diff --git "a/checkpoint-3880/trainer_state.json" "b/checkpoint-3880/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3880/trainer_state.json"
@@ -0,0 +1,6823 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 3880,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002061855670103093,
+      "grad_norm": 4.401242256164551,
+      "learning_rate": 2.061855670103093e-06,
+      "loss": 5.7831,
+      "step": 4
+    },
+    {
+      "epoch": 0.004123711340206186,
+      "grad_norm": 4.165046691894531,
+      "learning_rate": 4.123711340206186e-06,
+      "loss": 5.6527,
+      "step": 8
+    },
+    {
+      "epoch": 0.006185567010309278,
+      "grad_norm": 3.4843931198120117,
+      "learning_rate": 6.185567010309279e-06,
+      "loss": 5.5896,
+      "step": 12
+    },
+    {
+      "epoch": 0.008247422680412371,
+      "grad_norm": 3.29345965385437,
+      "learning_rate": 8.247422680412371e-06,
+      "loss": 5.3939,
+      "step": 16
+    },
+    {
+      "epoch": 0.010309278350515464,
+      "grad_norm": 3.1640310287475586,
+      "learning_rate": 1.0309278350515464e-05,
+      "loss": 5.0285,
+      "step": 20
+    },
+    {
+      "epoch": 0.012371134020618556,
+      "grad_norm": 3.238795757293701,
+      "learning_rate": 1.2371134020618558e-05,
+      "loss": 5.3506,
+      "step": 24
+    },
+    {
+      "epoch": 0.01443298969072165,
+      "grad_norm": 2.786606550216675,
+      "learning_rate": 1.4432989690721649e-05,
+      "loss": 4.8622,
+      "step": 28
+    },
+    {
+      "epoch": 0.016494845360824743,
+      "grad_norm": 2.689506769180298,
+      "learning_rate": 1.6494845360824743e-05,
+      "loss": 4.7998,
+      "step": 32
+    },
+    {
+      "epoch": 0.018556701030927835,
+      "grad_norm": 3.6669228076934814,
+      "learning_rate": 1.8556701030927837e-05,
+      "loss": 4.8852,
+      "step": 36
+    },
+    {
+      "epoch": 0.020618556701030927,
+      "grad_norm": 2.614325761795044,
+      "learning_rate": 2.0618556701030927e-05,
+      "loss": 4.5481,
+      "step": 40
+    },
+    {
+      "epoch": 0.02268041237113402,
+      "grad_norm": 2.6969540119171143,
+      "learning_rate": 2.268041237113402e-05,
+      "loss": 4.4287,
+      "step": 44
+    },
+    {
+      "epoch": 0.024742268041237112,
+      "grad_norm": 2.5721373558044434,
+      "learning_rate": 2.4742268041237116e-05,
+      "loss": 4.2744,
+      "step": 48
+    },
+    {
+      "epoch": 0.026804123711340205,
+      "grad_norm": 2.7030205726623535,
+      "learning_rate": 2.6804123711340206e-05,
+      "loss": 4.2424,
+      "step": 52
+    },
+    {
+      "epoch": 0.0288659793814433,
+      "grad_norm": 2.5551116466522217,
+      "learning_rate": 2.8865979381443297e-05,
+      "loss": 4.1868,
+      "step": 56
+    },
+    {
+      "epoch": 0.030927835051546393,
+      "grad_norm": 2.6516504287719727,
+      "learning_rate": 3.0927835051546395e-05,
+      "loss": 3.9569,
+      "step": 60
+    },
+    {
+      "epoch": 0.032989690721649485,
+      "grad_norm": 2.516420602798462,
+      "learning_rate": 3.2989690721649485e-05,
+      "loss": 4.0263,
+      "step": 64
+    },
+    {
+      "epoch": 0.03505154639175258,
+      "grad_norm": 2.4037609100341797,
+      "learning_rate": 3.5051546391752576e-05,
+      "loss": 4.001,
+      "step": 68
+    },
+    {
+      "epoch": 0.03711340206185567,
+      "grad_norm": 2.3336949348449707,
+      "learning_rate": 3.7113402061855674e-05,
+      "loss": 3.7654,
+      "step": 72
+    },
+    {
+      "epoch": 0.03917525773195876,
+      "grad_norm": 2.5022168159484863,
+      "learning_rate": 3.9175257731958764e-05,
+      "loss": 3.7868,
+      "step": 76
+    },
+    {
+      "epoch": 0.041237113402061855,
+      "grad_norm": 2.496216297149658,
+      "learning_rate": 4.1237113402061855e-05,
+      "loss": 3.7257,
+      "step": 80
+    },
+    {
+      "epoch": 0.04329896907216495,
+      "grad_norm": 2.3523142337799072,
+      "learning_rate": 4.329896907216495e-05,
+      "loss": 3.6547,
+      "step": 84
+    },
+    {
+      "epoch": 0.04536082474226804,
+      "grad_norm": 2.4950451850891113,
+      "learning_rate": 4.536082474226804e-05,
+      "loss": 3.5265,
+      "step": 88
+    },
+    {
+      "epoch": 0.04742268041237113,
+      "grad_norm": 2.5210957527160645,
+      "learning_rate": 4.7422680412371134e-05,
+      "loss": 3.5772,
+      "step": 92
+    },
+    {
+      "epoch": 0.049484536082474224,
+      "grad_norm": 2.3704540729522705,
+      "learning_rate": 4.948453608247423e-05,
+      "loss": 3.6708,
+      "step": 96
+    },
+    {
+      "epoch": 0.05154639175257732,
+      "grad_norm": 2.361452102661133,
+      "learning_rate": 5.1546391752577315e-05,
+      "loss": 3.4935,
+      "step": 100
+    },
+    {
+      "epoch": 0.05360824742268041,
+      "grad_norm": 2.4770519733428955,
+      "learning_rate": 5.360824742268041e-05,
+      "loss": 3.4953,
+      "step": 104
+    },
+    {
+      "epoch": 0.05567010309278351,
+      "grad_norm": 2.6673684120178223,
+      "learning_rate": 5.567010309278351e-05,
+      "loss": 3.4362,
+      "step": 108
+    },
+    {
+      "epoch": 0.0577319587628866,
+      "grad_norm": 2.3836777210235596,
+      "learning_rate": 5.7731958762886594e-05,
+      "loss": 3.4341,
+      "step": 112
+    },
+    {
+      "epoch": 0.05979381443298969,
+      "grad_norm": 2.4742355346679688,
+      "learning_rate": 5.979381443298969e-05,
+      "loss": 3.4443,
+      "step": 116
+    },
+    {
+      "epoch": 0.061855670103092786,
+      "grad_norm": 2.513885974884033,
+      "learning_rate": 6.185567010309279e-05,
+      "loss": 3.3867,
+      "step": 120
+    },
+    {
+      "epoch": 0.06391752577319587,
+      "grad_norm": 2.633009433746338,
+      "learning_rate": 6.391752577319587e-05,
+      "loss": 3.3804,
+      "step": 124
+    },
+    {
+      "epoch": 0.06597938144329897,
+      "grad_norm": 2.7852959632873535,
+      "learning_rate": 6.597938144329897e-05,
+      "loss": 3.4522,
+      "step": 128
+    },
+    {
+      "epoch": 0.06804123711340206,
+      "grad_norm": 2.9940905570983887,
+      "learning_rate": 6.804123711340207e-05,
+      "loss": 3.3606,
+      "step": 132
+    },
+    {
+      "epoch": 0.07010309278350516,
+      "grad_norm": 2.4730873107910156,
+      "learning_rate": 7.010309278350515e-05,
+      "loss": 3.2392,
+      "step": 136
+    },
+    {
+      "epoch": 0.07216494845360824,
+      "grad_norm": 2.1951076984405518,
+      "learning_rate": 7.216494845360825e-05,
+      "loss": 3.4019,
+      "step": 140
+    },
+    {
+      "epoch": 0.07422680412371134,
+      "grad_norm": 1.9874905347824097,
+      "learning_rate": 7.422680412371135e-05,
+      "loss": 3.1746,
+      "step": 144
+    },
+    {
+      "epoch": 0.07628865979381444,
+      "grad_norm": 2.2700319290161133,
+      "learning_rate": 7.628865979381443e-05,
+      "loss": 3.1944,
+      "step": 148
+    },
+    {
+      "epoch": 0.07835051546391752,
+      "grad_norm": 2.2276201248168945,
+      "learning_rate": 7.835051546391753e-05,
+      "loss": 3.2753,
+      "step": 152
+    },
+    {
+      "epoch": 0.08041237113402062,
+      "grad_norm": 2.150624990463257,
+      "learning_rate": 8.041237113402063e-05,
+      "loss": 3.2761,
+      "step": 156
+    },
+    {
+      "epoch": 0.08247422680412371,
+      "grad_norm": 2.2154808044433594,
+      "learning_rate": 8.247422680412371e-05,
+      "loss": 3.1575,
+      "step": 160
+    },
+    {
+      "epoch": 0.08453608247422681,
+      "grad_norm": 2.6334950923919678,
+      "learning_rate": 8.453608247422681e-05,
+      "loss": 3.185,
+      "step": 164
+    },
+    {
+      "epoch": 0.0865979381443299,
+      "grad_norm": 2.70011305809021,
+      "learning_rate": 8.65979381443299e-05,
+      "loss": 3.1316,
+      "step": 168
+    },
+    {
+      "epoch": 0.088659793814433,
+      "grad_norm": 2.533686637878418,
+      "learning_rate": 8.865979381443299e-05,
+      "loss": 3.1387,
+      "step": 172
+    },
+    {
+      "epoch": 0.09072164948453608,
+      "grad_norm": 2.154860019683838,
+      "learning_rate": 9.072164948453609e-05,
+      "loss": 3.1592,
+      "step": 176
+    },
+    {
+      "epoch": 0.09278350515463918,
+      "grad_norm": 2.1940603256225586,
+      "learning_rate": 9.278350515463918e-05,
+      "loss": 3.1773,
+      "step": 180
+    },
+    {
+      "epoch": 0.09484536082474226,
+      "grad_norm": 2.257709264755249,
+      "learning_rate": 9.484536082474227e-05,
+      "loss": 3.1052,
+      "step": 184
+    },
+    {
+      "epoch": 0.09690721649484536,
+      "grad_norm": 2.08589768409729,
+      "learning_rate": 9.690721649484537e-05,
+      "loss": 3.0373,
+      "step": 188
+    },
+    {
+      "epoch": 0.09896907216494845,
+      "grad_norm": 2.060812473297119,
+      "learning_rate": 9.896907216494846e-05,
+      "loss": 3.112,
+      "step": 192
+    },
+    {
+      "epoch": 0.10103092783505155,
+      "grad_norm": 2.3649418354034424,
+      "learning_rate": 9.999992735780168e-05,
+      "loss": 3.198,
+      "step": 196
+    },
+    {
+      "epoch": 0.10309278350515463,
+      "grad_norm": 2.339231252670288,
+      "learning_rate": 9.999934622148157e-05,
+      "loss": 3.1736,
+      "step": 200
+    },
+    {
+      "epoch": 0.10515463917525773,
+      "grad_norm": 2.037454128265381,
+      "learning_rate": 9.999818395559577e-05,
+      "loss": 3.1006,
+      "step": 204
+    },
+    {
+      "epoch": 0.10721649484536082,
+      "grad_norm": 2.343181848526001,
+      "learning_rate": 9.999644057365295e-05,
+      "loss": 3.1055,
+      "step": 208
+    },
+    {
+      "epoch": 0.10927835051546392,
+      "grad_norm": 2.3469326496124268,
+      "learning_rate": 9.999411609591603e-05,
+      "loss": 3.0211,
+      "step": 212
+    },
+    {
+      "epoch": 0.11134020618556702,
+      "grad_norm": 2.2599899768829346,
+      "learning_rate": 9.999121054940182e-05,
+      "loss": 3.0236,
+      "step": 216
+    },
+    {
+      "epoch": 0.1134020618556701,
+      "grad_norm": 2.114220142364502,
+      "learning_rate": 9.998772396788072e-05,
+      "loss": 3.0895,
+      "step": 220
+    },
+    {
+      "epoch": 0.1154639175257732,
+      "grad_norm": 2.0050344467163086,
+      "learning_rate": 9.998365639187638e-05,
+      "loss": 3.0542,
+      "step": 224
+    },
+    {
+      "epoch": 0.11752577319587629,
+      "grad_norm": 2.1245343685150146,
+      "learning_rate": 9.997900786866519e-05,
+      "loss": 3.0631,
+      "step": 228
+    },
+    {
+      "epoch": 0.11958762886597939,
+      "grad_norm": 2.0271108150482178,
+      "learning_rate": 9.997377845227576e-05,
+      "loss": 3.1621,
+      "step": 232
+    },
+    {
+      "epoch": 0.12164948453608247,
+      "grad_norm": 2.075441837310791,
+      "learning_rate": 9.996796820348822e-05,
+      "loss": 2.9906,
+      "step": 236
+    },
+    {
+      "epoch": 0.12371134020618557,
+      "grad_norm": 2.118668556213379,
+      "learning_rate": 9.996157718983362e-05,
+      "loss": 3.0718,
+      "step": 240
+    },
+    {
+      "epoch": 0.12577319587628866,
+      "grad_norm": 2.0540637969970703,
+      "learning_rate": 9.995460548559307e-05,
+      "loss": 3.077,
+      "step": 244
+    },
+    {
+      "epoch": 0.12783505154639174,
+      "grad_norm": 2.2028098106384277,
+      "learning_rate": 9.99470531717969e-05,
+      "loss": 3.1064,
+      "step": 248
+    },
+    {
+      "epoch": 0.12989690721649486,
+      "grad_norm": 2.0607352256774902,
+      "learning_rate": 9.993892033622374e-05,
+      "loss": 3.0582,
+      "step": 252
+    },
+    {
+      "epoch": 0.13195876288659794,
+      "grad_norm": 2.054838180541992,
+      "learning_rate": 9.993020707339939e-05,
+      "loss": 2.991,
+      "step": 256
+    },
+    {
+      "epoch": 0.13402061855670103,
+      "grad_norm": 1.9367263317108154,
+      "learning_rate": 9.992091348459591e-05,
+      "loss": 2.9583,
+      "step": 260
+    },
+    {
+      "epoch": 0.1360824742268041,
+      "grad_norm": 1.9889065027236938,
+      "learning_rate": 9.99110396778303e-05,
+      "loss": 2.9845,
+      "step": 264
+    },
+    {
+      "epoch": 0.13814432989690723,
+      "grad_norm": 2.249467134475708,
+      "learning_rate": 9.990058576786325e-05,
+      "loss": 3.0511,
+      "step": 268
+    },
+    {
+      "epoch": 0.1402061855670103,
+      "grad_norm": 2.011195659637451,
+      "learning_rate": 9.98895518761979e-05,
+      "loss": 2.9925,
+      "step": 272
+    },
+    {
+      "epoch": 0.1422680412371134,
+      "grad_norm": 2.0568485260009766,
+      "learning_rate": 9.987793813107833e-05,
+      "loss": 2.9491,
+      "step": 276
+    },
+    {
+      "epoch": 0.14432989690721648,
+      "grad_norm": 1.9748948812484741,
+      "learning_rate": 9.986574466748812e-05,
+      "loss": 3.0667,
+      "step": 280
+    },
+    {
+      "epoch": 0.1463917525773196,
+      "grad_norm": 2.234513282775879,
+      "learning_rate": 9.985297162714877e-05,
+      "loss": 2.8514,
+      "step": 284
+    },
+    {
+      "epoch": 0.14845360824742268,
+      "grad_norm": 1.9163928031921387,
+      "learning_rate": 9.983961915851804e-05,
+      "loss": 3.1231,
+      "step": 288
+    },
+    {
+      "epoch": 0.15051546391752577,
+      "grad_norm": 1.9212491512298584,
+      "learning_rate": 9.982568741678823e-05,
+      "loss": 2.9029,
+      "step": 292
+    },
+    {
+      "epoch": 0.15257731958762888,
+      "grad_norm": 2.065171241760254,
+      "learning_rate": 9.981117656388445e-05,
+      "loss": 2.9171,
+      "step": 296
+    },
+    {
+      "epoch": 0.15463917525773196,
+      "grad_norm": 1.979385495185852,
+      "learning_rate": 9.979608676846258e-05,
+      "loss": 3.0767,
+      "step": 300
+    },
+    {
+      "epoch": 0.15670103092783505,
+      "grad_norm": 1.9221059083938599,
+      "learning_rate": 9.978041820590743e-05,
+      "loss": 3.0627,
+      "step": 304
+    },
+    {
+      "epoch": 0.15876288659793814,
+      "grad_norm": 2.0437042713165283,
+      "learning_rate": 9.97641710583307e-05,
+      "loss": 2.938,
+      "step": 308
+    },
+    {
+      "epoch": 0.16082474226804125,
+      "grad_norm": 1.9110627174377441,
+      "learning_rate": 9.974734551456881e-05,
+      "loss": 2.9462,
+      "step": 312
+    },
+    {
+      "epoch": 0.16288659793814433,
+      "grad_norm": 1.8492796421051025,
+      "learning_rate": 9.972994177018074e-05,
+      "loss": 3.0006,
+      "step": 316
+    },
+    {
+      "epoch": 0.16494845360824742,
+      "grad_norm": 1.9782150983810425,
+      "learning_rate": 9.971196002744575e-05,
+      "loss": 2.9449,
+      "step": 320
+    },
+    {
+      "epoch": 0.1670103092783505,
+      "grad_norm": 1.8963398933410645,
+      "learning_rate": 9.969340049536099e-05,
+      "loss": 2.9932,
+      "step": 324
+    },
+    {
+      "epoch": 0.16907216494845362,
+      "grad_norm": 2.0604023933410645,
+      "learning_rate": 9.967426338963917e-05,
+      "loss": 2.8904,
+      "step": 328
+    },
+    {
+      "epoch": 0.1711340206185567,
+      "grad_norm": 2.0546135902404785,
+      "learning_rate": 9.965454893270592e-05,
+      "loss": 3.0603,
+      "step": 332
+    },
+    {
+      "epoch": 0.1731958762886598,
+      "grad_norm": 1.9946964979171753,
+      "learning_rate": 9.963425735369736e-05,
+      "loss": 2.917,
+      "step": 336
+    },
+    {
+      "epoch": 0.17525773195876287,
+      "grad_norm": 1.8940315246582031,
+      "learning_rate": 9.961338888845725e-05,
+      "loss": 2.9194,
+      "step": 340
+    },
+    {
+      "epoch": 0.177319587628866,
+      "grad_norm": 1.9344205856323242,
+      "learning_rate": 9.959194377953447e-05,
+      "loss": 2.9884,
+      "step": 344
+    },
+    {
+      "epoch": 0.17938144329896907,
+      "grad_norm": 1.9542304277420044,
+      "learning_rate": 9.956992227617995e-05,
+      "loss": 2.9051,
+      "step": 348
+    },
+    {
+      "epoch": 0.18144329896907216,
+      "grad_norm": 1.9588041305541992,
+      "learning_rate": 9.954732463434402e-05,
+      "loss": 2.7897,
+      "step": 352
+    },
+    {
+      "epoch": 0.18350515463917524,
+      "grad_norm": 1.7251778841018677,
+      "learning_rate": 9.952415111667324e-05,
+      "loss": 2.7511,
+      "step": 356
+    },
+    {
+      "epoch": 0.18556701030927836,
+      "grad_norm": 1.930525302886963,
+      "learning_rate": 9.950040199250746e-05,
+      "loss": 2.8544,
+      "step": 360
+    },
+    {
+      "epoch": 0.18762886597938144,
+      "grad_norm": 2.0370941162109375,
+      "learning_rate": 9.947607753787667e-05,
+      "loss": 3.0097,
+      "step": 364
+    },
+    {
+      "epoch": 0.18969072164948453,
+      "grad_norm": 2.0568158626556396,
+      "learning_rate": 9.945117803549774e-05,
+      "loss": 2.8412,
+      "step": 368
+    },
+    {
+      "epoch": 0.19175257731958764,
+      "grad_norm": 1.9520659446716309,
+      "learning_rate": 9.942570377477121e-05,
+      "loss": 2.8904,
+      "step": 372
+    },
+    {
+      "epoch": 0.19381443298969073,
+      "grad_norm": 1.9777973890304565,
+      "learning_rate": 9.939965505177786e-05,
+      "loss": 2.9093,
+      "step": 376
+    },
+    {
+      "epoch": 0.1958762886597938,
+      "grad_norm": 2.0709102153778076,
+      "learning_rate": 9.937303216927534e-05,
+      "loss": 2.7659,
+      "step": 380
+    },
+    {
+      "epoch": 0.1979381443298969,
+      "grad_norm": 1.919057846069336,
+      "learning_rate": 9.934583543669453e-05,
+      "loss": 2.8857,
+      "step": 384
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9447135925292969,
+      "learning_rate": 9.931806517013612e-05,
+      "loss": 2.7969,
+      "step": 388
+    },
+    {
+      "epoch": 0.2020618556701031,
+      "grad_norm": 1.9707754850387573,
+      "learning_rate": 9.928972169236676e-05,
+      "loss": 2.9706,
+      "step": 392
+    },
+    {
+      "epoch": 0.20412371134020618,
+      "grad_norm": 2.1095566749572754,
+      "learning_rate": 9.926080533281543e-05,
+      "loss": 2.8079,
+      "step": 396
+    },
+    {
+      "epoch": 0.20618556701030927,
+      "grad_norm": 1.9369230270385742,
+      "learning_rate": 9.923131642756954e-05,
+      "loss": 2.759,
+      "step": 400
+    },
+    {
+      "epoch": 0.20824742268041238,
+      "grad_norm": 1.8451541662216187,
+      "learning_rate": 9.920125531937107e-05,
+      "loss": 2.8428,
+      "step": 404
+    },
+    {
+      "epoch": 0.21030927835051547,
+      "grad_norm": 1.8733794689178467,
+      "learning_rate": 9.917062235761259e-05,
+      "loss": 2.7717,
+      "step": 408
+    },
+    {
+      "epoch": 0.21237113402061855,
+      "grad_norm": 1.8459867238998413,
+      "learning_rate": 9.913941789833311e-05,
+      "loss": 2.8008,
+      "step": 412
+    },
+    {
+      "epoch": 0.21443298969072164,
+      "grad_norm": 1.8867549896240234,
+      "learning_rate": 9.910764230421406e-05,
+      "loss": 2.7071,
+      "step": 416
+    },
+    {
+      "epoch": 0.21649484536082475,
+      "grad_norm": 1.8722797632217407,
+      "learning_rate": 9.907529594457504e-05,
+      "loss": 2.8076,
+      "step": 420
+    },
+    {
+      "epoch": 0.21855670103092784,
+      "grad_norm": 1.9473061561584473,
+      "learning_rate": 9.904237919536945e-05,
+      "loss": 2.8613,
+      "step": 424
+    },
+    {
+      "epoch": 0.22061855670103092,
+      "grad_norm": 1.8926331996917725,
+      "learning_rate": 9.900889243918024e-05,
+      "loss": 2.7264,
+      "step": 428
+    },
+    {
+      "epoch": 0.22268041237113403,
+      "grad_norm": 2.013500928878784,
+      "learning_rate": 9.897483606521536e-05,
+      "loss": 2.7307,
+      "step": 432
+    },
+    {
+      "epoch": 0.22474226804123712,
+      "grad_norm": 1.7798408269882202,
+      "learning_rate": 9.894021046930333e-05,
+      "loss": 2.7089,
+      "step": 436
+    },
+    {
+      "epoch": 0.2268041237113402,
+      "grad_norm": 1.8650320768356323,
+      "learning_rate": 9.890501605388853e-05,
+      "loss": 2.812,
+      "step": 440
+    },
+    {
+      "epoch": 0.2288659793814433,
+      "grad_norm": 1.8471648693084717,
+      "learning_rate": 9.886925322802663e-05,
+      "loss": 2.8819,
+      "step": 444
+    },
+    {
+      "epoch": 0.2309278350515464,
+      "grad_norm": 1.7936201095581055,
+      "learning_rate": 9.883292240737978e-05,
+      "loss": 2.8319,
+      "step": 448
+    },
+    {
+      "epoch": 0.2329896907216495,
+      "grad_norm": 1.89383065700531,
+      "learning_rate": 9.87960240142118e-05,
+      "loss": 2.7643,
+      "step": 452
+    },
+    {
+      "epoch": 0.23505154639175257,
+      "grad_norm": 1.7932506799697876,
+      "learning_rate": 9.875855847738319e-05,
+      "loss": 2.7604,
+      "step": 456
+    },
+    {
+      "epoch": 0.23711340206185566,
+      "grad_norm": 1.8939586877822876,
+      "learning_rate": 9.872052623234632e-05,
+      "loss": 2.6743,
+      "step": 460
+    },
+    {
+      "epoch": 0.23917525773195877,
+      "grad_norm": 1.8440996408462524,
+      "learning_rate": 9.868192772114016e-05,
+      "loss": 2.6986,
+      "step": 464
+    },
+    {
+      "epoch": 0.24123711340206186,
+      "grad_norm": 2.047183036804199,
+      "learning_rate": 9.864276339238534e-05,
+      "loss": 2.7008,
+      "step": 468
+    },
+    {
+      "epoch": 0.24329896907216494,
+      "grad_norm": 1.7253282070159912,
+      "learning_rate": 9.860303370127876e-05,
+      "loss": 2.7944,
+      "step": 472
+    },
+    {
+      "epoch": 0.24536082474226803,
+      "grad_norm": 1.8298146724700928,
+      "learning_rate": 9.856273910958847e-05,
+      "loss": 2.637,
+      "step": 476
+    },
+    {
+      "epoch": 0.24742268041237114,
+      "grad_norm": 1.7853765487670898,
+      "learning_rate": 9.852188008564813e-05,
+      "loss": 2.7162,
+      "step": 480
+    },
+    {
+      "epoch": 0.24948453608247423,
+      "grad_norm": 1.8834904432296753,
+      "learning_rate": 9.84804571043517e-05,
+      "loss": 2.8762,
+      "step": 484
+    },
+    {
+      "epoch": 0.2515463917525773,
+      "grad_norm": 1.9652595520019531,
+      "learning_rate": 9.843847064714785e-05,
+      "loss": 2.7794,
+      "step": 488
+    },
+    {
+      "epoch": 0.2536082474226804,
+      "grad_norm": 1.983886957168579,
+      "learning_rate": 9.839592120203441e-05,
+      "loss": 2.8331,
+      "step": 492
+    },
+    {
+      "epoch": 0.2556701030927835,
+      "grad_norm": 1.9896190166473389,
+      "learning_rate": 9.835280926355261e-05,
+      "loss": 2.8015,
+      "step": 496
+    },
+    {
+      "epoch": 0.25773195876288657,
+      "grad_norm": 1.70455002784729,
+      "learning_rate": 9.83091353327815e-05,
+      "loss": 2.7135,
+      "step": 500
+    },
+    {
+      "epoch": 0.2597938144329897,
+      "grad_norm": 2.0131478309631348,
+      "learning_rate": 9.826489991733194e-05,
+      "loss": 2.7678,
+      "step": 504
+    },
+    {
+      "epoch": 0.2618556701030928,
+      "grad_norm": 2.189298152923584,
+      "learning_rate": 9.822010353134081e-05,
+      "loss": 2.8352,
+      "step": 508
+    },
+    {
+      "epoch": 0.2639175257731959,
+      "grad_norm": 2.022023916244507,
+      "learning_rate": 9.817474669546501e-05,
+      "loss": 2.7118,
+      "step": 512
+    },
+    {
+      "epoch": 0.26597938144329897,
+      "grad_norm": 2.1792008876800537,
+      "learning_rate": 9.812882993687539e-05,
+      "loss": 2.8367,
+      "step": 516
+    },
+    {
+      "epoch": 0.26804123711340205,
+      "grad_norm": 1.8317099809646606,
+      "learning_rate": 9.808235378925066e-05,
+      "loss": 2.7326,
+      "step": 520
+    },
+    {
+      "epoch": 0.27010309278350514,
+      "grad_norm": 1.7824715375900269,
+      "learning_rate": 9.803531879277113e-05,
+      "loss": 2.7147,
+      "step": 524
+    },
+    {
+      "epoch": 0.2721649484536082,
+      "grad_norm": 1.7565789222717285,
+      "learning_rate": 9.798772549411252e-05,
+      "loss": 2.8029,
+      "step": 528
+    },
+    {
+      "epoch": 0.27422680412371137,
+      "grad_norm": 1.805213212966919,
+      "learning_rate": 9.793957444643951e-05,
+      "loss": 2.6505,
+      "step": 532
+    },
+    {
+      "epoch": 0.27628865979381445,
+      "grad_norm": 1.8401093482971191,
+      "learning_rate": 9.789086620939936e-05,
+      "loss": 2.7262,
+      "step": 536
+    },
+    {
+      "epoch": 0.27835051546391754,
+      "grad_norm": 1.7650929689407349,
+      "learning_rate": 9.784160134911541e-05,
+      "loss": 2.6988,
+      "step": 540
+    },
+    {
+      "epoch": 0.2804123711340206,
+      "grad_norm": 1.902664065361023,
+      "learning_rate": 9.77917804381805e-05,
+      "loss": 2.7028,
+      "step": 544
+    },
+    {
+      "epoch": 0.2824742268041237,
+      "grad_norm": 1.6840468645095825,
+      "learning_rate": 9.774140405565024e-05,
+      "loss": 2.7141,
+      "step": 548
+    },
+    {
+      "epoch": 0.2845360824742268,
+      "grad_norm": 1.8542104959487915,
+      "learning_rate": 9.769047278703644e-05,
+      "loss": 2.7512,
+      "step": 552
+    },
+    {
+      "epoch": 0.2865979381443299,
+      "grad_norm": 1.9419035911560059,
+      "learning_rate": 9.763898722430015e-05,
+      "loss": 2.7085,
+      "step": 556
+    },
+    {
+      "epoch": 0.28865979381443296,
+      "grad_norm": 1.8914318084716797,
+      "learning_rate": 9.758694796584483e-05,
+      "loss": 2.7119,
+      "step": 560
+    },
+    {
+      "epoch": 0.2907216494845361,
+      "grad_norm": 2.028513193130493,
+      "learning_rate": 9.753435561650946e-05,
+      "loss": 2.7293,
+      "step": 564
+    },
+    {
+      "epoch": 0.2927835051546392,
+      "grad_norm": 1.7596354484558105,
+      "learning_rate": 9.748121078756137e-05,
+      "loss": 2.6624,
+      "step": 568
+    },
+    {
+      "epoch": 0.2948453608247423,
+      "grad_norm": 1.955361247062683,
+      "learning_rate": 9.742751409668929e-05,
+      "loss": 2.6958,
+      "step": 572
+    },
+    {
+      "epoch": 0.29690721649484536,
+      "grad_norm": 1.980281949043274,
+      "learning_rate": 9.737326616799605e-05,
+      "loss": 2.6921,
+      "step": 576
+    },
+    {
+      "epoch": 0.29896907216494845,
+      "grad_norm": 2.0537736415863037,
+      "learning_rate": 9.731846763199144e-05,
+      "loss": 2.7734,
+      "step": 580
+    },
+    {
+      "epoch": 0.30103092783505153,
+      "grad_norm": 2.0544698238372803,
+      "learning_rate": 9.726311912558474e-05,
+      "loss": 2.622,
+      "step": 584
+    },
+    {
+      "epoch": 0.3030927835051546,
+      "grad_norm": 1.8842084407806396,
+      "learning_rate": 9.720722129207746e-05,
+      "loss": 2.6809,
+      "step": 588
+    },
+    {
+      "epoch": 0.30515463917525776,
+      "grad_norm": 1.9250359535217285,
+      "learning_rate": 9.715077478115574e-05,
+      "loss": 2.7359,
+      "step": 592
+    },
+    {
+      "epoch": 0.30721649484536084,
+      "grad_norm": 1.7907671928405762,
+      "learning_rate": 9.709378024888292e-05,
+      "loss": 2.7575,
+      "step": 596
+    },
+    {
+      "epoch": 0.30927835051546393,
+      "grad_norm": 1.8203668594360352,
+      "learning_rate": 9.703623835769178e-05,
+      "loss": 2.7181,
+      "step": 600
+    },
+    {
+      "epoch": 0.311340206185567,
+      "grad_norm": 1.837925672531128,
+      "learning_rate": 9.697814977637696e-05,
+      "loss": 2.743,
+      "step": 604
+    },
+    {
+      "epoch": 0.3134020618556701,
+      "grad_norm": 1.7200803756713867,
+      "learning_rate": 9.691951518008715e-05,
+      "loss": 2.6844,
+      "step": 608
+    },
+    {
+      "epoch": 0.3154639175257732,
+      "grad_norm": 1.9397820234298706,
+      "learning_rate": 9.686033525031719e-05,
+      "loss": 2.5767,
+      "step": 612
+    },
+    {
+      "epoch": 0.31752577319587627,
+      "grad_norm": 1.9038435220718384,
+      "learning_rate": 9.680061067490021e-05,
+      "loss": 2.6503,
+      "step": 616
+    },
+    {
+      "epoch": 0.31958762886597936,
+      "grad_norm": 1.9289252758026123,
+      "learning_rate": 9.674034214799964e-05,
+      "loss": 2.6878,
+      "step": 620
+    },
+    {
+      "epoch": 0.3216494845360825,
+      "grad_norm": 1.7512463331222534,
+      "learning_rate": 9.667953037010108e-05,
+      "loss": 2.7551,
+      "step": 624
+    },
+    {
+      "epoch": 0.3237113402061856,
+      "grad_norm": 1.9182549715042114,
+      "learning_rate": 9.661817604800421e-05,
+      "loss": 2.7986,
+      "step": 628
+    },
+    {
+      "epoch": 0.32577319587628867,
+      "grad_norm": 1.8200740814208984,
+      "learning_rate": 9.655627989481458e-05,
+      "loss": 2.6923,
+      "step": 632
+    },
+    {
+      "epoch": 0.32783505154639175,
+      "grad_norm": 1.8738125562667847,
+      "learning_rate": 9.649384262993525e-05,
+      "loss": 2.7518,
+      "step": 636
+    },
+    {
+      "epoch": 0.32989690721649484,
+      "grad_norm": 1.9261345863342285,
+      "learning_rate": 9.64308649790586e-05,
+      "loss": 2.6037,
+      "step": 640
+    },
+    {
+      "epoch": 0.3319587628865979,
+      "grad_norm": 1.7359293699264526,
+      "learning_rate": 9.636734767415763e-05,
+      "loss": 2.7102,
+      "step": 644
+    },
+    {
+      "epoch": 0.334020618556701,
+      "grad_norm": 2.001077175140381,
+      "learning_rate": 9.630329145347767e-05,
+      "loss": 2.7068,
+      "step": 648
+    },
+    {
+      "epoch": 0.33608247422680415,
+      "grad_norm": 1.7505935430526733,
+      "learning_rate": 9.623869706152777e-05,
+      "loss": 2.6348,
+      "step": 652
+    },
+    {
+      "epoch": 0.33814432989690724,
+      "grad_norm": 1.9153965711593628,
+      "learning_rate": 9.617356524907193e-05,
+      "loss": 2.6874,
+      "step": 656
+    },
+    {
+      "epoch": 0.3402061855670103,
+      "grad_norm": 1.805038332939148,
+      "learning_rate": 9.61078967731205e-05,
+      "loss": 2.5985,
+      "step": 660
+    },
+    {
+      "epoch": 0.3422680412371134,
+      "grad_norm": 1.7926819324493408,
+      "learning_rate": 9.604169239692133e-05,
+      "loss": 2.7203,
+      "step": 664
+    },
+    {
+      "epoch": 0.3443298969072165,
+      "grad_norm": 1.8701540231704712,
+      "learning_rate": 9.597495288995089e-05,
+      "loss": 2.6819,
+      "step": 668
+    },
+    {
+      "epoch": 0.3463917525773196,
+      "grad_norm": 2.0685527324676514,
+      "learning_rate": 9.590767902790529e-05,
+      "loss": 2.7401,
+      "step": 672
+    },
+    {
+      "epoch": 0.34845360824742266,
+      "grad_norm": 1.7549872398376465,
+      "learning_rate": 9.583987159269143e-05,
+      "loss": 2.5833,
+      "step": 676
+    },
+    {
+      "epoch": 0.35051546391752575,
+      "grad_norm": 1.8161414861679077,
+      "learning_rate": 9.577153137241765e-05,
+      "loss": 2.7274,
+      "step": 680
+    },
+    {
+      "epoch": 0.3525773195876289,
+      "grad_norm": 1.7566990852355957,
+      "learning_rate": 9.570265916138484e-05,
+      "loss": 2.7536,
+      "step": 684
+    },
+    {
+      "epoch": 0.354639175257732,
+      "grad_norm": 1.8565185070037842,
+      "learning_rate": 9.563325576007701e-05,
+      "loss": 2.6785,
+      "step": 688
+    },
+    {
+      "epoch": 0.35670103092783506,
+      "grad_norm": 1.7326369285583496,
+      "learning_rate": 9.556332197515207e-05,
+      "loss": 2.7096,
+      "step": 692
+    },
+    {
+      "epoch": 0.35876288659793815,
+      "grad_norm": 1.705234408378601,
+      "learning_rate": 9.549285861943247e-05,
+      "loss": 2.7314,
+      "step": 696
+    },
+    {
+      "epoch": 0.36082474226804123,
+      "grad_norm": 1.9541505575180054,
+      "learning_rate": 9.542186651189569e-05,
+      "loss": 2.6204,
+      "step": 700
+    },
+    {
+      "epoch": 0.3628865979381443,
+      "grad_norm": 1.8346929550170898,
+      "learning_rate": 9.535034647766476e-05,
+      "loss": 2.6913,
+      "step": 704
+    },
+    {
+      "epoch": 0.3649484536082474,
+      "grad_norm": 1.9132497310638428,
+      "learning_rate": 9.527829934799869e-05,
+      "loss": 2.7041,
+      "step": 708
+    },
+    {
+      "epoch": 0.3670103092783505,
+      "grad_norm": 1.763213038444519,
+      "learning_rate": 9.520572596028278e-05,
+      "loss": 2.6216,
+      "step": 712
+    },
+    {
+      "epoch": 0.36907216494845363,
+      "grad_norm": 1.9224257469177246,
+      "learning_rate": 9.513262715801887e-05,
+      "loss": 2.6632,
+      "step": 716
+    },
+    {
+      "epoch": 0.3711340206185567,
+      "grad_norm": 1.8817148208618164,
+      "learning_rate": 9.505900379081559e-05,
+      "loss": 2.6868,
+      "step": 720
+    },
+    {
+      "epoch": 0.3731958762886598,
+      "grad_norm": 1.7395341396331787,
+      "learning_rate": 9.498485671437842e-05,
+      "loss": 2.5171,
+      "step": 724
+    },
+    {
+      "epoch": 0.3752577319587629,
+      "grad_norm": 1.7571380138397217,
+      "learning_rate": 9.491018679049981e-05,
+      "loss": 2.6194,
+      "step": 728
+    },
+    {
+      "epoch": 0.37731958762886597,
+      "grad_norm": 1.8208237886428833,
+      "learning_rate": 9.48349948870491e-05,
+      "loss": 2.6101,
+      "step": 732
+    },
+    {
+      "epoch": 0.37938144329896906,
+      "grad_norm": 1.888909935951233,
+      "learning_rate": 9.47592818779625e-05,
+      "loss": 2.6569,
+      "step": 736
+    },
+    {
+      "epoch": 0.38144329896907214,
+      "grad_norm": 1.7806968688964844,
+      "learning_rate": 9.468304864323288e-05,
+      "loss": 2.5329,
+      "step": 740
+    },
+    {
+      "epoch": 0.3835051546391753,
+      "grad_norm": 1.8010936975479126,
+      "learning_rate": 9.460629606889952e-05,
+      "loss": 2.6971,
+      "step": 744
+    },
+    {
+      "epoch": 0.38556701030927837,
+      "grad_norm": 1.7562525272369385,
+      "learning_rate": 9.452902504703793e-05,
+      "loss": 2.676,
+      "step": 748
+    },
+    {
+      "epoch": 0.38762886597938145,
+      "grad_norm": 1.828466534614563,
+      "learning_rate": 9.445123647574936e-05,
+      "loss": 2.5952,
+      "step": 752
+    },
+    {
+      "epoch": 0.38969072164948454,
+      "grad_norm": 1.7644858360290527,
+      "learning_rate": 9.437293125915037e-05,
+      "loss": 2.6475,
+      "step": 756
+    },
+    {
+      "epoch": 0.3917525773195876,
+      "grad_norm": 1.6723216772079468,
+      "learning_rate": 9.429411030736242e-05,
+      "loss": 2.6823,
+      "step": 760
+    },
+    {
+      "epoch": 0.3938144329896907,
+      "grad_norm": 1.7757784128189087,
+      "learning_rate": 9.421477453650118e-05,
+      "loss": 2.5711,
+      "step": 764
+    },
+    {
+      "epoch": 0.3958762886597938,
+      "grad_norm": 1.647575855255127,
+      "learning_rate": 9.413492486866598e-05,
+      "loss": 2.6141,
+      "step": 768
+    },
+    {
+      "epoch": 0.3979381443298969,
+      "grad_norm": 1.9347645044326782,
+      "learning_rate": 9.405456223192897e-05,
+      "loss": 2.6708,
+      "step": 772
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.838582158088684,
+      "learning_rate": 9.397368756032445e-05,
+      "loss": 2.7496,
+      "step": 776
+    },
+    {
+      "epoch": 0.4020618556701031,
+      "grad_norm": 1.8382269144058228,
+      "learning_rate": 9.389230179383801e-05,
+      "loss": 2.5654,
+      "step": 780
+    },
+    {
+      "epoch": 0.4041237113402062,
+      "grad_norm": 1.8177680969238281,
+      "learning_rate": 9.381040587839548e-05,
+      "loss": 2.6553,
+      "step": 784
+    },
+    {
+      "epoch": 0.4061855670103093,
+      "grad_norm": 1.7643057107925415,
+      "learning_rate": 9.372800076585207e-05,
+      "loss": 2.5951,
+      "step": 788
+    },
+    {
+      "epoch": 0.40824742268041236,
+      "grad_norm": 1.8181430101394653,
+      "learning_rate": 9.364508741398127e-05,
+      "loss": 2.6452,
+      "step": 792
+    },
+    {
+      "epoch": 0.41030927835051545,
+      "grad_norm": 1.8445159196853638,
+      "learning_rate": 9.356166678646366e-05,
+      "loss": 2.6172,
+      "step": 796
+    },
+    {
+      "epoch": 0.41237113402061853,
+      "grad_norm": 1.6617608070373535,
+      "learning_rate": 9.347773985287578e-05,
+      "loss": 2.5781,
+      "step": 800
+    },
+    {
+      "epoch": 0.4144329896907217,
+      "grad_norm": 1.9522396326065063,
+      "learning_rate": 9.339330758867883e-05,
+      "loss": 2.6309,
+      "step": 804
+    },
+    {
+      "epoch": 0.41649484536082476,
+      "grad_norm": 2.5656485557556152,
+      "learning_rate": 9.330837097520738e-05,
+      "loss": 2.6967,
+      "step": 808
+    },
+    {
+      "epoch": 0.41855670103092785,
+      "grad_norm": 1.6885048151016235,
+      "learning_rate": 9.322293099965784e-05,
+      "loss": 2.617,
+      "step": 812
+    },
+    {
+      "epoch": 0.42061855670103093,
+      "grad_norm": 1.7637752294540405,
+      "learning_rate": 9.313698865507713e-05,
+      "loss": 2.6538,
+      "step": 816
+    },
+    {
+      "epoch": 0.422680412371134,
+      "grad_norm": 1.7737189531326294,
+      "learning_rate": 9.305054494035106e-05,
+      "loss": 2.6113,
+      "step": 820
+    },
+    {
+      "epoch": 0.4247422680412371,
+      "grad_norm": 1.8070309162139893,
+      "learning_rate": 9.296360086019272e-05,
+      "loss": 2.5118,
+      "step": 824
+    },
+    {
+      "epoch": 0.4268041237113402,
+      "grad_norm": 1.73598313331604,
+      "learning_rate": 9.287615742513086e-05,
+      "loss": 2.5826,
+      "step": 828
+    },
+    {
+      "epoch": 0.4288659793814433,
+      "grad_norm": 1.688819169998169,
+      "learning_rate": 9.278821565149806e-05,
+      "loss": 2.5375,
+      "step": 832
+    },
+    {
+      "epoch": 0.4309278350515464,
+      "grad_norm": 1.7314485311508179,
+      "learning_rate": 9.269977656141898e-05,
+      "loss": 2.5819,
+      "step": 836
+    },
+    {
+      "epoch": 0.4329896907216495,
+      "grad_norm": 1.657240390777588,
+      "learning_rate": 9.261084118279847e-05,
+      "loss": 2.6643,
+      "step": 840
+    },
+    {
+      "epoch": 0.4350515463917526,
+      "grad_norm": 1.7966091632843018,
+      "learning_rate": 9.25214105493096e-05,
+      "loss": 2.7022,
+      "step": 844
+    },
+    {
+      "epoch": 0.43711340206185567,
+      "grad_norm": 1.732354760169983,
+      "learning_rate": 9.243148570038164e-05,
+      "loss": 2.6098,
+      "step": 848
+    },
+    {
+      "epoch": 0.43917525773195876,
+      "grad_norm": 1.737281322479248,
+      "learning_rate": 9.234106768118809e-05,
+      "loss": 2.5879,
+      "step": 852
+    },
+    {
+      "epoch": 0.44123711340206184,
+      "grad_norm": 1.7843495607376099,
+      "learning_rate": 9.225015754263431e-05,
+      "loss": 2.5793,
+      "step": 856
+    },
+    {
+      "epoch": 0.44329896907216493,
+      "grad_norm": 1.7821077108383179,
+      "learning_rate": 9.215875634134552e-05,
+      "loss": 2.6578,
+      "step": 860
+    },
+    {
+      "epoch": 0.44536082474226807,
+      "grad_norm": 1.8063935041427612,
+      "learning_rate": 9.206686513965445e-05,
+      "loss": 2.7075,
+      "step": 864
+    },
+    {
+      "epoch": 0.44742268041237115,
+      "grad_norm": 1.5605168342590332,
+      "learning_rate": 9.19744850055889e-05,
+      "loss": 2.5988,
+      "step": 868
+    },
+    {
+      "epoch": 0.44948453608247424,
+      "grad_norm": 2.078798294067383,
+      "learning_rate": 9.188161701285949e-05,
+      "loss": 2.6126,
+      "step": 872
+    },
+    {
+      "epoch": 0.4515463917525773,
+      "grad_norm": 1.7260762453079224,
+      "learning_rate": 9.178826224084705e-05,
+      "loss": 2.6352,
+      "step": 876
+    },
+    {
+      "epoch": 0.4536082474226804,
+      "grad_norm": 1.7088130712509155,
+      "learning_rate": 9.169442177459011e-05,
+      "loss": 2.6324,
+      "step": 880
+    },
+    {
+      "epoch": 0.4556701030927835,
+      "grad_norm": 1.8762470483779907,
+      "learning_rate": 9.160009670477234e-05,
+      "loss": 2.517,
+      "step": 884
+    },
+    {
+      "epoch": 0.4577319587628866,
+      "grad_norm": 1.9028972387313843,
+      "learning_rate": 9.150528812770981e-05,
+      "loss": 2.5386,
+      "step": 888
+    },
+    {
+      "epoch": 0.45979381443298967,
+      "grad_norm": 1.5713077783584595,
+      "learning_rate": 9.140999714533827e-05,
+      "loss": 2.6347,
+      "step": 892
+    },
+    {
+      "epoch": 0.4618556701030928,
+      "grad_norm": 1.603164792060852,
+      "learning_rate": 9.131422486520034e-05,
+      "loss": 2.5138,
+      "step": 896
+    },
+    {
+      "epoch": 0.4639175257731959,
+      "grad_norm": 1.7028288841247559,
+      "learning_rate": 9.121797240043267e-05,
+      "loss": 2.5366,
+      "step": 900
+    },
+    {
+      "epoch": 0.465979381443299,
+      "grad_norm": 1.7099529504776,
+      "learning_rate": 9.11212408697529e-05,
+      "loss": 2.6465,
+      "step": 904
+    },
+    {
+      "epoch": 0.46804123711340206,
+      "grad_norm": 1.7134875059127808,
+      "learning_rate": 9.102403139744683e-05,
+      "loss": 2.6331,
+      "step": 908
+    },
+    {
+      "epoch": 0.47010309278350515,
+      "grad_norm": 1.6669530868530273,
+      "learning_rate": 9.092634511335519e-05,
+      "loss": 2.5381,
+      "step": 912
+    },
+    {
+      "epoch": 0.47216494845360824,
+      "grad_norm": 1.7473549842834473,
+      "learning_rate": 9.082818315286055e-05,
+      "loss": 2.584,
+      "step": 916
+    },
+    {
+      "epoch": 0.4742268041237113,
+      "grad_norm": 1.673309087753296,
+      "learning_rate": 9.07295466568742e-05,
+      "loss": 2.4649,
+      "step": 920
+    },
+    {
+      "epoch": 0.4762886597938144,
+      "grad_norm": 1.8044261932373047,
+      "learning_rate": 9.063043677182283e-05,
+      "loss": 2.6512,
+      "step": 924
+    },
+    {
+      "epoch": 0.47835051546391755,
+      "grad_norm": 1.7258899211883545,
+      "learning_rate": 9.053085464963518e-05,
+      "loss": 2.5977,
+      "step": 928
+    },
+    {
+      "epoch": 0.48041237113402063,
+      "grad_norm": 1.7187895774841309,
+      "learning_rate": 9.043080144772868e-05,
+      "loss": 2.5948,
+      "step": 932
+    },
+    {
+      "epoch": 0.4824742268041237,
+      "grad_norm": 1.797176718711853,
+      "learning_rate": 9.033027832899601e-05,
+      "loss": 2.659,
+      "step": 936
+    },
+    {
+      "epoch": 0.4845360824742268,
+      "grad_norm": 1.6816799640655518,
+      "learning_rate": 9.022928646179159e-05,
+      "loss": 2.6123,
+      "step": 940
+    },
+    {
+      "epoch": 0.4865979381443299,
+      "grad_norm": 1.7294647693634033,
+      "learning_rate": 9.012782701991795e-05,
+      "loss": 2.5824,
+      "step": 944
+    },
+    {
+      "epoch": 0.488659793814433,
+      "grad_norm": 1.6819491386413574,
+      "learning_rate": 9.002590118261216e-05,
+      "loss": 2.5717,
+      "step": 948
+    },
+    {
+      "epoch": 0.49072164948453606,
+      "grad_norm": 1.667133092880249,
+      "learning_rate": 8.992351013453204e-05,
+      "loss": 2.6282,
+      "step": 952
+    },
+    {
+      "epoch": 0.4927835051546392,
+      "grad_norm": 1.6454589366912842,
+      "learning_rate": 8.982065506574247e-05,
+      "loss": 2.5939,
+      "step": 956
+    },
+    {
+      "epoch": 0.4948453608247423,
+      "grad_norm": 1.8332362174987793,
+      "learning_rate": 8.971733717170148e-05,
+      "loss": 2.4949,
+      "step": 960
+    },
+    {
+      "epoch": 0.49690721649484537,
+      "grad_norm": 6.3988037109375,
+      "learning_rate": 8.961355765324648e-05,
+      "loss": 2.6143,
+      "step": 964
+    },
+    {
+      "epoch": 0.49896907216494846,
+      "grad_norm": 1.6663529872894287,
+      "learning_rate": 8.950931771658014e-05,
+      "loss": 2.5308,
+      "step": 968
+    },
+    {
+      "epoch": 0.5010309278350515,
+      "grad_norm": 1.6272313594818115,
+      "learning_rate": 8.940461857325647e-05,
+      "loss": 2.4458,
+      "step": 972
+    },
+    {
+      "epoch": 0.5030927835051546,
+      "grad_norm": 1.758596658706665,
+      "learning_rate": 8.929946144016677e-05,
+      "loss": 2.517,
+      "step": 976
+    },
+    {
+      "epoch": 0.5051546391752577,
+      "grad_norm": 1.6491873264312744,
+      "learning_rate": 8.919384753952538e-05,
+      "loss": 2.5408,
+      "step": 980
+    },
+    {
+      "epoch": 0.5072164948453608,
+      "grad_norm": 1.5984055995941162,
+      "learning_rate": 8.908777809885557e-05,
+      "loss": 2.5359,
+      "step": 984
+    },
+    {
+      "epoch": 0.5092783505154639,
+      "grad_norm": 1.651795744895935,
+      "learning_rate": 8.898125435097521e-05,
+      "loss": 2.5151,
+      "step": 988
+    },
+    {
+      "epoch": 0.511340206185567,
+      "grad_norm": 1.6190180778503418,
+      "learning_rate": 8.887427753398248e-05,
+      "loss": 2.6733,
+      "step": 992
+    },
+    {
+      "epoch": 0.51340206185567,
+      "grad_norm": 1.6881654262542725,
+      "learning_rate": 8.876684889124145e-05,
+      "loss": 2.5783,
+      "step": 996
+    },
+    {
+      "epoch": 0.5154639175257731,
+      "grad_norm": 1.786693811416626,
+      "learning_rate": 8.865896967136766e-05,
+      "loss": 2.6061,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5175257731958763,
+      "grad_norm": 1.841579556465149,
+      "learning_rate": 8.855064112821361e-05,
+      "loss": 2.6615,
+      "step": 1004
+    },
+    {
+      "epoch": 0.5195876288659794,
+      "grad_norm": 1.5715938806533813,
+      "learning_rate": 8.844186452085412e-05,
+      "loss": 2.6044,
+      "step": 1008
+    },
+    {
+      "epoch": 0.5216494845360825,
+      "grad_norm": 1.616853952407837,
+      "learning_rate": 8.83326411135718e-05,
+      "loss": 2.538,
+      "step": 1012
+    },
+    {
+      "epoch": 0.5237113402061856,
+      "grad_norm": 1.7380198240280151,
+      "learning_rate": 8.822297217584225e-05,
+      "loss": 2.5838,
+      "step": 1016
+    },
+    {
+      "epoch": 0.5257731958762887,
+      "grad_norm": 1.703635334968567,
+      "learning_rate": 8.81128589823194e-05,
+      "loss": 2.5035,
+      "step": 1020
+    },
+    {
+      "epoch": 0.5278350515463918,
+      "grad_norm": 1.9964070320129395,
+      "learning_rate": 8.80023028128206e-05,
+      "loss": 2.544,
+      "step": 1024
+    },
+    {
+      "epoch": 0.5298969072164949,
+      "grad_norm": 1.5779422521591187,
+      "learning_rate": 8.789130495231186e-05,
+      "loss": 2.6569,
+      "step": 1028
+    },
+    {
+      "epoch": 0.5319587628865979,
+      "grad_norm": 1.771633505821228,
+      "learning_rate": 8.77798666908928e-05,
+      "loss": 2.5315,
+      "step": 1032
+    },
+    {
+      "epoch": 0.534020618556701,
+      "grad_norm": 1.7763731479644775,
+      "learning_rate": 8.766798932378172e-05,
+      "loss": 2.5138,
+      "step": 1036
+    },
+    {
+      "epoch": 0.5360824742268041,
+      "grad_norm": 1.699804425239563,
+      "learning_rate": 8.755567415130058e-05,
+      "loss": 2.5168,
+      "step": 1040
+    },
+    {
+      "epoch": 0.5381443298969072,
+      "grad_norm": 1.8998210430145264,
+      "learning_rate": 8.744292247885975e-05,
+      "loss": 2.4795,
+      "step": 1044
+    },
+    {
+      "epoch": 0.5402061855670103,
+      "grad_norm": 1.7309194803237915,
+      "learning_rate": 8.732973561694297e-05,
+      "loss": 2.5866,
+      "step": 1048
+    },
+    {
+      "epoch": 0.5422680412371134,
+      "grad_norm": 1.667391061782837,
+      "learning_rate": 8.721611488109212e-05,
+      "loss": 2.5016,
+      "step": 1052
+    },
+    {
+      "epoch": 0.5443298969072164,
+      "grad_norm": 1.7690322399139404,
+      "learning_rate": 8.710206159189182e-05,
+      "loss": 2.6189,
+      "step": 1056
+    },
+    {
+      "epoch": 0.5463917525773195,
+      "grad_norm": 1.6563546657562256,
+      "learning_rate": 8.698757707495417e-05,
+      "loss": 2.4559,
+      "step": 1060
+    },
+    {
+      "epoch": 0.5484536082474227,
+      "grad_norm": 1.8325282335281372,
+      "learning_rate": 8.687266266090333e-05,
+      "loss": 2.546,
+      "step": 1064
+    },
+    {
+      "epoch": 0.5505154639175258,
+      "grad_norm": 1.6227245330810547,
+      "learning_rate": 8.675731968536002e-05,
+      "loss": 2.5709,
+      "step": 1068
+    },
+    {
+      "epoch": 0.5525773195876289,
+      "grad_norm": 1.6123030185699463,
+      "learning_rate": 8.664154948892607e-05,
+      "loss": 2.5047,
+      "step": 1072
+    },
+    {
+      "epoch": 0.554639175257732,
+      "grad_norm": 1.631011724472046,
+      "learning_rate": 8.65253534171687e-05,
+      "loss": 2.4568,
+      "step": 1076
+    },
+    {
+      "epoch": 0.5567010309278351,
+      "grad_norm": 1.746801495552063,
+      "learning_rate": 8.640873282060506e-05,
+      "loss": 2.5825,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5587628865979382,
+      "grad_norm": 1.727461338043213,
+      "learning_rate": 8.629168905468641e-05,
+      "loss": 2.5724,
+      "step": 1084
+    },
+    {
+      "epoch": 0.5608247422680412,
+      "grad_norm": 1.7059168815612793,
+      "learning_rate": 8.617422347978239e-05,
+      "loss": 2.5469,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5628865979381443,
+      "grad_norm": 1.6282740831375122,
+      "learning_rate": 8.605633746116519e-05,
+      "loss": 2.5503,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5649484536082474,
+      "grad_norm": 1.5870437622070312,
+      "learning_rate": 8.593803236899379e-05,
+      "loss": 2.5664,
+      "step": 1096
+    },
+    {
+      "epoch": 0.5670103092783505,
+      "grad_norm": 1.7030541896820068,
+      "learning_rate": 8.581930957829786e-05,
+      "loss": 2.4574,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5690721649484536,
+      "grad_norm": 1.6299339532852173,
+      "learning_rate": 8.570017046896197e-05,
+      "loss": 2.5025,
+      "step": 1104
+    },
+    {
+      "epoch": 0.5711340206185567,
+      "grad_norm": 1.648901104927063,
+      "learning_rate": 8.558061642570936e-05,
+      "loss": 2.5028,
+      "step": 1108
+    },
+    {
+      "epoch": 0.5731958762886598,
+      "grad_norm": 1.5893465280532837,
+      "learning_rate": 8.5460648838086e-05,
+      "loss": 2.4432,
+      "step": 1112
+    },
+    {
+      "epoch": 0.5752577319587628,
+      "grad_norm": 1.686439871788025,
+      "learning_rate": 8.534026910044435e-05,
+      "loss": 2.4683,
+      "step": 1116
+    },
+    {
+      "epoch": 0.5773195876288659,
+      "grad_norm": 1.687406301498413,
+      "learning_rate": 8.521947861192723e-05,
+      "loss": 2.4717,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5793814432989691,
+      "grad_norm": 1.7713639736175537,
+      "learning_rate": 8.509827877645144e-05,
+      "loss": 2.6082,
+      "step": 1124
+    },
+    {
+      "epoch": 0.5814432989690722,
+      "grad_norm": 1.7374292612075806,
+      "learning_rate": 8.49766710026916e-05,
+      "loss": 2.4267,
+      "step": 1128
+    },
+    {
+      "epoch": 0.5835051546391753,
+      "grad_norm": 1.5629856586456299,
+      "learning_rate": 8.48546567040636e-05,
+      "loss": 2.4698,
+      "step": 1132
+    },
+    {
+      "epoch": 0.5855670103092784,
+      "grad_norm": 1.6608953475952148,
+      "learning_rate": 8.473223729870834e-05,
+      "loss": 2.4221,
+      "step": 1136
+    },
+    {
+      "epoch": 0.5876288659793815,
+      "grad_norm": 1.6824475526809692,
+      "learning_rate": 8.460941420947514e-05,
+      "loss": 2.5411,
+      "step": 1140
+    },
+    {
+      "epoch": 0.5896907216494846,
+      "grad_norm": 1.590063214302063,
+      "learning_rate": 8.448618886390522e-05,
+      "loss": 2.4957,
+      "step": 1144
+    },
+    {
+      "epoch": 0.5917525773195876,
+      "grad_norm": 1.6329950094223022,
+      "learning_rate": 8.436256269421515e-05,
+      "loss": 2.489,
+      "step": 1148
+    },
+    {
+      "epoch": 0.5938144329896907,
+      "grad_norm": 1.7138408422470093,
+      "learning_rate": 8.423853713728016e-05,
+      "loss": 2.4355,
+      "step": 1152
+    },
+    {
+      "epoch": 0.5958762886597938,
+      "grad_norm": 1.707245111465454,
+      "learning_rate": 8.411411363461745e-05,
+      "loss": 2.4351,
+      "step": 1156
+    },
+    {
+      "epoch": 0.5979381443298969,
+      "grad_norm": 1.6063274145126343,
+      "learning_rate": 8.398929363236948e-05,
+      "loss": 2.4771,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.62711763381958,
+      "learning_rate": 8.386407858128706e-05,
+      "loss": 2.4136,
+      "step": 1164
+    },
+    {
+      "epoch": 0.6020618556701031,
+      "grad_norm": 1.7263373136520386,
+      "learning_rate": 8.373846993671261e-05,
+      "loss": 2.6003,
+      "step": 1168
+    },
+    {
+      "epoch": 0.6041237113402061,
+      "grad_norm": 1.6953507661819458,
+      "learning_rate": 8.361246915856314e-05,
+      "loss": 2.5567,
+      "step": 1172
+    },
+    {
+      "epoch": 0.6061855670103092,
+      "grad_norm": 1.6107468605041504,
+      "learning_rate": 8.348607771131336e-05,
+      "loss": 2.3889,
+      "step": 1176
+    },
+    {
+      "epoch": 0.6082474226804123,
+      "grad_norm": 1.5828678607940674,
+      "learning_rate": 8.335929706397863e-05,
+      "loss": 2.4063,
+      "step": 1180
+    },
+    {
+      "epoch": 0.6103092783505155,
+      "grad_norm": 1.6693129539489746,
+      "learning_rate": 8.323212869009782e-05,
+      "loss": 2.4474,
+      "step": 1184
+    },
+    {
+      "epoch": 0.6123711340206186,
+      "grad_norm": 1.575134515762329,
+      "learning_rate": 8.310457406771635e-05,
+      "loss": 2.5156,
+      "step": 1188
+    },
+    {
+      "epoch": 0.6144329896907217,
+      "grad_norm": 1.600196123123169,
+      "learning_rate": 8.297663467936882e-05,
+      "loss": 2.4151,
+      "step": 1192
+    },
+    {
+      "epoch": 0.6164948453608248,
+      "grad_norm": 1.6348146200180054,
+      "learning_rate": 8.28483120120619e-05,
+      "loss": 2.4336,
+      "step": 1196
+    },
+    {
+      "epoch": 0.6185567010309279,
+      "grad_norm": 1.6424046754837036,
+      "learning_rate": 8.271960755725702e-05,
+      "loss": 2.45,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6206185567010309,
+      "grad_norm": 1.6451953649520874,
+      "learning_rate": 8.2590522810853e-05,
+      "loss": 2.4414,
+      "step": 1204
+    },
+    {
+      "epoch": 0.622680412371134,
+      "grad_norm": 1.6768877506256104,
+      "learning_rate": 8.246105927316874e-05,
+      "loss": 2.459,
+      "step": 1208
+    },
+    {
+      "epoch": 0.6247422680412371,
+      "grad_norm": 1.7277884483337402,
+      "learning_rate": 8.233121844892568e-05,
+      "loss": 2.4398,
+      "step": 1212
+    },
+    {
+      "epoch": 0.6268041237113402,
+      "grad_norm": 1.6326000690460205,
+      "learning_rate": 8.220100184723038e-05,
+      "loss": 2.5278,
+      "step": 1216
+    },
+    {
+      "epoch": 0.6288659793814433,
+      "grad_norm": 1.6363636255264282,
+      "learning_rate": 8.2070410981557e-05,
+      "loss": 2.466,
+      "step": 1220
+    },
+    {
+      "epoch": 0.6309278350515464,
+      "grad_norm": 1.7234989404678345,
+      "learning_rate": 8.193944736972963e-05,
+      "loss": 2.4129,
+      "step": 1224
+    },
+    {
+      "epoch": 0.6329896907216495,
+      "grad_norm": 1.6150864362716675,
+      "learning_rate": 8.180811253390472e-05,
+      "loss": 2.4872,
+      "step": 1228
+    },
+    {
+      "epoch": 0.6350515463917525,
+      "grad_norm": 1.7305302619934082,
+      "learning_rate": 8.167640800055335e-05,
+      "loss": 2.5246,
+      "step": 1232
+    },
+    {
+      "epoch": 0.6371134020618556,
+      "grad_norm": 1.7068026065826416,
+      "learning_rate": 8.15443353004435e-05,
+      "loss": 2.4428,
+      "step": 1236
+    },
+    {
+      "epoch": 0.6391752577319587,
+      "grad_norm": 1.7080657482147217,
+      "learning_rate": 8.141189596862225e-05,
+      "loss": 2.3974,
+      "step": 1240
+    },
+    {
+      "epoch": 0.6412371134020619,
+      "grad_norm": 1.690262794494629,
+      "learning_rate": 8.127909154439796e-05,
+      "loss": 2.5146,
+      "step": 1244
+    },
+    {
+      "epoch": 0.643298969072165,
+      "grad_norm": 1.7586865425109863,
+      "learning_rate": 8.114592357132236e-05,
+      "loss": 2.459,
+      "step": 1248
+    },
+    {
+      "epoch": 0.6453608247422681,
+      "grad_norm": 1.6623564958572388,
+      "learning_rate": 8.101239359717263e-05,
+      "loss": 2.4779,
+      "step": 1252
+    },
+    {
+      "epoch": 0.6474226804123712,
+      "grad_norm": 1.694476842880249,
+      "learning_rate": 8.087850317393335e-05,
+      "loss": 2.4863,
+      "step": 1256
+    },
+    {
+      "epoch": 0.6494845360824743,
+      "grad_norm": 1.7273415327072144,
+      "learning_rate": 8.074425385777857e-05,
+      "loss": 2.5528,
+      "step": 1260
+    },
+    {
+      "epoch": 0.6515463917525773,
+      "grad_norm": 1.61915922164917,
+      "learning_rate": 8.06096472090536e-05,
+      "loss": 2.5224,
+      "step": 1264
+    },
+    {
+      "epoch": 0.6536082474226804,
+      "grad_norm": 1.64993417263031,
+      "learning_rate": 8.047468479225699e-05,
+      "loss": 2.475,
+      "step": 1268
+    },
+    {
+      "epoch": 0.6556701030927835,
+      "grad_norm": 1.6789292097091675,
+      "learning_rate": 8.033936817602225e-05,
+      "loss": 2.5414,
+      "step": 1272
+    },
+    {
+      "epoch": 0.6577319587628866,
+      "grad_norm": 1.534625768661499,
+      "learning_rate": 8.020369893309969e-05,
+      "loss": 2.5186,
+      "step": 1276
+    },
+    {
+      "epoch": 0.6597938144329897,
+      "grad_norm": 1.6226372718811035,
+      "learning_rate": 8.006767864033805e-05,
+      "loss": 2.5079,
+      "step": 1280
+    },
+    {
+      "epoch": 0.6618556701030928,
+      "grad_norm": 1.6352249383926392,
+      "learning_rate": 7.993130887866631e-05,
+      "loss": 2.4257,
+      "step": 1284
+    },
+    {
+      "epoch": 0.6639175257731958,
+      "grad_norm": 1.6394164562225342,
+      "learning_rate": 7.97945912330752e-05,
+      "loss": 2.4692,
+      "step": 1288
+    },
+    {
+      "epoch": 0.6659793814432989,
+      "grad_norm": 1.7224717140197754,
+      "learning_rate": 7.965752729259881e-05,
+      "loss": 2.5426,
+      "step": 1292
+    },
+    {
+      "epoch": 0.668041237113402,
+      "grad_norm": 1.662453532218933,
+      "learning_rate": 7.952011865029614e-05,
+      "loss": 2.5418,
+      "step": 1296
+    },
+    {
+      "epoch": 0.6701030927835051,
+      "grad_norm": 1.5507217645645142,
+      "learning_rate": 7.938236690323255e-05,
+      "loss": 2.3688,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6721649484536083,
+      "grad_norm": 1.6199603080749512,
+      "learning_rate": 7.924427365246125e-05,
+      "loss": 2.4501,
+      "step": 1304
+    },
+    {
+      "epoch": 0.6742268041237114,
+      "grad_norm": 1.5297261476516724,
+      "learning_rate": 7.910584050300465e-05,
+      "loss": 2.4709,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6762886597938145,
+      "grad_norm": 1.5516424179077148,
+      "learning_rate": 7.896706906383568e-05,
+      "loss": 2.4583,
+      "step": 1312
+    },
+    {
+      "epoch": 0.6783505154639176,
+      "grad_norm": 1.5144871473312378,
+      "learning_rate": 7.882796094785918e-05,
+      "loss": 2.4077,
+      "step": 1316
+    },
+    {
+      "epoch": 0.6804123711340206,
+      "grad_norm": 1.6823511123657227,
+      "learning_rate": 7.868851777189306e-05,
+      "loss": 2.4394,
+      "step": 1320
+    },
+    {
+      "epoch": 0.6824742268041237,
+      "grad_norm": 1.5894042253494263,
+      "learning_rate": 7.854874115664957e-05,
+      "loss": 2.4256,
+      "step": 1324
+    },
+    {
+      "epoch": 0.6845360824742268,
+      "grad_norm": 1.682361125946045,
+      "learning_rate": 7.84086327267164e-05,
+      "loss": 2.4082,
+      "step": 1328
+    },
+    {
+      "epoch": 0.6865979381443299,
+      "grad_norm": 1.6730499267578125,
+      "learning_rate": 7.826819411053787e-05,
+      "loss": 2.43,
+      "step": 1332
+    },
+    {
+      "epoch": 0.688659793814433,
+      "grad_norm": 1.5576122999191284,
+      "learning_rate": 7.812742694039599e-05,
+      "loss": 2.4109,
+      "step": 1336
+    },
+    {
+      "epoch": 0.6907216494845361,
+      "grad_norm": 1.735579013824463,
+      "learning_rate": 7.798633285239141e-05,
+      "loss": 2.5218,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6927835051546392,
+      "grad_norm": 1.5843349695205688,
+      "learning_rate": 7.784491348642452e-05,
+      "loss": 2.4067,
+      "step": 1344
+    },
+    {
+      "epoch": 0.6948453608247422,
+      "grad_norm": 1.5844374895095825,
+      "learning_rate": 7.770317048617631e-05,
+      "loss": 2.4879,
+      "step": 1348
+    },
+    {
+      "epoch": 0.6969072164948453,
+      "grad_norm": 1.6641557216644287,
+      "learning_rate": 7.756110549908924e-05,
+      "loss": 2.4427,
+      "step": 1352
+    },
+    {
+      "epoch": 0.6989690721649484,
+      "grad_norm": 1.567859172821045,
+      "learning_rate": 7.741872017634824e-05,
+      "loss": 2.458,
+      "step": 1356
+    },
+    {
+      "epoch": 0.7010309278350515,
+      "grad_norm": 1.649179220199585,
+      "learning_rate": 7.727601617286132e-05,
+      "loss": 2.4483,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7030927835051546,
+      "grad_norm": 1.6399873495101929,
+      "learning_rate": 7.713299514724051e-05,
+      "loss": 2.5019,
+      "step": 1364
+    },
+    {
+      "epoch": 0.7051546391752578,
+      "grad_norm": 1.724725365638733,
+      "learning_rate": 7.698965876178246e-05,
+      "loss": 2.4014,
+      "step": 1368
+    },
+    {
+      "epoch": 0.7072164948453609,
+      "grad_norm": 1.734067440032959,
+      "learning_rate": 7.68460086824492e-05,
+      "loss": 2.3332,
+      "step": 1372
+    },
+    {
+      "epoch": 0.709278350515464,
+      "grad_norm": 1.5960577726364136,
+      "learning_rate": 7.67020465788487e-05,
+      "loss": 2.4735,
+      "step": 1376
+    },
+    {
+      "epoch": 0.711340206185567,
+      "grad_norm": 1.6822943687438965,
+      "learning_rate": 7.655777412421554e-05,
+      "loss": 2.515,
+      "step": 1380
+    },
+    {
+      "epoch": 0.7134020618556701,
+      "grad_norm": 1.7459498643875122,
+      "learning_rate": 7.641319299539145e-05,
+      "loss": 2.4009,
+      "step": 1384
+    },
+    {
+      "epoch": 0.7154639175257732,
+      "grad_norm": 1.5728472471237183,
+      "learning_rate": 7.626830487280573e-05,
+      "loss": 2.3888,
+      "step": 1388
+    },
+    {
+      "epoch": 0.7175257731958763,
+      "grad_norm": 1.621801495552063,
+      "learning_rate": 7.612311144045588e-05,
+      "loss": 2.4036,
+      "step": 1392
+    },
+    {
+      "epoch": 0.7195876288659794,
+      "grad_norm": 1.6353496313095093,
+      "learning_rate": 7.597761438588784e-05,
+      "loss": 2.4504,
+      "step": 1396
+    },
+    {
+      "epoch": 0.7216494845360825,
+      "grad_norm": 1.5395163297653198,
+      "learning_rate": 7.583181540017656e-05,
+      "loss": 2.3526,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7237113402061855,
+      "grad_norm": 1.6341197490692139,
+      "learning_rate": 7.568571617790624e-05,
+      "loss": 2.4689,
+      "step": 1404
+    },
+    {
+      "epoch": 0.7257731958762886,
+      "grad_norm": 1.5934208631515503,
+      "learning_rate": 7.553931841715061e-05,
+      "loss": 2.5371,
+      "step": 1408
+    },
+    {
+      "epoch": 0.7278350515463917,
+      "grad_norm": 1.6311752796173096,
+      "learning_rate": 7.539262381945323e-05,
+      "loss": 2.4693,
+      "step": 1412
+    },
+    {
+      "epoch": 0.7298969072164948,
+      "grad_norm": 1.6778818368911743,
+      "learning_rate": 7.52456340898078e-05,
+      "loss": 2.4041,
+      "step": 1416
+    },
+    {
+      "epoch": 0.7319587628865979,
+      "grad_norm": 1.5758112668991089,
+      "learning_rate": 7.509835093663821e-05,
+      "loss": 2.4328,
+      "step": 1420
+    },
+    {
+      "epoch": 0.734020618556701,
+      "grad_norm": 1.6166560649871826,
+      "learning_rate": 7.495077607177872e-05,
+      "loss": 2.386,
+      "step": 1424
+    },
+    {
+      "epoch": 0.7360824742268042,
+      "grad_norm": 1.4999052286148071,
+      "learning_rate": 7.48029112104541e-05,
+      "loss": 2.3724,
+      "step": 1428
+    },
+    {
+      "epoch": 0.7381443298969073,
+      "grad_norm": 1.6186021566390991,
+      "learning_rate": 7.465475807125968e-05,
+      "loss": 2.3181,
+      "step": 1432
+    },
+    {
+      "epoch": 0.7402061855670103,
+      "grad_norm": 1.7516733407974243,
+      "learning_rate": 7.450631837614138e-05,
+      "loss": 2.4565,
+      "step": 1436
+    },
+    {
+      "epoch": 0.7422680412371134,
+      "grad_norm": 1.647589087486267,
+      "learning_rate": 7.435759385037565e-05,
+      "loss": 2.4047,
+      "step": 1440
+    },
+    {
+      "epoch": 0.7443298969072165,
+      "grad_norm": 1.5819942951202393,
+      "learning_rate": 7.420858622254946e-05,
+      "loss": 2.455,
+      "step": 1444
+    },
+    {
+      "epoch": 0.7463917525773196,
+      "grad_norm": 1.5764845609664917,
+      "learning_rate": 7.405929722454026e-05,
+      "loss": 2.3826,
+      "step": 1448
+    },
+    {
+      "epoch": 0.7484536082474227,
+      "grad_norm": 1.6311144828796387,
+      "learning_rate": 7.39097285914957e-05,
+      "loss": 2.3991,
+      "step": 1452
+    },
+    {
+      "epoch": 0.7505154639175258,
+      "grad_norm": 1.5981923341751099,
+      "learning_rate": 7.375988206181365e-05,
+      "loss": 2.4517,
+      "step": 1456
+    },
+    {
+      "epoch": 0.7525773195876289,
+      "grad_norm": 1.5369350910186768,
+      "learning_rate": 7.360975937712185e-05,
+      "loss": 2.4181,
+      "step": 1460
+    },
+    {
+      "epoch": 0.7546391752577319,
+      "grad_norm": 1.6119554042816162,
+      "learning_rate": 7.345936228225769e-05,
+      "loss": 2.4652,
+      "step": 1464
+    },
+    {
+      "epoch": 0.756701030927835,
+      "grad_norm": 1.7482960224151611,
+      "learning_rate": 7.330869252524804e-05,
+      "loss": 2.4213,
+      "step": 1468
+    },
+    {
+      "epoch": 0.7587628865979381,
+      "grad_norm": 1.676172137260437,
+      "learning_rate": 7.315775185728877e-05,
+      "loss": 2.3816,
+      "step": 1472
+    },
+    {
+      "epoch": 0.7608247422680412,
+      "grad_norm": 4.144953727722168,
+      "learning_rate": 7.300654203272454e-05,
+      "loss": 2.412,
+      "step": 1476
+    },
+    {
+      "epoch": 0.7628865979381443,
+      "grad_norm": 1.5259236097335815,
+      "learning_rate": 7.285506480902831e-05,
+      "loss": 2.4274,
+      "step": 1480
+    },
+    {
+      "epoch": 0.7649484536082474,
+      "grad_norm": 1.5517241954803467,
+      "learning_rate": 7.270332194678097e-05,
+      "loss": 2.3633,
+      "step": 1484
+    },
+    {
+      "epoch": 0.7670103092783506,
+      "grad_norm": 1.583932638168335,
+      "learning_rate": 7.255131520965087e-05,
+      "loss": 2.3964,
+      "step": 1488
+    },
+    {
+      "epoch": 0.7690721649484537,
+      "grad_norm": 1.5848530530929565,
+      "learning_rate": 7.239904636437325e-05,
+      "loss": 2.42,
+      "step": 1492
+    },
+    {
+      "epoch": 0.7711340206185567,
+      "grad_norm": 1.5512683391571045,
+      "learning_rate": 7.224651718072984e-05,
+      "loss": 2.4654,
+      "step": 1496
+    },
+    {
+      "epoch": 0.7731958762886598,
+      "grad_norm": 1.5761094093322754,
+      "learning_rate": 7.20937294315282e-05,
+      "loss": 2.4011,
+      "step": 1500
+    },
+    {
+      "epoch": 0.7752577319587629,
+      "grad_norm": 1.519158124923706,
+      "learning_rate": 7.194068489258109e-05,
+      "loss": 2.4315,
+      "step": 1504
+    },
+    {
+      "epoch": 0.777319587628866,
+      "grad_norm": 1.5774180889129639,
+      "learning_rate": 7.178738534268591e-05,
+      "loss": 2.4667,
+      "step": 1508
+    },
+    {
+      "epoch": 0.7793814432989691,
+      "grad_norm": 1.6394808292388916,
+      "learning_rate": 7.163383256360398e-05,
+      "loss": 2.4861,
+      "step": 1512
+    },
+    {
+      "epoch": 0.7814432989690722,
+      "grad_norm": 1.549655795097351,
+      "learning_rate": 7.14800283400398e-05,
+      "loss": 2.3323,
+      "step": 1516
+    },
+    {
+      "epoch": 0.7835051546391752,
+      "grad_norm": 1.6154853105545044,
+      "learning_rate": 7.132597445962042e-05,
+      "loss": 2.5378,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7855670103092783,
+      "grad_norm": 1.6021648645401,
+      "learning_rate": 7.117167271287453e-05,
+      "loss": 2.4249,
+      "step": 1524
+    },
+    {
+      "epoch": 0.7876288659793814,
+      "grad_norm": 1.665974736213684,
+      "learning_rate": 7.101712489321169e-05,
+      "loss": 2.4566,
+      "step": 1528
+    },
+    {
+      "epoch": 0.7896907216494845,
+      "grad_norm": 1.6290616989135742,
+      "learning_rate": 7.086233279690158e-05,
+      "loss": 2.3565,
+      "step": 1532
+    },
+    {
+      "epoch": 0.7917525773195876,
+      "grad_norm": 1.5475256443023682,
+      "learning_rate": 7.070729822305298e-05,
+      "loss": 2.4542,
+      "step": 1536
+    },
+    {
+      "epoch": 0.7938144329896907,
+      "grad_norm": 1.6090917587280273,
+      "learning_rate": 7.055202297359293e-05,
+      "loss": 2.4498,
+      "step": 1540
+    },
+    {
+      "epoch": 0.7958762886597938,
+      "grad_norm": 1.5616388320922852,
+      "learning_rate": 7.039650885324582e-05,
+      "loss": 2.3708,
+      "step": 1544
+    },
+    {
+      "epoch": 0.797938144329897,
+      "grad_norm": 1.711113452911377,
+      "learning_rate": 7.024075766951233e-05,
+      "loss": 2.5234,
+      "step": 1548
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6016921997070312,
+      "learning_rate": 7.008477123264848e-05,
+      "loss": 2.4755,
+      "step": 1552
+    },
+    {
+      "epoch": 0.8020618556701031,
+      "grad_norm": 1.5505585670471191,
+      "learning_rate": 6.99285513556446e-05,
+      "loss": 2.3435,
+      "step": 1556
+    },
+    {
+      "epoch": 0.8041237113402062,
+      "grad_norm": 1.532699704170227,
+      "learning_rate": 6.977209985420419e-05,
+      "loss": 2.4878,
+      "step": 1560
+    },
+    {
+      "epoch": 0.8061855670103093,
+      "grad_norm": 2.197500705718994,
+      "learning_rate": 6.961541854672293e-05,
+      "loss": 2.3707,
+      "step": 1564
+    },
+    {
+      "epoch": 0.8082474226804124,
+      "grad_norm": 1.6112059354782104,
+      "learning_rate": 6.945850925426742e-05,
+      "loss": 2.4803,
+      "step": 1568
+    },
+    {
+      "epoch": 0.8103092783505155,
+      "grad_norm": 1.6318986415863037,
+      "learning_rate": 6.930137380055403e-05,
+      "loss": 2.315,
+      "step": 1572
+    },
+    {
+      "epoch": 0.8123711340206186,
+      "grad_norm": 1.607735514640808,
+      "learning_rate": 6.914401401192789e-05,
+      "loss": 2.3804,
+      "step": 1576
+    },
+    {
+      "epoch": 0.8144329896907216,
+      "grad_norm": 1.474547266960144,
+      "learning_rate": 6.898643171734137e-05,
+      "loss": 2.3736,
+      "step": 1580
+    },
+    {
+      "epoch": 0.8164948453608247,
+      "grad_norm": 1.9736120700836182,
+      "learning_rate": 6.882862874833305e-05,
+      "loss": 2.3804,
+      "step": 1584
+    },
+    {
+      "epoch": 0.8185567010309278,
+      "grad_norm": 1.5962013006210327,
+      "learning_rate": 6.867060693900631e-05,
+      "loss": 2.4503,
+      "step": 1588
+    },
+    {
+      "epoch": 0.8206185567010309,
+      "grad_norm": 1.565443754196167,
+      "learning_rate": 6.851236812600808e-05,
+      "loss": 2.4805,
+      "step": 1592
+    },
+    {
+      "epoch": 0.822680412371134,
+      "grad_norm": 1.6651315689086914,
+      "learning_rate": 6.835391414850748e-05,
+      "loss": 2.4496,
+      "step": 1596
+    },
+    {
+      "epoch": 0.8247422680412371,
+      "grad_norm": 1.5796141624450684,
+      "learning_rate": 6.819524684817438e-05,
+      "loss": 2.4143,
+      "step": 1600
+    },
+    {
+      "epoch": 0.8268041237113402,
+      "grad_norm": 1.513759732246399,
+      "learning_rate": 6.803636806915812e-05,
+      "loss": 2.4383,
+      "step": 1604
+    },
+    {
+      "epoch": 0.8288659793814434,
+      "grad_norm": 1.556221842765808,
+      "learning_rate": 6.787727965806591e-05,
+      "loss": 2.4026,
+      "step": 1608
+    },
+    {
+      "epoch": 0.8309278350515464,
+      "grad_norm": 1.5572640895843506,
+      "learning_rate": 6.771798346394157e-05,
+      "loss": 2.4022,
+      "step": 1612
+    },
+    {
+      "epoch": 0.8329896907216495,
+      "grad_norm": 1.602418065071106,
+      "learning_rate": 6.755848133824383e-05,
+      "loss": 2.4555,
+      "step": 1616
+    },
+    {
+      "epoch": 0.8350515463917526,
+      "grad_norm": 1.676102638244629,
+      "learning_rate": 6.739877513482497e-05,
+      "loss": 2.3695,
+      "step": 1620
+    },
+    {
+      "epoch": 0.8371134020618557,
+      "grad_norm": 1.748834252357483,
+      "learning_rate": 6.72388667099092e-05,
+      "loss": 2.2801,
+      "step": 1624
+    },
+    {
+      "epoch": 0.8391752577319588,
+      "grad_norm": 1.5271378755569458,
+      "learning_rate": 6.707875792207108e-05,
+      "loss": 2.2845,
+      "step": 1628
+    },
+    {
+      "epoch": 0.8412371134020619,
+      "grad_norm": 1.7049288749694824,
+      "learning_rate": 6.691845063221393e-05,
+      "loss": 2.4126,
+      "step": 1632
+    },
+    {
+      "epoch": 0.843298969072165,
+      "grad_norm": 1.590855360031128,
+      "learning_rate": 6.675794670354826e-05,
+      "loss": 2.3571,
+      "step": 1636
+    },
+    {
+      "epoch": 0.845360824742268,
+      "grad_norm": 1.7120413780212402,
+      "learning_rate": 6.659724800157002e-05,
+      "loss": 2.4111,
+      "step": 1640
+    },
+    {
+      "epoch": 0.8474226804123711,
+      "grad_norm": 1.5834614038467407,
+      "learning_rate": 6.643635639403897e-05,
+      "loss": 2.4045,
+      "step": 1644
+    },
+    {
+      "epoch": 0.8494845360824742,
+      "grad_norm": 1.4899585247039795,
+      "learning_rate": 6.627527375095697e-05,
+      "loss": 2.2914,
+      "step": 1648
+    },
+    {
+      "epoch": 0.8515463917525773,
+      "grad_norm": 1.573357105255127,
+      "learning_rate": 6.611400194454619e-05,
+      "loss": 2.3746,
+      "step": 1652
+    },
+    {
+      "epoch": 0.8536082474226804,
+      "grad_norm": 1.528746247291565,
+      "learning_rate": 6.595254284922748e-05,
+      "loss": 2.3568,
+      "step": 1656
+    },
+    {
+      "epoch": 0.8556701030927835,
+      "grad_norm": 1.6635682582855225,
+      "learning_rate": 6.579089834159844e-05,
+      "loss": 2.3868,
+      "step": 1660
+    },
+    {
+      "epoch": 0.8577319587628865,
+      "grad_norm": 1.5613417625427246,
+      "learning_rate": 6.562907030041168e-05,
+      "loss": 2.377,
+      "step": 1664
+    },
+    {
+      "epoch": 0.8597938144329897,
+      "grad_norm": 1.6135239601135254,
+      "learning_rate": 6.546706060655298e-05,
+      "loss": 2.4542,
+      "step": 1668
+    },
+    {
+      "epoch": 0.8618556701030928,
+      "grad_norm": 1.5151220560073853,
+      "learning_rate": 6.530487114301944e-05,
+      "loss": 2.2057,
+      "step": 1672
+    },
+    {
+      "epoch": 0.8639175257731959,
+      "grad_norm": 1.6280299425125122,
+      "learning_rate": 6.514250379489753e-05,
+      "loss": 2.3594,
+      "step": 1676
+    },
+    {
+      "epoch": 0.865979381443299,
+      "grad_norm": 1.5538045167922974,
+      "learning_rate": 6.49799604493413e-05,
+      "loss": 2.4155,
+      "step": 1680
+    },
+    {
+      "epoch": 0.8680412371134021,
+      "grad_norm": 1.5412131547927856,
+      "learning_rate": 6.481724299555029e-05,
+      "loss": 2.3285,
+      "step": 1684
+    },
+    {
+      "epoch": 0.8701030927835052,
+      "grad_norm": 1.7468520402908325,
+      "learning_rate": 6.465435332474768e-05,
+      "loss": 2.2616,
+      "step": 1688
+    },
+    {
+      "epoch": 0.8721649484536083,
+      "grad_norm": 1.4656267166137695,
+      "learning_rate": 6.449129333015834e-05,
+      "loss": 2.2932,
+      "step": 1692
+    },
+    {
+      "epoch": 0.8742268041237113,
+      "grad_norm": 1.554957628250122,
+      "learning_rate": 6.432806490698671e-05,
+      "loss": 2.4188,
+      "step": 1696
+    },
+    {
+      "epoch": 0.8762886597938144,
+      "grad_norm": 1.6073874235153198,
+      "learning_rate": 6.416466995239485e-05,
+      "loss": 2.2739,
+      "step": 1700
+    },
+    {
+      "epoch": 0.8783505154639175,
+      "grad_norm": 1.6184226274490356,
+      "learning_rate": 6.400111036548037e-05,
+      "loss": 2.3879,
+      "step": 1704
+    },
+    {
+      "epoch": 0.8804123711340206,
+      "grad_norm": 1.5735539197921753,
+      "learning_rate": 6.383738804725438e-05,
+      "loss": 2.3191,
+      "step": 1708
+    },
+    {
+      "epoch": 0.8824742268041237,
+      "grad_norm": 1.6026270389556885,
+      "learning_rate": 6.36735049006193e-05,
+      "loss": 2.5497,
+      "step": 1712
+    },
+    {
+      "epoch": 0.8845360824742268,
+      "grad_norm": 1.6282131671905518,
+      "learning_rate": 6.350946283034696e-05,
+      "loss": 2.4152,
+      "step": 1716
+    },
+    {
+      "epoch": 0.8865979381443299,
+      "grad_norm": 1.553600549697876,
+      "learning_rate": 6.334526374305615e-05,
+      "loss": 2.3417,
+      "step": 1720
+    },
+    {
+      "epoch": 0.8886597938144329,
+      "grad_norm": 1.5668585300445557,
+      "learning_rate": 6.318090954719074e-05,
+      "loss": 2.3811,
+      "step": 1724
+    },
+    {
+      "epoch": 0.8907216494845361,
+      "grad_norm": 1.7151823043823242,
+      "learning_rate": 6.301640215299735e-05,
+      "loss": 2.2925,
+      "step": 1728
+    },
+    {
+      "epoch": 0.8927835051546392,
+      "grad_norm": 1.5615097284317017,
+      "learning_rate": 6.285174347250322e-05,
+      "loss": 2.3591,
+      "step": 1732
+    },
+    {
+      "epoch": 0.8948453608247423,
+      "grad_norm": 1.6054131984710693,
+      "learning_rate": 6.26869354194939e-05,
+      "loss": 2.3084,
+      "step": 1736
+    },
+    {
+      "epoch": 0.8969072164948454,
+      "grad_norm": 1.606719732284546,
+      "learning_rate": 6.252197990949108e-05,
+      "loss": 2.4081,
+      "step": 1740
+    },
+    {
+      "epoch": 0.8989690721649485,
+      "grad_norm": 1.6772600412368774,
+      "learning_rate": 6.235687885973032e-05,
+      "loss": 2.3733,
+      "step": 1744
+    },
+    {
+      "epoch": 0.9010309278350516,
+      "grad_norm": 1.508259654045105,
+      "learning_rate": 6.219163418913872e-05,
+      "loss": 2.3318,
+      "step": 1748
+    },
+    {
+      "epoch": 0.9030927835051547,
+      "grad_norm": 1.5906274318695068,
+      "learning_rate": 6.202624781831268e-05,
+      "loss": 2.297,
+      "step": 1752
+    },
+    {
+      "epoch": 0.9051546391752577,
+      "grad_norm": 1.5747168064117432,
+      "learning_rate": 6.186072166949552e-05,
+      "loss": 2.4498,
+      "step": 1756
+    },
+    {
+      "epoch": 0.9072164948453608,
+      "grad_norm": 1.5119340419769287,
+      "learning_rate": 6.16950576665552e-05,
+      "loss": 2.3045,
+      "step": 1760
+    },
+    {
+      "epoch": 0.9092783505154639,
+      "grad_norm": 1.5936625003814697,
+      "learning_rate": 6.152925773496189e-05,
+      "loss": 2.3989,
+      "step": 1764
+    },
+    {
+      "epoch": 0.911340206185567,
+      "grad_norm": 1.6385735273361206,
+      "learning_rate": 6.13633238017656e-05,
+      "loss": 2.3415,
+      "step": 1768
+    },
+    {
+      "epoch": 0.9134020618556701,
+      "grad_norm": 1.6806796789169312,
+      "learning_rate": 6.119725779557386e-05,
+      "loss": 2.37,
+      "step": 1772
+    },
+    {
+      "epoch": 0.9154639175257732,
+      "grad_norm": 1.5404062271118164,
+      "learning_rate": 6.103106164652924e-05,
+      "loss": 2.476,
+      "step": 1776
+    },
+    {
+      "epoch": 0.9175257731958762,
+      "grad_norm": 1.5629997253417969,
+      "learning_rate": 6.086473728628691e-05,
+      "loss": 2.3843,
+      "step": 1780
+    },
+    {
+      "epoch": 0.9195876288659793,
+      "grad_norm": 1.5114765167236328,
+      "learning_rate": 6.069828664799221e-05,
+      "loss": 2.3972,
+      "step": 1784
+    },
+    {
+      "epoch": 0.9216494845360824,
+      "grad_norm": 1.5878251791000366,
+      "learning_rate": 6.053171166625817e-05,
+      "loss": 2.3686,
+      "step": 1788
+    },
+    {
+      "epoch": 0.9237113402061856,
+      "grad_norm": 1.6494956016540527,
+      "learning_rate": 6.036501427714304e-05,
+      "loss": 2.4627,
+      "step": 1792
+    },
+    {
+      "epoch": 0.9257731958762887,
+      "grad_norm": 1.5250719785690308,
+      "learning_rate": 6.0198196418127804e-05,
+      "loss": 2.332,
+      "step": 1796
+    },
+    {
+      "epoch": 0.9278350515463918,
+      "grad_norm": 1.5805870294570923,
+      "learning_rate": 6.0031260028093596e-05,
+      "loss": 2.2881,
+      "step": 1800
+    },
+    {
+      "epoch": 0.9298969072164949,
+      "grad_norm": 1.6690597534179688,
+      "learning_rate": 5.98642070472992e-05,
+      "loss": 2.2946,
+      "step": 1804
+    },
+    {
+      "epoch": 0.931958762886598,
+      "grad_norm": 1.5214745998382568,
+      "learning_rate": 5.969703941735858e-05,
+      "loss": 2.3984,
+      "step": 1808
+    },
+    {
+      "epoch": 0.934020618556701,
+      "grad_norm": 1.6358650922775269,
+      "learning_rate": 5.952975908121815e-05,
+      "loss": 2.3758,
+      "step": 1812
+    },
+    {
+      "epoch": 0.9360824742268041,
+      "grad_norm": 1.9057590961456299,
+      "learning_rate": 5.936236798313431e-05,
+      "loss": 2.3347,
+      "step": 1816
+    },
+    {
+      "epoch": 0.9381443298969072,
+      "grad_norm": 1.6489713191986084,
+      "learning_rate": 5.9194868068650845e-05,
+      "loss": 2.2552,
+      "step": 1820
+    },
+    {
+      "epoch": 0.9402061855670103,
+      "grad_norm": 1.5380582809448242,
+      "learning_rate": 5.902726128457625e-05,
+      "loss": 2.2898,
+      "step": 1824
+    },
+    {
+      "epoch": 0.9422680412371134,
+      "grad_norm": 1.5303738117218018,
+      "learning_rate": 5.885954957896115e-05,
+      "loss": 2.2896,
+      "step": 1828
+    },
+    {
+      "epoch": 0.9443298969072165,
+      "grad_norm": 1.577243685722351,
+      "learning_rate": 5.8691734901075634e-05,
+      "loss": 2.3281,
+      "step": 1832
+    },
+    {
+      "epoch": 0.9463917525773196,
+      "grad_norm": 1.545732021331787,
+      "learning_rate": 5.852381920138663e-05,
+      "loss": 2.2197,
+      "step": 1836
+    },
+    {
+      "epoch": 0.9484536082474226,
+      "grad_norm": 1.5541272163391113,
+      "learning_rate": 5.835580443153522e-05,
+      "loss": 2.2776,
+      "step": 1840
+    },
+    {
+      "epoch": 0.9505154639175257,
+      "grad_norm": 1.5543440580368042,
+      "learning_rate": 5.818769254431395e-05,
+      "loss": 2.3597,
+      "step": 1844
+    },
+    {
+      "epoch": 0.9525773195876288,
+      "grad_norm": 1.5079783201217651,
+      "learning_rate": 5.80194854936441e-05,
+      "loss": 2.3476,
+      "step": 1848
+    },
+    {
+      "epoch": 0.954639175257732,
+      "grad_norm": 1.503811240196228,
+      "learning_rate": 5.7851185234553064e-05,
+      "loss": 2.3318,
+      "step": 1852
+    },
+    {
+      "epoch": 0.9567010309278351,
+      "grad_norm": 1.532036542892456,
+      "learning_rate": 5.768279372315153e-05,
+      "loss": 2.2278,
+      "step": 1856
+    },
+    {
+      "epoch": 0.9587628865979382,
+      "grad_norm": 1.5316864252090454,
+      "learning_rate": 5.7514312916610814e-05,
+      "loss": 2.3835,
+      "step": 1860
+    },
+    {
+      "epoch": 0.9608247422680413,
+      "grad_norm": 1.5309964418411255,
+      "learning_rate": 5.7345744773140086e-05,
+      "loss": 2.3779,
+      "step": 1864
+    },
+    {
+      "epoch": 0.9628865979381444,
+      "grad_norm": 1.577561855316162,
+      "learning_rate": 5.71770912519636e-05,
+      "loss": 2.3334,
+      "step": 1868
+    },
+    {
+      "epoch": 0.9649484536082474,
+      "grad_norm": 1.5817878246307373,
+      "learning_rate": 5.7008354313297926e-05,
+      "loss": 2.28,
+      "step": 1872
+    },
+    {
+      "epoch": 0.9670103092783505,
+      "grad_norm": 1.5670493841171265,
+      "learning_rate": 5.683953591832922e-05,
+      "loss": 2.304,
+      "step": 1876
+    },
+    {
+      "epoch": 0.9690721649484536,
+      "grad_norm": 1.5366458892822266,
+      "learning_rate": 5.667063802919032e-05,
+      "loss": 2.4577,
+      "step": 1880
+    },
+    {
+      "epoch": 0.9711340206185567,
+      "grad_norm": 1.5330809354782104,
+      "learning_rate": 5.6501662608938014e-05,
+      "loss": 2.4178,
+      "step": 1884
+    },
+    {
+      "epoch": 0.9731958762886598,
+      "grad_norm": 1.4714388847351074,
+      "learning_rate": 5.633261162153027e-05,
+      "loss": 2.3252,
+      "step": 1888
+    },
+    {
+      "epoch": 0.9752577319587629,
+      "grad_norm": 1.5558996200561523,
+      "learning_rate": 5.6163487031803305e-05,
+      "loss": 2.3597,
+      "step": 1892
+    },
+    {
+      "epoch": 0.977319587628866,
+      "grad_norm": 1.5362873077392578,
+      "learning_rate": 5.5994290805448826e-05,
+      "loss": 2.2979,
+      "step": 1896
+    },
+    {
+      "epoch": 0.979381443298969,
+      "grad_norm": 1.5119233131408691,
+      "learning_rate": 5.582502490899111e-05,
+      "loss": 2.3582,
+      "step": 1900
+    },
+    {
+      "epoch": 0.9814432989690721,
+      "grad_norm": 1.6467958688735962,
+      "learning_rate": 5.565569130976422e-05,
+      "loss": 2.3366,
+      "step": 1904
+    },
+    {
+      "epoch": 0.9835051546391752,
+      "grad_norm": 1.6580411195755005,
+      "learning_rate": 5.548629197588913e-05,
+      "loss": 2.2901,
+      "step": 1908
+    },
+    {
+      "epoch": 0.9855670103092784,
+      "grad_norm": 1.4866169691085815,
+      "learning_rate": 5.5316828876250795e-05,
+      "loss": 2.3313,
+      "step": 1912
+    },
+    {
+      "epoch": 0.9876288659793815,
+      "grad_norm": 1.6460485458374023,
+      "learning_rate": 5.514730398047533e-05,
+      "loss": 2.2917,
+      "step": 1916
+    },
+    {
+      "epoch": 0.9896907216494846,
+      "grad_norm": 1.4895657300949097,
+      "learning_rate": 5.497771925890706e-05,
+      "loss": 2.3709,
+      "step": 1920
+    },
+    {
+      "epoch": 0.9917525773195877,
+      "grad_norm": 1.4158916473388672,
+      "learning_rate": 5.48080766825857e-05,
+      "loss": 2.1582,
+      "step": 1924
+    },
+    {
+      "epoch": 0.9938144329896907,
+      "grad_norm": 1.611509919166565,
+      "learning_rate": 5.463837822322333e-05,
+      "loss": 2.5048,
+      "step": 1928
+    },
+    {
+      "epoch": 0.9958762886597938,
+      "grad_norm": 1.6175845861434937,
+      "learning_rate": 5.446862585318161e-05,
+      "loss": 2.3836,
+      "step": 1932
+    },
+    {
+      "epoch": 0.9979381443298969,
+      "grad_norm": 1.6315233707427979,
+      "learning_rate": 5.429882154544875e-05,
+      "loss": 2.2926,
+      "step": 1936
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.427371025085449,
+      "learning_rate": 5.4128967273616625e-05,
+      "loss": 2.3215,
+      "step": 1940
+    },
+    {
+      "epoch": 1.002061855670103,
+      "grad_norm": 1.421981692314148,
+      "learning_rate": 5.395906501185783e-05,
+      "loss": 1.8189,
+      "step": 1944
+    },
+    {
+      "epoch": 1.0041237113402062,
+      "grad_norm": 1.4998725652694702,
+      "learning_rate": 5.3789116734902746e-05,
+      "loss": 1.9157,
+      "step": 1948
+    },
+    {
+      "epoch": 1.0061855670103093,
+      "grad_norm": 1.4507402181625366,
+      "learning_rate": 5.361912441801654e-05,
+      "loss": 1.9084,
+      "step": 1952
+    },
+    {
+      "epoch": 1.0082474226804123,
+      "grad_norm": 1.5483083724975586,
+      "learning_rate": 5.344909003697634e-05,
+      "loss": 1.824,
+      "step": 1956
+    },
+    {
+      "epoch": 1.0103092783505154,
+      "grad_norm": 1.579430103302002,
+      "learning_rate": 5.327901556804803e-05,
+      "loss": 1.8283,
+      "step": 1960
+    },
+    {
+      "epoch": 1.0123711340206185,
+      "grad_norm": 1.5646827220916748,
+      "learning_rate": 5.310890298796353e-05,
+      "loss": 1.8367,
+      "step": 1964
+    },
+    {
+      "epoch": 1.0144329896907216,
+      "grad_norm": 1.6156511306762695,
+      "learning_rate": 5.293875427389771e-05,
+      "loss": 1.9887,
+      "step": 1968
+    },
+    {
+      "epoch": 1.0164948453608247,
+      "grad_norm": 1.416155457496643,
+      "learning_rate": 5.2768571403445366e-05,
+      "loss": 1.7009,
+      "step": 1972
+    },
+    {
+      "epoch": 1.0185567010309278,
+      "grad_norm": 1.4358748197555542,
+      "learning_rate": 5.259835635459833e-05,
+      "loss": 1.7943,
+      "step": 1976
+    },
+    {
+      "epoch": 1.0206185567010309,
+      "grad_norm": 1.567254900932312,
+      "learning_rate": 5.242811110572242e-05,
+      "loss": 1.8559,
+      "step": 1980
+    },
+    {
+      "epoch": 1.022680412371134,
+      "grad_norm": 1.6107474565505981,
+      "learning_rate": 5.2257837635534446e-05,
+      "loss": 1.8663,
+      "step": 1984
+    },
+    {
+      "epoch": 1.024742268041237,
+      "grad_norm": 1.8279513120651245,
+      "learning_rate": 5.208753792307923e-05,
+      "loss": 1.85,
+      "step": 1988
+    },
+    {
+      "epoch": 1.02680412371134,
+      "grad_norm": 1.7236356735229492,
+      "learning_rate": 5.191721394770667e-05,
+      "loss": 1.806,
+      "step": 1992
+    },
+    {
+      "epoch": 1.0288659793814432,
+      "grad_norm": 1.4513978958129883,
+      "learning_rate": 5.174686768904855e-05,
+      "loss": 1.881,
+      "step": 1996
+    },
+    {
+      "epoch": 1.0309278350515463,
+      "grad_norm": 1.497959852218628,
+      "learning_rate": 5.15765011269957e-05,
+      "loss": 1.776,
+      "step": 2000
+    },
+    {
+      "epoch": 1.0329896907216496,
+      "grad_norm": 1.4081003665924072,
+      "learning_rate": 5.140611624167497e-05,
+      "loss": 1.7823,
+      "step": 2004
+    },
+    {
+      "epoch": 1.0350515463917527,
+      "grad_norm": 1.4467487335205078,
+      "learning_rate": 5.1235715013426146e-05,
+      "loss": 1.7799,
+      "step": 2008
+    },
+    {
+      "epoch": 1.0371134020618558,
+      "grad_norm": 1.518298625946045,
+      "learning_rate": 5.1065299422778936e-05,
+      "loss": 1.8807,
+      "step": 2012
+    },
+    {
+      "epoch": 1.0391752577319588,
+      "grad_norm": 1.3984616994857788,
+      "learning_rate": 5.0894871450430004e-05,
+      "loss": 1.8241,
+      "step": 2016
+    },
+    {
+      "epoch": 1.041237113402062,
+      "grad_norm": 1.5757567882537842,
+      "learning_rate": 5.072443307721992e-05,
+      "loss": 1.8525,
+      "step": 2020
+    },
+    {
+      "epoch": 1.043298969072165,
+      "grad_norm": 1.7042030096054077,
+      "learning_rate": 5.0553986284110164e-05,
+      "loss": 1.7814,
+      "step": 2024
+    },
+    {
+      "epoch": 1.045360824742268,
+      "grad_norm": 1.5050081014633179,
+      "learning_rate": 5.0383533052160045e-05,
+      "loss": 1.8751,
+      "step": 2028
+    },
+    {
+      "epoch": 1.0474226804123712,
+      "grad_norm": 1.522011160850525,
+      "learning_rate": 5.02130753625037e-05,
+      "loss": 1.8369,
+      "step": 2032
+    },
+    {
+      "epoch": 1.0494845360824743,
+      "grad_norm": 1.5450692176818848,
+      "learning_rate": 5.004261519632713e-05,
+      "loss": 1.8428,
+      "step": 2036
+    },
+    {
+      "epoch": 1.0515463917525774,
+      "grad_norm": 2.67858624458313,
+      "learning_rate": 4.987215453484507e-05,
+      "loss": 1.7921,
+      "step": 2040
+    },
+    {
+      "epoch": 1.0536082474226804,
+      "grad_norm": 1.628353476524353,
+      "learning_rate": 4.9701695359278035e-05,
+      "loss": 1.862,
+      "step": 2044
+    },
+    {
+      "epoch": 1.0556701030927835,
+      "grad_norm": 1.5472396612167358,
+      "learning_rate": 4.953123965082927e-05,
+      "loss": 1.8028,
+      "step": 2048
+    },
+    {
+      "epoch": 1.0577319587628866,
+      "grad_norm": 1.4193578958511353,
+      "learning_rate": 4.936078939066169e-05,
+      "loss": 1.7504,
+      "step": 2052
+    },
+    {
+      "epoch": 1.0597938144329897,
+      "grad_norm": 1.6026136875152588,
+      "learning_rate": 4.919034655987493e-05,
+      "loss": 1.8197,
+      "step": 2056
+    },
+    {
+      "epoch": 1.0618556701030928,
+      "grad_norm": 1.494009256362915,
+      "learning_rate": 4.901991313948227e-05,
+      "loss": 1.813,
+      "step": 2060
+    },
+    {
+      "epoch": 1.0639175257731959,
+      "grad_norm": 1.569420337677002,
+      "learning_rate": 4.884949111038757e-05,
+      "loss": 1.8011,
+      "step": 2064
+    },
+    {
+      "epoch": 1.065979381443299,
+      "grad_norm": 1.5701764822006226,
+      "learning_rate": 4.8679082453362354e-05,
+      "loss": 1.793,
+      "step": 2068
+    },
+    {
+      "epoch": 1.068041237113402,
+      "grad_norm": 1.4102658033370972,
+      "learning_rate": 4.850868914902269e-05,
+      "loss": 1.7566,
+      "step": 2072
+    },
+    {
+      "epoch": 1.0701030927835051,
+      "grad_norm": 1.5660536289215088,
+      "learning_rate": 4.833831317780621e-05,
+      "loss": 1.8408,
+      "step": 2076
+    },
+    {
+      "epoch": 1.0721649484536082,
+      "grad_norm": 1.5886646509170532,
+      "learning_rate": 4.81679565199491e-05,
+      "loss": 1.8117,
+      "step": 2080
+    },
+    {
+      "epoch": 1.0742268041237113,
+      "grad_norm": 1.531605839729309,
+      "learning_rate": 4.799762115546304e-05,
+      "loss": 1.8381,
+      "step": 2084
+    },
+    {
+      "epoch": 1.0762886597938144,
+      "grad_norm": 1.4816917181015015,
+      "learning_rate": 4.782730906411225e-05,
+      "loss": 1.8265,
+      "step": 2088
+    },
+    {
+      "epoch": 1.0783505154639175,
+      "grad_norm": 1.535863995552063,
+      "learning_rate": 4.7657022225390454e-05,
+      "loss": 1.7533,
+      "step": 2092
+    },
+    {
+      "epoch": 1.0804123711340206,
+      "grad_norm": 1.5383597612380981,
+      "learning_rate": 4.748676261849787e-05,
+      "loss": 1.8174,
+      "step": 2096
+    },
+    {
+      "epoch": 1.0824742268041236,
+      "grad_norm": 1.5802308320999146,
+      "learning_rate": 4.731653222231819e-05,
+      "loss": 1.814,
+      "step": 2100
+    },
+    {
+      "epoch": 1.0845360824742267,
+      "grad_norm": 1.6186000108718872,
+      "learning_rate": 4.7146333015395606e-05,
+      "loss": 1.836,
+      "step": 2104
+    },
+    {
+      "epoch": 1.0865979381443298,
+      "grad_norm": 1.4844691753387451,
+      "learning_rate": 4.6976166975911815e-05,
+      "loss": 1.7533,
+      "step": 2108
+    },
+    {
+      "epoch": 1.088659793814433,
+      "grad_norm": 1.4020297527313232,
+      "learning_rate": 4.6806036081663026e-05,
+      "loss": 1.7584,
+      "step": 2112
+    },
+    {
+      "epoch": 1.090721649484536,
+      "grad_norm": 1.5726157426834106,
+      "learning_rate": 4.663594231003692e-05,
+      "loss": 1.78,
+      "step": 2116
+    },
+    {
+      "epoch": 1.0927835051546393,
+      "grad_norm": 1.5943951606750488,
+      "learning_rate": 4.646588763798976e-05,
+      "loss": 1.8729,
+      "step": 2120
+    },
+    {
+      "epoch": 1.0948453608247424,
+      "grad_norm": 1.499718427658081,
+      "learning_rate": 4.629587404202331e-05,
+      "loss": 1.7568,
+      "step": 2124
+    },
+    {
+      "epoch": 1.0969072164948455,
+      "grad_norm": 1.5948656797409058,
+      "learning_rate": 4.6125903498162e-05,
+      "loss": 1.7359,
+      "step": 2128
+    },
+    {
+      "epoch": 1.0989690721649485,
+      "grad_norm": 1.4080160856246948,
+      "learning_rate": 4.59559779819298e-05,
+      "loss": 1.691,
+      "step": 2132
+    },
+    {
+      "epoch": 1.1010309278350516,
+      "grad_norm": 1.509520411491394,
+      "learning_rate": 4.5786099468327365e-05,
+      "loss": 1.8097,
+      "step": 2136
+    },
+    {
+      "epoch": 1.1030927835051547,
+      "grad_norm": 1.4405332803726196,
+      "learning_rate": 4.561626993180904e-05,
+      "loss": 1.7825,
+      "step": 2140
+    },
+    {
+      "epoch": 1.1051546391752578,
+      "grad_norm": 1.5002368688583374,
+      "learning_rate": 4.5446491346259975e-05,
+      "loss": 1.7178,
+      "step": 2144
+    },
+    {
+      "epoch": 1.1072164948453609,
+      "grad_norm": 1.4368022680282593,
+      "learning_rate": 4.5276765684973024e-05,
+      "loss": 1.7831,
+      "step": 2148
+    },
+    {
+      "epoch": 1.109278350515464,
+      "grad_norm": 1.4888081550598145,
+      "learning_rate": 4.510709492062601e-05,
+      "loss": 1.7647,
+      "step": 2152
+    },
+    {
+      "epoch": 1.111340206185567,
+      "grad_norm": 1.5033565759658813,
+      "learning_rate": 4.4937481025258674e-05,
+      "loss": 1.8698,
+      "step": 2156
+    },
+    {
+      "epoch": 1.1134020618556701,
+      "grad_norm": 1.4232312440872192,
+      "learning_rate": 4.476792597024976e-05,
+      "loss": 1.9078,
+      "step": 2160
+    },
+    {
+      "epoch": 1.1154639175257732,
+      "grad_norm": 1.4969874620437622,
+      "learning_rate": 4.4598431726294175e-05,
+      "loss": 1.7898,
+      "step": 2164
+    },
+    {
+      "epoch": 1.1175257731958763,
+      "grad_norm": 1.5777771472930908,
+      "learning_rate": 4.442900026337998e-05,
+      "loss": 1.7817,
+      "step": 2168
+    },
+    {
+      "epoch": 1.1195876288659794,
+      "grad_norm": 2.387895345687866,
+      "learning_rate": 4.425963355076558e-05,
+      "loss": 1.7712,
+      "step": 2172
+    },
+    {
+      "epoch": 1.1216494845360825,
+      "grad_norm": 1.4322453737258911,
+      "learning_rate": 4.409033355695683e-05,
+      "loss": 1.7369,
+      "step": 2176
+    },
+    {
+      "epoch": 1.1237113402061856,
+      "grad_norm": 1.4894506931304932,
+      "learning_rate": 4.392110224968403e-05,
+      "loss": 1.7687,
+      "step": 2180
+    },
+    {
+      "epoch": 1.1257731958762887,
+      "grad_norm": 1.5217411518096924,
+      "learning_rate": 4.3751941595879265e-05,
+      "loss": 1.8199,
+      "step": 2184
+    },
+    {
+      "epoch": 1.1278350515463917,
+      "grad_norm": 1.5253269672393799,
+      "learning_rate": 4.358285356165338e-05,
+      "loss": 1.8539,
+      "step": 2188
+    },
+    {
+      "epoch": 1.1298969072164948,
+      "grad_norm": 1.4793150424957275,
+      "learning_rate": 4.34138401122732e-05,
+      "loss": 1.805,
+      "step": 2192
+    },
+    {
+      "epoch": 1.131958762886598,
+      "grad_norm": 1.5564053058624268,
+      "learning_rate": 4.3244903212138634e-05,
+      "loss": 1.7565,
+      "step": 2196
+    },
+    {
+      "epoch": 1.134020618556701,
+      "grad_norm": 1.4678430557250977,
+      "learning_rate": 4.307604482475993e-05,
+      "loss": 1.6859,
+      "step": 2200
+    },
+    {
+      "epoch": 1.136082474226804,
+      "grad_norm": 1.4880309104919434,
+      "learning_rate": 4.290726691273477e-05,
+      "loss": 1.7329,
+      "step": 2204
+    },
+    {
+      "epoch": 1.1381443298969072,
+      "grad_norm": 1.8314374685287476,
+      "learning_rate": 4.27385714377255e-05,
+      "loss": 1.7669,
+      "step": 2208
+    },
+    {
+      "epoch": 1.1402061855670103,
+      "grad_norm": 1.5519520044326782,
+      "learning_rate": 4.256996036043631e-05,
+      "loss": 1.7761,
+      "step": 2212
+    },
+    {
+      "epoch": 1.1422680412371133,
+      "grad_norm": 1.4087693691253662,
+      "learning_rate": 4.240143564059045e-05,
+      "loss": 1.743,
+      "step": 2216
+    },
+    {
+      "epoch": 1.1443298969072164,
+      "grad_norm": 1.5219017267227173,
+      "learning_rate": 4.2232999236907524e-05,
+      "loss": 1.8374,
+      "step": 2220
+    },
+    {
+      "epoch": 1.1463917525773195,
+      "grad_norm": 1.6253538131713867,
+      "learning_rate": 4.206465310708058e-05,
+      "loss": 1.841,
+      "step": 2224
+    },
+    {
+      "epoch": 1.1484536082474226,
+      "grad_norm": 1.4791438579559326,
+      "learning_rate": 4.189639920775346e-05,
+      "loss": 1.6623,
+      "step": 2228
+    },
+    {
+      "epoch": 1.1505154639175257,
+      "grad_norm": 1.4935940504074097,
+      "learning_rate": 4.172823949449807e-05,
+      "loss": 1.7644,
+      "step": 2232
+    },
+    {
+      "epoch": 1.1525773195876288,
+      "grad_norm": 1.6122525930404663,
+      "learning_rate": 4.156017592179158e-05,
+      "loss": 1.8102,
+      "step": 2236
+    },
+    {
+      "epoch": 1.1546391752577319,
+      "grad_norm": 1.6047066450119019,
+      "learning_rate": 4.139221044299376e-05,
+      "loss": 1.8315,
+      "step": 2240
+    },
+    {
+      "epoch": 1.156701030927835,
+      "grad_norm": 1.4906078577041626,
+      "learning_rate": 4.122434501032423e-05,
+      "loss": 1.7467,
+      "step": 2244
+    },
+    {
+      "epoch": 1.158762886597938,
+      "grad_norm": 2.4341137409210205,
+      "learning_rate": 4.105658157483983e-05,
+      "loss": 1.828,
+      "step": 2248
+    },
+    {
+      "epoch": 1.1608247422680413,
+      "grad_norm": 1.490577220916748,
+      "learning_rate": 4.088892208641188e-05,
+      "loss": 1.792,
+      "step": 2252
+    },
+    {
+      "epoch": 1.1628865979381444,
+      "grad_norm": 1.4628297090530396,
+      "learning_rate": 4.072136849370363e-05,
+      "loss": 1.765,
+      "step": 2256
+    },
+    {
+      "epoch": 1.1649484536082475,
+      "grad_norm": 1.4456610679626465,
+      "learning_rate": 4.055392274414743e-05,
+      "loss": 1.8382,
+      "step": 2260
+    },
+    {
+      "epoch": 1.1670103092783506,
+      "grad_norm": 1.5569915771484375,
+      "learning_rate": 4.0386586783922256e-05,
+      "loss": 1.8598,
+      "step": 2264
+    },
+    {
+      "epoch": 1.1690721649484537,
+      "grad_norm": 1.5139881372451782,
+      "learning_rate": 4.021936255793101e-05,
+      "loss": 1.7757,
+      "step": 2268
+    },
+    {
+      "epoch": 1.1711340206185568,
+      "grad_norm": 1.5133105516433716,
+      "learning_rate": 4.005225200977796e-05,
+      "loss": 1.7901,
+      "step": 2272
+    },
+    {
+      "epoch": 1.1731958762886598,
+      "grad_norm": 1.4687427282333374,
+      "learning_rate": 3.9885257081746104e-05,
+      "loss": 1.8429,
+      "step": 2276
+    },
+    {
+      "epoch": 1.175257731958763,
+      "grad_norm": 1.58542001247406,
+      "learning_rate": 3.97183797147746e-05,
+      "loss": 1.724,
+      "step": 2280
+    },
+    {
+      "epoch": 1.177319587628866,
+      "grad_norm": 1.4827735424041748,
+      "learning_rate": 3.955162184843625e-05,
+      "loss": 1.7716,
+      "step": 2284
+    },
+    {
+      "epoch": 1.179381443298969,
+      "grad_norm": 1.5209972858428955,
+      "learning_rate": 3.938498542091494e-05,
+      "loss": 1.8294,
+      "step": 2288
+    },
+    {
+      "epoch": 1.1814432989690722,
+      "grad_norm": 1.5304073095321655,
+      "learning_rate": 3.9218472368983084e-05,
+      "loss": 1.8146,
+      "step": 2292
+    },
+    {
+      "epoch": 1.1835051546391753,
+      "grad_norm": 1.4535797834396362,
+      "learning_rate": 3.905208462797914e-05,
+      "loss": 1.6667,
+      "step": 2296
+    },
+    {
+      "epoch": 1.1855670103092784,
+      "grad_norm": 1.5424301624298096,
+      "learning_rate": 3.888582413178509e-05,
+      "loss": 1.8117,
+      "step": 2300
+    },
+    {
+      "epoch": 1.1876288659793814,
+      "grad_norm": 1.5533775091171265,
+      "learning_rate": 3.8719692812804034e-05,
+      "loss": 1.779,
+      "step": 2304
+    },
+    {
+      "epoch": 1.1896907216494845,
+      "grad_norm": 1.5786696672439575,
+      "learning_rate": 3.8553692601937636e-05,
+      "loss": 1.8124,
+      "step": 2308
+    },
+    {
+      "epoch": 1.1917525773195876,
+      "grad_norm": 1.9105570316314697,
+      "learning_rate": 3.8387825428563706e-05,
+      "loss": 1.7474,
+      "step": 2312
+    },
+    {
+      "epoch": 1.1938144329896907,
+      "grad_norm": 1.4653416872024536,
+      "learning_rate": 3.822209322051384e-05,
+      "loss": 1.7107,
+      "step": 2316
+    },
+    {
+      "epoch": 1.1958762886597938,
+      "grad_norm": 1.6114842891693115,
+      "learning_rate": 3.805649790405094e-05,
+      "loss": 1.8546,
+      "step": 2320
+    },
+    {
+      "epoch": 1.1979381443298969,
+      "grad_norm": 1.525752305984497,
+      "learning_rate": 3.789104140384686e-05,
+      "loss": 1.7314,
+      "step": 2324
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.5032085180282593,
+      "learning_rate": 3.772572564296005e-05,
+      "loss": 1.884,
+      "step": 2328
+    },
+    {
+      "epoch": 1.202061855670103,
+      "grad_norm": 1.60364830493927,
+      "learning_rate": 3.7560552542813125e-05,
+      "loss": 1.7848,
+      "step": 2332
+    },
+    {
+      "epoch": 1.2041237113402061,
+      "grad_norm": 1.538434624671936,
+      "learning_rate": 3.739552402317064e-05,
+      "loss": 1.7257,
+      "step": 2336
+    },
+    {
+      "epoch": 1.2061855670103092,
+      "grad_norm": 1.4421312808990479,
+      "learning_rate": 3.723064200211674e-05,
+      "loss": 1.6767,
+      "step": 2340
+    },
+    {
+      "epoch": 1.2082474226804123,
+      "grad_norm": 1.5462045669555664,
+      "learning_rate": 3.706590839603278e-05,
+      "loss": 1.7877,
+      "step": 2344
+    },
+    {
+      "epoch": 1.2103092783505154,
+      "grad_norm": 1.4567534923553467,
+      "learning_rate": 3.6901325119575194e-05,
+      "loss": 1.7842,
+      "step": 2348
+    },
+    {
+      "epoch": 1.2123711340206185,
+      "grad_norm": 2.0819544792175293,
+      "learning_rate": 3.673689408565317e-05,
+      "loss": 1.7632,
+      "step": 2352
+    },
+    {
+      "epoch": 1.2144329896907216,
+      "grad_norm": 1.5570515394210815,
+      "learning_rate": 3.657261720540639e-05,
+      "loss": 1.7674,
+      "step": 2356
+    },
+    {
+      "epoch": 1.2164948453608249,
+      "grad_norm": 1.6284637451171875,
+      "learning_rate": 3.640849638818286e-05,
+      "loss": 1.7434,
+      "step": 2360
+    },
+    {
+      "epoch": 1.218556701030928,
+      "grad_norm": 1.4900153875350952,
+      "learning_rate": 3.6244533541516713e-05,
+      "loss": 1.7535,
+      "step": 2364
+    },
+    {
+      "epoch": 1.220618556701031,
+      "grad_norm": 1.5037250518798828,
+      "learning_rate": 3.608073057110601e-05,
+      "loss": 1.7759,
+      "step": 2368
+    },
+    {
+      "epoch": 1.2226804123711341,
+      "grad_norm": 5.1282219886779785,
+      "learning_rate": 3.591708938079068e-05,
+      "loss": 1.7459,
+      "step": 2372
+    },
+    {
+      "epoch": 1.2247422680412372,
+      "grad_norm": 1.5764257907867432,
+      "learning_rate": 3.575361187253019e-05,
+      "loss": 1.8166,
+      "step": 2376
+    },
+    {
+      "epoch": 1.2268041237113403,
+      "grad_norm": 1.4848092794418335,
+      "learning_rate": 3.55902999463817e-05,
+      "loss": 1.7265,
+      "step": 2380
+    },
+    {
+      "epoch": 1.2288659793814434,
+      "grad_norm": 1.583311915397644,
+      "learning_rate": 3.5427155500477795e-05,
+      "loss": 1.7755,
+      "step": 2384
+    },
+    {
+      "epoch": 1.2309278350515465,
+      "grad_norm": 1.9810250997543335,
+      "learning_rate": 3.526418043100448e-05,
+      "loss": 1.8337,
+      "step": 2388
+    },
+    {
+      "epoch": 1.2329896907216495,
+      "grad_norm": 1.5246814489364624,
+      "learning_rate": 3.510137663217915e-05,
+      "loss": 1.7465,
+      "step": 2392
+    },
+    {
+      "epoch": 1.2350515463917526,
+      "grad_norm": 1.4694105386734009,
+      "learning_rate": 3.4938745996228586e-05,
+      "loss": 1.7879,
+      "step": 2396
+    },
+    {
+      "epoch": 1.2371134020618557,
+      "grad_norm": 1.551598072052002,
+      "learning_rate": 3.47762904133669e-05,
+      "loss": 1.7552,
+      "step": 2400
+    },
+    {
+      "epoch": 1.2391752577319588,
+      "grad_norm": 1.4880276918411255,
+      "learning_rate": 3.4614011771773644e-05,
+      "loss": 1.7289,
+      "step": 2404
+    },
+    {
+      "epoch": 1.2412371134020619,
+      "grad_norm": 1.4992886781692505,
+      "learning_rate": 3.445191195757179e-05,
+      "loss": 1.7777,
+      "step": 2408
+    },
+    {
+      "epoch": 1.243298969072165,
+      "grad_norm": 1.5845531225204468,
+      "learning_rate": 3.428999285480586e-05,
+      "loss": 1.7689,
+      "step": 2412
+    },
+    {
+      "epoch": 1.245360824742268,
+      "grad_norm": 1.5566850900650024,
+      "learning_rate": 3.412825634542005e-05,
+      "loss": 1.7837,
+      "step": 2416
+    },
+    {
+      "epoch": 1.2474226804123711,
+      "grad_norm": 1.5015497207641602,
+      "learning_rate": 3.396670430923628e-05,
+      "loss": 1.7094,
+      "step": 2420
+    },
+    {
+      "epoch": 1.2494845360824742,
+      "grad_norm": 1.4368141889572144,
+      "learning_rate": 3.380533862393236e-05,
+      "loss": 1.7203,
+      "step": 2424
+    },
+    {
+      "epoch": 1.2515463917525773,
+      "grad_norm": 1.4907705783843994,
+      "learning_rate": 3.364416116502027e-05,
+      "loss": 1.7477,
+      "step": 2428
+    },
+    {
+      "epoch": 1.2536082474226804,
+      "grad_norm": 2.8708791732788086,
+      "learning_rate": 3.348317380582423e-05,
+      "loss": 1.7573,
+      "step": 2432
+    },
+    {
+      "epoch": 1.2556701030927835,
+      "grad_norm": 1.5418363809585571,
+      "learning_rate": 3.332237841745898e-05,
+      "loss": 1.7518,
+      "step": 2436
+    },
+    {
+      "epoch": 1.2577319587628866,
+      "grad_norm": 1.4115269184112549,
+      "learning_rate": 3.3161776868808036e-05,
+      "loss": 1.7183,
+      "step": 2440
+    },
+    {
+      "epoch": 1.2597938144329897,
+      "grad_norm": 1.5407153367996216,
+      "learning_rate": 3.300137102650197e-05,
+      "loss": 1.7318,
+      "step": 2444
+    },
+    {
+      "epoch": 1.2618556701030927,
+      "grad_norm": 1.5344724655151367,
+      "learning_rate": 3.284116275489672e-05,
+      "loss": 1.7652,
+      "step": 2448
+    },
+    {
+      "epoch": 1.2639175257731958,
+      "grad_norm": 1.4697588682174683,
+      "learning_rate": 3.268115391605189e-05,
+      "loss": 1.7609,
+      "step": 2452
+    },
+    {
+      "epoch": 1.265979381443299,
+      "grad_norm": 1.5398411750793457,
+      "learning_rate": 3.2521346369709135e-05,
+      "loss": 1.7362,
+      "step": 2456
+    },
+    {
+      "epoch": 1.268041237113402,
+      "grad_norm": 1.5244160890579224,
+      "learning_rate": 3.236174197327055e-05,
+      "loss": 1.7365,
+      "step": 2460
+    },
+    {
+      "epoch": 1.270103092783505,
+      "grad_norm": 1.5290905237197876,
+      "learning_rate": 3.220234258177707e-05,
+      "loss": 1.7776,
+      "step": 2464
+    },
+    {
+      "epoch": 1.2721649484536082,
+      "grad_norm": 1.4930541515350342,
+      "learning_rate": 3.2043150047886894e-05,
+      "loss": 1.7694,
+      "step": 2468
+    },
+    {
+      "epoch": 1.2742268041237113,
+      "grad_norm": 1.489328384399414,
+      "learning_rate": 3.188416622185397e-05,
+      "loss": 1.8005,
+      "step": 2472
+    },
+    {
+      "epoch": 1.2762886597938143,
+      "grad_norm": 1.4238892793655396,
+      "learning_rate": 3.1725392951506504e-05,
+      "loss": 1.6492,
+      "step": 2476
+    },
+    {
+      "epoch": 1.2783505154639174,
+      "grad_norm": 1.566266417503357,
+      "learning_rate": 3.156683208222544e-05,
+      "loss": 1.719,
+      "step": 2480
+    },
+    {
+      "epoch": 1.2804123711340205,
+      "grad_norm": 1.5726135969161987,
+      "learning_rate": 3.140848545692309e-05,
+      "loss": 1.7275,
+      "step": 2484
+    },
+    {
+      "epoch": 1.2824742268041236,
+      "grad_norm": 1.5072052478790283,
+      "learning_rate": 3.12503549160216e-05,
+      "loss": 1.7749,
+      "step": 2488
+    },
+    {
+      "epoch": 1.2845360824742267,
+      "grad_norm": 1.540624976158142,
+      "learning_rate": 3.109244229743167e-05,
+      "loss": 1.8149,
+      "step": 2492
+    },
+    {
+      "epoch": 1.2865979381443298,
+      "grad_norm": 1.9997202157974243,
+      "learning_rate": 3.0934749436531116e-05,
+      "loss": 1.7481,
+      "step": 2496
+    },
+    {
+      "epoch": 1.2886597938144329,
+      "grad_norm": 1.4298206567764282,
+      "learning_rate": 3.0777278166143616e-05,
+      "loss": 1.7337,
+      "step": 2500
+    },
+    {
+      "epoch": 1.2907216494845362,
+      "grad_norm": 1.5551060438156128,
+      "learning_rate": 3.062003031651727e-05,
+      "loss": 1.8913,
+      "step": 2504
+    },
+    {
+      "epoch": 1.2927835051546392,
+      "grad_norm": 2.138140916824341,
+      "learning_rate": 3.0463007715303483e-05,
+      "loss": 1.7596,
+      "step": 2508
+    },
+    {
+      "epoch": 1.2948453608247423,
+      "grad_norm": 1.5046266317367554,
+      "learning_rate": 3.0306212187535653e-05,
+      "loss": 1.8142,
+      "step": 2512
+    },
+    {
+      "epoch": 1.2969072164948454,
+      "grad_norm": 1.426291823387146,
+      "learning_rate": 3.014964555560792e-05,
+      "loss": 1.7556,
+      "step": 2516
+    },
+    {
+      "epoch": 1.2989690721649485,
+      "grad_norm": 1.474760890007019,
+      "learning_rate": 2.9993309639254092e-05,
+      "loss": 1.8039,
+      "step": 2520
+    },
+    {
+      "epoch": 1.3010309278350516,
+      "grad_norm": 1.5093271732330322,
+      "learning_rate": 2.9837206255526373e-05,
+      "loss": 1.7842,
+      "step": 2524
+    },
+    {
+      "epoch": 1.3030927835051547,
+      "grad_norm": 1.5908783674240112,
+      "learning_rate": 2.9681337218774313e-05,
+      "loss": 1.846,
+      "step": 2528
+    },
+    {
+      "epoch": 1.3051546391752578,
+      "grad_norm": 1.6266577243804932,
+      "learning_rate": 2.952570434062371e-05,
+      "loss": 1.7245,
+      "step": 2532
+    },
+    {
+      "epoch": 1.3072164948453608,
+      "grad_norm": 1.5125893354415894,
+      "learning_rate": 2.9370309429955595e-05,
+      "loss": 1.749,
+      "step": 2536
+    },
+    {
+      "epoch": 1.309278350515464,
+      "grad_norm": 1.4698082208633423,
+      "learning_rate": 2.9215154292885062e-05,
+      "loss": 1.8698,
+      "step": 2540
+    },
+    {
+      "epoch": 1.311340206185567,
+      "grad_norm": 1.5628530979156494,
+      "learning_rate": 2.9060240732740467e-05,
+      "loss": 1.7998,
+      "step": 2544
+    },
+    {
+      "epoch": 1.31340206185567,
+      "grad_norm": 1.476222276687622,
+      "learning_rate": 2.890557055004236e-05,
+      "loss": 1.6143,
+      "step": 2548
+    },
+    {
+      "epoch": 1.3154639175257732,
+      "grad_norm": 1.5117274522781372,
+      "learning_rate": 2.875114554248256e-05,
+      "loss": 1.6858,
+      "step": 2552
+    },
+    {
+      "epoch": 1.3175257731958763,
+      "grad_norm": 1.5859644412994385,
+      "learning_rate": 2.8596967504903283e-05,
+      "loss": 1.8203,
+      "step": 2556
+    },
+    {
+      "epoch": 1.3195876288659794,
+      "grad_norm": 1.470045566558838,
+      "learning_rate": 2.844303822927627e-05,
+      "loss": 1.6019,
+      "step": 2560
+    },
+    {
+      "epoch": 1.3216494845360824,
+      "grad_norm": 1.5127053260803223,
+      "learning_rate": 2.828935950468201e-05,
+      "loss": 1.7554,
+      "step": 2564
+    },
+    {
+      "epoch": 1.3237113402061855,
+      "grad_norm": 1.5124480724334717,
+      "learning_rate": 2.8135933117288872e-05,
+      "loss": 1.7442,
+      "step": 2568
+    },
+    {
+      "epoch": 1.3257731958762886,
+      "grad_norm": 1.54438316822052,
+      "learning_rate": 2.7982760850332316e-05,
+      "loss": 1.7874,
+      "step": 2572
+    },
+    {
+      "epoch": 1.3278350515463917,
+      "grad_norm": 1.460144281387329,
+      "learning_rate": 2.7829844484094318e-05,
+      "loss": 1.6712,
+      "step": 2576
+    },
+    {
+      "epoch": 1.3298969072164948,
+      "grad_norm": 1.5282338857650757,
+      "learning_rate": 2.767718579588252e-05,
+      "loss": 1.7423,
+      "step": 2580
+    },
+    {
+      "epoch": 1.3319587628865979,
+      "grad_norm": 1.6081974506378174,
+      "learning_rate": 2.7524786560009707e-05,
+      "loss": 1.715,
+      "step": 2584
+    },
+    {
+      "epoch": 1.334020618556701,
+      "grad_norm": 2.33564829826355,
+      "learning_rate": 2.737264854777306e-05,
+      "loss": 1.769,
+      "step": 2588
+    },
+    {
+      "epoch": 1.3360824742268043,
+      "grad_norm": 1.538969874382019,
+      "learning_rate": 2.7220773527433625e-05,
+      "loss": 1.732,
+      "step": 2592
+    },
+    {
+      "epoch": 1.3381443298969073,
+      "grad_norm": 1.4337162971496582,
+      "learning_rate": 2.7069163264195784e-05,
+      "loss": 1.7052,
+      "step": 2596
+    },
+    {
+      "epoch": 1.3402061855670104,
+      "grad_norm": 1.4599614143371582,
+      "learning_rate": 2.691781952018671e-05,
+      "loss": 1.7344,
+      "step": 2600
+    },
+    {
+      "epoch": 1.3422680412371135,
+      "grad_norm": 1.4979077577590942,
+      "learning_rate": 2.6766744054435883e-05,
+      "loss": 1.7518,
+      "step": 2604
+    },
+    {
+      "epoch": 1.3443298969072166,
+      "grad_norm": 1.5297064781188965,
+      "learning_rate": 2.6615938622854664e-05,
+      "loss": 1.733,
+      "step": 2608
+    },
+    {
+      "epoch": 1.3463917525773197,
+      "grad_norm": 1.50015127658844,
+      "learning_rate": 2.6465404978215857e-05,
+      "loss": 1.6939,
+      "step": 2612
+    },
+    {
+      "epoch": 1.3484536082474228,
+      "grad_norm": 1.5101268291473389,
+      "learning_rate": 2.6315144870133384e-05,
+      "loss": 1.6588,
+      "step": 2616
+    },
+    {
+      "epoch": 1.3505154639175259,
+      "grad_norm": 1.506935715675354,
+      "learning_rate": 2.6165160045041904e-05,
+      "loss": 1.7295,
+      "step": 2620
+    },
+    {
+      "epoch": 1.352577319587629,
+      "grad_norm": 1.9392681121826172,
+      "learning_rate": 2.6015452246176537e-05,
+      "loss": 1.7802,
+      "step": 2624
+    },
+    {
+      "epoch": 1.354639175257732,
+      "grad_norm": 1.5243134498596191,
+      "learning_rate": 2.586602321355257e-05,
+      "loss": 1.6937,
+      "step": 2628
+    },
+    {
+      "epoch": 1.3567010309278351,
+      "grad_norm": 1.5403796434402466,
+      "learning_rate": 2.571687468394536e-05,
+      "loss": 1.7298,
+      "step": 2632
+    },
+    {
+      "epoch": 1.3587628865979382,
+      "grad_norm": 1.4517806768417358,
+      "learning_rate": 2.5568008390869915e-05,
+      "loss": 1.6566,
+      "step": 2636
+    },
+    {
+      "epoch": 1.3608247422680413,
+      "grad_norm": 1.4871841669082642,
+      "learning_rate": 2.5419426064560957e-05,
+      "loss": 1.7271,
+      "step": 2640
+    },
+    {
+      "epoch": 1.3628865979381444,
+      "grad_norm": 1.4928479194641113,
+      "learning_rate": 2.5271129431952727e-05,
+      "loss": 1.7903,
+      "step": 2644
+    },
+    {
+      "epoch": 1.3649484536082475,
+      "grad_norm": 1.517552137374878,
+      "learning_rate": 2.5123120216658903e-05,
+      "loss": 1.803,
+      "step": 2648
+    },
+    {
+      "epoch": 1.3670103092783505,
+      "grad_norm": 1.4770328998565674,
+      "learning_rate": 2.497540013895262e-05,
+      "loss": 1.7156,
+      "step": 2652
+    },
+    {
+      "epoch": 1.3690721649484536,
+      "grad_norm": 1.518353819847107,
+      "learning_rate": 2.482797091574642e-05,
+      "loss": 1.6648,
+      "step": 2656
+    },
+    {
+      "epoch": 1.3711340206185567,
+      "grad_norm": 1.5015162229537964,
+      "learning_rate": 2.4680834260572304e-05,
+      "loss": 1.7288,
+      "step": 2660
+    },
+    {
+      "epoch": 1.3731958762886598,
+      "grad_norm": 1.5462713241577148,
+      "learning_rate": 2.4533991883561868e-05,
+      "loss": 1.8064,
+      "step": 2664
+    },
+    {
+      "epoch": 1.3752577319587629,
+      "grad_norm": 1.4880030155181885,
+      "learning_rate": 2.4387445491426327e-05,
+      "loss": 1.7886,
+      "step": 2668
+    },
+    {
+      "epoch": 1.377319587628866,
+      "grad_norm": 1.5615061521530151,
+      "learning_rate": 2.424119678743677e-05,
+      "loss": 1.643,
+      "step": 2672
+    },
+    {
+      "epoch": 1.379381443298969,
+      "grad_norm": 1.5558855533599854,
+      "learning_rate": 2.4095247471404387e-05,
+      "loss": 1.7335,
+      "step": 2676
+    },
+    {
+      "epoch": 1.3814432989690721,
+      "grad_norm": 1.5481622219085693,
+      "learning_rate": 2.3949599239660585e-05,
+      "loss": 1.7311,
+      "step": 2680
+    },
+    {
+      "epoch": 1.3835051546391752,
+      "grad_norm": 1.471251130104065,
+      "learning_rate": 2.380425378503738e-05,
+      "loss": 1.711,
+      "step": 2684
+    },
+    {
+      "epoch": 1.3855670103092783,
+      "grad_norm": 1.4730710983276367,
+      "learning_rate": 2.365921279684767e-05,
+      "loss": 1.6672,
+      "step": 2688
+    },
+    {
+      "epoch": 1.3876288659793814,
+      "grad_norm": 1.4712542295455933,
+      "learning_rate": 2.3514477960865627e-05,
+      "loss": 1.628,
+      "step": 2692
+    },
+    {
+      "epoch": 1.3896907216494845,
+      "grad_norm": 1.4011591672897339,
+      "learning_rate": 2.33700509593071e-05,
+      "loss": 1.6386,
+      "step": 2696
+    },
+    {
+      "epoch": 1.3917525773195876,
+      "grad_norm": 1.4686334133148193,
+      "learning_rate": 2.322593347081005e-05,
+      "loss": 1.6377,
+      "step": 2700
+    },
+    {
+      "epoch": 1.3938144329896907,
+      "grad_norm": 1.5750575065612793,
+      "learning_rate": 2.308212717041505e-05,
+      "loss": 1.7587,
+      "step": 2704
+    },
+    {
+      "epoch": 1.3958762886597937,
+      "grad_norm": 1.5493422746658325,
+      "learning_rate": 2.2938633729545816e-05,
+      "loss": 1.7323,
+      "step": 2708
+    },
+    {
+      "epoch": 1.3979381443298968,
+      "grad_norm": 1.4504088163375854,
+      "learning_rate": 2.279545481598977e-05,
+      "loss": 1.7473,
+      "step": 2712
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.5989000797271729,
+      "learning_rate": 2.2652592093878666e-05,
+      "loss": 1.7669,
+      "step": 2716
+    },
+    {
+      "epoch": 1.402061855670103,
+      "grad_norm": 1.5275282859802246,
+      "learning_rate": 2.2510047223669233e-05,
+      "loss": 1.8059,
+      "step": 2720
+    },
+    {
+      "epoch": 1.404123711340206,
+      "grad_norm": 1.5826573371887207,
+      "learning_rate": 2.2367821862123944e-05,
+      "loss": 1.7529,
+      "step": 2724
+    },
+    {
+      "epoch": 1.4061855670103092,
+      "grad_norm": 1.5159717798233032,
+      "learning_rate": 2.2225917662291663e-05,
+      "loss": 1.6796,
+      "step": 2728
+    },
+    {
+      "epoch": 1.4082474226804123,
+      "grad_norm": 1.537111759185791,
+      "learning_rate": 2.2084336273488443e-05,
+      "loss": 1.6562,
+      "step": 2732
+    },
+    {
+      "epoch": 1.4103092783505153,
+      "grad_norm": 1.5044008493423462,
+      "learning_rate": 2.1943079341278427e-05,
+      "loss": 1.6649,
+      "step": 2736
+    },
+    {
+      "epoch": 1.4123711340206184,
+      "grad_norm": 1.5891298055648804,
+      "learning_rate": 2.180214850745467e-05,
+      "loss": 1.7776,
+      "step": 2740
+    },
+    {
+      "epoch": 1.4144329896907217,
+      "grad_norm": 1.5218358039855957,
+      "learning_rate": 2.166154541002011e-05,
+      "loss": 1.6714,
+      "step": 2744
+    },
+    {
+      "epoch": 1.4164948453608248,
+      "grad_norm": 1.467380404472351,
+      "learning_rate": 2.152127168316843e-05,
+      "loss": 1.6669,
+      "step": 2748
+    },
+    {
+      "epoch": 1.418556701030928,
+      "grad_norm": 1.4052174091339111,
+      "learning_rate": 2.1381328957265152e-05,
+      "loss": 1.5786,
+      "step": 2752
+    },
+    {
+      "epoch": 1.420618556701031,
+      "grad_norm": 1.505531907081604,
+      "learning_rate": 2.1241718858828653e-05,
+      "loss": 1.7733,
+      "step": 2756
+    },
+    {
+      "epoch": 1.422680412371134,
+      "grad_norm": 1.9099241495132446,
+      "learning_rate": 2.110244301051128e-05,
+      "loss": 1.6561,
+      "step": 2760
+    },
+    {
+      "epoch": 1.4247422680412372,
+      "grad_norm": 1.6298562288284302,
+      "learning_rate": 2.0963503031080415e-05,
+      "loss": 1.8327,
+      "step": 2764
+    },
+    {
+      "epoch": 1.4268041237113402,
+      "grad_norm": 1.4871844053268433,
+      "learning_rate": 2.0824900535399834e-05,
+      "loss": 1.726,
+      "step": 2768
+    },
+    {
+      "epoch": 1.4288659793814433,
+      "grad_norm": 1.4929816722869873,
+      "learning_rate": 2.068663713441073e-05,
+      "loss": 1.7089,
+      "step": 2772
+    },
+    {
+      "epoch": 1.4309278350515464,
+      "grad_norm": 1.5244849920272827,
+      "learning_rate": 2.054871443511313e-05,
+      "loss": 1.724,
+      "step": 2776
+    },
+    {
+      "epoch": 1.4329896907216495,
+      "grad_norm": 1.3867377042770386,
+      "learning_rate": 2.0411134040547154e-05,
+      "loss": 1.6098,
+      "step": 2780
+    },
+    {
+      "epoch": 1.4350515463917526,
+      "grad_norm": 1.552032470703125,
+      "learning_rate": 2.0273897549774416e-05,
+      "loss": 1.7457,
+      "step": 2784
+    },
+    {
+      "epoch": 1.4371134020618557,
+      "grad_norm": 1.4442955255508423,
+      "learning_rate": 2.0137006557859396e-05,
+      "loss": 1.7092,
+      "step": 2788
+    },
+    {
+      "epoch": 1.4391752577319588,
+      "grad_norm": 1.504491925239563,
+      "learning_rate": 2.000046265585099e-05,
+      "loss": 1.6646,
+      "step": 2792
+    },
+    {
+      "epoch": 1.4412371134020618,
+      "grad_norm": 1.5086458921432495,
+      "learning_rate": 1.986426743076391e-05,
+      "loss": 1.738,
+      "step": 2796
+    },
+    {
+      "epoch": 1.443298969072165,
+      "grad_norm": 1.4446042776107788,
+      "learning_rate": 1.9728422465560275e-05,
+      "loss": 1.6802,
+      "step": 2800
+    },
+    {
+      "epoch": 1.445360824742268,
+      "grad_norm": 1.5059337615966797,
+      "learning_rate": 1.9592929339131243e-05,
+      "loss": 1.5487,
+      "step": 2804
+    },
+    {
+      "epoch": 1.447422680412371,
+      "grad_norm": 1.4459666013717651,
+      "learning_rate": 1.9457789626278657e-05,
+      "loss": 1.6508,
+      "step": 2808
+    },
+    {
+      "epoch": 1.4494845360824742,
+      "grad_norm": 1.757739543914795,
+      "learning_rate": 1.9323004897696673e-05,
+      "loss": 1.78,
+      "step": 2812
+    },
+    {
+      "epoch": 1.4515463917525773,
+      "grad_norm": 1.5750166177749634,
+      "learning_rate": 1.9188576719953633e-05,
+      "loss": 1.7814,
+      "step": 2816
+    },
+    {
+      "epoch": 1.4536082474226804,
+      "grad_norm": 1.4686044454574585,
+      "learning_rate": 1.9054506655473724e-05,
+      "loss": 1.6798,
+      "step": 2820
+    },
+    {
+      "epoch": 1.4556701030927834,
+      "grad_norm": 1.4810421466827393,
+      "learning_rate": 1.892079626251888e-05,
+      "loss": 1.653,
+      "step": 2824
+    },
+    {
+      "epoch": 1.4577319587628865,
+      "grad_norm": 1.4591325521469116,
+      "learning_rate": 1.8787447095170702e-05,
+      "loss": 1.7042,
+      "step": 2828
+    },
+    {
+      "epoch": 1.4597938144329896,
+      "grad_norm": 1.4883931875228882,
+      "learning_rate": 1.8654460703312266e-05,
+      "loss": 1.7581,
+      "step": 2832
+    },
+    {
+      "epoch": 1.461855670103093,
+      "grad_norm": 1.4359487295150757,
+      "learning_rate": 1.852183863261032e-05,
+      "loss": 1.6085,
+      "step": 2836
+    },
+    {
+      "epoch": 1.463917525773196,
+      "grad_norm": 1.4442452192306519,
+      "learning_rate": 1.8389582424497136e-05,
+      "loss": 1.6518,
+      "step": 2840
+    },
+    {
+      "epoch": 1.465979381443299,
+      "grad_norm": 1.5850920677185059,
+      "learning_rate": 1.825769361615266e-05,
+      "loss": 1.7989,
+      "step": 2844
+    },
+    {
+      "epoch": 1.4680412371134022,
+      "grad_norm": 1.4874738454818726,
+      "learning_rate": 1.8126173740486664e-05,
+      "loss": 1.7408,
+      "step": 2848
+    },
+    {
+      "epoch": 1.4701030927835053,
+      "grad_norm": 1.5400047302246094,
+      "learning_rate": 1.799502432612089e-05,
+      "loss": 1.7256,
+      "step": 2852
+    },
+    {
+      "epoch": 1.4721649484536083,
+      "grad_norm": 1.4548730850219727,
+      "learning_rate": 1.7864246897371316e-05,
+      "loss": 1.7029,
+      "step": 2856
+    },
+    {
+      "epoch": 1.4742268041237114,
+      "grad_norm": 1.4862723350524902,
+      "learning_rate": 1.773384297423043e-05,
+      "loss": 1.6857,
+      "step": 2860
+    },
+    {
+      "epoch": 1.4762886597938145,
+      "grad_norm": 1.4751530885696411,
+      "learning_rate": 1.760381407234955e-05,
+      "loss": 1.6826,
+      "step": 2864
+    },
+    {
+      "epoch": 1.4783505154639176,
+      "grad_norm": 1.4540187120437622,
+      "learning_rate": 1.7474161703021218e-05,
+      "loss": 1.6407,
+      "step": 2868
+    },
+    {
+      "epoch": 1.4804123711340207,
+      "grad_norm": 1.55220365524292,
+      "learning_rate": 1.734488737316165e-05,
+      "loss": 1.7539,
+      "step": 2872
+    },
+    {
+      "epoch": 1.4824742268041238,
+      "grad_norm": 1.4643796682357788,
+      "learning_rate": 1.721599258529319e-05,
+      "loss": 1.7111,
+      "step": 2876
+    },
+    {
+      "epoch": 1.4845360824742269,
+      "grad_norm": 1.6646052598953247,
+      "learning_rate": 1.708747883752685e-05,
+      "loss": 1.7609,
+      "step": 2880
+    },
+    {
+      "epoch": 1.48659793814433,
+      "grad_norm": 1.3996634483337402,
+      "learning_rate": 1.695934762354497e-05,
+      "loss": 1.6767,
+      "step": 2884
+    },
+    {
+      "epoch": 1.488659793814433,
+      "grad_norm": 1.4679023027420044,
+      "learning_rate": 1.6831600432583727e-05,
+      "loss": 1.6345,
+      "step": 2888
+    },
+    {
+      "epoch": 1.4907216494845361,
+      "grad_norm": 1.4191251993179321,
+      "learning_rate": 1.6704238749415957e-05,
+      "loss": 1.689,
+      "step": 2892
+    },
+    {
+      "epoch": 1.4927835051546392,
+      "grad_norm": 1.6434361934661865,
+      "learning_rate": 1.6577264054333768e-05,
+      "loss": 1.7641,
+      "step": 2896
+    },
+    {
+      "epoch": 1.4948453608247423,
+      "grad_norm": 1.5503323078155518,
+      "learning_rate": 1.6450677823131472e-05,
+      "loss": 1.6991,
+      "step": 2900
+    },
+    {
+      "epoch": 1.4969072164948454,
+      "grad_norm": 1.7964752912521362,
+      "learning_rate": 1.6324481527088302e-05,
+      "loss": 1.707,
+      "step": 2904
+    },
+    {
+      "epoch": 1.4989690721649485,
+      "grad_norm": 1.470279574394226,
+      "learning_rate": 1.6198676632951464e-05,
+      "loss": 1.7253,
+      "step": 2908
+    },
+    {
+      "epoch": 1.5010309278350515,
+      "grad_norm": 1.5459221601486206,
+      "learning_rate": 1.6073264602918918e-05,
+      "loss": 1.6298,
+      "step": 2912
+    },
+    {
+      "epoch": 1.5030927835051546,
+      "grad_norm": 1.4339414834976196,
+      "learning_rate": 1.5948246894622497e-05,
+      "loss": 1.6155,
+      "step": 2916
+    },
+    {
+      "epoch": 1.5051546391752577,
+      "grad_norm": 1.5148845911026,
+      "learning_rate": 1.582362496111094e-05,
+      "loss": 1.7269,
+      "step": 2920
+    },
+    {
+      "epoch": 1.5072164948453608,
+      "grad_norm": 1.8622448444366455,
+      "learning_rate": 1.569940025083297e-05,
+      "loss": 1.6962,
+      "step": 2924
+    },
+    {
+      "epoch": 1.5092783505154639,
+      "grad_norm": 1.5076379776000977,
+      "learning_rate": 1.5575574207620517e-05,
+      "loss": 1.6369,
+      "step": 2928
+    },
+    {
+      "epoch": 1.511340206185567,
+      "grad_norm": 1.511597752571106,
+      "learning_rate": 1.5452148270671894e-05,
+      "loss": 1.7466,
+      "step": 2932
+    },
+    {
+      "epoch": 1.51340206185567,
+      "grad_norm": 1.4678515195846558,
+      "learning_rate": 1.5329123874535085e-05,
+      "loss": 1.7594,
+      "step": 2936
+    },
+    {
+      "epoch": 1.5154639175257731,
+      "grad_norm": 1.545318841934204,
+      "learning_rate": 1.5206502449091054e-05,
+      "loss": 1.7206,
+      "step": 2940
+    },
+    {
+      "epoch": 1.5175257731958762,
+      "grad_norm": 1.458013653755188,
+      "learning_rate": 1.5084285419537153e-05,
+      "loss": 1.7171,
+      "step": 2944
+    },
+    {
+      "epoch": 1.5195876288659793,
+      "grad_norm": 1.4504350423812866,
+      "learning_rate": 1.4962474206370541e-05,
+      "loss": 1.7554,
+      "step": 2948
+    },
+    {
+      "epoch": 1.5216494845360824,
+      "grad_norm": 1.4534108638763428,
+      "learning_rate": 1.4841070225371673e-05,
+      "loss": 1.6591,
+      "step": 2952
+    },
+    {
+      "epoch": 1.5237113402061855,
+      "grad_norm": 1.5048794746398926,
+      "learning_rate": 1.4720074887587887e-05,
+      "loss": 1.6382,
+      "step": 2956
+    },
+    {
+      "epoch": 1.5257731958762886,
+      "grad_norm": 1.566748857498169,
+      "learning_rate": 1.4599489599316896e-05,
+      "loss": 1.73,
+      "step": 2960
+    },
+    {
+      "epoch": 1.5278350515463917,
+      "grad_norm": 1.5701326131820679,
+      "learning_rate": 1.4479315762090562e-05,
+      "loss": 1.694,
+      "step": 2964
+    },
+    {
+      "epoch": 1.5298969072164947,
+      "grad_norm": 1.5283926725387573,
+      "learning_rate": 1.4359554772658552e-05,
+      "loss": 1.6844,
+      "step": 2968
+    },
+    {
+      "epoch": 1.5319587628865978,
+      "grad_norm": 1.5629454851150513,
+      "learning_rate": 1.4240208022972074e-05,
+      "loss": 1.7401,
+      "step": 2972
+    },
+    {
+      "epoch": 1.534020618556701,
+      "grad_norm": 1.537468671798706,
+      "learning_rate": 1.4121276900167796e-05,
+      "loss": 1.7075,
+      "step": 2976
+    },
+    {
+      "epoch": 1.536082474226804,
+      "grad_norm": 1.488130807876587,
+      "learning_rate": 1.400276278655162e-05,
+      "loss": 1.7203,
+      "step": 2980
+    },
+    {
+      "epoch": 1.538144329896907,
+      "grad_norm": 1.385069727897644,
+      "learning_rate": 1.3884667059582656e-05,
+      "loss": 1.6419,
+      "step": 2984
+    },
+    {
+      "epoch": 1.5402061855670102,
+      "grad_norm": 1.4615826606750488,
+      "learning_rate": 1.3766991091857246e-05,
+      "loss": 1.7021,
+      "step": 2988
+    },
+    {
+      "epoch": 1.5422680412371133,
+      "grad_norm": 1.5640617609024048,
+      "learning_rate": 1.3649736251092898e-05,
+      "loss": 1.7353,
+      "step": 2992
+    },
+    {
+      "epoch": 1.5443298969072163,
+      "grad_norm": 1.561741590499878,
+      "learning_rate": 1.353290390011258e-05,
+      "loss": 1.6911,
+      "step": 2996
+    },
+    {
+      "epoch": 1.5463917525773194,
+      "grad_norm": 1.5131316184997559,
+      "learning_rate": 1.3416495396828694e-05,
+      "loss": 1.6645,
+      "step": 3000
+    },
+    {
+      "epoch": 1.5484536082474227,
+      "grad_norm": 1.5184184312820435,
+      "learning_rate": 1.3300512094227391e-05,
+      "loss": 1.6424,
+      "step": 3004
+    },
+    {
+      "epoch": 1.5505154639175258,
+      "grad_norm": 1.4697967767715454,
+      "learning_rate": 1.3184955340352828e-05,
+      "loss": 1.7499,
+      "step": 3008
+    },
+    {
+      "epoch": 1.552577319587629,
+      "grad_norm": 1.62870192527771,
+      "learning_rate": 1.3069826478291486e-05,
+      "loss": 1.676,
+      "step": 3012
+    },
+    {
+      "epoch": 1.554639175257732,
+      "grad_norm": 1.531766653060913,
+      "learning_rate": 1.2955126846156573e-05,
+      "loss": 1.6913,
+      "step": 3016
+    },
+    {
+      "epoch": 1.556701030927835,
+      "grad_norm": 2.1785354614257812,
+      "learning_rate": 1.2840857777072468e-05,
+      "loss": 1.6899,
+      "step": 3020
+    },
+    {
+      "epoch": 1.5587628865979382,
+      "grad_norm": 1.403219223022461,
+      "learning_rate": 1.2727020599159235e-05,
+      "loss": 1.6918,
+      "step": 3024
+    },
+    {
+      "epoch": 1.5608247422680412,
+      "grad_norm": 1.5640937089920044,
+      "learning_rate": 1.2613616635517161e-05,
+      "loss": 1.6787,
+      "step": 3028
+    },
+    {
+      "epoch": 1.5628865979381443,
+      "grad_norm": 1.4356787204742432,
+      "learning_rate": 1.2500647204211402e-05,
+      "loss": 1.6923,
+      "step": 3032
+    },
+    {
+      "epoch": 1.5649484536082474,
+      "grad_norm": 1.4457017183303833,
+      "learning_rate": 1.238811361825667e-05,
+      "loss": 1.6461,
+      "step": 3036
+    },
+    {
+      "epoch": 1.5670103092783505,
+      "grad_norm": 2.227945566177368,
+      "learning_rate": 1.2276017185601912e-05,
+      "loss": 1.7129,
+      "step": 3040
+    },
+    {
+      "epoch": 1.5690721649484536,
+      "grad_norm": 1.4856458902359009,
+      "learning_rate": 1.2164359209115234e-05,
+      "loss": 1.6834,
+      "step": 3044
+    },
+    {
+      "epoch": 1.5711340206185567,
+      "grad_norm": 1.5406701564788818,
+      "learning_rate": 1.2053140986568612e-05,
+      "loss": 1.6185,
+      "step": 3048
+    },
+    {
+      "epoch": 1.5731958762886598,
+      "grad_norm": 1.554215908050537,
+      "learning_rate": 1.1942363810622909e-05,
+      "loss": 1.6404,
+      "step": 3052
+    },
+    {
+      "epoch": 1.5752577319587628,
+      "grad_norm": 1.6283259391784668,
+      "learning_rate": 1.1832028968812774e-05,
+      "loss": 1.6818,
+      "step": 3056
+    },
+    {
+      "epoch": 1.577319587628866,
+      "grad_norm": 1.4719226360321045,
+      "learning_rate": 1.172213774353177e-05,
+      "loss": 1.6619,
+      "step": 3060
+    },
+    {
+      "epoch": 1.5793814432989692,
+      "grad_norm": 1.5907206535339355,
+      "learning_rate": 1.1612691412017373e-05,
+      "loss": 1.6963,
+      "step": 3064
+    },
+    {
+      "epoch": 1.5814432989690723,
+      "grad_norm": 1.426649570465088,
+      "learning_rate": 1.1503691246336234e-05,
+      "loss": 1.7719,
+      "step": 3068
+    },
+    {
+      "epoch": 1.5835051546391754,
+      "grad_norm": 1.5006170272827148,
+      "learning_rate": 1.1395138513369286e-05,
+      "loss": 1.632,
+      "step": 3072
+    },
+    {
+      "epoch": 1.5855670103092785,
+      "grad_norm": 1.4917073249816895,
+      "learning_rate": 1.1287034474797065e-05,
+      "loss": 1.7659,
+      "step": 3076
+    },
+    {
+      "epoch": 1.5876288659793816,
+      "grad_norm": 1.4859620332717896,
+      "learning_rate": 1.117938038708507e-05,
+      "loss": 1.7446,
+      "step": 3080
+    },
+    {
+      "epoch": 1.5896907216494847,
+      "grad_norm": 1.4105993509292603,
+      "learning_rate": 1.1072177501469128e-05,
+      "loss": 1.6825,
+      "step": 3084
+    },
+    {
+      "epoch": 1.5917525773195877,
+      "grad_norm": 1.499107837677002,
+      "learning_rate": 1.0965427063940853e-05,
+      "loss": 1.7547,
+      "step": 3088
+    },
+    {
+      "epoch": 1.5938144329896908,
+      "grad_norm": 1.5192339420318604,
+      "learning_rate": 1.0859130315233174e-05,
+      "loss": 1.6612,
+      "step": 3092
+    },
+    {
+      "epoch": 1.595876288659794,
+      "grad_norm": 1.4649794101715088,
+      "learning_rate": 1.0753288490805918e-05,
+      "loss": 1.6028,
+      "step": 3096
+    },
+    {
+      "epoch": 1.597938144329897,
+      "grad_norm": 1.538676381111145,
+      "learning_rate": 1.064790282083144e-05,
+      "loss": 1.5495,
+      "step": 3100
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.5038882493972778,
+      "learning_rate": 1.0542974530180327e-05,
+      "loss": 1.7314,
+      "step": 3104
+    },
+    {
+      "epoch": 1.6020618556701032,
+      "grad_norm": 1.5052063465118408,
+      "learning_rate": 1.0438504838407165e-05,
+      "loss": 1.6695,
+      "step": 3108
+    },
+    {
+      "epoch": 1.6041237113402063,
+      "grad_norm": 1.5474481582641602,
+      "learning_rate": 1.0334494959736347e-05,
+      "loss": 1.6846,
+      "step": 3112
+    },
+    {
+      "epoch": 1.6061855670103093,
+      "grad_norm": 1.4491521120071411,
+      "learning_rate": 1.0230946103048022e-05,
+      "loss": 1.601,
+      "step": 3116
+    },
+    {
+      "epoch": 1.6082474226804124,
+      "grad_norm": 1.4010319709777832,
+      "learning_rate": 1.012785947186397e-05,
+      "loss": 1.701,
+      "step": 3120
+    },
+    {
+      "epoch": 1.6103092783505155,
+      "grad_norm": 1.4654611349105835,
+      "learning_rate": 1.002523626433361e-05,
+      "loss": 1.6416,
+      "step": 3124
+    },
+    {
+      "epoch": 1.6123711340206186,
+      "grad_norm": 1.5731595754623413,
+      "learning_rate": 9.92307767322016e-06,
+      "loss": 1.7372,
+      "step": 3128
+    },
+    {
+      "epoch": 1.6144329896907217,
+      "grad_norm": 1.546984076499939,
+      "learning_rate": 9.821384885886676e-06,
+      "loss": 1.7713,
+      "step": 3132
+    },
+    {
+      "epoch": 1.6164948453608248,
+      "grad_norm": 1.4809237718582153,
+      "learning_rate": 9.720159084282355e-06,
+      "loss": 1.6343,
+      "step": 3136
+    },
+    {
+      "epoch": 1.6185567010309279,
+      "grad_norm": 1.392882227897644,
+      "learning_rate": 9.619401444928683e-06,
+      "loss": 1.6566,
+      "step": 3140
+    },
+    {
+      "epoch": 1.620618556701031,
+      "grad_norm": 1.4652302265167236,
+      "learning_rate": 9.519113138905838e-06,
+      "loss": 1.7887,
+      "step": 3144
+    },
+    {
+      "epoch": 1.622680412371134,
+      "grad_norm": 1.5458775758743286,
+      "learning_rate": 9.419295331839061e-06,
+      "loss": 1.731,
+      "step": 3148
+    },
+    {
+      "epoch": 1.6247422680412371,
+      "grad_norm": 1.5418813228607178,
+      "learning_rate": 9.319949183885108e-06,
+      "loss": 1.6952,
+      "step": 3152
+    },
+    {
+      "epoch": 1.6268041237113402,
+      "grad_norm": 1.5179048776626587,
+      "learning_rate": 9.221075849718713e-06,
+      "loss": 1.7358,
+      "step": 3156
+    },
+    {
+      "epoch": 1.6288659793814433,
+      "grad_norm": 1.53363037109375,
+      "learning_rate": 9.1226764785193e-06,
+      "loss": 1.6322,
+      "step": 3160
+    },
+    {
+      "epoch": 1.6309278350515464,
+      "grad_norm": 1.4504185914993286,
+      "learning_rate": 9.024752213957482e-06,
+      "loss": 1.6023,
+      "step": 3164
+    },
+    {
+      "epoch": 1.6329896907216495,
+      "grad_norm": 1.5878132581710815,
+      "learning_rate": 8.927304194181857e-06,
+      "loss": 1.7398,
+      "step": 3168
+    },
+    {
+      "epoch": 1.6350515463917525,
+      "grad_norm": 1.533153772354126,
+      "learning_rate": 8.83033355180573e-06,
+      "loss": 1.766,
+      "step": 3172
+    },
+    {
+      "epoch": 1.6371134020618556,
+      "grad_norm": 1.4958655834197998,
+      "learning_rate": 8.73384141389399e-06,
+      "loss": 1.6772,
+      "step": 3176
+    },
+    {
+      "epoch": 1.6391752577319587,
+      "grad_norm": 1.4422835111618042,
+      "learning_rate": 8.63782890194998e-06,
+      "loss": 1.6434,
+      "step": 3180
+    },
+    {
+      "epoch": 1.6412371134020618,
+      "grad_norm": 1.6030116081237793,
+      "learning_rate": 8.5422971319025e-06,
+      "loss": 1.5916,
+      "step": 3184
+    },
+    {
+      "epoch": 1.6432989690721649,
+      "grad_norm": 1.4988054037094116,
+      "learning_rate": 8.447247214092768e-06,
+      "loss": 1.7199,
+      "step": 3188
+    },
+    {
+      "epoch": 1.645360824742268,
+      "grad_norm": 1.5074199438095093,
+      "learning_rate": 8.35268025326158e-06,
+      "loss": 1.6741,
+      "step": 3192
+    },
+    {
+      "epoch": 1.647422680412371,
+      "grad_norm": 1.4830085039138794,
+      "learning_rate": 8.25859734853645e-06,
+      "loss": 1.637,
+      "step": 3196
+    },
+    {
+      "epoch": 1.6494845360824741,
+      "grad_norm": 1.6016579866409302,
+      "learning_rate": 8.164999593418826e-06,
+      "loss": 1.6371,
+      "step": 3200
+    },
+    {
+      "epoch": 1.6515463917525772,
+      "grad_norm": 1.4918121099472046,
+      "learning_rate": 8.071888075771378e-06,
+      "loss": 1.6303,
+      "step": 3204
+    },
+    {
+      "epoch": 1.6536082474226803,
+      "grad_norm": 1.42795991897583,
+      "learning_rate": 7.979263877805394e-06,
+      "loss": 1.6543,
+      "step": 3208
+    },
+    {
+      "epoch": 1.6556701030927834,
+      "grad_norm": 1.5172784328460693,
+      "learning_rate": 7.887128076068134e-06,
+      "loss": 1.71,
+      "step": 3212
+    },
+    {
+      "epoch": 1.6577319587628865,
+      "grad_norm": 1.493607521057129,
+      "learning_rate": 7.795481741430393e-06,
+      "loss": 1.6678,
+      "step": 3216
+    },
+    {
+      "epoch": 1.6597938144329896,
+      "grad_norm": 1.5664525032043457,
+      "learning_rate": 7.704325939073958e-06,
+      "loss": 1.7857,
+      "step": 3220
+    },
+    {
+      "epoch": 1.6618556701030927,
+      "grad_norm": 1.5040996074676514,
+      "learning_rate": 7.613661728479321e-06,
+      "loss": 1.8031,
+      "step": 3224
+    },
+    {
+      "epoch": 1.6639175257731957,
+      "grad_norm": 1.462308645248413,
+      "learning_rate": 7.52349016341335e-06,
+      "loss": 1.7017,
+      "step": 3228
+    },
+    {
+      "epoch": 1.6659793814432988,
+      "grad_norm": 1.4423810243606567,
+      "learning_rate": 7.433812291916992e-06,
+      "loss": 1.7352,
+      "step": 3232
+    },
+    {
+      "epoch": 1.668041237113402,
+      "grad_norm": 1.4877485036849976,
+      "learning_rate": 7.344629156293114e-06,
+      "loss": 1.6151,
+      "step": 3236
+    },
+    {
+      "epoch": 1.670103092783505,
+      "grad_norm": 1.4941354990005493,
+      "learning_rate": 7.255941793094423e-06,
+      "loss": 1.6128,
+      "step": 3240
+    },
+    {
+      "epoch": 1.6721649484536083,
+      "grad_norm": 1.5927449464797974,
+      "learning_rate": 7.167751233111358e-06,
+      "loss": 1.706,
+      "step": 3244
+    },
+    {
+      "epoch": 1.6742268041237114,
+      "grad_norm": 1.6141644716262817,
+      "learning_rate": 7.080058501360171e-06,
+      "loss": 1.6318,
+      "step": 3248
+    },
+    {
+      "epoch": 1.6762886597938145,
+      "grad_norm": 1.5904043912887573,
+      "learning_rate": 6.9928646170709656e-06,
+      "loss": 1.6387,
+      "step": 3252
+    },
+    {
+      "epoch": 1.6783505154639176,
+      "grad_norm": 1.4306963682174683,
+      "learning_rate": 6.906170593675876e-06,
+      "loss": 1.6231,
+      "step": 3256
+    },
+    {
+      "epoch": 1.6804123711340206,
+      "grad_norm": 1.6362873315811157,
+      "learning_rate": 6.819977438797281e-06,
+      "loss": 1.7401,
+      "step": 3260
+    },
+    {
+      "epoch": 1.6824742268041237,
+      "grad_norm": 1.4902920722961426,
+      "learning_rate": 6.734286154236091e-06,
+      "loss": 1.676,
+      "step": 3264
+    },
+    {
+      "epoch": 1.6845360824742268,
+      "grad_norm": 1.499423623085022,
+      "learning_rate": 6.649097735960108e-06,
+      "loss": 1.643,
+      "step": 3268
+    },
+    {
+      "epoch": 1.68659793814433,
+      "grad_norm": 1.4293397665023804,
+      "learning_rate": 6.564413174092443e-06,
+      "loss": 1.6775,
+      "step": 3272
+    },
+    {
+      "epoch": 1.688659793814433,
+      "grad_norm": 1.5098601579666138,
+      "learning_rate": 6.480233452900036e-06,
+      "loss": 1.5725,
+      "step": 3276
+    },
+    {
+      "epoch": 1.690721649484536,
+      "grad_norm": 1.6427370309829712,
+      "learning_rate": 6.396559550782177e-06,
+      "loss": 1.7431,
+      "step": 3280
+    },
+    {
+      "epoch": 1.6927835051546392,
+      "grad_norm": 1.4890693426132202,
+      "learning_rate": 6.313392440259136e-06,
+      "loss": 1.6801,
+      "step": 3284
+    },
+    {
+      "epoch": 1.6948453608247422,
+      "grad_norm": 1.4871695041656494,
+      "learning_rate": 6.230733087960888e-06,
+      "loss": 1.7111,
+      "step": 3288
+    },
+    {
+      "epoch": 1.6969072164948453,
+      "grad_norm": 1.6153513193130493,
+      "learning_rate": 6.14858245461587e-06,
+      "loss": 1.7601,
+      "step": 3292
+    },
+    {
+      "epoch": 1.6989690721649484,
+      "grad_norm": 1.5377050638198853,
+      "learning_rate": 6.066941495039796e-06,
+      "loss": 1.6788,
+      "step": 3296
+    },
+    {
+      "epoch": 1.7010309278350515,
+      "grad_norm": 1.4935017824172974,
+      "learning_rate": 5.985811158124599e-06,
+      "loss": 1.7559,
+      "step": 3300
+    },
+    {
+      "epoch": 1.7030927835051546,
+      "grad_norm": 1.4736721515655518,
+      "learning_rate": 5.905192386827352e-06,
+      "loss": 1.731,
+      "step": 3304
+    },
+    {
+      "epoch": 1.705154639175258,
+      "grad_norm": 1.5431056022644043,
+      "learning_rate": 5.825086118159329e-06,
+      "loss": 1.6675,
+      "step": 3308
+    },
+    {
+      "epoch": 1.707216494845361,
+      "grad_norm": 1.7205324172973633,
+      "learning_rate": 5.745493283175146e-06,
+      "loss": 1.6615,
+      "step": 3312
+    },
+    {
+      "epoch": 1.709278350515464,
+      "grad_norm": 1.5387599468231201,
+      "learning_rate": 5.666414806961856e-06,
+      "loss": 1.7063,
+      "step": 3316
+    },
+    {
+      "epoch": 1.7113402061855671,
+      "grad_norm": 1.43263578414917,
+      "learning_rate": 5.58785160862832e-06,
+      "loss": 1.6548,
+      "step": 3320
+    },
+    {
+      "epoch": 1.7134020618556702,
+      "grad_norm": 1.5297513008117676,
+      "learning_rate": 5.50980460129441e-06,
+      "loss": 1.6402,
+      "step": 3324
+    },
+    {
+      "epoch": 1.7154639175257733,
+      "grad_norm": 1.521336555480957,
+      "learning_rate": 5.432274692080464e-06,
+      "loss": 1.6973,
+      "step": 3328
+    },
+    {
+      "epoch": 1.7175257731958764,
+      "grad_norm": 1.4514061212539673,
+      "learning_rate": 5.355262782096709e-06,
+      "loss": 1.6918,
+      "step": 3332
+    },
+    {
+      "epoch": 1.7195876288659795,
+      "grad_norm": 1.48029363155365,
+      "learning_rate": 5.278769766432801e-06,
+      "loss": 1.7185,
+      "step": 3336
+    },
+    {
+      "epoch": 1.7216494845360826,
+      "grad_norm": 1.4712656736373901,
+      "learning_rate": 5.202796534147436e-06,
+      "loss": 1.6631,
+      "step": 3340
+    },
+    {
+      "epoch": 1.7237113402061857,
+      "grad_norm": 1.4765143394470215,
+      "learning_rate": 5.127343968257969e-06,
+      "loss": 1.6347,
+      "step": 3344
+    },
+    {
+      "epoch": 1.7257731958762887,
+      "grad_norm": 1.5265347957611084,
+      "learning_rate": 5.05241294573024e-06,
+      "loss": 1.7666,
+      "step": 3348
+    },
+    {
+      "epoch": 1.7278350515463918,
+      "grad_norm": 1.5441725254058838,
+      "learning_rate": 4.978004337468256e-06,
+      "loss": 1.6875,
+      "step": 3352
+    },
+    {
+      "epoch": 1.729896907216495,
+      "grad_norm": 1.4827260971069336,
+      "learning_rate": 4.904119008304175e-06,
+      "loss": 1.63,
+      "step": 3356
+    },
+    {
+      "epoch": 1.731958762886598,
+      "grad_norm": 1.488405466079712,
+      "learning_rate": 4.830757816988219e-06,
+      "loss": 1.6433,
+      "step": 3360
+    },
+    {
+      "epoch": 1.734020618556701,
+      "grad_norm": 1.5134429931640625,
+      "learning_rate": 4.757921616178662e-06,
+      "loss": 1.5377,
+      "step": 3364
+    },
+    {
+      "epoch": 1.7360824742268042,
+      "grad_norm": 1.469539999961853,
+      "learning_rate": 4.685611252431998e-06,
+      "loss": 1.6067,
+      "step": 3368
+    },
+    {
+      "epoch": 1.7381443298969073,
+      "grad_norm": 1.5114245414733887,
+      "learning_rate": 4.6138275661930075e-06,
+      "loss": 1.6444,
+      "step": 3372
+    },
+    {
+      "epoch": 1.7402061855670103,
+      "grad_norm": 1.5059422254562378,
+      "learning_rate": 4.54257139178505e-06,
+      "loss": 1.7052,
+      "step": 3376
+    },
+    {
+      "epoch": 1.7422680412371134,
+      "grad_norm": 1.5071382522583008,
+      "learning_rate": 4.4718435574003514e-06,
+      "loss": 1.7082,
+      "step": 3380
+    },
+    {
+      "epoch": 1.7443298969072165,
+      "grad_norm": 1.500165581703186,
+      "learning_rate": 4.401644885090356e-06,
+      "loss": 1.6129,
+      "step": 3384
+    },
+    {
+      "epoch": 1.7463917525773196,
+      "grad_norm": 1.5666110515594482,
+      "learning_rate": 4.331976190756226e-06,
+      "loss": 1.7796,
+      "step": 3388
+    },
+    {
+      "epoch": 1.7484536082474227,
+      "grad_norm": 1.4586031436920166,
+      "learning_rate": 4.2628382841393145e-06,
+      "loss": 1.6676,
+      "step": 3392
+    },
+    {
+      "epoch": 1.7505154639175258,
+      "grad_norm": 1.5573093891143799,
+      "learning_rate": 4.194231968811757e-06,
+      "loss": 1.7022,
+      "step": 3396
+    },
+    {
+      "epoch": 1.7525773195876289,
+      "grad_norm": 1.5714795589447021,
+      "learning_rate": 4.126158042167139e-06,
+      "loss": 1.7316,
+      "step": 3400
+    },
+    {
+      "epoch": 1.754639175257732,
+      "grad_norm": 1.4940961599349976,
+      "learning_rate": 4.058617295411243e-06,
+      "loss": 1.6096,
+      "step": 3404
+    },
+    {
+      "epoch": 1.756701030927835,
+      "grad_norm": 1.6426243782043457,
+      "learning_rate": 3.991610513552829e-06,
+      "loss": 1.6747,
+      "step": 3408
+    },
+    {
+      "epoch": 1.7587628865979381,
+      "grad_norm": 1.5700215101242065,
+      "learning_rate": 3.925138475394513e-06,
+      "loss": 1.7568,
+      "step": 3412
+    },
+    {
+      "epoch": 1.7608247422680412,
+      "grad_norm": 7.2929534912109375,
+      "learning_rate": 3.859201953523739e-06,
+      "loss": 1.8217,
+      "step": 3416
+    },
+    {
+      "epoch": 1.7628865979381443,
+      "grad_norm": 1.4354559183120728,
+      "learning_rate": 3.7938017143037685e-06,
+      "loss": 1.6345,
+      "step": 3420
+    },
+    {
+      "epoch": 1.7649484536082474,
+      "grad_norm": 1.439919114112854,
+      "learning_rate": 3.728938517864794e-06,
+      "loss": 1.6007,
+      "step": 3424
+    },
+    {
+      "epoch": 1.7670103092783505,
+      "grad_norm": 1.496739387512207,
+      "learning_rate": 3.6646131180951006e-06,
+      "loss": 1.7252,
+      "step": 3428
+    },
+    {
+      "epoch": 1.7690721649484535,
+      "grad_norm": 1.4635599851608276,
+      "learning_rate": 3.6008262626322843e-06,
+      "loss": 1.647,
+      "step": 3432
+    },
+    {
+      "epoch": 1.7711340206185566,
+      "grad_norm": 1.4412612915039062,
+      "learning_rate": 3.5375786928546097e-06,
+      "loss": 1.6424,
+      "step": 3436
+    },
+    {
+      "epoch": 1.7731958762886597,
+      "grad_norm": 1.5021198987960815,
+      "learning_rate": 3.474871143872338e-06,
+      "loss": 1.6989,
+      "step": 3440
+    },
+    {
+      "epoch": 1.7752577319587628,
+      "grad_norm": 1.5215860605239868,
+      "learning_rate": 3.412704344519213e-06,
+      "loss": 1.59,
+      "step": 3444
+    },
+    {
+      "epoch": 1.7773195876288659,
+      "grad_norm": 1.4903206825256348,
+      "learning_rate": 3.3510790173439775e-06,
+      "loss": 1.6253,
+      "step": 3448
+    },
+    {
+      "epoch": 1.779381443298969,
+      "grad_norm": 1.4466136693954468,
+      "learning_rate": 3.2899958786019926e-06,
+      "loss": 1.6756,
+      "step": 3452
+    },
+    {
+      "epoch": 1.781443298969072,
+      "grad_norm": 1.7129707336425781,
+      "learning_rate": 3.229455638246898e-06,
+      "loss": 1.6763,
+      "step": 3456
+    },
+    {
+      "epoch": 1.7835051546391751,
+      "grad_norm": 1.4847286939620972,
+      "learning_rate": 3.16945899992237e-06,
+      "loss": 1.7183,
+      "step": 3460
+    },
+    {
+      "epoch": 1.7855670103092782,
+      "grad_norm": 1.4779984951019287,
+      "learning_rate": 3.1100066609539347e-06,
+      "loss": 1.7174,
+      "step": 3464
+    },
+    {
+      "epoch": 1.7876288659793813,
+      "grad_norm": 1.5085519552230835,
+      "learning_rate": 3.0510993123408604e-06,
+      "loss": 1.6961,
+      "step": 3468
+    },
+    {
+      "epoch": 1.7896907216494844,
+      "grad_norm": 1.5164207220077515,
+      "learning_rate": 2.992737638748144e-06,
+      "loss": 1.5474,
+      "step": 3472
+    },
+    {
+      "epoch": 1.7917525773195875,
+      "grad_norm": 1.5061603784561157,
+      "learning_rate": 2.9349223184985288e-06,
+      "loss": 1.6408,
+      "step": 3476
+    },
+    {
+      "epoch": 1.7938144329896906,
+      "grad_norm": 1.4625122547149658,
+      "learning_rate": 2.8776540235646463e-06,
+      "loss": 1.6354,
+      "step": 3480
+    },
+    {
+      "epoch": 1.7958762886597937,
+      "grad_norm": 1.4752477407455444,
+      "learning_rate": 2.8209334195611836e-06,
+      "loss": 1.7005,
+      "step": 3484
+    },
+    {
+      "epoch": 1.797938144329897,
+      "grad_norm": 1.5524309873580933,
+      "learning_rate": 2.764761165737162e-06,
+      "loss": 1.6028,
+      "step": 3488
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.4997243881225586,
+      "learning_rate": 2.7091379149682685e-06,
+      "loss": 1.6429,
+      "step": 3492
+    },
+    {
+      "epoch": 1.8020618556701031,
+      "grad_norm": 1.5430649518966675,
+      "learning_rate": 2.654064313749266e-06,
+      "loss": 1.7119,
+      "step": 3496
+    },
+    {
+      "epoch": 1.8041237113402062,
+      "grad_norm": 1.5093491077423096,
+      "learning_rate": 2.5995410021864787e-06,
+      "loss": 1.6618,
+      "step": 3500
+    },
+    {
+      "epoch": 1.8061855670103093,
+      "grad_norm": 1.492345929145813,
+      "learning_rate": 2.545568613990362e-06,
+      "loss": 1.6848,
+      "step": 3504
+    },
+    {
+      "epoch": 1.8082474226804124,
+      "grad_norm": 1.542244553565979,
+      "learning_rate": 2.4921477764681457e-06,
+      "loss": 1.7365,
+      "step": 3508
+    },
+    {
+      "epoch": 1.8103092783505155,
+      "grad_norm": 1.4747071266174316,
+      "learning_rate": 2.4392791105164846e-06,
+      "loss": 1.6802,
+      "step": 3512
+    },
+    {
+      "epoch": 1.8123711340206186,
+      "grad_norm": 1.4944431781768799,
+      "learning_rate": 2.386963230614325e-06,
+      "loss": 1.6634,
+      "step": 3516
+    },
+    {
+      "epoch": 1.8144329896907216,
+      "grad_norm": 1.4797273874282837,
+      "learning_rate": 2.335200744815708e-06,
+      "loss": 1.7111,
+      "step": 3520
+    },
+    {
+      "epoch": 1.8164948453608247,
+      "grad_norm": 1.4942731857299805,
+      "learning_rate": 2.283992254742706e-06,
+      "loss": 1.7994,
+      "step": 3524
+    },
+    {
+      "epoch": 1.8185567010309278,
+      "grad_norm": 1.5504451990127563,
+      "learning_rate": 2.2333383555784737e-06,
+      "loss": 1.7724,
+      "step": 3528
+    },
+    {
+      "epoch": 1.820618556701031,
+      "grad_norm": 1.5781549215316772,
+      "learning_rate": 2.1832396360602693e-06,
+      "loss": 1.695,
+      "step": 3532
+    },
+    {
+      "epoch": 1.822680412371134,
+      "grad_norm": 1.7646689414978027,
+      "learning_rate": 2.1336966784726465e-06,
+      "loss": 1.7736,
+      "step": 3536
+    },
+    {
+      "epoch": 1.824742268041237,
+      "grad_norm": 1.4293087720870972,
+      "learning_rate": 2.0847100586406877e-06,
+      "loss": 1.608,
+      "step": 3540
+    },
+    {
+      "epoch": 1.8268041237113402,
+      "grad_norm": 1.5219712257385254,
+      "learning_rate": 2.0362803459232914e-06,
+      "loss": 1.6738,
+      "step": 3544
+    },
+    {
+      "epoch": 1.8288659793814435,
+      "grad_norm": 1.4095622301101685,
+      "learning_rate": 1.9884081032065737e-06,
+      "loss": 1.6249,
+      "step": 3548
+    },
+    {
+      "epoch": 1.8309278350515465,
+      "grad_norm": 3.822401285171509,
+      "learning_rate": 1.9410938868973327e-06,
+      "loss": 1.7157,
+      "step": 3552
+    },
+    {
+      "epoch": 1.8329896907216496,
+      "grad_norm": 1.7696354389190674,
+      "learning_rate": 1.894338246916555e-06,
+      "loss": 1.6029,
+      "step": 3556
+    },
+    {
+      "epoch": 1.8350515463917527,
+      "grad_norm": 1.427482008934021,
+      "learning_rate": 1.848141726693048e-06,
+      "loss": 1.5971,
+      "step": 3560
+    },
+    {
+      "epoch": 1.8371134020618558,
+      "grad_norm": 1.6495184898376465,
+      "learning_rate": 1.802504863157095e-06,
+      "loss": 1.6344,
+      "step": 3564
+    },
+    {
+      "epoch": 1.839175257731959,
+      "grad_norm": 1.4954864978790283,
+      "learning_rate": 1.7574281867342502e-06,
+      "loss": 1.7968,
+      "step": 3568
+    },
+    {
+      "epoch": 1.841237113402062,
+      "grad_norm": 1.578550934791565,
+      "learning_rate": 1.7129122213391523e-06,
+      "loss": 1.5697,
+      "step": 3572
+    },
+    {
+      "epoch": 1.843298969072165,
+      "grad_norm": 1.8335249423980713,
+      "learning_rate": 1.6689574843694433e-06,
+      "loss": 1.6689,
+      "step": 3576
+    },
+    {
+      "epoch": 1.8453608247422681,
+      "grad_norm": 1.4485599994659424,
+      "learning_rate": 1.6255644866997378e-06,
+      "loss": 1.606,
+      "step": 3580
+    },
+    {
+      "epoch": 1.8474226804123712,
+      "grad_norm": 1.5148478746414185,
+      "learning_rate": 1.5827337326757174e-06,
+      "loss": 1.7588,
+      "step": 3584
+    },
+    {
+      "epoch": 1.8494845360824743,
+      "grad_norm": 1.4764747619628906,
+      "learning_rate": 1.5404657201082363e-06,
+      "loss": 1.6629,
+      "step": 3588
+    },
+    {
+      "epoch": 1.8515463917525774,
+      "grad_norm": 1.4880666732788086,
+      "learning_rate": 1.4987609402675573e-06,
+      "loss": 1.6173,
+      "step": 3592
+    },
+    {
+      "epoch": 1.8536082474226805,
+      "grad_norm": 1.5709367990493774,
+      "learning_rate": 1.4576198778776195e-06,
+      "loss": 1.6901,
+      "step": 3596
+    },
+    {
+      "epoch": 1.8556701030927836,
+      "grad_norm": 1.4975337982177734,
+      "learning_rate": 1.4170430111104472e-06,
+      "loss": 1.7681,
+      "step": 3600
+    },
+    {
+      "epoch": 1.8577319587628867,
+      "grad_norm": 1.6037880182266235,
+      "learning_rate": 1.377030811580543e-06,
+      "loss": 1.7217,
+      "step": 3604
+    },
+    {
+      "epoch": 1.8597938144329897,
+      "grad_norm": 1.5378186702728271,
+      "learning_rate": 1.3375837443394157e-06,
+      "loss": 1.7365,
+      "step": 3608
+    },
+    {
+      "epoch": 1.8618556701030928,
+      "grad_norm": 1.527868628501892,
+      "learning_rate": 1.298702267870211e-06,
+      "loss": 1.6275,
+      "step": 3612
+    },
+    {
+      "epoch": 1.863917525773196,
+      "grad_norm": 1.530971646308899,
+      "learning_rate": 1.2603868340823444e-06,
+      "loss": 1.7038,
+      "step": 3616
+    },
+    {
+      "epoch": 1.865979381443299,
+      "grad_norm": 1.5488992929458618,
+      "learning_rate": 1.2226378883062716e-06,
+      "loss": 1.6488,
+      "step": 3620
+    },
+    {
+      "epoch": 1.868041237113402,
+      "grad_norm": 1.4633854627609253,
+      "learning_rate": 1.185455869288299e-06,
+      "loss": 1.7151,
+      "step": 3624
+    },
+    {
+      "epoch": 1.8701030927835052,
+      "grad_norm": 1.463667631149292,
+      "learning_rate": 1.1488412091854917e-06,
+      "loss": 1.7143,
+      "step": 3628
+    },
+    {
+      "epoch": 1.8721649484536083,
+      "grad_norm": 1.5287612676620483,
+      "learning_rate": 1.112794333560635e-06,
+      "loss": 1.7644,
+      "step": 3632
+    },
+    {
+      "epoch": 1.8742268041237113,
+      "grad_norm": 1.5180222988128662,
+      "learning_rate": 1.0773156613773206e-06,
+      "loss": 1.6957,
+      "step": 3636
+    },
+    {
+      "epoch": 1.8762886597938144,
+      "grad_norm": 1.490254282951355,
+      "learning_rate": 1.0424056049950392e-06,
+      "loss": 1.6732,
+      "step": 3640
+    },
+    {
+      "epoch": 1.8783505154639175,
+      "grad_norm": 1.6682502031326294,
+      "learning_rate": 1.008064570164413e-06,
+      "loss": 1.6909,
+      "step": 3644
+    },
+    {
+      "epoch": 1.8804123711340206,
+      "grad_norm": 1.5792622566223145,
+      "learning_rate": 9.742929560224879e-07,
+      "loss": 1.7305,
+      "step": 3648
+    },
+    {
+      "epoch": 1.8824742268041237,
+      "grad_norm": 1.510890245437622,
+      "learning_rate": 9.410911550880475e-07,
+      "loss": 1.5359,
+      "step": 3652
+    },
+    {
+      "epoch": 1.8845360824742268,
+      "grad_norm": 1.5118955373764038,
+      "learning_rate": 9.084595532571127e-07,
+      "loss": 1.7005,
+      "step": 3656
+    },
+    {
+      "epoch": 1.8865979381443299,
+      "grad_norm": 1.5225536823272705,
+      "learning_rate": 8.763985297984167e-07,
+      "loss": 1.699,
+      "step": 3660
+    },
+    {
+      "epoch": 1.888659793814433,
+      "grad_norm": 1.539660930633545,
+      "learning_rate": 8.449084573489918e-07,
+      "loss": 1.7721,
+      "step": 3664
+    },
+    {
+      "epoch": 1.890721649484536,
+      "grad_norm": 1.3900866508483887,
+      "learning_rate": 8.13989701909873e-07,
+      "loss": 1.53,
+      "step": 3668
+    },
+    {
+      "epoch": 1.8927835051546391,
+      "grad_norm": 1.519671082496643,
+      "learning_rate": 7.836426228418292e-07,
+      "loss": 1.6317,
+      "step": 3672
+    },
+    {
+      "epoch": 1.8948453608247422,
+      "grad_norm": 1.4589074850082397,
+      "learning_rate": 7.538675728611555e-07,
+      "loss": 1.6417,
+      "step": 3676
+    },
+    {
+      "epoch": 1.8969072164948453,
+      "grad_norm": 1.5633991956710815,
+      "learning_rate": 7.246648980356153e-07,
+      "loss": 1.6138,
+      "step": 3680
+    },
+    {
+      "epoch": 1.8989690721649484,
+      "grad_norm": 1.5480862855911255,
+      "learning_rate": 6.960349377804043e-07,
+      "loss": 1.6623,
+      "step": 3684
+    },
+    {
+      "epoch": 1.9010309278350515,
+      "grad_norm": 1.5118290185928345,
+      "learning_rate": 6.679780248541933e-07,
+      "loss": 1.8283,
+      "step": 3688
+    },
+    {
+      "epoch": 1.9030927835051545,
+      "grad_norm": 1.479387879371643,
+      "learning_rate": 6.404944853552974e-07,
+      "loss": 1.7512,
+      "step": 3692
+    },
+    {
+      "epoch": 1.9051546391752576,
+      "grad_norm": 1.4344489574432373,
+      "learning_rate": 6.135846387178235e-07,
+      "loss": 1.6923,
+      "step": 3696
+    },
+    {
+      "epoch": 1.9072164948453607,
+      "grad_norm": 1.4317504167556763,
+      "learning_rate": 5.872487977080232e-07,
+      "loss": 1.6294,
+      "step": 3700
+    },
+    {
+      "epoch": 1.9092783505154638,
+      "grad_norm": 1.4593110084533691,
+      "learning_rate": 5.614872684206074e-07,
+      "loss": 1.6398,
+      "step": 3704
+    },
+    {
+      "epoch": 1.9113402061855669,
+      "grad_norm": 1.5218007564544678,
+      "learning_rate": 5.36300350275215e-07,
+      "loss": 1.7071,
+      "step": 3708
+    },
+    {
+      "epoch": 1.91340206185567,
+      "grad_norm": 1.5086115598678589,
+      "learning_rate": 5.116883360129388e-07,
+      "loss": 1.626,
+      "step": 3712
+    },
+    {
+      "epoch": 1.915463917525773,
+      "grad_norm": 1.75813627243042,
+      "learning_rate": 4.876515116928993e-07,
+      "loss": 1.7443,
+      "step": 3716
+    },
+    {
+      "epoch": 1.9175257731958761,
+      "grad_norm": 1.46547269821167,
+      "learning_rate": 4.641901566889317e-07,
+      "loss": 1.5604,
+      "step": 3720
+    },
+    {
+      "epoch": 1.9195876288659792,
+      "grad_norm": 1.5154309272766113,
+      "learning_rate": 4.413045436863439e-07,
+      "loss": 1.6684,
+      "step": 3724
+    },
+    {
+      "epoch": 1.9216494845360823,
+      "grad_norm": 1.4802361726760864,
+      "learning_rate": 4.189949386787462e-07,
+      "loss": 1.6636,
+      "step": 3728
+    },
+    {
+      "epoch": 1.9237113402061856,
+      "grad_norm": 1.56373131275177,
+      "learning_rate": 3.9726160096494325e-07,
+      "loss": 1.6667,
+      "step": 3732
+    },
+    {
+      "epoch": 1.9257731958762887,
+      "grad_norm": 1.4294252395629883,
+      "learning_rate": 3.761047831459419e-07,
+      "loss": 1.6603,
+      "step": 3736
+    },
+    {
+      "epoch": 1.9278350515463918,
+      "grad_norm": 1.5142799615859985,
+      "learning_rate": 3.555247311220089e-07,
+      "loss": 1.6348,
+      "step": 3740
+    },
+    {
+      "epoch": 1.9298969072164949,
+      "grad_norm": 1.4750471115112305,
+      "learning_rate": 3.355216840898012e-07,
+      "loss": 1.6561,
+      "step": 3744
+    },
+    {
+      "epoch": 1.931958762886598,
+      "grad_norm": 1.4013311862945557,
+      "learning_rate": 3.160958745395959e-07,
+      "loss": 1.6969,
+      "step": 3748
+    },
+    {
+      "epoch": 1.934020618556701,
+      "grad_norm": 1.6206867694854736,
+      "learning_rate": 2.9724752825259774e-07,
+      "loss": 1.6877,
+      "step": 3752
+    },
+    {
+      "epoch": 1.9360824742268041,
+      "grad_norm": 1.5444731712341309,
+      "learning_rate": 2.7897686429829703e-07,
+      "loss": 1.7525,
+      "step": 3756
+    },
+    {
+      "epoch": 1.9381443298969072,
+      "grad_norm": 1.6207637786865234,
+      "learning_rate": 2.612840950319273e-07,
+      "loss": 1.6376,
+      "step": 3760
+    },
+    {
+      "epoch": 1.9402061855670103,
+      "grad_norm": 1.4988189935684204,
+      "learning_rate": 2.441694260920002e-07,
+      "loss": 1.6437,
+      "step": 3764
+    },
+    {
+      "epoch": 1.9422680412371134,
+      "grad_norm": 1.4861098527908325,
+      "learning_rate": 2.276330563979301e-07,
+      "loss": 1.636,
+      "step": 3768
+    },
+    {
+      "epoch": 1.9443298969072165,
+      "grad_norm": 1.4013590812683105,
+      "learning_rate": 2.1167517814768557e-07,
+      "loss": 1.5653,
+      "step": 3772
+    },
+    {
+      "epoch": 1.9463917525773196,
+      "grad_norm": 1.5663115978240967,
+      "learning_rate": 1.9629597681559698e-07,
+      "loss": 1.7384,
+      "step": 3776
+    },
+    {
+      "epoch": 1.9484536082474226,
+      "grad_norm": 1.4937258958816528,
+      "learning_rate": 1.814956311501692e-07,
+      "loss": 1.6718,
+      "step": 3780
+    },
+    {
+      "epoch": 1.9505154639175257,
+      "grad_norm": 1.5502136945724487,
+      "learning_rate": 1.6727431317202224e-07,
+      "loss": 1.6496,
+      "step": 3784
+    },
+    {
+      "epoch": 1.9525773195876288,
+      "grad_norm": 1.40053391456604,
+      "learning_rate": 1.5363218817188162e-07,
+      "loss": 1.6056,
+      "step": 3788
+    },
+    {
+      "epoch": 1.9546391752577321,
+      "grad_norm": 1.4303723573684692,
+      "learning_rate": 1.405694147086689e-07,
+      "loss": 1.6602,
+      "step": 3792
+    },
+    {
+      "epoch": 1.9567010309278352,
+      "grad_norm": 1.4497610330581665,
+      "learning_rate": 1.28086144607642e-07,
+      "loss": 1.6338,
+      "step": 3796
+    },
+    {
+      "epoch": 1.9587628865979383,
+      "grad_norm": 1.5628520250320435,
+      "learning_rate": 1.1618252295864107e-07,
+      "loss": 1.7174,
+      "step": 3800
+    },
+    {
+      "epoch": 1.9608247422680414,
+      "grad_norm": 1.4401839971542358,
+      "learning_rate": 1.0485868811441757e-07,
+      "loss": 1.6945,
+      "step": 3804
+    },
+    {
+      "epoch": 1.9628865979381445,
+      "grad_norm": 1.5296087265014648,
+      "learning_rate": 9.411477168898563e-08,
+      "loss": 1.7328,
+      "step": 3808
+    },
+    {
+      "epoch": 1.9649484536082475,
+      "grad_norm": 1.556199312210083,
+      "learning_rate": 8.395089855613437e-08,
+      "loss": 1.7358,
+      "step": 3812
+    },
+    {
+      "epoch": 1.9670103092783506,
+      "grad_norm": 1.5304512977600098,
+      "learning_rate": 7.436718684794563e-08,
+      "loss": 1.7665,
+      "step": 3816
+    },
+    {
+      "epoch": 1.9690721649484537,
+      "grad_norm": 1.4749436378479004,
+      "learning_rate": 6.536374795344524e-08,
+      "loss": 1.6267,
+      "step": 3820
+    },
+    {
+      "epoch": 1.9711340206185568,
+      "grad_norm": 1.5322349071502686,
+      "learning_rate": 5.694068651729279e-08,
+      "loss": 1.789,
+      "step": 3824
+    },
+    {
+      "epoch": 1.97319587628866,
+      "grad_norm": 1.553791880607605,
+      "learning_rate": 4.9098100438566e-08,
+      "loss": 1.6623,
+      "step": 3828
+    },
+    {
+      "epoch": 1.975257731958763,
+      "grad_norm": 1.543352484703064,
+      "learning_rate": 4.183608086963386e-08,
+      "loss": 1.7421,
+      "step": 3832
+    },
+    {
+      "epoch": 1.977319587628866,
+      "grad_norm": 1.5451158285140991,
+      "learning_rate": 3.515471221507416e-08,
+      "loss": 1.7402,
+      "step": 3836
+    },
+    {
+      "epoch": 1.9793814432989691,
+      "grad_norm": 1.4942976236343384,
+      "learning_rate": 2.90540721307353e-08,
+      "loss": 1.6397,
+      "step": 3840
+    },
+    {
+      "epoch": 1.9814432989690722,
+      "grad_norm": 1.4658242464065552,
+      "learning_rate": 2.3534231522781557e-08,
+      "loss": 1.6033,
+      "step": 3844
+    },
+    {
+      "epoch": 1.9835051546391753,
+      "grad_norm": 1.4824728965759277,
+      "learning_rate": 1.8595254546904806e-08,
+      "loss": 1.6109,
+      "step": 3848
+    },
+    {
+      "epoch": 1.9855670103092784,
+      "grad_norm": 1.3989909887313843,
+      "learning_rate": 1.4237198607564007e-08,
+      "loss": 1.599,
+      "step": 3852
+    },
+    {
+      "epoch": 1.9876288659793815,
+      "grad_norm": 1.497368574142456,
+      "learning_rate": 1.0460114357330187e-08,
+      "loss": 1.6775,
+      "step": 3856
+    },
+    {
+      "epoch": 1.9896907216494846,
+      "grad_norm": 1.4998079538345337,
+      "learning_rate": 7.264045696281363e-09,
+      "loss": 1.6132,
+      "step": 3860
+    },
+    {
+      "epoch": 1.9917525773195877,
+      "grad_norm": 1.5205719470977783,
+      "learning_rate": 4.649029771502944e-09,
+      "loss": 1.6882,
+      "step": 3864
+    },
+    {
+      "epoch": 1.9938144329896907,
+      "grad_norm": 1.5549403429031372,
+      "learning_rate": 2.6150969766547406e-09,
+      "loss": 1.7583,
+      "step": 3868
+    },
+    {
+      "epoch": 1.9958762886597938,
+      "grad_norm": 1.5483360290527344,
+      "learning_rate": 1.162270951615696e-09,
+      "loss": 1.6231,
+      "step": 3872
+    },
+    {
+      "epoch": 1.997938144329897,
+      "grad_norm": 1.444972276687622,
+      "learning_rate": 2.9056858220632974e-10,
+      "loss": 1.7232,
+      "step": 3876
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 4.258011817932129,
+      "learning_rate": 0.0,
+      "loss": 1.6374,
+      "step": 3880
+    }
+  ],
+  "logging_steps": 4,
+  "max_steps": 3880,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1940,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.564802121202401e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}