+      "loss": 2.9952,
+      "step": 2490
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0303627252578735,
+      "learning_rate": 4.868309743225256e-05,
+      "loss": 3.0636,
+      "step": 2495
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.1090822219848633,
+      "learning_rate": 4.851644709003233e-05,
+      "loss": 3.0876,
+      "step": 2500
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0675830841064453,
+      "learning_rate": 4.834981324143964e-05,
+      "loss": 3.0573,
+      "step": 2505
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0036590099334717,
+      "learning_rate": 4.818319773905191e-05,
+      "loss": 3.0297,
+      "step": 2510
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.041754126548767,
+      "learning_rate": 4.801660243524261e-05,
+      "loss": 2.9507,
+      "step": 2515
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0789331197738647,
+      "learning_rate": 4.7850029182160626e-05,
+      "loss": 3.0633,
+      "step": 2520
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.015762209892273,
+      "learning_rate": 4.768347983170973e-05,
+      "loss": 3.0388,
+      "step": 2525
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0645495653152466,
+      "learning_rate": 4.7516956235527884e-05,
+      "loss": 3.1631,
+      "step": 2530
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0723949670791626,
+      "learning_rate": 4.735046024496682e-05,
+      "loss": 2.9642,
+      "step": 2535
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.036794900894165,
+      "learning_rate": 4.7183993711071286e-05,
+      "loss": 3.0467,
+      "step": 2540
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.1749889850616455,
+      "learning_rate": 4.7017558484558554e-05,
+      "loss": 3.0597,
+      "step": 2545
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0650322437286377,
+      "learning_rate": 4.6851156415797844e-05,
+      "loss": 3.0598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0720289945602417,
+      "learning_rate": 4.6684789354789746e-05,
+      "loss": 2.9977,
+      "step": 2555
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0925712585449219,
+      "learning_rate": 4.651845915114563e-05,
+      "loss": 3.1103,
+      "step": 2560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.063539981842041,
+      "learning_rate": 4.6352167654067095e-05,
+      "loss": 3.0225,
+      "step": 2565
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0431323051452637,
+      "learning_rate": 4.618591671232544e-05,
+      "loss": 3.0463,
+      "step": 2570
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.0726792812347412,
+      "learning_rate": 4.601970817424106e-05,
+      "loss": 3.0796,
+      "step": 2575
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.1471610069274902,
+      "learning_rate": 4.585354388766292e-05,
+      "loss": 3.0079,
+      "step": 2580
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.1070219278335571,
+      "learning_rate": 4.568742569994802e-05,
+      "loss": 3.1182,
+      "step": 2585
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.158220648765564,
+      "learning_rate": 4.552135545794086e-05,
+      "loss": 3.0372,
+      "step": 2590
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.0492193698883057,
+      "learning_rate": 4.535533500795288e-05,
+      "loss": 3.0207,
+      "step": 2595
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.2903916835784912,
+      "learning_rate": 4.5189366195741953e-05,
+      "loss": 3.0064,
+      "step": 2600
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.0783137083053589,
+      "learning_rate": 4.502345086649186e-05,
+      "loss": 3.0026,
+      "step": 2605
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.0791529417037964,
+      "learning_rate": 4.485759086479179e-05,
+      "loss": 3.033,
+      "step": 2610
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.0538674592971802,
+      "learning_rate": 4.469178803461579e-05,
+      "loss": 3.0676,
+      "step": 2615
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.1873081922531128,
+      "learning_rate": 4.4526044219302326e-05,
+      "loss": 2.9884,
+      "step": 2620
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.1096831560134888,
+      "learning_rate": 4.4360361261533745e-05,
+      "loss": 3.0778,
+      "step": 2625
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0487239360809326,
+      "learning_rate": 4.419474100331579e-05,
+      "loss": 3.0849,
+      "step": 2630
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0631368160247803,
+      "learning_rate": 4.402918528595715e-05,
+      "loss": 3.0974,
+      "step": 2635
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0593897104263306,
+      "learning_rate": 4.386369595004896e-05,
+      "loss": 3.0487,
+      "step": 2640
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9948830604553223,
+      "learning_rate": 4.3698274835444354e-05,
+      "loss": 3.0793,
+      "step": 2645
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.033443570137024,
+      "learning_rate": 4.3532923781238e-05,
+      "loss": 3.101,
+      "step": 2650
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.073686957359314,
+      "learning_rate": 4.336764462574566e-05,
+      "loss": 3.082,
+      "step": 2655
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.032832145690918,
+      "learning_rate": 4.320243920648376e-05,
+      "loss": 3.006,
+      "step": 2660
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.1529659032821655,
+      "learning_rate": 4.303730936014894e-05,
+      "loss": 2.9774,
+      "step": 2665
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0000460147857666,
+      "learning_rate": 4.287225692259765e-05,
+      "loss": 2.9237,
+      "step": 2670
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.9973639249801636,
+      "learning_rate": 4.270728372882575e-05,
+      "loss": 3.0115,
+      "step": 2675
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.05452299118042,
+      "learning_rate": 4.254239161294804e-05,
+      "loss": 3.0474,
+      "step": 2680
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0591851472854614,
+      "learning_rate": 4.237758240817802e-05,
+      "loss": 3.0432,
+      "step": 2685
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0358829498291016,
+      "learning_rate": 4.2212857946807336e-05,
+      "loss": 3.0401,
+      "step": 2690
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.175520658493042,
+      "learning_rate": 4.2048220060185516e-05,
+      "loss": 3.0245,
+      "step": 2695
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0333977937698364,
+      "learning_rate": 4.188367057869957e-05,
+      "loss": 2.8782,
+      "step": 2700
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.07869553565979,
+      "learning_rate": 4.171921133175365e-05,
+      "loss": 3.0724,
+      "step": 2705
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0302010774612427,
+      "learning_rate": 4.155484414774872e-05,
+      "loss": 2.9962,
+      "step": 2710
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0263460874557495,
+      "learning_rate": 4.139057085406221e-05,
+      "loss": 3.0747,
+      "step": 2715
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.1280380487442017,
+      "learning_rate": 4.1226393277027726e-05,
+      "loss": 3.0475,
+      "step": 2720
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0348632335662842,
+      "learning_rate": 4.106231324191471e-05,
+      "loss": 3.0591,
+      "step": 2725
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0165739059448242,
+      "learning_rate": 4.089833257290817e-05,
+      "loss": 3.1233,
+      "step": 2730
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0293173789978027,
+      "learning_rate": 4.073445309308842e-05,
+      "loss": 3.0282,
+      "step": 2735
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0683913230895996,
+      "learning_rate": 4.0570676624410756e-05,
+      "loss": 3.0318,
+      "step": 2740
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0587563514709473,
+      "learning_rate": 4.040700498768525e-05,
+      "loss": 3.0647,
+      "step": 2745
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0239886045455933,
+      "learning_rate": 4.024344000255648e-05,
+      "loss": 2.9992,
+      "step": 2750
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0538274049758911,
+      "learning_rate": 4.0079983487483313e-05,
+      "loss": 3.0528,
+      "step": 2755
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0498000383377075,
+      "learning_rate": 3.9916637259718683e-05,
+      "loss": 3.105,
+      "step": 2760
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.211574912071228,
+      "learning_rate": 3.9753403135289396e-05,
+      "loss": 2.9811,
+      "step": 2765
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.0358808040618896,
+      "learning_rate": 3.9590282928975914e-05,
+      "loss": 2.9964,
+      "step": 2770
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.0570118427276611,
+      "learning_rate": 3.942727845429221e-05,
+      "loss": 2.906,
+      "step": 2775
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.2332627773284912,
+      "learning_rate": 3.926439152346558e-05,
+      "loss": 3.022,
+      "step": 2780
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.1530132293701172,
+      "learning_rate": 3.910162394741653e-05,
+      "loss": 2.9988,
+      "step": 2785
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.0132982730865479,
+      "learning_rate": 3.893897753573861e-05,
+      "loss": 2.9654,
+      "step": 2790
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.0742069482803345,
+      "learning_rate": 3.877645409667829e-05,
+      "loss": 3.0135,
+      "step": 2795
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.0182029008865356,
+      "learning_rate": 3.861405543711491e-05,
+      "loss": 2.9888,
+      "step": 2800
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.0430015325546265,
+      "learning_rate": 3.8451783362540507e-05,
+      "loss": 3.0477,
+      "step": 2805
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.1208815574645996,
+      "learning_rate": 3.828963967703983e-05,
+      "loss": 3.0401,
+      "step": 2810
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.1161303520202637,
+      "learning_rate": 3.8127626183270223e-05,
+      "loss": 2.9986,
+      "step": 2815
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.0346485376358032,
+      "learning_rate": 3.796574468244161e-05,
+      "loss": 3.0071,
+      "step": 2820
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.0584827661514282,
+      "learning_rate": 3.7803996974296444e-05,
+      "loss": 3.0357,
+      "step": 2825
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.0474509000778198,
+      "learning_rate": 3.7642384857089776e-05,
+      "loss": 2.9715,
+      "step": 2830
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.0421720743179321,
+      "learning_rate": 3.748091012756915e-05,
+      "loss": 2.9982,
+      "step": 2835
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.0171830654144287,
+      "learning_rate": 3.731957458095467e-05,
+      "loss": 3.0238,
+      "step": 2840
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.1705058813095093,
+      "learning_rate": 3.71583800109191e-05,
+      "loss": 2.9714,
+      "step": 2845
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.02668297290802,
+      "learning_rate": 3.699732820956784e-05,
+      "loss": 3.0152,
+      "step": 2850
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.1950023174285889,
+      "learning_rate": 3.6836420967419057e-05,
+      "loss": 2.958,
+      "step": 2855
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.0784318447113037,
+      "learning_rate": 3.6675660073383745e-05,
+      "loss": 2.9889,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0552057027816772,
+      "learning_rate": 3.6515047314745856e-05,
+      "loss": 2.9629,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.1507591009140015,
+      "learning_rate": 3.6354584477142437e-05,
+      "loss": 3.0797,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0158036947250366,
+      "learning_rate": 3.6194273344543736e-05,
+      "loss": 3.0202,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9868055582046509,
+      "learning_rate": 3.6034115699233425e-05,
+      "loss": 2.9834,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0302212238311768,
+      "learning_rate": 3.5874113321788736e-05,
+      "loss": 3.047,
+      "step": 2885
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0958445072174072,
+      "learning_rate": 3.571426799106071e-05,
+      "loss": 2.9674,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9823729395866394,
+      "learning_rate": 3.555458148415437e-05,
+      "loss": 3.0503,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0195577144622803,
+      "learning_rate": 3.539505557640901e-05,
+      "loss": 2.9481,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0259385108947754,
+      "learning_rate": 3.523569204137843e-05,
+      "loss": 2.9699,
+      "step": 2905
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.059617519378662,
+      "learning_rate": 3.5076492650811246e-05,
+      "loss": 2.9348,
+      "step": 2910
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.0996873378753662,
+      "learning_rate": 3.491745917463113e-05,
+      "loss": 3.0041,
+      "step": 2915
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.0191603899002075,
+      "learning_rate": 3.475859338091721e-05,
+      "loss": 2.9241,
+      "step": 2920
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.084431529045105,
+      "learning_rate": 3.4599897035884374e-05,
+      "loss": 3.1061,
+      "step": 2925
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.0117292404174805,
+      "learning_rate": 3.444137190386363e-05,
+      "loss": 3.0229,
+      "step": 2930
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.034151315689087,
+      "learning_rate": 3.4283019747282514e-05,
+      "loss": 2.9843,
+      "step": 2935
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.1034634113311768,
+      "learning_rate": 3.412484232664545e-05,
+      "loss": 3.0178,
+      "step": 2940
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.256437063217163,
+      "learning_rate": 3.396684140051424e-05,
+      "loss": 3.0308,
+      "step": 2945
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.1831390857696533,
+      "learning_rate": 3.3809018725488466e-05,
+      "loss": 3.0132,
+      "step": 2950
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.0726600885391235,
+      "learning_rate": 3.365137605618598e-05,
+      "loss": 2.9744,
+      "step": 2955
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0197064876556396,
+      "learning_rate": 3.3493915145223395e-05,
+      "loss": 2.9329,
+      "step": 2960
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0443058013916016,
+      "learning_rate": 3.3336637743196584e-05,
+      "loss": 3.0242,
+      "step": 2965
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0014925003051758,
+      "learning_rate": 3.317954559866126e-05,
+      "loss": 2.9349,
+      "step": 2970
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0623462200164795,
+      "learning_rate": 3.302264045811344e-05,
+      "loss": 2.9755,
+      "step": 2975
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0433576107025146,
+      "learning_rate": 3.286592406597021e-05,
+      "loss": 2.9424,
+      "step": 2980
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0413775444030762,
+      "learning_rate": 3.270939816455012e-05,
+      "loss": 2.988,
+      "step": 2985
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0432535409927368,
+      "learning_rate": 3.255306449405395e-05,
+      "loss": 2.9819,
+      "step": 2990
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.1115838289260864,
+      "learning_rate": 3.2396924792545304e-05,
+      "loss": 2.9621,
+      "step": 2995
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0652596950531006,
+      "learning_rate": 3.224098079593132e-05,
+      "loss": 2.9722,
+      "step": 3000
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0429493188858032,
+      "learning_rate": 3.2085234237943354e-05,
+      "loss": 2.9245,
+      "step": 3005
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.1402490139007568,
+      "learning_rate": 3.19296868501177e-05,
+      "loss": 2.9569,
+      "step": 3010
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.0204291343688965,
+      "learning_rate": 3.177434036177636e-05,
+      "loss": 3.0853,
+      "step": 3015
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.1746922731399536,
+      "learning_rate": 3.1619196500007804e-05,
+      "loss": 2.9717,
+      "step": 3020
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.0246679782867432,
+      "learning_rate": 3.146425698964776e-05,
+      "loss": 2.994,
+      "step": 3025
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.0718449354171753,
+      "learning_rate": 3.1309523553260046e-05,
+      "loss": 3.082,
+      "step": 3030
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.0518536567687988,
+      "learning_rate": 3.115499791111743e-05,
+      "loss": 3.0246,
+      "step": 3035
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.0678455829620361,
+      "learning_rate": 3.10006817811825e-05,
+      "loss": 2.8946,
+      "step": 3040
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.118604302406311,
+      "learning_rate": 3.084657687908855e-05,
+      "loss": 3.0081,
+      "step": 3045
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.168013334274292,
+      "learning_rate": 3.069268491812052e-05,
+      "loss": 2.9994,
+      "step": 3050
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.0308539867401123,
+      "learning_rate": 3.0539007609195934e-05,
+      "loss": 3.0219,
+      "step": 3055
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0488004684448242,
+      "learning_rate": 3.0385546660845908e-05,
+      "loss": 2.9881,
+      "step": 3060
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0688416957855225,
+      "learning_rate": 3.0232303779196132e-05,
+      "loss": 3.0163,
+      "step": 3065
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0394408702850342,
+      "learning_rate": 3.0079280667947885e-05,
+      "loss": 2.986,
+      "step": 3070
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0241698026657104,
+      "learning_rate": 2.9926479028359132e-05,
+      "loss": 2.906,
+      "step": 3075
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0117335319519043,
+      "learning_rate": 2.97739005592256e-05,
+      "loss": 3.0282,
+      "step": 3080
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.05054771900177,
+      "learning_rate": 2.962154695686187e-05,
+      "loss": 2.9591,
+      "step": 3085
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0562255382537842,
+      "learning_rate": 2.9469419915082536e-05,
+      "loss": 3.0694,
+      "step": 3090
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.039699673652649,
+      "learning_rate": 2.9317521125183368e-05,
+      "loss": 2.9505,
+      "step": 3095
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.013299822807312,
+      "learning_rate": 2.9165852275922524e-05,
+      "loss": 2.946,
+      "step": 3100
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0763413906097412,
+      "learning_rate": 2.901441505350174e-05,
+      "loss": 2.9782,
+      "step": 3105
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0640761852264404,
+      "learning_rate": 2.886321114154762e-05,
+      "loss": 2.9114,
+      "step": 3110
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0359145402908325,
+      "learning_rate": 2.87122422210929e-05,
+      "loss": 2.9279,
+      "step": 3115
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0430670976638794,
+      "learning_rate": 2.8561509970557736e-05,
+      "loss": 2.9701,
+      "step": 3120
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0799660682678223,
+      "learning_rate": 2.8411016065731146e-05,
+      "loss": 2.989,
+      "step": 3125
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0684157609939575,
+      "learning_rate": 2.826076217975222e-05,
+      "loss": 2.9469,
+      "step": 3130
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.1316529512405396,
+      "learning_rate": 2.8110749983091632e-05,
+      "loss": 2.9968,
+      "step": 3135
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9836276173591614,
+      "learning_rate": 2.7960981143533053e-05,
+      "loss": 3.0318,
+      "step": 3140
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0165281295776367,
+      "learning_rate": 2.781145732615457e-05,
+      "loss": 2.9819,
+      "step": 3145
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0563925504684448,
+      "learning_rate": 2.7662180193310218e-05,
+      "loss": 2.9495,
+      "step": 3150
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.9946759343147278,
+      "learning_rate": 2.751315140461145e-05,
+      "loss": 2.9372,
+      "step": 3155
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.0857210159301758,
+      "learning_rate": 2.7364372616908744e-05,
+      "loss": 2.9728,
+      "step": 3160
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.0404404401779175,
+      "learning_rate": 2.7215845484273152e-05,
+      "loss": 2.9315,
+      "step": 3165
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.0203680992126465,
+      "learning_rate": 2.7067571657977893e-05,
+      "loss": 2.9986,
+      "step": 3170
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.0924718379974365,
+      "learning_rate": 2.691955278648003e-05,
+      "loss": 2.9762,
+      "step": 3175
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.0304733514785767,
+      "learning_rate": 2.6771790515402112e-05,
+      "loss": 3.059,
+      "step": 3180
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.2347314357757568,
+      "learning_rate": 2.6624286487513916e-05,
+      "loss": 2.9933,
+      "step": 3185
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.1086734533309937,
+      "learning_rate": 2.6477042342714137e-05,
+      "loss": 3.002,
+      "step": 3190
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.0901292562484741,
+      "learning_rate": 2.633005971801219e-05,
+      "loss": 3.0097,
+      "step": 3195
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.1064410209655762,
+      "learning_rate": 2.6183340247510013e-05,
+      "loss": 2.9945,
+      "step": 3200
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.0292625427246094,
+      "learning_rate": 2.6036885562383856e-05,
+      "loss": 3.0133,
+      "step": 3205
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.1139415502548218,
+      "learning_rate": 2.5890697290866206e-05,
+      "loss": 2.9208,
+      "step": 3210
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.0623078346252441,
+      "learning_rate": 2.5744777058227642e-05,
+      "loss": 2.9424,
+      "step": 3215
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.0108777284622192,
+      "learning_rate": 2.5599126486758777e-05,
+      "loss": 2.9031,
+      "step": 3220
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.0001078844070435,
+      "learning_rate": 2.5453747195752243e-05,
+      "loss": 2.8608,
+      "step": 3225
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.1540250778198242,
+      "learning_rate": 2.530864080148464e-05,
+      "loss": 2.9748,
+      "step": 3230
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.124299168586731,
+      "learning_rate": 2.5163808917198615e-05,
+      "loss": 2.9228,
+      "step": 3235
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.0335966348648071,
+      "learning_rate": 2.501925315308492e-05,
+      "loss": 2.9867,
+      "step": 3240
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.0575364828109741,
+      "learning_rate": 2.4874975116264477e-05,
+      "loss": 2.972,
+      "step": 3245
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.044665813446045,
+      "learning_rate": 2.4730976410770534e-05,
+      "loss": 2.979,
+      "step": 3250
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.0527253150939941,
+      "learning_rate": 2.458725863753084e-05,
+      "loss": 2.9601,
+      "step": 3255
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.0238512754440308,
+      "learning_rate": 2.4443823394349834e-05,
+      "loss": 2.8508,
+      "step": 3260
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.101250410079956,
+      "learning_rate": 2.430067227589088e-05,
+      "loss": 2.9144,
+      "step": 3265
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9972196817398071,
+      "learning_rate": 2.4157806873658517e-05,
+      "loss": 2.9406,
+      "step": 3270
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.162981629371643,
+      "learning_rate": 2.401522877598087e-05,
+      "loss": 2.9642,
+      "step": 3275
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.107654094696045,
+      "learning_rate": 2.3872939567991827e-05,
+      "loss": 2.9408,
+      "step": 3280
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.0248068571090698,
+      "learning_rate": 2.373094083161353e-05,
+      "loss": 2.8816,
+      "step": 3285
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.029221534729004,
+      "learning_rate": 2.358923414553877e-05,
+      "loss": 2.9401,
+      "step": 3290
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.0091698169708252,
+      "learning_rate": 2.3447821085213405e-05,
+      "loss": 2.8775,
+      "step": 3295
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0124790668487549,
+      "learning_rate": 2.3306703222818878e-05,
+      "loss": 2.9208,
+      "step": 3300
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0469365119934082,
+      "learning_rate": 2.3165882127254705e-05,
+      "loss": 2.9814,
+      "step": 3305
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.2536821365356445,
+      "learning_rate": 2.302535936412108e-05,
+      "loss": 2.9304,
+      "step": 3310
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0279680490493774,
+      "learning_rate": 2.2885136495701415e-05,
+      "loss": 2.8435,
+      "step": 3315
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0763942003250122,
+      "learning_rate": 2.274521508094501e-05,
+      "loss": 2.9261,
+      "step": 3320
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.030404806137085,
+      "learning_rate": 2.2605596675449698e-05,
+      "loss": 3.0172,
+      "step": 3325
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0439190864562988,
+      "learning_rate": 2.246628283144457e-05,
+      "loss": 2.8992,
+      "step": 3330
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0290809869766235,
+      "learning_rate": 2.232727509777269e-05,
+      "loss": 3.0806,
+      "step": 3335
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.058119773864746,
+      "learning_rate": 2.2188575019873932e-05,
+      "loss": 2.9167,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.0898820161819458,
+      "learning_rate": 2.2050184139767704e-05,
+      "loss": 2.8946,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.0208537578582764,
+      "learning_rate": 2.191210399603591e-05,
+      "loss": 2.931,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.017960786819458,
+      "learning_rate": 2.1774336123805772e-05,
+      "loss": 2.9429,
+      "step": 3355
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.1384835243225098,
+      "learning_rate": 2.1636882054732776e-05,
+      "loss": 2.9701,
+      "step": 3360
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.1678301095962524,
+      "learning_rate": 2.1499743316983684e-05,
+      "loss": 2.9414,
+      "step": 3365
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.090770959854126,
+      "learning_rate": 2.1362921435219473e-05,
+      "loss": 2.9129,
+      "step": 3370
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.0095657110214233,
+      "learning_rate": 2.1226417930578464e-05,
+      "loss": 2.8908,
+      "step": 3375
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.0220203399658203,
+      "learning_rate": 2.109023432065935e-05,
+      "loss": 2.9649,
+      "step": 3380
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.0544750690460205,
+      "learning_rate": 2.095437211950434e-05,
+      "loss": 2.9183,
+      "step": 3385
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.0294827222824097,
+      "learning_rate": 2.0818832837582352e-05,
+      "loss": 2.9431,
+      "step": 3390
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0410583019256592,
+      "learning_rate": 2.068361798177218e-05,
+      "loss": 2.9157,
+      "step": 3395
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.1043765544891357,
+      "learning_rate": 2.0548729055345778e-05,
+      "loss": 2.9759,
+      "step": 3400
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0321478843688965,
+      "learning_rate": 2.0414167557951514e-05,
+      "loss": 2.8631,
+      "step": 3405
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0459082126617432,
+      "learning_rate": 2.0279934985597527e-05,
+      "loss": 3.0045,
+      "step": 3410
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0183168649673462,
+      "learning_rate": 2.0146032830635054e-05,
+      "loss": 2.9016,
+      "step": 3415
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.006795883178711,
+      "learning_rate": 2.001246258174192e-05,
+      "loss": 2.9133,
+      "step": 3420
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0646164417266846,
+      "learning_rate": 1.9879225723905886e-05,
+      "loss": 2.9729,
+      "step": 3425
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0548053979873657,
+      "learning_rate": 1.9746323738408203e-05,
+      "loss": 2.9754,
+      "step": 3430
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.073244571685791,
+      "learning_rate": 1.9613758102807117e-05,
+      "loss": 2.9281,
+      "step": 3435
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0305603742599487,
+      "learning_rate": 1.9481530290921474e-05,
+      "loss": 3.0633,
+      "step": 3440
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0235049724578857,
+      "learning_rate": 1.934964177281428e-05,
+      "loss": 2.9378,
+      "step": 3445
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0278953313827515,
+      "learning_rate": 1.9218094014776434e-05,
+      "loss": 2.9178,
+      "step": 3450
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0318273305892944,
+      "learning_rate": 1.9086888479310333e-05,
+      "loss": 3.0008,
+      "step": 3455
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.027692198753357,
+      "learning_rate": 1.895602662511371e-05,
+      "loss": 2.9887,
+      "step": 3460
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0255205631256104,
+      "learning_rate": 1.8825509907063327e-05,
+      "loss": 2.9067,
+      "step": 3465
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.032726526260376,
+      "learning_rate": 1.8695339776198872e-05,
+      "loss": 2.8492,
+      "step": 3470
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0367424488067627,
+      "learning_rate": 1.8565517679706783e-05,
+      "loss": 2.9587,
+      "step": 3475
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0441900491714478,
+      "learning_rate": 1.8436045060904174e-05,
+      "loss": 2.9499,
+      "step": 3480
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.090521216392517,
+      "learning_rate": 1.830692335922279e-05,
+      "loss": 2.9482,
+      "step": 3485
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.057732105255127,
+      "learning_rate": 1.8178154010192994e-05,
+      "loss": 2.9731,
+      "step": 3490
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.021804928779602,
+      "learning_rate": 1.8049738445427822e-05,
+      "loss": 2.978,
+      "step": 3495
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.0467053651809692,
+      "learning_rate": 1.7921678092607052e-05,
+      "loss": 2.9755,
+      "step": 3500
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.0179669857025146,
+      "learning_rate": 1.7793974375461352e-05,
+      "loss": 2.938,
+      "step": 3505
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.0066503286361694,
+      "learning_rate": 1.7666628713756417e-05,
+      "loss": 2.8577,
+      "step": 3510
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.0826014280319214,
+      "learning_rate": 1.7539642523277228e-05,
+      "loss": 2.9725,
+      "step": 3515
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.091303825378418,
+      "learning_rate": 1.7413017215812273e-05,
+      "loss": 2.9723,
+      "step": 3520
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.063717246055603,
+      "learning_rate": 1.728675419913788e-05,
+      "loss": 2.9279,
+      "step": 3525
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.069888949394226,
+      "learning_rate": 1.716085487700253e-05,
+      "loss": 2.9628,
+      "step": 3530
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.1069707870483398,
+      "learning_rate": 1.703532064911131e-05,
+      "loss": 2.9247,
+      "step": 3535
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.064904808998108,
+      "learning_rate": 1.6910152911110283e-05,
+      "loss": 2.9383,
+      "step": 3540
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.0969486236572266,
+      "learning_rate": 1.6785353054571024e-05,
+      "loss": 2.9503,
+      "step": 3545
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.0619415044784546,
+      "learning_rate": 1.666092246697512e-05,
+      "loss": 2.8936,
+      "step": 3550
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.1957061290740967,
+      "learning_rate": 1.6536862531698766e-05,
+      "loss": 2.9335,
+      "step": 3555
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.1700977087020874,
+      "learning_rate": 1.6413174627997328e-05,
+      "loss": 3.0092,
+      "step": 3560
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.0224504470825195,
+      "learning_rate": 1.6289860130990147e-05,
+      "loss": 2.9349,
+      "step": 3565
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.0252118110656738,
+      "learning_rate": 1.6166920411645064e-05,
+      "loss": 2.9296,
+      "step": 3570
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.0665481090545654,
+      "learning_rate": 1.6044356836763315e-05,
+      "loss": 2.9078,
+      "step": 3575
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.0700303316116333,
+      "learning_rate": 1.5922170768964285e-05,
+      "loss": 2.9363,
+      "step": 3580
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.104561448097229,
+      "learning_rate": 1.5800363566670362e-05,
+      "loss": 3.0238,
+      "step": 3585
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.0166821479797363,
+      "learning_rate": 1.5678936584091852e-05,
+      "loss": 2.8869,
+      "step": 3590
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.0269566774368286,
+      "learning_rate": 1.5557891171211892e-05,
+      "loss": 3.0122,
+      "step": 3595
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.0524886846542358,
+      "learning_rate": 1.5437228673771465e-05,
+      "loss": 2.9004,
+      "step": 3600
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.018042802810669,
+      "learning_rate": 1.5316950433254445e-05,
+      "loss": 2.9853,
+      "step": 3605
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.0478148460388184,
+      "learning_rate": 1.5197057786872649e-05,
+      "loss": 2.8939,
+      "step": 3610
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.0328203439712524,
+      "learning_rate": 1.5077552067551015e-05,
+      "loss": 2.9921,
+      "step": 3615
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.0347979068756104,
+      "learning_rate": 1.4958434603912747e-05,
+      "loss": 2.9324,
+      "step": 3620
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.0036380290985107,
+      "learning_rate": 1.4839706720264546e-05,
+      "loss": 2.9383,
+      "step": 3625
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.990047037601471,
+      "learning_rate": 1.4721369736581924e-05,
+      "loss": 2.9131,
+      "step": 3630
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0028934478759766,
+      "learning_rate": 1.4603424968494484e-05,
+      "loss": 2.9047,
+      "step": 3635
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0312615633010864,
+      "learning_rate": 1.448587372727132e-05,
+      "loss": 2.8849,
+      "step": 3640
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0701656341552734,
+      "learning_rate": 1.4368717319806419e-05,
+      "loss": 2.9952,
+      "step": 3645
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0108087062835693,
+      "learning_rate": 1.4251957048604152e-05,
+      "loss": 2.8067,
+      "step": 3650
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0179967880249023,
+      "learning_rate": 1.413559421176479e-05,
+      "loss": 2.9415,
+      "step": 3655
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0271110534667969,
+      "learning_rate": 1.4019630102970056e-05,
+      "loss": 2.9605,
+      "step": 3660
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0323963165283203,
+      "learning_rate": 1.3904066011468753e-05,
+      "loss": 2.9647,
+      "step": 3665
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0108588933944702,
+      "learning_rate": 1.3788903222062433e-05,
+      "loss": 2.8892,
+      "step": 3670
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0895428657531738,
+      "learning_rate": 1.3674143015091118e-05,
+      "loss": 2.9541,
+      "step": 3675
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.055143117904663,
+      "learning_rate": 1.355978666641905e-05,
+      "loss": 2.967,
+      "step": 3680
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.0495216846466064,
+      "learning_rate": 1.3445835447420507e-05,
+      "loss": 2.9434,
+      "step": 3685
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.1741347312927246,
+      "learning_rate": 1.3332290624965688e-05,
+      "loss": 2.8592,
+      "step": 3690
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.0490764379501343,
+      "learning_rate": 1.3219153461406609e-05,
+      "loss": 2.8898,
+      "step": 3695
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.0156313180923462,
+      "learning_rate": 1.3106425214563078e-05,
+      "loss": 2.8871,
+      "step": 3700
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.0351041555404663,
+      "learning_rate": 1.2994107137708716e-05,
+      "loss": 2.9649,
+      "step": 3705
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9821464419364929,
+      "learning_rate": 1.2882200479556988e-05,
+      "loss": 2.7567,
+      "step": 3710
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.0293828248977661,
+      "learning_rate": 1.2770706484247397e-05,
+      "loss": 3.044,
+      "step": 3715
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.0418314933776855,
+      "learning_rate": 1.2659626391331564e-05,
+      "loss": 2.9845,
+      "step": 3720
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.1662086248397827,
+      "learning_rate": 1.2548961435759493e-05,
+      "loss": 2.9251,
+      "step": 3725
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.004215955734253,
+      "learning_rate": 1.2438712847865846e-05,
+      "loss": 2.7973,
+      "step": 3730
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.051112174987793,
+      "learning_rate": 1.2328881853356244e-05,
+      "loss": 2.879,
+      "step": 3735
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0411102771759033,
+      "learning_rate": 1.221946967329365e-05,
+      "loss": 2.9298,
+      "step": 3740
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.110374927520752,
+      "learning_rate": 1.2110477524084796e-05,
+      "loss": 2.8414,
+      "step": 3745
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0748777389526367,
+      "learning_rate": 1.2001906617466657e-05,
+      "loss": 2.9641,
+      "step": 3750
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0381336212158203,
+      "learning_rate": 1.1893758160492978e-05,
+      "loss": 2.9166,
+      "step": 3755
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0083729028701782,
+      "learning_rate": 1.1786033355520859e-05,
+      "loss": 2.7783,
+      "step": 3760
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0652658939361572,
+      "learning_rate": 1.1678733400197373e-05,
+      "loss": 2.9004,
+      "step": 3765
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.041797161102295,
+      "learning_rate": 1.1571859487446263e-05,
+      "loss": 2.955,
+      "step": 3770
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0103260278701782,
+      "learning_rate": 1.1465412805454695e-05,
+      "loss": 2.9241,
+      "step": 3775
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.046828031539917,
+      "learning_rate": 1.1359394537660011e-05,
+      "loss": 2.924,
+      "step": 3780
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.0475406646728516,
+      "learning_rate": 1.125380586273661e-05,
+      "loss": 2.9699,
+      "step": 3785
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.046965479850769,
+      "learning_rate": 1.1148647954582808e-05,
+      "loss": 2.8634,
+      "step": 3790
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.0719529390335083,
+      "learning_rate": 1.1043921982307819e-05,
+      "loss": 2.9003,
+      "step": 3795
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.020414113998413,
+      "learning_rate": 1.0939629110218735e-05,
+      "loss": 2.8621,
+      "step": 3800
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.0194061994552612,
+      "learning_rate": 1.0835770497807596e-05,
+      "loss": 2.9386,
+      "step": 3805
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.0563113689422607,
+      "learning_rate": 1.0732347299738493e-05,
+      "loss": 2.9796,
+      "step": 3810
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.0426850318908691,
+      "learning_rate": 1.0629360665834732e-05,
+      "loss": 2.9268,
+      "step": 3815
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.037277340888977,
+      "learning_rate": 1.052681174106604e-05,
+      "loss": 2.9021,
+      "step": 3820
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.0325355529785156,
+      "learning_rate": 1.0424701665535852e-05,
+      "loss": 2.9348,
+      "step": 3825
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.0245922803878784,
+      "learning_rate": 1.0323031574468638e-05,
+      "loss": 2.8854,
+      "step": 3830
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.025760293006897,
+      "learning_rate": 1.0221802598197261e-05,
+      "loss": 2.894,
+      "step": 3835
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.1027809381484985,
+      "learning_rate": 1.0121015862150423e-05,
+      "loss": 2.8603,
+      "step": 3840
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.08948814868927,
+      "learning_rate": 1.0020672486840154e-05,
+      "loss": 2.966,
+      "step": 3845
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.0438921451568604,
+      "learning_rate": 9.920773587849364e-06,
+      "loss": 2.885,
+      "step": 3850
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.0353776216506958,
+      "learning_rate": 9.821320275819401e-06,
+      "loss": 2.8779,
+      "step": 3855
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.0220935344696045,
+      "learning_rate": 9.72231365643777e-06,
+      "loss": 2.864,
+      "step": 3860
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.0436768531799316,
+      "learning_rate": 9.623754830425779e-06,
+      "loss": 2.991,
+      "step": 3865
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.072954773902893,
+      "learning_rate": 9.52564489352632e-06,
+      "loss": 2.9132,
+      "step": 3870
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0170408487319946,
+      "learning_rate": 9.427984936491702e-06,
+      "loss": 2.9115,
+      "step": 3875
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.992523729801178,
+      "learning_rate": 9.330776045071509e-06,
+      "loss": 2.9234,
+      "step": 3880
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0199633836746216,
+      "learning_rate": 9.23401930000054e-06,
+      "loss": 2.966,
+      "step": 3885
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0794175863265991,
+      "learning_rate": 9.137715776986772e-06,
+      "loss": 2.9374,
+      "step": 3890
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0142548084259033,
+      "learning_rate": 9.041866546699434e-06,
+      "loss": 2.9269,
+      "step": 3895
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0138798952102661,
+      "learning_rate": 8.946472674757078e-06,
+      "loss": 2.847,
+      "step": 3900
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9981939196586609,
+      "learning_rate": 8.851535221715735e-06,
+      "loss": 2.9043,
+      "step": 3905
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0097284317016602,
+      "learning_rate": 8.757055243057132e-06,
+      "loss": 2.9587,
+      "step": 3910
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0881823301315308,
+      "learning_rate": 8.663033789176967e-06,
+      "loss": 2.8831,
+      "step": 3915
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.1520594358444214,
+      "learning_rate": 8.5694719053732e-06,
+      "loss": 2.8992,
+      "step": 3920
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.141189694404602,
+      "learning_rate": 8.476370631834458e-06,
+      "loss": 2.8394,
+      "step": 3925
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.0872468948364258,
+      "learning_rate": 8.383731003628452e-06,
+      "loss": 2.8789,
+      "step": 3930
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.0820999145507812,
+      "learning_rate": 8.291554050690508e-06,
+      "loss": 2.8336,
+      "step": 3935
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.0161839723587036,
+      "learning_rate": 8.199840797812058e-06,
+      "loss": 2.8955,
+      "step": 3940
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.0762035846710205,
+      "learning_rate": 8.108592264629295e-06,
+      "loss": 2.9941,
+      "step": 3945
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.0508668422698975,
+      "learning_rate": 8.017809465611803e-06,
+      "loss": 2.797,
+      "step": 3950
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.0316884517669678,
+      "learning_rate": 7.927493410051324e-06,
+      "loss": 2.8648,
+      "step": 3955
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.9886409044265747,
+      "learning_rate": 7.837645102050473e-06,
+      "loss": 2.8788,
+      "step": 3960
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.0440611839294434,
+      "learning_rate": 7.748265540511635e-06,
+      "loss": 2.9606,
+      "step": 3965
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.057845950126648,
+      "learning_rate": 7.65935571912582e-06,
+      "loss": 2.8397,
+      "step": 3970
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.1614924669265747,
+      "learning_rate": 7.5709166263616405e-06,
+      "loss": 2.9892,
+      "step": 3975
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.0376337766647339,
+      "learning_rate": 7.482949245454302e-06,
+      "loss": 2.8739,
+      "step": 3980
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.079857587814331,
+      "learning_rate": 7.3954545543946876e-06,
+      "loss": 2.8968,
+      "step": 3985
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.051576018333435,
+      "learning_rate": 7.308433525918468e-06,
+      "loss": 2.9745,
+      "step": 3990
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.0478307008743286,
+      "learning_rate": 7.221887127495313e-06,
+      "loss": 2.9211,
+      "step": 3995
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.063901424407959,
+      "learning_rate": 7.1358163213181114e-06,
+      "loss": 2.911,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4811,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.1658574908358656e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}