diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4418 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 3125,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0016,
+      "grad_norm": 9.999629974365234,
+      "learning_rate": 4.25531914893617e-08,
+      "loss": 1.0338,
+      "step": 5
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": 7.134084224700928,
+      "learning_rate": 9.574468085106382e-08,
+      "loss": 1.1826,
+      "step": 10
+    },
+    {
+      "epoch": 0.0048,
+      "grad_norm": 5.787372589111328,
+      "learning_rate": 1.4893617021276595e-07,
+      "loss": 1.145,
+      "step": 15
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 13.603924751281738,
+      "learning_rate": 2.0212765957446807e-07,
+      "loss": 1.1501,
+      "step": 20
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 13.660343170166016,
+      "learning_rate": 2.5531914893617016e-07,
+      "loss": 1.1397,
+      "step": 25
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 5.391168594360352,
+      "learning_rate": 3.085106382978723e-07,
+      "loss": 1.0871,
+      "step": 30
+    },
+    {
+      "epoch": 0.0112,
+      "grad_norm": 9.299227714538574,
+      "learning_rate": 3.617021276595745e-07,
+      "loss": 1.2235,
+      "step": 35
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 4.13670539855957,
+      "learning_rate": 4.148936170212766e-07,
+      "loss": 1.0978,
+      "step": 40
+    },
+    {
+      "epoch": 0.0144,
+      "grad_norm": 5.579084396362305,
+      "learning_rate": 4.6808510638297873e-07,
+      "loss": 1.2814,
+      "step": 45
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 13.503495216369629,
+      "learning_rate": 5.212765957446809e-07,
+      "loss": 0.9962,
+      "step": 50
+    },
+    {
+      "epoch": 0.0176,
+      "grad_norm": 16.958608627319336,
+      "learning_rate": 5.74468085106383e-07,
+      "loss": 1.0797,
+      "step": 55
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 8.709441184997559,
+      "learning_rate": 6.276595744680851e-07,
+      "loss": 1.1269,
+      "step": 60
+    },
+    {
+      "epoch": 0.0208,
+      "grad_norm": 6.347010612487793,
+      "learning_rate": 6.808510638297872e-07,
+      "loss": 1.1201,
+      "step": 65
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 12.77128791809082,
+      "learning_rate": 7.340425531914893e-07,
+      "loss": 1.0846,
+      "step": 70
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 8.114184379577637,
+      "learning_rate": 7.872340425531915e-07,
+      "loss": 1.151,
+      "step": 75
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 17.945985794067383,
+      "learning_rate": 8.404255319148936e-07,
+      "loss": 1.0711,
+      "step": 80
+    },
+    {
+      "epoch": 0.0272,
+      "grad_norm": 5.469882965087891,
+      "learning_rate": 8.936170212765957e-07,
+      "loss": 0.9901,
+      "step": 85
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 10.689380645751953,
+      "learning_rate": 9.468085106382978e-07,
+      "loss": 1.2635,
+      "step": 90
+    },
+    {
+      "epoch": 0.0304,
+      "grad_norm": 9.217676162719727,
+      "learning_rate": 1e-06,
+      "loss": 1.0604,
+      "step": 95
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 4.253987789154053,
+      "learning_rate": 9.98350379412735e-07,
+      "loss": 1.1865,
+      "step": 100
+    },
+    {
+      "epoch": 0.0336,
+      "grad_norm": 10.984722137451172,
+      "learning_rate": 9.967007588254702e-07,
+      "loss": 1.1504,
+      "step": 105
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 6.79901647567749,
+      "learning_rate": 9.950511382382052e-07,
+      "loss": 1.0338,
+      "step": 110
+    },
+    {
+      "epoch": 0.0368,
+      "grad_norm": 10.826879501342773,
+      "learning_rate": 9.934015176509404e-07,
+      "loss": 1.1058,
+      "step": 115
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 6.437893390655518,
+      "learning_rate": 9.917518970636754e-07,
+      "loss": 1.0317,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.419424295425415,
+      "learning_rate": 9.901022764764103e-07,
+      "loss": 1.0573,
+      "step": 125
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 14.196756362915039,
+      "learning_rate": 9.884526558891456e-07,
+      "loss": 1.0892,
+      "step": 130
+    },
+    {
+      "epoch": 0.0432,
+      "grad_norm": 4.492501258850098,
+      "learning_rate": 9.868030353018806e-07,
+      "loss": 0.9772,
+      "step": 135
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 8.752493858337402,
+      "learning_rate": 9.851534147146155e-07,
+      "loss": 0.984,
+      "step": 140
+    },
+    {
+      "epoch": 0.0464,
+      "grad_norm": 3.8549039363861084,
+      "learning_rate": 9.835037941273505e-07,
+      "loss": 0.9935,
+      "step": 145
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 5.623340129852295,
+      "learning_rate": 9.818541735400857e-07,
+      "loss": 1.1332,
+      "step": 150
+    },
+    {
+      "epoch": 0.0496,
+      "grad_norm": 5.848334789276123,
+      "learning_rate": 9.802045529528207e-07,
+      "loss": 0.912,
+      "step": 155
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 7.2841877937316895,
+      "learning_rate": 9.78554932365556e-07,
+      "loss": 1.1562,
+      "step": 160
+    },
+    {
+      "epoch": 0.0528,
+      "grad_norm": 5.735965728759766,
+      "learning_rate": 9.76905311778291e-07,
+      "loss": 0.9819,
+      "step": 165
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 3.197845220565796,
+      "learning_rate": 9.75255691191026e-07,
+      "loss": 0.9843,
+      "step": 170
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 8.21580696105957,
+      "learning_rate": 9.736060706037611e-07,
+      "loss": 0.9533,
+      "step": 175
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 5.551872730255127,
+      "learning_rate": 9.719564500164961e-07,
+      "loss": 1.0205,
+      "step": 180
+    },
+    {
+      "epoch": 0.0592,
+      "grad_norm": 5.322945594787598,
+      "learning_rate": 9.703068294292313e-07,
+      "loss": 1.0036,
+      "step": 185
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 4.90363883972168,
+      "learning_rate": 9.686572088419663e-07,
+      "loss": 1.0135,
+      "step": 190
+    },
+    {
+      "epoch": 0.0624,
+      "grad_norm": 8.087169647216797,
+      "learning_rate": 9.670075882547013e-07,
+      "loss": 1.0293,
+      "step": 195
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 4.1587910652160645,
+      "learning_rate": 9.653579676674365e-07,
+      "loss": 1.0719,
+      "step": 200
+    },
+    {
+      "epoch": 0.0656,
+      "grad_norm": 3.2837698459625244,
+      "learning_rate": 9.637083470801715e-07,
+      "loss": 1.0107,
+      "step": 205
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 5.359975814819336,
+      "learning_rate": 9.620587264929065e-07,
+      "loss": 0.9036,
+      "step": 210
+    },
+    {
+      "epoch": 0.0688,
+      "grad_norm": 5.2580037117004395,
+      "learning_rate": 9.604091059056415e-07,
+      "loss": 0.9704,
+      "step": 215
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 9.796117782592773,
+      "learning_rate": 9.587594853183767e-07,
+      "loss": 0.9456,
+      "step": 220
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 10.24465560913086,
+      "learning_rate": 9.571098647311117e-07,
+      "loss": 1.0502,
+      "step": 225
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 7.899555683135986,
+      "learning_rate": 9.55460244143847e-07,
+      "loss": 0.9849,
+      "step": 230
+    },
+    {
+      "epoch": 0.0752,
+      "grad_norm": 6.677064418792725,
+      "learning_rate": 9.53810623556582e-07,
+      "loss": 1.103,
+      "step": 235
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 8.85185718536377,
+      "learning_rate": 9.52161002969317e-07,
+      "loss": 0.9351,
+      "step": 240
+    },
+    {
+      "epoch": 0.0784,
+      "grad_norm": 7.826456069946289,
+      "learning_rate": 9.505113823820521e-07,
+      "loss": 0.9369,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.929803371429443,
+      "learning_rate": 9.488617617947871e-07,
+      "loss": 0.8399,
+      "step": 250
+    },
+    {
+      "epoch": 0.0816,
+      "grad_norm": 4.857858180999756,
+      "learning_rate": 9.472121412075222e-07,
+      "loss": 0.9738,
+      "step": 255
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 7.609739303588867,
+      "learning_rate": 9.455625206202573e-07,
+      "loss": 1.0397,
+      "step": 260
+    },
+    {
+      "epoch": 0.0848,
+      "grad_norm": 4.687152862548828,
+      "learning_rate": 9.439129000329924e-07,
+      "loss": 1.0207,
+      "step": 265
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 3.707120180130005,
+      "learning_rate": 9.422632794457274e-07,
+      "loss": 1.0404,
+      "step": 270
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 7.183946132659912,
+      "learning_rate": 9.406136588584625e-07,
+      "loss": 0.9329,
+      "step": 275
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 4.107320785522461,
+      "learning_rate": 9.389640382711976e-07,
+      "loss": 0.9674,
+      "step": 280
+    },
+    {
+      "epoch": 0.0912,
+      "grad_norm": 2.987569808959961,
+      "learning_rate": 9.373144176839326e-07,
+      "loss": 1.0431,
+      "step": 285
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 7.492343902587891,
+      "learning_rate": 9.356647970966677e-07,
+      "loss": 0.8819,
+      "step": 290
+    },
+    {
+      "epoch": 0.0944,
+      "grad_norm": 3.460360050201416,
+      "learning_rate": 9.340151765094027e-07,
+      "loss": 1.1553,
+      "step": 295
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 4.670125961303711,
+      "learning_rate": 9.323655559221378e-07,
+      "loss": 0.9511,
+      "step": 300
+    },
+    {
+      "epoch": 0.0976,
+      "grad_norm": 6.766526699066162,
+      "learning_rate": 9.307159353348729e-07,
+      "loss": 0.9625,
+      "step": 305
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 3.6760966777801514,
+      "learning_rate": 9.29066314747608e-07,
+      "loss": 0.9503,
+      "step": 310
+    },
+    {
+      "epoch": 0.1008,
+      "grad_norm": 9.794249534606934,
+      "learning_rate": 9.27416694160343e-07,
+      "loss": 0.9533,
+      "step": 315
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 8.894811630249023,
+      "learning_rate": 9.257670735730781e-07,
+      "loss": 0.8668,
+      "step": 320
+    },
+    {
+      "epoch": 0.104,
+      "grad_norm": 3.3411269187927246,
+      "learning_rate": 9.241174529858132e-07,
+      "loss": 0.9171,
+      "step": 325
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 9.227668762207031,
+      "learning_rate": 9.224678323985483e-07,
+      "loss": 0.9065,
+      "step": 330
+    },
+    {
+      "epoch": 0.1072,
+      "grad_norm": 3.217501640319824,
+      "learning_rate": 9.208182118112834e-07,
+      "loss": 0.945,
+      "step": 335
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 3.9283738136291504,
+      "learning_rate": 9.191685912240184e-07,
+      "loss": 0.9166,
+      "step": 340
+    },
+    {
+      "epoch": 0.1104,
+      "grad_norm": 7.905593395233154,
+      "learning_rate": 9.175189706367535e-07,
+      "loss": 1.0471,
+      "step": 345
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 3.8356964588165283,
+      "learning_rate": 9.158693500494886e-07,
+      "loss": 0.9593,
+      "step": 350
+    },
+    {
+      "epoch": 0.1136,
+      "grad_norm": 4.25161600112915,
+      "learning_rate": 9.142197294622237e-07,
+      "loss": 0.9251,
+      "step": 355
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 2.885007381439209,
+      "learning_rate": 9.125701088749587e-07,
+      "loss": 0.9124,
+      "step": 360
+    },
+    {
+      "epoch": 0.1168,
+      "grad_norm": 6.2251763343811035,
+      "learning_rate": 9.109204882876937e-07,
+      "loss": 1.0204,
+      "step": 365
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 5.769200801849365,
+      "learning_rate": 9.092708677004288e-07,
+      "loss": 0.8877,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.167758941650391,
+      "learning_rate": 9.076212471131639e-07,
+      "loss": 0.9371,
+      "step": 375
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 4.101468563079834,
+      "learning_rate": 9.05971626525899e-07,
+      "loss": 0.9571,
+      "step": 380
+    },
+    {
+      "epoch": 0.1232,
+      "grad_norm": 3.352560043334961,
+      "learning_rate": 9.04322005938634e-07,
+      "loss": 0.9988,
+      "step": 385
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 4.724895000457764,
+      "learning_rate": 9.026723853513691e-07,
+      "loss": 0.972,
+      "step": 390
+    },
+    {
+      "epoch": 0.1264,
+      "grad_norm": 5.613089561462402,
+      "learning_rate": 9.010227647641042e-07,
+      "loss": 0.8806,
+      "step": 395
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 4.489896774291992,
+      "learning_rate": 8.993731441768393e-07,
+      "loss": 0.9119,
+      "step": 400
+    },
+    {
+      "epoch": 0.1296,
+      "grad_norm": 6.169275760650635,
+      "learning_rate": 8.977235235895744e-07,
+      "loss": 0.9942,
+      "step": 405
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 4.484738826751709,
+      "learning_rate": 8.960739030023094e-07,
+      "loss": 0.8567,
+      "step": 410
+    },
+    {
+      "epoch": 0.1328,
+      "grad_norm": 3.1110870838165283,
+      "learning_rate": 8.944242824150445e-07,
+      "loss": 0.912,
+      "step": 415
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 7.398248672485352,
+      "learning_rate": 8.927746618277796e-07,
+      "loss": 0.9913,
+      "step": 420
+    },
+    {
+      "epoch": 0.136,
+      "grad_norm": 6.651194095611572,
+      "learning_rate": 8.911250412405147e-07,
+      "loss": 0.8492,
+      "step": 425
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 6.037364482879639,
+      "learning_rate": 8.894754206532498e-07,
+      "loss": 0.9839,
+      "step": 430
+    },
+    {
+      "epoch": 0.1392,
+      "grad_norm": 3.9697699546813965,
+      "learning_rate": 8.878258000659847e-07,
+      "loss": 0.9343,
+      "step": 435
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 4.4155497550964355,
+      "learning_rate": 8.861761794787198e-07,
+      "loss": 0.9474,
+      "step": 440
+    },
+    {
+      "epoch": 0.1424,
+      "grad_norm": 4.292988300323486,
+      "learning_rate": 8.845265588914549e-07,
+      "loss": 1.0596,
+      "step": 445
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 2.772756338119507,
+      "learning_rate": 8.8287693830419e-07,
+      "loss": 0.9765,
+      "step": 450
+    },
+    {
+      "epoch": 0.1456,
+      "grad_norm": 7.738980770111084,
+      "learning_rate": 8.81227317716925e-07,
+      "loss": 0.8599,
+      "step": 455
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 9.246415138244629,
+      "learning_rate": 8.795776971296601e-07,
+      "loss": 0.9711,
+      "step": 460
+    },
+    {
+      "epoch": 0.1488,
+      "grad_norm": 5.940875053405762,
+      "learning_rate": 8.779280765423952e-07,
+      "loss": 0.9433,
+      "step": 465
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 6.259022235870361,
+      "learning_rate": 8.762784559551303e-07,
+      "loss": 0.9859,
+      "step": 470
+    },
+    {
+      "epoch": 0.152,
+      "grad_norm": 7.941705226898193,
+      "learning_rate": 8.746288353678654e-07,
+      "loss": 0.8696,
+      "step": 475
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 3.571704626083374,
+      "learning_rate": 8.729792147806004e-07,
+      "loss": 0.943,
+      "step": 480
+    },
+    {
+      "epoch": 0.1552,
+      "grad_norm": 4.129303455352783,
+      "learning_rate": 8.713295941933355e-07,
+      "loss": 0.8251,
+      "step": 485
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 8.326216697692871,
+      "learning_rate": 8.696799736060706e-07,
+      "loss": 0.9393,
+      "step": 490
+    },
+    {
+      "epoch": 0.1584,
+      "grad_norm": 2.903012275695801,
+      "learning_rate": 8.680303530188057e-07,
+      "loss": 0.8944,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.4961628913879395,
+      "learning_rate": 8.663807324315408e-07,
+      "loss": 1.0191,
+      "step": 500
+    },
+    {
+      "epoch": 0.1616,
+      "grad_norm": 17.958810806274414,
+      "learning_rate": 8.647311118442758e-07,
+      "loss": 0.8689,
+      "step": 505
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 7.708248138427734,
+      "learning_rate": 8.630814912570108e-07,
+      "loss": 0.9344,
+      "step": 510
+    },
+    {
+      "epoch": 0.1648,
+      "grad_norm": 3.0089898109436035,
+      "learning_rate": 8.614318706697459e-07,
+      "loss": 0.9085,
+      "step": 515
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 3.333603858947754,
+      "learning_rate": 8.59782250082481e-07,
+      "loss": 0.9389,
+      "step": 520
+    },
+    {
+      "epoch": 0.168,
+      "grad_norm": 4.273075580596924,
+      "learning_rate": 8.58132629495216e-07,
+      "loss": 0.9545,
+      "step": 525
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 3.9365367889404297,
+      "learning_rate": 8.564830089079511e-07,
+      "loss": 1.044,
+      "step": 530
+    },
+    {
+      "epoch": 0.1712,
+      "grad_norm": 8.090559959411621,
+      "learning_rate": 8.548333883206862e-07,
+      "loss": 0.8949,
+      "step": 535
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 3.675675868988037,
+      "learning_rate": 8.531837677334213e-07,
+      "loss": 0.8984,
+      "step": 540
+    },
+    {
+      "epoch": 0.1744,
+      "grad_norm": 4.396546840667725,
+      "learning_rate": 8.515341471461564e-07,
+      "loss": 0.8276,
+      "step": 545
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 5.8129706382751465,
+      "learning_rate": 8.498845265588914e-07,
+      "loss": 0.9485,
+      "step": 550
+    },
+    {
+      "epoch": 0.1776,
+      "grad_norm": 4.243994235992432,
+      "learning_rate": 8.482349059716265e-07,
+      "loss": 0.8613,
+      "step": 555
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 3.252338409423828,
+      "learning_rate": 8.465852853843616e-07,
+      "loss": 0.9014,
+      "step": 560
+    },
+    {
+      "epoch": 0.1808,
+      "grad_norm": 2.9563217163085938,
+      "learning_rate": 8.449356647970967e-07,
+      "loss": 0.9632,
+      "step": 565
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 4.2105302810668945,
+      "learning_rate": 8.432860442098317e-07,
+      "loss": 0.8758,
+      "step": 570
+    },
+    {
+      "epoch": 0.184,
+      "grad_norm": 4.756136417388916,
+      "learning_rate": 8.416364236225668e-07,
+      "loss": 0.9102,
+      "step": 575
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 3.9406795501708984,
+      "learning_rate": 8.399868030353019e-07,
+      "loss": 0.9286,
+      "step": 580
+    },
+    {
+      "epoch": 0.1872,
+      "grad_norm": 5.04837703704834,
+      "learning_rate": 8.383371824480369e-07,
+      "loss": 1.0353,
+      "step": 585
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 2.621293306350708,
+      "learning_rate": 8.36687561860772e-07,
+      "loss": 0.9047,
+      "step": 590
+    },
+    {
+      "epoch": 0.1904,
+      "grad_norm": 4.60697078704834,
+      "learning_rate": 8.35037941273507e-07,
+      "loss": 0.9403,
+      "step": 595
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 3.5132675170898438,
+      "learning_rate": 8.333883206862421e-07,
+      "loss": 0.9576,
+      "step": 600
+    },
+    {
+      "epoch": 0.1936,
+      "grad_norm": 6.349253177642822,
+      "learning_rate": 8.317387000989772e-07,
+      "loss": 0.9443,
+      "step": 605
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 4.456959247589111,
+      "learning_rate": 8.300890795117123e-07,
+      "loss": 0.9476,
+      "step": 610
+    },
+    {
+      "epoch": 0.1968,
+      "grad_norm": 3.2254133224487305,
+      "learning_rate": 8.284394589244474e-07,
+      "loss": 0.9401,
+      "step": 615
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 4.775053977966309,
+      "learning_rate": 8.267898383371824e-07,
+      "loss": 1.0189,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.755006790161133,
+      "learning_rate": 8.251402177499175e-07,
+      "loss": 0.8521,
+      "step": 625
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 4.0426716804504395,
+      "learning_rate": 8.234905971626526e-07,
+      "loss": 0.9075,
+      "step": 630
+    },
+    {
+      "epoch": 0.2032,
+      "grad_norm": 4.348297119140625,
+      "learning_rate": 8.218409765753877e-07,
+      "loss": 0.9625,
+      "step": 635
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 6.654307842254639,
+      "learning_rate": 8.201913559881227e-07,
+      "loss": 0.8582,
+      "step": 640
+    },
+    {
+      "epoch": 0.2064,
+      "grad_norm": 4.090488910675049,
+      "learning_rate": 8.185417354008578e-07,
+      "loss": 0.9359,
+      "step": 645
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 5.286700248718262,
+      "learning_rate": 8.168921148135929e-07,
+      "loss": 0.9783,
+      "step": 650
+    },
+    {
+      "epoch": 0.2096,
+      "grad_norm": 4.438581466674805,
+      "learning_rate": 8.15242494226328e-07,
+      "loss": 0.9526,
+      "step": 655
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 3.836512565612793,
+      "learning_rate": 8.13592873639063e-07,
+      "loss": 0.8768,
+      "step": 660
+    },
+    {
+      "epoch": 0.2128,
+      "grad_norm": 3.1037371158599854,
+      "learning_rate": 8.11943253051798e-07,
+      "loss": 0.963,
+      "step": 665
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 5.103884696960449,
+      "learning_rate": 8.102936324645331e-07,
+      "loss": 0.9322,
+      "step": 670
+    },
+    {
+      "epoch": 0.216,
+      "grad_norm": 3.707827091217041,
+      "learning_rate": 8.086440118772682e-07,
+      "loss": 0.8691,
+      "step": 675
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 3.7925477027893066,
+      "learning_rate": 8.069943912900033e-07,
+      "loss": 0.9636,
+      "step": 680
+    },
+    {
+      "epoch": 0.2192,
+      "grad_norm": 4.857919692993164,
+      "learning_rate": 8.053447707027383e-07,
+      "loss": 0.8867,
+      "step": 685
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 3.513091564178467,
+      "learning_rate": 8.036951501154734e-07,
+      "loss": 0.9646,
+      "step": 690
+    },
+    {
+      "epoch": 0.2224,
+      "grad_norm": 3.4893958568573,
+      "learning_rate": 8.020455295282085e-07,
+      "loss": 0.9276,
+      "step": 695
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 3.087334156036377,
+      "learning_rate": 8.003959089409436e-07,
+      "loss": 0.89,
+      "step": 700
+    },
+    {
+      "epoch": 0.2256,
+      "grad_norm": 4.584767818450928,
+      "learning_rate": 7.987462883536787e-07,
+      "loss": 1.0421,
+      "step": 705
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 3.7277486324310303,
+      "learning_rate": 7.970966677664137e-07,
+      "loss": 0.8757,
+      "step": 710
+    },
+    {
+      "epoch": 0.2288,
+      "grad_norm": 4.989358425140381,
+      "learning_rate": 7.954470471791488e-07,
+      "loss": 0.8421,
+      "step": 715
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 4.230960845947266,
+      "learning_rate": 7.937974265918839e-07,
+      "loss": 0.883,
+      "step": 720
+    },
+    {
+      "epoch": 0.232,
+      "grad_norm": 7.118892192840576,
+      "learning_rate": 7.92147806004619e-07,
+      "loss": 0.9549,
+      "step": 725
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 10.041189193725586,
+      "learning_rate": 7.904981854173541e-07,
+      "loss": 0.9897,
+      "step": 730
+    },
+    {
+      "epoch": 0.2352,
+      "grad_norm": 3.822767734527588,
+      "learning_rate": 7.88848564830089e-07,
+      "loss": 0.8711,
+      "step": 735
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 4.475744724273682,
+      "learning_rate": 7.871989442428241e-07,
+      "loss": 0.9175,
+      "step": 740
+    },
+    {
+      "epoch": 0.2384,
+      "grad_norm": 5.1056976318359375,
+      "learning_rate": 7.855493236555592e-07,
+      "loss": 0.9022,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.522522926330566,
+      "learning_rate": 7.838997030682943e-07,
+      "loss": 1.0007,
+      "step": 750
+    },
+    {
+      "epoch": 0.2416,
+      "grad_norm": 4.390966415405273,
+      "learning_rate": 7.822500824810293e-07,
+      "loss": 0.9074,
+      "step": 755
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 7.068999290466309,
+      "learning_rate": 7.806004618937644e-07,
+      "loss": 0.8454,
+      "step": 760
+    },
+    {
+      "epoch": 0.2448,
+      "grad_norm": 3.558549642562866,
+      "learning_rate": 7.789508413064995e-07,
+      "loss": 0.88,
+      "step": 765
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 4.7729949951171875,
+      "learning_rate": 7.773012207192346e-07,
+      "loss": 0.8577,
+      "step": 770
+    },
+    {
+      "epoch": 0.248,
+      "grad_norm": 3.512878894805908,
+      "learning_rate": 7.756516001319697e-07,
+      "loss": 1.0939,
+      "step": 775
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 2.6263558864593506,
+      "learning_rate": 7.740019795447047e-07,
+      "loss": 0.8807,
+      "step": 780
+    },
+    {
+      "epoch": 0.2512,
+      "grad_norm": 2.518568992614746,
+      "learning_rate": 7.723523589574398e-07,
+      "loss": 0.7706,
+      "step": 785
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 5.156455993652344,
+      "learning_rate": 7.707027383701749e-07,
+      "loss": 0.8809,
+      "step": 790
+    },
+    {
+      "epoch": 0.2544,
+      "grad_norm": 7.734130382537842,
+      "learning_rate": 7.6905311778291e-07,
+      "loss": 0.967,
+      "step": 795
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 2.6134588718414307,
+      "learning_rate": 7.674034971956451e-07,
+      "loss": 0.7965,
+      "step": 800
+    },
+    {
+      "epoch": 0.2576,
+      "grad_norm": 9.95977783203125,
+      "learning_rate": 7.657538766083801e-07,
+      "loss": 0.9095,
+      "step": 805
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 3.093651533126831,
+      "learning_rate": 7.64104256021115e-07,
+      "loss": 0.9707,
+      "step": 810
+    },
+    {
+      "epoch": 0.2608,
+      "grad_norm": 3.039573907852173,
+      "learning_rate": 7.624546354338501e-07,
+      "loss": 0.9317,
+      "step": 815
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 4.40300989151001,
+      "learning_rate": 7.608050148465853e-07,
+      "loss": 0.8223,
+      "step": 820
+    },
+    {
+      "epoch": 0.264,
+      "grad_norm": 5.527564525604248,
+      "learning_rate": 7.591553942593202e-07,
+      "loss": 0.9234,
+      "step": 825
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 5.654271602630615,
+      "learning_rate": 7.575057736720553e-07,
+      "loss": 0.9437,
+      "step": 830
+    },
+    {
+      "epoch": 0.2672,
+      "grad_norm": 5.316553115844727,
+      "learning_rate": 7.558561530847904e-07,
+      "loss": 0.8862,
+      "step": 835
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 2.8125505447387695,
+      "learning_rate": 7.542065324975255e-07,
+      "loss": 0.8235,
+      "step": 840
+    },
+    {
+      "epoch": 0.2704,
+      "grad_norm": 5.254530429840088,
+      "learning_rate": 7.525569119102606e-07,
+      "loss": 0.8986,
+      "step": 845
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 5.69275426864624,
+      "learning_rate": 7.509072913229956e-07,
+      "loss": 0.9043,
+      "step": 850
+    },
+    {
+      "epoch": 0.2736,
+      "grad_norm": 4.995587348937988,
+      "learning_rate": 7.492576707357307e-07,
+      "loss": 0.8242,
+      "step": 855
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 8.568499565124512,
+      "learning_rate": 7.476080501484658e-07,
+      "loss": 0.9116,
+      "step": 860
+    },
+    {
+      "epoch": 0.2768,
+      "grad_norm": 5.804699420928955,
+      "learning_rate": 7.45958429561201e-07,
+      "loss": 0.863,
+      "step": 865
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 7.921741962432861,
+      "learning_rate": 7.44308808973936e-07,
+      "loss": 0.9719,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.888009071350098,
+      "learning_rate": 7.42659188386671e-07,
+      "loss": 0.8999,
+      "step": 875
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 3.9009482860565186,
+      "learning_rate": 7.410095677994061e-07,
+      "loss": 0.9543,
+      "step": 880
+    },
+    {
+      "epoch": 0.2832,
+      "grad_norm": 3.502060651779175,
+      "learning_rate": 7.393599472121411e-07,
+      "loss": 0.8367,
+      "step": 885
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 4.675789833068848,
+      "learning_rate": 7.377103266248762e-07,
+      "loss": 0.8843,
+      "step": 890
+    },
+    {
+      "epoch": 0.2864,
+      "grad_norm": 3.246445417404175,
+      "learning_rate": 7.360607060376112e-07,
+      "loss": 0.9663,
+      "step": 895
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 3.168081521987915,
+      "learning_rate": 7.344110854503463e-07,
+      "loss": 0.8186,
+      "step": 900
+    },
+    {
+      "epoch": 0.2896,
+      "grad_norm": 5.229098320007324,
+      "learning_rate": 7.327614648630814e-07,
+      "loss": 0.9686,
+      "step": 905
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 6.258688926696777,
+      "learning_rate": 7.311118442758165e-07,
+      "loss": 0.8538,
+      "step": 910
+    },
+    {
+      "epoch": 0.2928,
+      "grad_norm": 5.021489143371582,
+      "learning_rate": 7.294622236885516e-07,
+      "loss": 0.7937,
+      "step": 915
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 3.9986860752105713,
+      "learning_rate": 7.278126031012866e-07,
+      "loss": 0.8503,
+      "step": 920
+    },
+    {
+      "epoch": 0.296,
+      "grad_norm": 3.62813663482666,
+      "learning_rate": 7.261629825140217e-07,
+      "loss": 0.8199,
+      "step": 925
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 4.517638683319092,
+      "learning_rate": 7.245133619267568e-07,
+      "loss": 0.8521,
+      "step": 930
+    },
+    {
+      "epoch": 0.2992,
+      "grad_norm": 8.5663423538208,
+      "learning_rate": 7.228637413394919e-07,
+      "loss": 0.9132,
+      "step": 935
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 4.776329040527344,
+      "learning_rate": 7.212141207522269e-07,
+      "loss": 0.8566,
+      "step": 940
+    },
+    {
+      "epoch": 0.3024,
+      "grad_norm": 4.764708518981934,
+      "learning_rate": 7.19564500164962e-07,
+      "loss": 0.8181,
+      "step": 945
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 3.810011863708496,
+      "learning_rate": 7.179148795776971e-07,
+      "loss": 1.003,
+      "step": 950
+    },
+    {
+      "epoch": 0.3056,
+      "grad_norm": 4.385900020599365,
+      "learning_rate": 7.162652589904322e-07,
+      "loss": 0.8525,
+      "step": 955
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 8.322861671447754,
+      "learning_rate": 7.146156384031672e-07,
+      "loss": 0.9047,
+      "step": 960
+    },
+    {
+      "epoch": 0.3088,
+      "grad_norm": 6.5899224281311035,
+      "learning_rate": 7.129660178159022e-07,
+      "loss": 0.7103,
+      "step": 965
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 6.054207801818848,
+      "learning_rate": 7.113163972286373e-07,
+      "loss": 0.876,
+      "step": 970
+    },
+    {
+      "epoch": 0.312,
+      "grad_norm": 3.6956799030303955,
+      "learning_rate": 7.096667766413724e-07,
+      "loss": 0.9382,
+      "step": 975
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 4.055649757385254,
+      "learning_rate": 7.080171560541075e-07,
+      "loss": 0.8944,
+      "step": 980
+    },
+    {
+      "epoch": 0.3152,
+      "grad_norm": 2.8398051261901855,
+      "learning_rate": 7.063675354668426e-07,
+      "loss": 0.8233,
+      "step": 985
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 5.903645992279053,
+      "learning_rate": 7.047179148795776e-07,
+      "loss": 1.0067,
+      "step": 990
+    },
+    {
+      "epoch": 0.3184,
+      "grad_norm": 5.374630451202393,
+      "learning_rate": 7.030682942923127e-07,
+      "loss": 0.8423,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.729516506195068,
+      "learning_rate": 7.014186737050478e-07,
+      "loss": 1.0209,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3216,
+      "grad_norm": 3.4207727909088135,
+      "learning_rate": 6.997690531177829e-07,
+      "loss": 0.9196,
+      "step": 1005
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 5.797353744506836,
+      "learning_rate": 6.981194325305179e-07,
+      "loss": 0.921,
+      "step": 1010
+    },
+    {
+      "epoch": 0.3248,
+      "grad_norm": 4.802167892456055,
+      "learning_rate": 6.96469811943253e-07,
+      "loss": 0.8844,
+      "step": 1015
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 6.671936511993408,
+      "learning_rate": 6.948201913559881e-07,
+      "loss": 0.8399,
+      "step": 1020
+    },
+    {
+      "epoch": 0.328,
+      "grad_norm": 4.027926921844482,
+      "learning_rate": 6.931705707687232e-07,
+      "loss": 0.8733,
+      "step": 1025
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 7.0996575355529785,
+      "learning_rate": 6.915209501814583e-07,
+      "loss": 0.8886,
+      "step": 1030
+    },
+    {
+      "epoch": 0.3312,
+      "grad_norm": 3.9534752368927,
+      "learning_rate": 6.898713295941932e-07,
+      "loss": 0.9144,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 5.016911506652832,
+      "learning_rate": 6.882217090069283e-07,
+      "loss": 0.9936,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3344,
+      "grad_norm": 3.6231181621551514,
+      "learning_rate": 6.865720884196634e-07,
+      "loss": 0.7205,
+      "step": 1045
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 3.6556529998779297,
+      "learning_rate": 6.849224678323985e-07,
+      "loss": 0.914,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3376,
+      "grad_norm": 3.27970552444458,
+      "learning_rate": 6.832728472451335e-07,
+      "loss": 0.8306,
+      "step": 1055
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 3.816570997238159,
+      "learning_rate": 6.816232266578686e-07,
+      "loss": 0.9458,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3408,
+      "grad_norm": 7.391907215118408,
+      "learning_rate": 6.799736060706037e-07,
+      "loss": 0.8389,
+      "step": 1065
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 3.756998300552368,
+      "learning_rate": 6.783239854833388e-07,
+      "loss": 0.8889,
+      "step": 1070
+    },
+    {
+      "epoch": 0.344,
+      "grad_norm": 4.19740629196167,
+      "learning_rate": 6.766743648960739e-07,
+      "loss": 0.8942,
+      "step": 1075
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 4.351824760437012,
+      "learning_rate": 6.750247443088089e-07,
+      "loss": 0.9602,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3472,
+      "grad_norm": 3.371953010559082,
+      "learning_rate": 6.73375123721544e-07,
+      "loss": 0.9365,
+      "step": 1085
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 5.847254753112793,
+      "learning_rate": 6.717255031342791e-07,
+      "loss": 0.7705,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3504,
+      "grad_norm": 5.160730361938477,
+      "learning_rate": 6.700758825470142e-07,
+      "loss": 0.9221,
+      "step": 1095
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 3.4430952072143555,
+      "learning_rate": 6.684262619597493e-07,
+      "loss": 0.9484,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3536,
+      "grad_norm": 4.218683242797852,
+      "learning_rate": 6.667766413724843e-07,
+      "loss": 0.8446,
+      "step": 1105
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 5.120244026184082,
+      "learning_rate": 6.651270207852193e-07,
+      "loss": 0.9515,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3568,
+      "grad_norm": 5.609252452850342,
+      "learning_rate": 6.634774001979544e-07,
+      "loss": 0.8694,
+      "step": 1115
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 3.753680467605591,
+      "learning_rate": 6.618277796106895e-07,
+      "loss": 0.9242,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.665069580078125,
+      "learning_rate": 6.601781590234245e-07,
+      "loss": 0.9034,
+      "step": 1125
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 4.715619087219238,
+      "learning_rate": 6.585285384361596e-07,
+      "loss": 0.8842,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3632,
+      "grad_norm": 4.438577651977539,
+      "learning_rate": 6.568789178488947e-07,
+      "loss": 0.9432,
+      "step": 1135
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 3.8930490016937256,
+      "learning_rate": 6.552292972616298e-07,
+      "loss": 0.8113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.3664,
+      "grad_norm": 4.182096004486084,
+      "learning_rate": 6.535796766743649e-07,
+      "loss": 0.7999,
+      "step": 1145
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 5.353331089019775,
+      "learning_rate": 6.519300560870999e-07,
+      "loss": 0.8499,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3696,
+      "grad_norm": 3.641796588897705,
+      "learning_rate": 6.50280435499835e-07,
+      "loss": 0.9247,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 4.2418646812438965,
+      "learning_rate": 6.486308149125701e-07,
+      "loss": 1.0406,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3728,
+      "grad_norm": 2.948838233947754,
+      "learning_rate": 6.469811943253052e-07,
+      "loss": 0.8411,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 7.832685947418213,
+      "learning_rate": 6.453315737380403e-07,
+      "loss": 0.9339,
+      "step": 1170
+    },
+    {
+      "epoch": 0.376,
+      "grad_norm": 3.882305145263672,
+      "learning_rate": 6.436819531507753e-07,
+      "loss": 0.9936,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 6.916220664978027,
+      "learning_rate": 6.420323325635104e-07,
+      "loss": 1.0441,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3792,
+      "grad_norm": 6.770009517669678,
+      "learning_rate": 6.403827119762454e-07,
+      "loss": 0.9299,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 4.584465980529785,
+      "learning_rate": 6.387330913889805e-07,
+      "loss": 0.9169,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3824,
+      "grad_norm": 8.000226974487305,
+      "learning_rate": 6.370834708017155e-07,
+      "loss": 1.0345,
+      "step": 1195
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 13.314818382263184,
+      "learning_rate": 6.354338502144506e-07,
+      "loss": 0.919,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3856,
+      "grad_norm": 3.7661311626434326,
+      "learning_rate": 6.337842296271857e-07,
+      "loss": 0.9678,
+      "step": 1205
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 4.133317470550537,
+      "learning_rate": 6.321346090399208e-07,
+      "loss": 0.9533,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3888,
+      "grad_norm": 2.6607346534729004,
+      "learning_rate": 6.304849884526559e-07,
+      "loss": 0.9946,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 3.8332831859588623,
+      "learning_rate": 6.288353678653909e-07,
+      "loss": 0.8293,
+      "step": 1220
+    },
+    {
+      "epoch": 0.392,
+      "grad_norm": 3.6170003414154053,
+      "learning_rate": 6.27185747278126e-07,
+      "loss": 0.8952,
+      "step": 1225
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 5.026386737823486,
+      "learning_rate": 6.255361266908611e-07,
+      "loss": 0.917,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3952,
+      "grad_norm": 10.285544395446777,
+      "learning_rate": 6.238865061035962e-07,
+      "loss": 0.8624,
+      "step": 1235
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 2.899703025817871,
+      "learning_rate": 6.222368855163313e-07,
+      "loss": 0.9242,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3984,
+      "grad_norm": 6.087869167327881,
+      "learning_rate": 6.205872649290663e-07,
+      "loss": 0.8574,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.513827323913574,
+      "learning_rate": 6.189376443418014e-07,
+      "loss": 0.9205,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4016,
+      "grad_norm": 3.0381855964660645,
+      "learning_rate": 6.172880237545365e-07,
+      "loss": 0.8779,
+      "step": 1255
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 9.265677452087402,
+      "learning_rate": 6.156384031672715e-07,
+      "loss": 0.8645,
+      "step": 1260
+    },
+    {
+      "epoch": 0.4048,
+      "grad_norm": 5.159205436706543,
+      "learning_rate": 6.139887825800065e-07,
+      "loss": 0.7365,
+      "step": 1265
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 4.157783031463623,
+      "learning_rate": 6.123391619927416e-07,
+      "loss": 1.003,
+      "step": 1270
+    },
+    {
+      "epoch": 0.408,
+      "grad_norm": 4.129422187805176,
+      "learning_rate": 6.106895414054767e-07,
+      "loss": 1.0333,
+      "step": 1275
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 5.1480536460876465,
+      "learning_rate": 6.090399208182118e-07,
+      "loss": 0.9097,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4112,
+      "grad_norm": 6.21195650100708,
+      "learning_rate": 6.073903002309469e-07,
+      "loss": 0.9753,
+      "step": 1285
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 4.375741481781006,
+      "learning_rate": 6.057406796436819e-07,
+      "loss": 0.8613,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4144,
+      "grad_norm": 6.381781578063965,
+      "learning_rate": 6.04091059056417e-07,
+      "loss": 0.8447,
+      "step": 1295
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 6.689861297607422,
+      "learning_rate": 6.024414384691521e-07,
+      "loss": 0.9417,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4176,
+      "grad_norm": 7.573152542114258,
+      "learning_rate": 6.007918178818872e-07,
+      "loss": 0.8559,
+      "step": 1305
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 3.2965288162231445,
+      "learning_rate": 5.991421972946222e-07,
+      "loss": 0.8483,
+      "step": 1310
+    },
+    {
+      "epoch": 0.4208,
+      "grad_norm": 3.5138070583343506,
+      "learning_rate": 5.974925767073573e-07,
+      "loss": 0.7122,
+      "step": 1315
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 3.6955363750457764,
+      "learning_rate": 5.958429561200924e-07,
+      "loss": 0.8294,
+      "step": 1320
+    },
+    {
+      "epoch": 0.424,
+      "grad_norm": 3.6129462718963623,
+      "learning_rate": 5.941933355328275e-07,
+      "loss": 1.0534,
+      "step": 1325
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 4.549454212188721,
+      "learning_rate": 5.925437149455626e-07,
+      "loss": 0.8865,
+      "step": 1330
+    },
+    {
+      "epoch": 0.4272,
+      "grad_norm": 5.264537811279297,
+      "learning_rate": 5.908940943582975e-07,
+      "loss": 0.9219,
+      "step": 1335
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 5.2592267990112305,
+      "learning_rate": 5.892444737710326e-07,
+      "loss": 0.8414,
+      "step": 1340
+    },
+    {
+      "epoch": 0.4304,
+      "grad_norm": 7.621036529541016,
+      "learning_rate": 5.875948531837677e-07,
+      "loss": 0.8553,
+      "step": 1345
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 4.718931674957275,
+      "learning_rate": 5.859452325965028e-07,
+      "loss": 1.0172,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4336,
+      "grad_norm": 3.382890224456787,
+      "learning_rate": 5.842956120092379e-07,
+      "loss": 0.9208,
+      "step": 1355
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 5.720285415649414,
+      "learning_rate": 5.826459914219729e-07,
+      "loss": 0.9376,
+      "step": 1360
+    },
+    {
+      "epoch": 0.4368,
+      "grad_norm": 6.348506450653076,
+      "learning_rate": 5.80996370834708e-07,
+      "loss": 0.8526,
+      "step": 1365
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 3.639775276184082,
+      "learning_rate": 5.793467502474431e-07,
+      "loss": 0.9067,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.016887187957764,
+      "learning_rate": 5.776971296601782e-07,
+      "loss": 0.9526,
+      "step": 1375
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 6.822007656097412,
+      "learning_rate": 5.760475090729132e-07,
+      "loss": 0.776,
+      "step": 1380
+    },
+    {
+      "epoch": 0.4432,
+      "grad_norm": 4.3558125495910645,
+      "learning_rate": 5.743978884856483e-07,
+      "loss": 0.8932,
+      "step": 1385
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 6.4160475730896,
+      "learning_rate": 5.727482678983834e-07,
+      "loss": 0.9081,
+      "step": 1390
+    },
+    {
+      "epoch": 0.4464,
+      "grad_norm": 3.1769394874572754,
+      "learning_rate": 5.710986473111185e-07,
+      "loss": 0.8688,
+      "step": 1395
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 3.9851887226104736,
+      "learning_rate": 5.694490267238536e-07,
+      "loss": 0.877,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4496,
+      "grad_norm": 5.6506571769714355,
+      "learning_rate": 5.677994061365886e-07,
+      "loss": 0.8325,
+      "step": 1405
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 3.549743175506592,
+      "learning_rate": 5.661497855493236e-07,
+      "loss": 0.8987,
+      "step": 1410
+    },
+    {
+      "epoch": 0.4528,
+      "grad_norm": 3.143094062805176,
+      "learning_rate": 5.645001649620587e-07,
+      "loss": 0.9429,
+      "step": 1415
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 8.156094551086426,
+      "learning_rate": 5.628505443747938e-07,
+      "loss": 0.7903,
+      "step": 1420
+    },
+    {
+      "epoch": 0.456,
+      "grad_norm": 4.86202335357666,
+      "learning_rate": 5.612009237875289e-07,
+      "loss": 0.933,
+      "step": 1425
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 5.636049270629883,
+      "learning_rate": 5.595513032002639e-07,
+      "loss": 0.921,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4592,
+      "grad_norm": 3.5446226596832275,
+      "learning_rate": 5.57901682612999e-07,
+      "loss": 0.8239,
+      "step": 1435
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 3.516528606414795,
+      "learning_rate": 5.562520620257341e-07,
+      "loss": 0.894,
+      "step": 1440
+    },
+    {
+      "epoch": 0.4624,
+      "grad_norm": 3.1388487815856934,
+      "learning_rate": 5.546024414384692e-07,
+      "loss": 0.9401,
+      "step": 1445
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 3.378370761871338,
+      "learning_rate": 5.529528208512042e-07,
+      "loss": 0.8786,
+      "step": 1450
+    },
+    {
+      "epoch": 0.4656,
+      "grad_norm": 4.898928165435791,
+      "learning_rate": 5.513032002639393e-07,
+      "loss": 0.8457,
+      "step": 1455
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 8.320155143737793,
+      "learning_rate": 5.496535796766744e-07,
+      "loss": 0.9857,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4688,
+      "grad_norm": 12.393474578857422,
+      "learning_rate": 5.480039590894095e-07,
+      "loss": 1.0179,
+      "step": 1465
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 3.834761142730713,
+      "learning_rate": 5.463543385021446e-07,
+      "loss": 0.8355,
+      "step": 1470
+    },
+    {
+      "epoch": 0.472,
+      "grad_norm": 5.7657694816589355,
+      "learning_rate": 5.447047179148796e-07,
+      "loss": 0.906,
+      "step": 1475
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 2.89928936958313,
+      "learning_rate": 5.430550973276147e-07,
+      "loss": 0.8453,
+      "step": 1480
+    },
+    {
+      "epoch": 0.4752,
+      "grad_norm": 3.30023455619812,
+      "learning_rate": 5.414054767403497e-07,
+      "loss": 0.863,
+      "step": 1485
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 6.904449462890625,
+      "learning_rate": 5.397558561530848e-07,
+      "loss": 0.8722,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4784,
+      "grad_norm": 2.936325788497925,
+      "learning_rate": 5.381062355658197e-07,
+      "loss": 0.8565,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.707699775695801,
+      "learning_rate": 5.364566149785548e-07,
+      "loss": 0.827,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4816,
+      "grad_norm": 3.5800673961639404,
+      "learning_rate": 5.3480699439129e-07,
+      "loss": 0.8582,
+      "step": 1505
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 5.940330505371094,
+      "learning_rate": 5.33157373804025e-07,
+      "loss": 0.9983,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4848,
+      "grad_norm": 4.438694000244141,
+      "learning_rate": 5.315077532167602e-07,
+      "loss": 0.8307,
+      "step": 1515
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 6.149857044219971,
+      "learning_rate": 5.298581326294951e-07,
+      "loss": 0.9128,
+      "step": 1520
+    },
+    {
+      "epoch": 0.488,
+      "grad_norm": 3.874925136566162,
+      "learning_rate": 5.282085120422302e-07,
+      "loss": 0.966,
+      "step": 1525
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 17.836402893066406,
+      "learning_rate": 5.265588914549653e-07,
+      "loss": 0.8737,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4912,
+      "grad_norm": 5.488133430480957,
+      "learning_rate": 5.249092708677005e-07,
+      "loss": 0.9353,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 4.590605735778809,
+      "learning_rate": 5.232596502804356e-07,
+      "loss": 0.8757,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4944,
+      "grad_norm": 3.0040788650512695,
+      "learning_rate": 5.216100296931705e-07,
+      "loss": 0.8116,
+      "step": 1545
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 3.2542080879211426,
+      "learning_rate": 5.199604091059056e-07,
+      "loss": 0.9621,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4976,
+      "grad_norm": 4.786755561828613,
+      "learning_rate": 5.183107885186407e-07,
+      "loss": 0.9551,
+      "step": 1555
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 5.058788299560547,
+      "learning_rate": 5.166611679313757e-07,
+      "loss": 0.7788,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5008,
+      "grad_norm": 6.7137131690979,
+      "learning_rate": 5.150115473441107e-07,
+      "loss": 0.8605,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5024,
+      "grad_norm": 5.942770004272461,
+      "learning_rate": 5.133619267568458e-07,
+      "loss": 0.8731,
+      "step": 1570
+    },
+    {
+      "epoch": 0.504,
+      "grad_norm": 3.7935171127319336,
+      "learning_rate": 5.117123061695809e-07,
+      "loss": 0.8126,
+      "step": 1575
+    },
+    {
+      "epoch": 0.5056,
+      "grad_norm": 3.8675737380981445,
+      "learning_rate": 5.10062685582316e-07,
+      "loss": 0.8163,
+      "step": 1580
+    },
+    {
+      "epoch": 0.5072,
+      "grad_norm": 3.6353890895843506,
+      "learning_rate": 5.084130649950511e-07,
+      "loss": 0.8407,
+      "step": 1585
+    },
+    {
+      "epoch": 0.5088,
+      "grad_norm": 6.919312477111816,
+      "learning_rate": 5.067634444077861e-07,
+      "loss": 0.8906,
+      "step": 1590
+    },
+    {
+      "epoch": 0.5104,
+      "grad_norm": 4.731250286102295,
+      "learning_rate": 5.051138238205212e-07,
+      "loss": 0.7421,
+      "step": 1595
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 3.6495304107666016,
+      "learning_rate": 5.034642032332563e-07,
+      "loss": 0.8691,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5136,
+      "grad_norm": 3.6082992553710938,
+      "learning_rate": 5.018145826459914e-07,
+      "loss": 0.9252,
+      "step": 1605
+    },
+    {
+      "epoch": 0.5152,
+      "grad_norm": 2.5912933349609375,
+      "learning_rate": 5.001649620587265e-07,
+      "loss": 0.8538,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5168,
+      "grad_norm": 7.729884624481201,
+      "learning_rate": 4.985153414714615e-07,
+      "loss": 0.7578,
+      "step": 1615
+    },
+    {
+      "epoch": 0.5184,
+      "grad_norm": 4.614051342010498,
+      "learning_rate": 4.968657208841966e-07,
+      "loss": 0.8363,
+      "step": 1620
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.5848758220672607,
+      "learning_rate": 4.952161002969316e-07,
+      "loss": 0.902,
+      "step": 1625
+    },
+    {
+      "epoch": 0.5216,
+      "grad_norm": 4.744536399841309,
+      "learning_rate": 4.935664797096667e-07,
+      "loss": 0.8356,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5232,
+      "grad_norm": 6.719925880432129,
+      "learning_rate": 4.919168591224018e-07,
+      "loss": 0.8663,
+      "step": 1635
+    },
+    {
+      "epoch": 0.5248,
+      "grad_norm": 5.994638442993164,
+      "learning_rate": 4.902672385351369e-07,
+      "loss": 0.8854,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5264,
+      "grad_norm": 3.5340418815612793,
+      "learning_rate": 4.88617617947872e-07,
+      "loss": 0.738,
+      "step": 1645
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 4.414712905883789,
+      "learning_rate": 4.86967997360607e-07,
+      "loss": 0.8637,
+      "step": 1650
+    },
+    {
+      "epoch": 0.5296,
+      "grad_norm": 3.8119003772735596,
+      "learning_rate": 4.853183767733421e-07,
+      "loss": 0.8948,
+      "step": 1655
+    },
+    {
+      "epoch": 0.5312,
+      "grad_norm": 3.453695058822632,
+      "learning_rate": 4.836687561860771e-07,
+      "loss": 0.9324,
+      "step": 1660
+    },
+    {
+      "epoch": 0.5328,
+      "grad_norm": 8.695696830749512,
+      "learning_rate": 4.820191355988122e-07,
+      "loss": 0.8486,
+      "step": 1665
+    },
+    {
+      "epoch": 0.5344,
+      "grad_norm": 3.696438789367676,
+      "learning_rate": 4.803695150115473e-07,
+      "loss": 0.8801,
+      "step": 1670
+    },
+    {
+      "epoch": 0.536,
+      "grad_norm": 5.598580360412598,
+      "learning_rate": 4.787198944242824e-07,
+      "loss": 0.9598,
+      "step": 1675
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 7.447549819946289,
+      "learning_rate": 4.770702738370175e-07,
+      "loss": 0.7981,
+      "step": 1680
+    },
+    {
+      "epoch": 0.5392,
+      "grad_norm": 3.8933768272399902,
+      "learning_rate": 4.754206532497526e-07,
+      "loss": 0.7836,
+      "step": 1685
+    },
+    {
+      "epoch": 0.5408,
+      "grad_norm": 4.233343124389648,
+      "learning_rate": 4.737710326624876e-07,
+      "loss": 0.9953,
+      "step": 1690
+    },
+    {
+      "epoch": 0.5424,
+      "grad_norm": 4.121957302093506,
+      "learning_rate": 4.721214120752227e-07,
+      "loss": 0.8909,
+      "step": 1695
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 3.326876640319824,
+      "learning_rate": 4.704717914879577e-07,
+      "loss": 0.8207,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5456,
+      "grad_norm": 4.2965006828308105,
+      "learning_rate": 4.688221709006928e-07,
+      "loss": 0.7338,
+      "step": 1705
+    },
+    {
+      "epoch": 0.5472,
+      "grad_norm": 12.319995880126953,
+      "learning_rate": 4.6717255031342787e-07,
+      "loss": 0.8716,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5488,
+      "grad_norm": 3.4306647777557373,
+      "learning_rate": 4.6552292972616297e-07,
+      "loss": 0.8844,
+      "step": 1715
+    },
+    {
+      "epoch": 0.5504,
+      "grad_norm": 3.8839616775512695,
+      "learning_rate": 4.638733091388981e-07,
+      "loss": 0.6966,
+      "step": 1720
+    },
+    {
+      "epoch": 0.552,
+      "grad_norm": 4.802063941955566,
+      "learning_rate": 4.622236885516331e-07,
+      "loss": 1.0128,
+      "step": 1725
+    },
+    {
+      "epoch": 0.5536,
+      "grad_norm": 3.7047386169433594,
+      "learning_rate": 4.6057406796436817e-07,
+      "loss": 0.7706,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5552,
+      "grad_norm": 5.304298400878906,
+      "learning_rate": 4.589244473771032e-07,
+      "loss": 0.7627,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5568,
+      "grad_norm": 3.211620330810547,
+      "learning_rate": 4.572748267898383e-07,
+      "loss": 0.8409,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5584,
+      "grad_norm": 4.873741149902344,
+      "learning_rate": 4.5562520620257337e-07,
+      "loss": 0.9454,
+      "step": 1745
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.801036834716797,
+      "learning_rate": 4.5397558561530847e-07,
+      "loss": 0.8148,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5616,
+      "grad_norm": 4.238209247589111,
+      "learning_rate": 4.5232596502804357e-07,
+      "loss": 0.907,
+      "step": 1755
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 5.311016082763672,
+      "learning_rate": 4.506763444407786e-07,
+      "loss": 0.8865,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5648,
+      "grad_norm": 5.096076011657715,
+      "learning_rate": 4.4902672385351366e-07,
+      "loss": 0.8967,
+      "step": 1765
+    },
+    {
+      "epoch": 0.5664,
+      "grad_norm": 3.0391855239868164,
+      "learning_rate": 4.473771032662487e-07,
+      "loss": 0.8671,
+      "step": 1770
+    },
+    {
+      "epoch": 0.568,
+      "grad_norm": 3.2100040912628174,
+      "learning_rate": 4.457274826789838e-07,
+      "loss": 0.8144,
+      "step": 1775
+    },
+    {
+      "epoch": 0.5696,
+      "grad_norm": 4.312873840332031,
+      "learning_rate": 4.4407786209171886e-07,
+      "loss": 0.9743,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5712,
+      "grad_norm": 4.577536582946777,
+      "learning_rate": 4.4242824150445396e-07,
+      "loss": 0.8988,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5728,
+      "grad_norm": 6.181716442108154,
+      "learning_rate": 4.40778620917189e-07,
+      "loss": 0.8709,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5744,
+      "grad_norm": 3.878676176071167,
+      "learning_rate": 4.391290003299241e-07,
+      "loss": 0.9212,
+      "step": 1795
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 3.634641170501709,
+      "learning_rate": 4.374793797426592e-07,
+      "loss": 0.9252,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5776,
+      "grad_norm": 4.589493274688721,
+      "learning_rate": 4.358297591553942e-07,
+      "loss": 0.9442,
+      "step": 1805
+    },
+    {
+      "epoch": 0.5792,
+      "grad_norm": 3.5581719875335693,
+      "learning_rate": 4.341801385681293e-07,
+      "loss": 0.8068,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5808,
+      "grad_norm": 10.048519134521484,
+      "learning_rate": 4.3253051798086436e-07,
+      "loss": 0.9334,
+      "step": 1815
+    },
+    {
+      "epoch": 0.5824,
+      "grad_norm": 7.136456489562988,
+      "learning_rate": 4.3088089739359946e-07,
+      "loss": 0.8711,
+      "step": 1820
+    },
+    {
+      "epoch": 0.584,
+      "grad_norm": 12.951844215393066,
+      "learning_rate": 4.292312768063345e-07,
+      "loss": 0.8709,
+      "step": 1825
+    },
+    {
+      "epoch": 0.5856,
+      "grad_norm": 4.377983093261719,
+      "learning_rate": 4.275816562190696e-07,
+      "loss": 0.8509,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5872,
+      "grad_norm": 4.836514472961426,
+      "learning_rate": 4.259320356318047e-07,
+      "loss": 0.882,
+      "step": 1835
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 3.0401344299316406,
+      "learning_rate": 4.242824150445397e-07,
+      "loss": 0.9351,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5904,
+      "grad_norm": 4.854428768157959,
+      "learning_rate": 4.226327944572748e-07,
+      "loss": 0.9465,
+      "step": 1845
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 3.092222213745117,
+      "learning_rate": 4.2098317387000985e-07,
+      "loss": 0.8038,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5936,
+      "grad_norm": 5.498143196105957,
+      "learning_rate": 4.1933355328274495e-07,
+      "loss": 0.769,
+      "step": 1855
+    },
+    {
+      "epoch": 0.5952,
+      "grad_norm": 2.4063949584960938,
+      "learning_rate": 4.1768393269548e-07,
+      "loss": 0.8057,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5968,
+      "grad_norm": 5.123895168304443,
+      "learning_rate": 4.160343121082151e-07,
+      "loss": 0.9554,
+      "step": 1865
+    },
+    {
+      "epoch": 0.5984,
+      "grad_norm": 7.29245662689209,
+      "learning_rate": 4.143846915209502e-07,
+      "loss": 0.9079,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9312267303466797,
+      "learning_rate": 4.1273507093368525e-07,
+      "loss": 0.8765,
+      "step": 1875
+    },
+    {
+      "epoch": 0.6016,
+      "grad_norm": 3.0390522480010986,
+      "learning_rate": 4.110854503464203e-07,
+      "loss": 0.9539,
+      "step": 1880
+    },
+    {
+      "epoch": 0.6032,
+      "grad_norm": 3.8350090980529785,
+      "learning_rate": 4.0943582975915535e-07,
+      "loss": 0.8397,
+      "step": 1885
+    },
+    {
+      "epoch": 0.6048,
+      "grad_norm": 3.9119083881378174,
+      "learning_rate": 4.0778620917189045e-07,
+      "loss": 0.8769,
+      "step": 1890
+    },
+    {
+      "epoch": 0.6064,
+      "grad_norm": 3.361199378967285,
+      "learning_rate": 4.061365885846255e-07,
+      "loss": 0.9028,
+      "step": 1895
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 4.87637186050415,
+      "learning_rate": 4.044869679973606e-07,
+      "loss": 0.8022,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6096,
+      "grad_norm": 4.546545505523682,
+      "learning_rate": 4.028373474100957e-07,
+      "loss": 0.7343,
+      "step": 1905
+    },
+    {
+      "epoch": 0.6112,
+      "grad_norm": 2.975339651107788,
+      "learning_rate": 4.0118772682283075e-07,
+      "loss": 0.8335,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6128,
+      "grad_norm": 3.8709919452667236,
+      "learning_rate": 3.995381062355658e-07,
+      "loss": 0.7803,
+      "step": 1915
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 2.690919876098633,
+      "learning_rate": 3.9788848564830084e-07,
+      "loss": 1.0247,
+      "step": 1920
+    },
+    {
+      "epoch": 0.616,
+      "grad_norm": 4.163801193237305,
+      "learning_rate": 3.9623886506103594e-07,
+      "loss": 0.8764,
+      "step": 1925
+    },
+    {
+      "epoch": 0.6176,
+      "grad_norm": 5.445613384246826,
+      "learning_rate": 3.94589244473771e-07,
+      "loss": 0.9047,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6192,
+      "grad_norm": 3.3369109630584717,
+      "learning_rate": 3.929396238865061e-07,
+      "loss": 0.9644,
+      "step": 1935
+    },
+    {
+      "epoch": 0.6208,
+      "grad_norm": 2.8063957691192627,
+      "learning_rate": 3.912900032992412e-07,
+      "loss": 0.8411,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6224,
+      "grad_norm": 3.369598865509033,
+      "learning_rate": 3.8964038271197624e-07,
+      "loss": 0.8904,
+      "step": 1945
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 11.861967086791992,
+      "learning_rate": 3.8799076212471134e-07,
+      "loss": 0.8503,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6256,
+      "grad_norm": 3.746105670928955,
+      "learning_rate": 3.8634114153744634e-07,
+      "loss": 0.8549,
+      "step": 1955
+    },
+    {
+      "epoch": 0.6272,
+      "grad_norm": 4.717544078826904,
+      "learning_rate": 3.8469152095018144e-07,
+      "loss": 0.7668,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6288,
+      "grad_norm": 3.0035829544067383,
+      "learning_rate": 3.830419003629165e-07,
+      "loss": 0.7519,
+      "step": 1965
+    },
+    {
+      "epoch": 0.6304,
+      "grad_norm": 4.065003395080566,
+      "learning_rate": 3.813922797756516e-07,
+      "loss": 0.8116,
+      "step": 1970
+    },
+    {
+      "epoch": 0.632,
+      "grad_norm": 5.251111030578613,
+      "learning_rate": 3.7974265918838663e-07,
+      "loss": 0.8879,
+      "step": 1975
+    },
+    {
+      "epoch": 0.6336,
+      "grad_norm": 5.612459659576416,
+      "learning_rate": 3.7809303860112173e-07,
+      "loss": 0.7931,
+      "step": 1980
+    },
+    {
+      "epoch": 0.6352,
+      "grad_norm": 4.041755199432373,
+      "learning_rate": 3.7644341801385684e-07,
+      "loss": 0.7963,
+      "step": 1985
+    },
+    {
+      "epoch": 0.6368,
+      "grad_norm": 9.98974609375,
+      "learning_rate": 3.7479379742659183e-07,
+      "loss": 0.8962,
+      "step": 1990
+    },
+    {
+      "epoch": 0.6384,
+      "grad_norm": 3.949065685272217,
+      "learning_rate": 3.7314417683932693e-07,
+      "loss": 0.9132,
+      "step": 1995
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.0974297523498535,
+      "learning_rate": 3.71494556252062e-07,
+      "loss": 0.9125,
+      "step": 2000
+    },
+    {
+      "epoch": 0.6416,
+      "grad_norm": 3.70499324798584,
+      "learning_rate": 3.698449356647971e-07,
+      "loss": 0.7575,
+      "step": 2005
+    },
+    {
+      "epoch": 0.6432,
+      "grad_norm": 4.345754623413086,
+      "learning_rate": 3.6819531507753213e-07,
+      "loss": 0.7473,
+      "step": 2010
+    },
+    {
+      "epoch": 0.6448,
+      "grad_norm": 2.8242263793945312,
+      "learning_rate": 3.6654569449026723e-07,
+      "loss": 0.959,
+      "step": 2015
+    },
+    {
+      "epoch": 0.6464,
+      "grad_norm": 3.6714463233947754,
+      "learning_rate": 3.6489607390300233e-07,
+      "loss": 0.7814,
+      "step": 2020
+    },
+    {
+      "epoch": 0.648,
+      "grad_norm": 4.4022908210754395,
+      "learning_rate": 3.632464533157374e-07,
+      "loss": 0.8977,
+      "step": 2025
+    },
+    {
+      "epoch": 0.6496,
+      "grad_norm": 3.5451886653900146,
+      "learning_rate": 3.6159683272847243e-07,
+      "loss": 0.8835,
+      "step": 2030
+    },
+    {
+      "epoch": 0.6512,
+      "grad_norm": 5.699954509735107,
+      "learning_rate": 3.599472121412075e-07,
+      "loss": 0.8339,
+      "step": 2035
+    },
+    {
+      "epoch": 0.6528,
+      "grad_norm": 3.2886204719543457,
+      "learning_rate": 3.582975915539426e-07,
+      "loss": 0.85,
+      "step": 2040
+    },
+    {
+      "epoch": 0.6544,
+      "grad_norm": 2.8363375663757324,
+      "learning_rate": 3.566479709666776e-07,
+      "loss": 0.8195,
+      "step": 2045
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 3.734877824783325,
+      "learning_rate": 3.549983503794127e-07,
+      "loss": 0.8803,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6576,
+      "grad_norm": 2.6836330890655518,
+      "learning_rate": 3.533487297921478e-07,
+      "loss": 0.8574,
+      "step": 2055
+    },
+    {
+      "epoch": 0.6592,
+      "grad_norm": 3.9296648502349854,
+      "learning_rate": 3.516991092048829e-07,
+      "loss": 0.7938,
+      "step": 2060
+    },
+    {
+      "epoch": 0.6608,
+      "grad_norm": 2.973696231842041,
+      "learning_rate": 3.500494886176179e-07,
+      "loss": 0.8593,
+      "step": 2065
+    },
+    {
+      "epoch": 0.6624,
+      "grad_norm": 4.675530433654785,
+      "learning_rate": 3.4839986803035297e-07,
+      "loss": 0.8456,
+      "step": 2070
+    },
+    {
+      "epoch": 0.664,
+      "grad_norm": 4.891861915588379,
+      "learning_rate": 3.4675024744308807e-07,
+      "loss": 0.9814,
+      "step": 2075
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 3.9921982288360596,
+      "learning_rate": 3.451006268558231e-07,
+      "loss": 0.8416,
+      "step": 2080
+    },
+    {
+      "epoch": 0.6672,
+      "grad_norm": 3.1958041191101074,
+      "learning_rate": 3.434510062685582e-07,
+      "loss": 0.905,
+      "step": 2085
+    },
+    {
+      "epoch": 0.6688,
+      "grad_norm": 4.344924449920654,
+      "learning_rate": 3.418013856812933e-07,
+      "loss": 0.8202,
+      "step": 2090
+    },
+    {
+      "epoch": 0.6704,
+      "grad_norm": 7.5191569328308105,
+      "learning_rate": 3.4015176509402837e-07,
+      "loss": 0.9426,
+      "step": 2095
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 4.440326690673828,
+      "learning_rate": 3.3850214450676347e-07,
+      "loss": 0.8205,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6736,
+      "grad_norm": 6.4901123046875,
+      "learning_rate": 3.3685252391949846e-07,
+      "loss": 0.7936,
+      "step": 2105
+    },
+    {
+      "epoch": 0.6752,
+      "grad_norm": 3.9426374435424805,
+      "learning_rate": 3.3520290333223357e-07,
+      "loss": 0.793,
+      "step": 2110
+    },
+    {
+      "epoch": 0.6768,
+      "grad_norm": 5.018584728240967,
+      "learning_rate": 3.335532827449686e-07,
+      "loss": 0.7799,
+      "step": 2115
+    },
+    {
+      "epoch": 0.6784,
+      "grad_norm": 3.7835421562194824,
+      "learning_rate": 3.319036621577037e-07,
+      "loss": 0.8696,
+      "step": 2120
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.0190839767456055,
+      "learning_rate": 3.302540415704388e-07,
+      "loss": 0.9139,
+      "step": 2125
+    },
+    {
+      "epoch": 0.6816,
+      "grad_norm": 5.751317977905273,
+      "learning_rate": 3.2860442098317386e-07,
+      "loss": 0.8522,
+      "step": 2130
+    },
+    {
+      "epoch": 0.6832,
+      "grad_norm": 6.684688091278076,
+      "learning_rate": 3.2695480039590896e-07,
+      "loss": 0.8061,
+      "step": 2135
+    },
+    {
+      "epoch": 0.6848,
+      "grad_norm": 2.783705234527588,
+      "learning_rate": 3.2530517980864396e-07,
+      "loss": 0.7725,
+      "step": 2140
+    },
+    {
+      "epoch": 0.6864,
+      "grad_norm": 4.636482238769531,
+      "learning_rate": 3.2365555922137906e-07,
+      "loss": 0.8479,
+      "step": 2145
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 5.260950565338135,
+      "learning_rate": 3.220059386341141e-07,
+      "loss": 1.0717,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6896,
+      "grad_norm": 5.191953659057617,
+      "learning_rate": 3.203563180468492e-07,
+      "loss": 0.8714,
+      "step": 2155
+    },
+    {
+      "epoch": 0.6912,
+      "grad_norm": 7.279730796813965,
+      "learning_rate": 3.187066974595843e-07,
+      "loss": 0.8246,
+      "step": 2160
+    },
+    {
+      "epoch": 0.6928,
+      "grad_norm": 2.966627359390259,
+      "learning_rate": 3.1705707687231936e-07,
+      "loss": 0.9053,
+      "step": 2165
+    },
+    {
+      "epoch": 0.6944,
+      "grad_norm": 8.789515495300293,
+      "learning_rate": 3.1540745628505446e-07,
+      "loss": 0.7851,
+      "step": 2170
+    },
+    {
+      "epoch": 0.696,
+      "grad_norm": 2.929105520248413,
+      "learning_rate": 3.137578356977895e-07,
+      "loss": 0.956,
+      "step": 2175
+    },
+    {
+      "epoch": 0.6976,
+      "grad_norm": 5.6356096267700195,
+      "learning_rate": 3.1210821511052456e-07,
+      "loss": 0.7946,
+      "step": 2180
+    },
+    {
+      "epoch": 0.6992,
+      "grad_norm": 3.3033862113952637,
+      "learning_rate": 3.104585945232596e-07,
+      "loss": 0.884,
+      "step": 2185
+    },
+    {
+      "epoch": 0.7008,
+      "grad_norm": 5.996482849121094,
+      "learning_rate": 3.088089739359947e-07,
+      "loss": 0.8342,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7024,
+      "grad_norm": 7.644280910491943,
+      "learning_rate": 3.0715935334872975e-07,
+      "loss": 0.8575,
+      "step": 2195
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 5.780369281768799,
+      "learning_rate": 3.0550973276146485e-07,
+      "loss": 0.84,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7056,
+      "grad_norm": 3.7677314281463623,
+      "learning_rate": 3.0386011217419995e-07,
+      "loss": 0.8764,
+      "step": 2205
+    },
+    {
+      "epoch": 0.7072,
+      "grad_norm": 4.153870105743408,
+      "learning_rate": 3.02210491586935e-07,
+      "loss": 0.8447,
+      "step": 2210
+    },
+    {
+      "epoch": 0.7088,
+      "grad_norm": 6.395594120025635,
+      "learning_rate": 3.0056087099967005e-07,
+      "loss": 0.8499,
+      "step": 2215
+    },
+    {
+      "epoch": 0.7104,
+      "grad_norm": 3.4210963249206543,
+      "learning_rate": 2.989112504124051e-07,
+      "loss": 0.7547,
+      "step": 2220
+    },
+    {
+      "epoch": 0.712,
+      "grad_norm": 2.710740327835083,
+      "learning_rate": 2.972616298251402e-07,
+      "loss": 0.8286,
+      "step": 2225
+    },
+    {
+      "epoch": 0.7136,
+      "grad_norm": 5.014111042022705,
+      "learning_rate": 2.9561200923787525e-07,
+      "loss": 0.8706,
+      "step": 2230
+    },
+    {
+      "epoch": 0.7152,
+      "grad_norm": 3.8330109119415283,
+      "learning_rate": 2.9396238865061035e-07,
+      "loss": 0.8221,
+      "step": 2235
+    },
+    {
+      "epoch": 0.7168,
+      "grad_norm": 5.695978164672852,
+      "learning_rate": 2.9231276806334545e-07,
+      "loss": 0.8412,
+      "step": 2240
+    },
+    {
+      "epoch": 0.7184,
+      "grad_norm": 5.974388599395752,
+      "learning_rate": 2.906631474760805e-07,
+      "loss": 0.8235,
+      "step": 2245
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9334166049957275,
+      "learning_rate": 2.890135268888156e-07,
+      "loss": 0.8965,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7216,
+      "grad_norm": 8.407828330993652,
+      "learning_rate": 2.873639063015506e-07,
+      "loss": 0.8991,
+      "step": 2255
+    },
+    {
+      "epoch": 0.7232,
+      "grad_norm": 4.443752765655518,
+      "learning_rate": 2.857142857142857e-07,
+      "loss": 0.7565,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7248,
+      "grad_norm": 6.351187229156494,
+      "learning_rate": 2.8406466512702074e-07,
+      "loss": 0.8202,
+      "step": 2265
+    },
+    {
+      "epoch": 0.7264,
+      "grad_norm": 4.715820789337158,
+      "learning_rate": 2.8241504453975584e-07,
+      "loss": 0.877,
+      "step": 2270
+    },
+    {
+      "epoch": 0.728,
+      "grad_norm": 3.1347246170043945,
+      "learning_rate": 2.8076542395249094e-07,
+      "loss": 0.8227,
+      "step": 2275
+    },
+    {
+      "epoch": 0.7296,
+      "grad_norm": 3.8322551250457764,
+      "learning_rate": 2.79115803365226e-07,
+      "loss": 0.7839,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7312,
+      "grad_norm": 4.289877414703369,
+      "learning_rate": 2.774661827779611e-07,
+      "loss": 0.7952,
+      "step": 2285
+    },
+    {
+      "epoch": 0.7328,
+      "grad_norm": 3.775768995285034,
+      "learning_rate": 2.758165621906961e-07,
+      "loss": 0.7924,
+      "step": 2290
+    },
+    {
+      "epoch": 0.7344,
+      "grad_norm": 4.233770370483398,
+      "learning_rate": 2.741669416034312e-07,
+      "loss": 0.8008,
+      "step": 2295
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 5.131399154663086,
+      "learning_rate": 2.7251732101616624e-07,
+      "loss": 0.8103,
+      "step": 2300
+    },
+    {
+      "epoch": 0.7376,
+      "grad_norm": 7.184566497802734,
+      "learning_rate": 2.7086770042890134e-07,
+      "loss": 0.9024,
+      "step": 2305
+    },
+    {
+      "epoch": 0.7392,
+      "grad_norm": 3.6044952869415283,
+      "learning_rate": 2.6921807984163644e-07,
+      "loss": 0.8925,
+      "step": 2310
+    },
+    {
+      "epoch": 0.7408,
+      "grad_norm": 8.124959945678711,
+      "learning_rate": 2.675684592543715e-07,
+      "loss": 0.8227,
+      "step": 2315
+    },
+    {
+      "epoch": 0.7424,
+      "grad_norm": 5.050447463989258,
+      "learning_rate": 2.659188386671066e-07,
+      "loss": 0.87,
+      "step": 2320
+    },
+    {
+      "epoch": 0.744,
+      "grad_norm": 3.2532646656036377,
+      "learning_rate": 2.6426921807984164e-07,
+      "loss": 0.9024,
+      "step": 2325
+    },
+    {
+      "epoch": 0.7456,
+      "grad_norm": 7.244692325592041,
+      "learning_rate": 2.626195974925767e-07,
+      "loss": 0.7402,
+      "step": 2330
+    },
+    {
+      "epoch": 0.7472,
+      "grad_norm": 5.4176435470581055,
+      "learning_rate": 2.6096997690531173e-07,
+      "loss": 0.9476,
+      "step": 2335
+    },
+    {
+      "epoch": 0.7488,
+      "grad_norm": 13.94157886505127,
+      "learning_rate": 2.5932035631804683e-07,
+      "loss": 0.7621,
+      "step": 2340
+    },
+    {
+      "epoch": 0.7504,
+      "grad_norm": 2.509117603302002,
+      "learning_rate": 2.5767073573078193e-07,
+      "loss": 0.7259,
+      "step": 2345
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 3.073138952255249,
+      "learning_rate": 2.56021115143517e-07,
+      "loss": 0.8224,
+      "step": 2350
+    },
+    {
+      "epoch": 0.7536,
+      "grad_norm": 3.9155077934265137,
+      "learning_rate": 2.543714945562521e-07,
+      "loss": 0.8596,
+      "step": 2355
+    },
+    {
+      "epoch": 0.7552,
+      "grad_norm": 6.405920028686523,
+      "learning_rate": 2.5272187396898713e-07,
+      "loss": 0.8152,
+      "step": 2360
+    },
+    {
+      "epoch": 0.7568,
+      "grad_norm": 3.8203928470611572,
+      "learning_rate": 2.510722533817222e-07,
+      "loss": 0.8907,
+      "step": 2365
+    },
+    {
+      "epoch": 0.7584,
+      "grad_norm": 3.9368674755096436,
+      "learning_rate": 2.494226327944573e-07,
+      "loss": 0.8804,
+      "step": 2370
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.452835559844971,
+      "learning_rate": 2.4777301220719233e-07,
+      "loss": 0.8304,
+      "step": 2375
+    },
+    {
+      "epoch": 0.7616,
+      "grad_norm": 4.987030982971191,
+      "learning_rate": 2.461233916199274e-07,
+      "loss": 0.9112,
+      "step": 2380
+    },
+    {
+      "epoch": 0.7632,
+      "grad_norm": 7.84393310546875,
+      "learning_rate": 2.444737710326625e-07,
+      "loss": 0.8584,
+      "step": 2385
+    },
+    {
+      "epoch": 0.7648,
+      "grad_norm": 3.063011646270752,
+      "learning_rate": 2.428241504453976e-07,
+      "loss": 0.8648,
+      "step": 2390
+    },
+    {
+      "epoch": 0.7664,
+      "grad_norm": 5.494943618774414,
+      "learning_rate": 2.411745298581326e-07,
+      "loss": 0.8633,
+      "step": 2395
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 3.209425449371338,
+      "learning_rate": 2.3952490927086767e-07,
+      "loss": 0.7618,
+      "step": 2400
+    },
+    {
+      "epoch": 0.7696,
+      "grad_norm": 3.163612127304077,
+      "learning_rate": 2.3787528868360277e-07,
+      "loss": 0.7093,
+      "step": 2405
+    },
+    {
+      "epoch": 0.7712,
+      "grad_norm": 4.022956848144531,
+      "learning_rate": 2.3622566809633785e-07,
+      "loss": 0.7705,
+      "step": 2410
+    },
+    {
+      "epoch": 0.7728,
+      "grad_norm": 3.748598575592041,
+      "learning_rate": 2.345760475090729e-07,
+      "loss": 0.942,
+      "step": 2415
+    },
+    {
+      "epoch": 0.7744,
+      "grad_norm": 3.5295920372009277,
+      "learning_rate": 2.3292642692180797e-07,
+      "loss": 0.8094,
+      "step": 2420
+    },
+    {
+      "epoch": 0.776,
+      "grad_norm": 5.767539978027344,
+      "learning_rate": 2.3127680633454305e-07,
+      "loss": 0.8425,
+      "step": 2425
+    },
+    {
+      "epoch": 0.7776,
+      "grad_norm": 3.4246678352355957,
+      "learning_rate": 2.2962718574727812e-07,
+      "loss": 0.7777,
+      "step": 2430
+    },
+    {
+      "epoch": 0.7792,
+      "grad_norm": 3.2887349128723145,
+      "learning_rate": 2.2797756516001317e-07,
+      "loss": 0.8842,
+      "step": 2435
+    },
+    {
+      "epoch": 0.7808,
+      "grad_norm": 4.332666397094727,
+      "learning_rate": 2.2632794457274827e-07,
+      "loss": 0.8946,
+      "step": 2440
+    },
+    {
+      "epoch": 0.7824,
+      "grad_norm": 3.6099178791046143,
+      "learning_rate": 2.2467832398548334e-07,
+      "loss": 0.929,
+      "step": 2445
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 2.9944612979888916,
+      "learning_rate": 2.230287033982184e-07,
+      "loss": 0.9407,
+      "step": 2450
+    },
+    {
+      "epoch": 0.7856,
+      "grad_norm": 4.5248613357543945,
+      "learning_rate": 2.2137908281095347e-07,
+      "loss": 0.9392,
+      "step": 2455
+    },
+    {
+      "epoch": 0.7872,
+      "grad_norm": 2.7737669944763184,
+      "learning_rate": 2.1972946222368854e-07,
+      "loss": 0.627,
+      "step": 2460
+    },
+    {
+      "epoch": 0.7888,
+      "grad_norm": 2.95540452003479,
+      "learning_rate": 2.1807984163642361e-07,
+      "loss": 0.7346,
+      "step": 2465
+    },
+    {
+      "epoch": 0.7904,
+      "grad_norm": 8.727768898010254,
+      "learning_rate": 2.1643022104915866e-07,
+      "loss": 0.9009,
+      "step": 2470
+    },
+    {
+      "epoch": 0.792,
+      "grad_norm": 4.173437118530273,
+      "learning_rate": 2.1478060046189376e-07,
+      "loss": 0.8968,
+      "step": 2475
+    },
+    {
+      "epoch": 0.7936,
+      "grad_norm": 3.4933767318725586,
+      "learning_rate": 2.1313097987462884e-07,
+      "loss": 0.868,
+      "step": 2480
+    },
+    {
+      "epoch": 0.7952,
+      "grad_norm": 3.33754301071167,
+      "learning_rate": 2.114813592873639e-07,
+      "loss": 0.8185,
+      "step": 2485
+    },
+    {
+      "epoch": 0.7968,
+      "grad_norm": 7.453437328338623,
+      "learning_rate": 2.0983173870009896e-07,
+      "loss": 0.8654,
+      "step": 2490
+    },
+    {
+      "epoch": 0.7984,
+      "grad_norm": 9.287111282348633,
+      "learning_rate": 2.0818211811283404e-07,
+      "loss": 0.7744,
+      "step": 2495
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.195357322692871,
+      "learning_rate": 2.065324975255691e-07,
+      "loss": 0.8295,
+      "step": 2500
+    },
+    {
+      "epoch": 0.8016,
+      "grad_norm": 4.878857612609863,
+      "learning_rate": 2.0488287693830418e-07,
+      "loss": 0.7917,
+      "step": 2505
+    },
+    {
+      "epoch": 0.8032,
+      "grad_norm": 3.239182710647583,
+      "learning_rate": 2.0323325635103923e-07,
+      "loss": 0.8339,
+      "step": 2510
+    },
+    {
+      "epoch": 0.8048,
+      "grad_norm": 4.196946144104004,
+      "learning_rate": 2.0158363576377433e-07,
+      "loss": 0.9609,
+      "step": 2515
+    },
+    {
+      "epoch": 0.8064,
+      "grad_norm": 2.7448809146881104,
+      "learning_rate": 1.999340151765094e-07,
+      "loss": 0.8516,
+      "step": 2520
+    },
+    {
+      "epoch": 0.808,
+      "grad_norm": 4.893881320953369,
+      "learning_rate": 1.9828439458924446e-07,
+      "loss": 0.7908,
+      "step": 2525
+    },
+    {
+      "epoch": 0.8096,
+      "grad_norm": 3.318279266357422,
+      "learning_rate": 1.9663477400197953e-07,
+      "loss": 0.7966,
+      "step": 2530
+    },
+    {
+      "epoch": 0.8112,
+      "grad_norm": 2.901827573776245,
+      "learning_rate": 1.949851534147146e-07,
+      "loss": 0.7831,
+      "step": 2535
+    },
+    {
+      "epoch": 0.8128,
+      "grad_norm": 5.1762847900390625,
+      "learning_rate": 1.9333553282744968e-07,
+      "loss": 0.9897,
+      "step": 2540
+    },
+    {
+      "epoch": 0.8144,
+      "grad_norm": 6.721929550170898,
+      "learning_rate": 1.9168591224018473e-07,
+      "loss": 0.8774,
+      "step": 2545
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 3.2159762382507324,
+      "learning_rate": 1.9003629165291983e-07,
+      "loss": 0.8543,
+      "step": 2550
+    },
+    {
+      "epoch": 0.8176,
+      "grad_norm": 2.2698450088500977,
+      "learning_rate": 1.883866710656549e-07,
+      "loss": 0.7665,
+      "step": 2555
+    },
+    {
+      "epoch": 0.8192,
+      "grad_norm": 6.745720386505127,
+      "learning_rate": 1.8673705047838998e-07,
+      "loss": 0.8497,
+      "step": 2560
+    },
+    {
+      "epoch": 0.8208,
+      "grad_norm": 7.653261661529541,
+      "learning_rate": 1.8508742989112503e-07,
+      "loss": 0.8156,
+      "step": 2565
+    },
+    {
+      "epoch": 0.8224,
+      "grad_norm": 5.070962905883789,
+      "learning_rate": 1.834378093038601e-07,
+      "loss": 0.8599,
+      "step": 2570
+    },
+    {
+      "epoch": 0.824,
+      "grad_norm": 3.101536750793457,
+      "learning_rate": 1.8178818871659517e-07,
+      "loss": 0.9103,
+      "step": 2575
+    },
+    {
+      "epoch": 0.8256,
+      "grad_norm": 5.420032024383545,
+      "learning_rate": 1.8013856812933025e-07,
+      "loss": 0.8812,
+      "step": 2580
+    },
+    {
+      "epoch": 0.8272,
+      "grad_norm": 6.0531697273254395,
+      "learning_rate": 1.7848894754206532e-07,
+      "loss": 0.9167,
+      "step": 2585
+    },
+    {
+      "epoch": 0.8288,
+      "grad_norm": 7.951639175415039,
+      "learning_rate": 1.768393269548004e-07,
+      "loss": 0.958,
+      "step": 2590
+    },
+    {
+      "epoch": 0.8304,
+      "grad_norm": 3.8448524475097656,
+      "learning_rate": 1.7518970636753547e-07,
+      "loss": 0.9107,
+      "step": 2595
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 6.432617664337158,
+      "learning_rate": 1.7354008578027052e-07,
+      "loss": 0.9313,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8336,
+      "grad_norm": 6.240530967712402,
+      "learning_rate": 1.718904651930056e-07,
+      "loss": 0.8152,
+      "step": 2605
+    },
+    {
+      "epoch": 0.8352,
+      "grad_norm": 3.8396613597869873,
+      "learning_rate": 1.7024084460574067e-07,
+      "loss": 0.8105,
+      "step": 2610
+    },
+    {
+      "epoch": 0.8368,
+      "grad_norm": 5.357729434967041,
+      "learning_rate": 1.6859122401847574e-07,
+      "loss": 0.8994,
+      "step": 2615
+    },
+    {
+      "epoch": 0.8384,
+      "grad_norm": 3.711209535598755,
+      "learning_rate": 1.669416034312108e-07,
+      "loss": 0.8206,
+      "step": 2620
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 8.217768669128418,
+      "learning_rate": 1.652919828439459e-07,
+      "loss": 0.8569,
+      "step": 2625
+    },
+    {
+      "epoch": 0.8416,
+      "grad_norm": 3.3683290481567383,
+      "learning_rate": 1.6364236225668097e-07,
+      "loss": 0.8385,
+      "step": 2630
+    },
+    {
+      "epoch": 0.8432,
+      "grad_norm": 5.228672981262207,
+      "learning_rate": 1.6199274166941604e-07,
+      "loss": 0.9177,
+      "step": 2635
+    },
+    {
+      "epoch": 0.8448,
+      "grad_norm": 4.824789524078369,
+      "learning_rate": 1.603431210821511e-07,
+      "loss": 0.8195,
+      "step": 2640
+    },
+    {
+      "epoch": 0.8464,
+      "grad_norm": 3.6725375652313232,
+      "learning_rate": 1.5869350049488616e-07,
+      "loss": 0.8017,
+      "step": 2645
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 3.130878210067749,
+      "learning_rate": 1.5704387990762124e-07,
+      "loss": 0.8088,
+      "step": 2650
+    },
+    {
+      "epoch": 0.8496,
+      "grad_norm": 4.449658393859863,
+      "learning_rate": 1.553942593203563e-07,
+      "loss": 0.7835,
+      "step": 2655
+    },
+    {
+      "epoch": 0.8512,
+      "grad_norm": 4.97245454788208,
+      "learning_rate": 1.537446387330914e-07,
+      "loss": 0.837,
+      "step": 2660
+    },
+    {
+      "epoch": 0.8528,
+      "grad_norm": 16.94793128967285,
+      "learning_rate": 1.5209501814582646e-07,
+      "loss": 0.8113,
+      "step": 2665
+    },
+    {
+      "epoch": 0.8544,
+      "grad_norm": 4.743756294250488,
+      "learning_rate": 1.5044539755856154e-07,
+      "loss": 0.8042,
+      "step": 2670
+    },
+    {
+      "epoch": 0.856,
+      "grad_norm": 10.094191551208496,
+      "learning_rate": 1.4879577697129658e-07,
+      "loss": 0.8571,
+      "step": 2675
+    },
+    {
+      "epoch": 0.8576,
+      "grad_norm": 5.925148010253906,
+      "learning_rate": 1.4714615638403166e-07,
+      "loss": 0.8749,
+      "step": 2680
+    },
+    {
+      "epoch": 0.8592,
+      "grad_norm": 3.691056251525879,
+      "learning_rate": 1.4549653579676673e-07,
+      "loss": 0.8824,
+      "step": 2685
+    },
+    {
+      "epoch": 0.8608,
+      "grad_norm": 3.58223295211792,
+      "learning_rate": 1.438469152095018e-07,
+      "loss": 0.8761,
+      "step": 2690
+    },
+    {
+      "epoch": 0.8624,
+      "grad_norm": 2.6448755264282227,
+      "learning_rate": 1.4219729462223686e-07,
+      "loss": 0.8868,
+      "step": 2695
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 3.1470277309417725,
+      "learning_rate": 1.4054767403497196e-07,
+      "loss": 0.738,
+      "step": 2700
+    },
+    {
+      "epoch": 0.8656,
+      "grad_norm": 2.670072317123413,
+      "learning_rate": 1.3889805344770703e-07,
+      "loss": 0.834,
+      "step": 2705
+    },
+    {
+      "epoch": 0.8672,
+      "grad_norm": 3.6198477745056152,
+      "learning_rate": 1.372484328604421e-07,
+      "loss": 0.866,
+      "step": 2710
+    },
+    {
+      "epoch": 0.8688,
+      "grad_norm": 3.4495441913604736,
+      "learning_rate": 1.3559881227317715e-07,
+      "loss": 0.8393,
+      "step": 2715
+    },
+    {
+      "epoch": 0.8704,
+      "grad_norm": 4.266736030578613,
+      "learning_rate": 1.3394919168591223e-07,
+      "loss": 0.6669,
+      "step": 2720
+    },
+    {
+      "epoch": 0.872,
+      "grad_norm": 7.86549711227417,
+      "learning_rate": 1.322995710986473e-07,
+      "loss": 0.8842,
+      "step": 2725
+    },
+    {
+      "epoch": 0.8736,
+      "grad_norm": 9.990023612976074,
+      "learning_rate": 1.3064995051138238e-07,
+      "loss": 0.8435,
+      "step": 2730
+    },
+    {
+      "epoch": 0.8752,
+      "grad_norm": 5.904709815979004,
+      "learning_rate": 1.2900032992411745e-07,
+      "loss": 0.8569,
+      "step": 2735
+    },
+    {
+      "epoch": 0.8768,
+      "grad_norm": 3.9784185886383057,
+      "learning_rate": 1.2735070933685253e-07,
+      "loss": 0.8613,
+      "step": 2740
+    },
+    {
+      "epoch": 0.8784,
+      "grad_norm": 3.423604965209961,
+      "learning_rate": 1.257010887495876e-07,
+      "loss": 0.8349,
+      "step": 2745
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 7.008749961853027,
+      "learning_rate": 1.2405146816232267e-07,
+      "loss": 0.739,
+      "step": 2750
+    },
+    {
+      "epoch": 0.8816,
+      "grad_norm": 4.69740104675293,
+      "learning_rate": 1.2240184757505772e-07,
+      "loss": 0.8429,
+      "step": 2755
+    },
+    {
+      "epoch": 0.8832,
+      "grad_norm": 11.424933433532715,
+      "learning_rate": 1.207522269877928e-07,
+      "loss": 0.8223,
+      "step": 2760
+    },
+    {
+      "epoch": 0.8848,
+      "grad_norm": 9.028188705444336,
+      "learning_rate": 1.1910260640052787e-07,
+      "loss": 0.8795,
+      "step": 2765
+    },
+    {
+      "epoch": 0.8864,
+      "grad_norm": 4.878159999847412,
+      "learning_rate": 1.1745298581326295e-07,
+      "loss": 0.8503,
+      "step": 2770
+    },
+    {
+      "epoch": 0.888,
+      "grad_norm": 4.5786237716674805,
+      "learning_rate": 1.1580336522599801e-07,
+      "loss": 0.7894,
+      "step": 2775
+    },
+    {
+      "epoch": 0.8896,
+      "grad_norm": 5.562559127807617,
+      "learning_rate": 1.141537446387331e-07,
+      "loss": 0.9339,
+      "step": 2780
+    },
+    {
+      "epoch": 0.8912,
+      "grad_norm": 5.035255432128906,
+      "learning_rate": 1.1250412405146816e-07,
+      "loss": 0.9041,
+      "step": 2785
+    },
+    {
+      "epoch": 0.8928,
+      "grad_norm": 3.3735504150390625,
+      "learning_rate": 1.1085450346420323e-07,
+      "loss": 0.8033,
+      "step": 2790
+    },
+    {
+      "epoch": 0.8944,
+      "grad_norm": 3.345130443572998,
+      "learning_rate": 1.0920488287693829e-07,
+      "loss": 0.7649,
+      "step": 2795
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 2.9101784229278564,
+      "learning_rate": 1.0755526228967338e-07,
+      "loss": 0.7702,
+      "step": 2800
+    },
+    {
+      "epoch": 0.8976,
+      "grad_norm": 8.602867126464844,
+      "learning_rate": 1.0590564170240844e-07,
+      "loss": 0.8769,
+      "step": 2805
+    },
+    {
+      "epoch": 0.8992,
+      "grad_norm": 3.7691822052001953,
+      "learning_rate": 1.042560211151435e-07,
+      "loss": 0.7437,
+      "step": 2810
+    },
+    {
+      "epoch": 0.9008,
+      "grad_norm": 3.548344612121582,
+      "learning_rate": 1.0260640052787859e-07,
+      "loss": 0.7927,
+      "step": 2815
+    },
+    {
+      "epoch": 0.9024,
+      "grad_norm": 6.130397319793701,
+      "learning_rate": 1.0095677994061365e-07,
+      "loss": 0.8964,
+      "step": 2820
+    },
+    {
+      "epoch": 0.904,
+      "grad_norm": 6.0678935050964355,
+      "learning_rate": 9.930715935334873e-08,
+      "loss": 0.8846,
+      "step": 2825
+    },
+    {
+      "epoch": 0.9056,
+      "grad_norm": 3.202853202819824,
+      "learning_rate": 9.765753876608379e-08,
+      "loss": 0.7872,
+      "step": 2830
+    },
+    {
+      "epoch": 0.9072,
+      "grad_norm": 7.076948165893555,
+      "learning_rate": 9.600791817881887e-08,
+      "loss": 0.841,
+      "step": 2835
+    },
+    {
+      "epoch": 0.9088,
+      "grad_norm": 4.527687072753906,
+      "learning_rate": 9.435829759155394e-08,
+      "loss": 0.8376,
+      "step": 2840
+    },
+    {
+      "epoch": 0.9104,
+      "grad_norm": 3.8655998706817627,
+      "learning_rate": 9.270867700428901e-08,
+      "loss": 0.9163,
+      "step": 2845
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 4.189991474151611,
+      "learning_rate": 9.105905641702407e-08,
+      "loss": 0.9171,
+      "step": 2850
+    },
+    {
+      "epoch": 0.9136,
+      "grad_norm": 3.6782002449035645,
+      "learning_rate": 8.940943582975916e-08,
+      "loss": 0.8107,
+      "step": 2855
+    },
+    {
+      "epoch": 0.9152,
+      "grad_norm": 3.8793959617614746,
+      "learning_rate": 8.775981524249422e-08,
+      "loss": 0.8324,
+      "step": 2860
+    },
+    {
+      "epoch": 0.9168,
+      "grad_norm": 6.360919952392578,
+      "learning_rate": 8.61101946552293e-08,
+      "loss": 0.901,
+      "step": 2865
+    },
+    {
+      "epoch": 0.9184,
+      "grad_norm": 4.218044757843018,
+      "learning_rate": 8.446057406796437e-08,
+      "loss": 0.7871,
+      "step": 2870
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.177008628845215,
+      "learning_rate": 8.281095348069944e-08,
+      "loss": 0.841,
+      "step": 2875
+    },
+    {
+      "epoch": 0.9216,
+      "grad_norm": 2.739051103591919,
+      "learning_rate": 8.11613328934345e-08,
+      "loss": 0.7872,
+      "step": 2880
+    },
+    {
+      "epoch": 0.9232,
+      "grad_norm": 8.707544326782227,
+      "learning_rate": 7.951171230616957e-08,
+      "loss": 0.9274,
+      "step": 2885
+    },
+    {
+      "epoch": 0.9248,
+      "grad_norm": 4.484316825866699,
+      "learning_rate": 7.786209171890465e-08,
+      "loss": 0.9121,
+      "step": 2890
+    },
+    {
+      "epoch": 0.9264,
+      "grad_norm": 3.211519479751587,
+      "learning_rate": 7.621247113163972e-08,
+      "loss": 0.8737,
+      "step": 2895
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 5.020310878753662,
+      "learning_rate": 7.456285054437479e-08,
+      "loss": 0.8358,
+      "step": 2900
+    },
+    {
+      "epoch": 0.9296,
+      "grad_norm": 5.312314510345459,
+      "learning_rate": 7.291322995710985e-08,
+      "loss": 0.831,
+      "step": 2905
+    },
+    {
+      "epoch": 0.9312,
+      "grad_norm": 2.917203903198242,
+      "learning_rate": 7.126360936984494e-08,
+      "loss": 0.8756,
+      "step": 2910
+    },
+    {
+      "epoch": 0.9328,
+      "grad_norm": 3.924370288848877,
+      "learning_rate": 6.961398878258e-08,
+      "loss": 0.7825,
+      "step": 2915
+    },
+    {
+      "epoch": 0.9344,
+      "grad_norm": 3.571991205215454,
+      "learning_rate": 6.796436819531507e-08,
+      "loss": 0.807,
+      "step": 2920
+    },
+    {
+      "epoch": 0.936,
+      "grad_norm": 5.816591739654541,
+      "learning_rate": 6.631474760805014e-08,
+      "loss": 0.8491,
+      "step": 2925
+    },
+    {
+      "epoch": 0.9376,
+      "grad_norm": 2.8010520935058594,
+      "learning_rate": 6.466512702078522e-08,
+      "loss": 0.9383,
+      "step": 2930
+    },
+    {
+      "epoch": 0.9392,
+      "grad_norm": 4.16404914855957,
+      "learning_rate": 6.301550643352028e-08,
+      "loss": 0.9048,
+      "step": 2935
+    },
+    {
+      "epoch": 0.9408,
+      "grad_norm": 3.1094634532928467,
+      "learning_rate": 6.136588584625536e-08,
+      "loss": 0.8829,
+      "step": 2940
+    },
+    {
+      "epoch": 0.9424,
+      "grad_norm": 6.206966400146484,
+      "learning_rate": 5.971626525899043e-08,
+      "loss": 0.8568,
+      "step": 2945
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 3.327371120452881,
+      "learning_rate": 5.80666446717255e-08,
+      "loss": 0.8895,
+      "step": 2950
+    },
+    {
+      "epoch": 0.9456,
+      "grad_norm": 3.068650722503662,
+      "learning_rate": 5.641702408446057e-08,
+      "loss": 0.8451,
+      "step": 2955
+    },
+    {
+      "epoch": 0.9472,
+      "grad_norm": 18.251916885375977,
+      "learning_rate": 5.4767403497195644e-08,
+      "loss": 0.8896,
+      "step": 2960
+    },
+    {
+      "epoch": 0.9488,
+      "grad_norm": 6.762292385101318,
+      "learning_rate": 5.311778290993071e-08,
+      "loss": 0.775,
+      "step": 2965
+    },
+    {
+      "epoch": 0.9504,
+      "grad_norm": 3.4393362998962402,
+      "learning_rate": 5.1468162322665786e-08,
+      "loss": 0.8701,
+      "step": 2970
+    },
+    {
+      "epoch": 0.952,
+      "grad_norm": 3.25495982170105,
+      "learning_rate": 4.9818541735400854e-08,
+      "loss": 0.9626,
+      "step": 2975
+    },
+    {
+      "epoch": 0.9536,
+      "grad_norm": 3.750603437423706,
+      "learning_rate": 4.816892114813593e-08,
+      "loss": 0.8311,
+      "step": 2980
+    },
+    {
+      "epoch": 0.9552,
+      "grad_norm": 8.03615951538086,
+      "learning_rate": 4.6519300560871e-08,
+      "loss": 0.8256,
+      "step": 2985
+    },
+    {
+      "epoch": 0.9568,
+      "grad_norm": 3.2003724575042725,
+      "learning_rate": 4.486967997360607e-08,
+      "loss": 0.8626,
+      "step": 2990
+    },
+    {
+      "epoch": 0.9584,
+      "grad_norm": 4.359886169433594,
+      "learning_rate": 4.3220059386341145e-08,
+      "loss": 0.8756,
+      "step": 2995
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.897675514221191,
+      "learning_rate": 4.1570438799076207e-08,
+      "loss": 0.8767,
+      "step": 3000
+    },
+    {
+      "epoch": 0.9616,
+      "grad_norm": 3.0032472610473633,
+      "learning_rate": 3.992081821181128e-08,
+      "loss": 0.9283,
+      "step": 3005
+    },
+    {
+      "epoch": 0.9632,
+      "grad_norm": 4.568953037261963,
+      "learning_rate": 3.827119762454635e-08,
+      "loss": 0.6707,
+      "step": 3010
+    },
+    {
+      "epoch": 0.9648,
+      "grad_norm": 6.175788879394531,
+      "learning_rate": 3.6621577037281423e-08,
+      "loss": 0.7988,
+      "step": 3015
+    },
+    {
+      "epoch": 0.9664,
+      "grad_norm": 6.2108354568481445,
+      "learning_rate": 3.497195645001649e-08,
+      "loss": 0.8649,
+      "step": 3020
+    },
+    {
+      "epoch": 0.968,
+      "grad_norm": 5.617148399353027,
+      "learning_rate": 3.3322335862751566e-08,
+      "loss": 0.8005,
+      "step": 3025
+    },
+    {
+      "epoch": 0.9696,
+      "grad_norm": 2.560255765914917,
+      "learning_rate": 3.1672715275486634e-08,
+      "loss": 0.8828,
+      "step": 3030
+    },
+    {
+      "epoch": 0.9712,
+      "grad_norm": 3.895402669906616,
+      "learning_rate": 3.002309468822171e-08,
+      "loss": 0.9119,
+      "step": 3035
+    },
+    {
+      "epoch": 0.9728,
+      "grad_norm": 3.2274169921875,
+      "learning_rate": 2.837347410095678e-08,
+      "loss": 0.9227,
+      "step": 3040
+    },
+    {
+      "epoch": 0.9744,
+      "grad_norm": 3.9759535789489746,
+      "learning_rate": 2.672385351369185e-08,
+      "loss": 0.7694,
+      "step": 3045
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 5.453105926513672,
+      "learning_rate": 2.507423292642692e-08,
+      "loss": 0.8129,
+      "step": 3050
+    },
+    {
+      "epoch": 0.9776,
+      "grad_norm": 3.8954710960388184,
+      "learning_rate": 2.342461233916199e-08,
+      "loss": 0.8646,
+      "step": 3055
+    },
+    {
+      "epoch": 0.9792,
+      "grad_norm": 5.158627033233643,
+      "learning_rate": 2.177499175189706e-08,
+      "loss": 0.873,
+      "step": 3060
+    },
+    {
+      "epoch": 0.9808,
+      "grad_norm": 3.405482769012451,
+      "learning_rate": 2.0125371164632132e-08,
+      "loss": 0.6943,
+      "step": 3065
+    },
+    {
+      "epoch": 0.9824,
+      "grad_norm": 4.411757469177246,
+      "learning_rate": 1.8475750577367203e-08,
+      "loss": 0.8178,
+      "step": 3070
+    },
+    {
+      "epoch": 0.984,
+      "grad_norm": 3.8500680923461914,
+      "learning_rate": 1.6826129990102277e-08,
+      "loss": 0.8075,
+      "step": 3075
+    },
+    {
+      "epoch": 0.9856,
+      "grad_norm": 9.081463813781738,
+      "learning_rate": 1.517650940283735e-08,
+      "loss": 0.8583,
+      "step": 3080
+    },
+    {
+      "epoch": 0.9872,
+      "grad_norm": 3.0819270610809326,
+      "learning_rate": 1.3526888815572416e-08,
+      "loss": 0.7818,
+      "step": 3085
+    },
+    {
+      "epoch": 0.9888,
+      "grad_norm": 12.406457901000977,
+      "learning_rate": 1.187726822830749e-08,
+      "loss": 0.9238,
+      "step": 3090
+    },
+    {
+      "epoch": 0.9904,
+      "grad_norm": 4.571506977081299,
+      "learning_rate": 1.022764764104256e-08,
+      "loss": 0.788,
+      "step": 3095
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 2.9352900981903076,
+      "learning_rate": 8.578027053777632e-09,
+      "loss": 0.9279,
+      "step": 3100
+    },
+    {
+      "epoch": 0.9936,
+      "grad_norm": 4.189651966094971,
+      "learning_rate": 6.928406466512702e-09,
+      "loss": 0.8379,
+      "step": 3105
+    },
+    {
+      "epoch": 0.9952,
+      "grad_norm": 3.8705317974090576,
+      "learning_rate": 5.278785879247773e-09,
+      "loss": 0.778,
+      "step": 3110
+    },
+    {
+      "epoch": 0.9968,
+      "grad_norm": 2.701988458633423,
+      "learning_rate": 3.629165291982844e-09,
+      "loss": 0.9363,
+      "step": 3115
+    },
+    {
+      "epoch": 0.9984,
+      "grad_norm": 3.2893664836883545,
+      "learning_rate": 1.9795447047179146e-09,
+      "loss": 0.7902,
+      "step": 3120
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.953529357910156,
+      "learning_rate": 3.299241174529858e-10,
+      "loss": 0.7962,
+      "step": 3125
+    },
+    {
+      "epoch": 1.0,
+      "step": 3125,
+      "total_flos": 1.0275244834155397e+18,
+      "train_loss": 0.8923269732666016,
+      "train_runtime": 7058.0642,
+      "train_samples_per_second": 7.084,
+      "train_steps_per_second": 0.443
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 3125,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0275244834155397e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}