diff --git "a/checkpoint-3500/trainer_state.json" "b/checkpoint-3500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3500/trainer_state.json"
@@ -0,0 +1,6158 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.692160611854685,
+  "eval_steps": 500,
+  "global_step": 3500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0076481835564053535,
+      "grad_norm": 0.291015625,
+      "learning_rate": 3.243243243243243e-05,
+      "loss": 1.3011,
+      "step": 4
+    },
+    {
+      "epoch": 0.015296367112810707,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.486486486486486e-05,
+      "loss": 1.3104,
+      "step": 8
+    },
+    {
+      "epoch": 0.022944550669216062,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 9.72972972972973e-05,
+      "loss": 1.2309,
+      "step": 12
+    },
+    {
+      "epoch": 0.030592734225621414,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00012972972972972972,
+      "loss": 1.2051,
+      "step": 16
+    },
+    {
+      "epoch": 0.03824091778202677,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00016216216216216215,
+      "loss": 1.1622,
+      "step": 20
+    },
+    {
+      "epoch": 0.045889101338432124,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0001945945945945946,
+      "loss": 1.1749,
+      "step": 24
+    },
+    {
+      "epoch": 0.05353728489483748,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00022702702702702703,
+      "loss": 1.149,
+      "step": 28
+    },
+    {
+      "epoch": 0.06118546845124283,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00025945945945945944,
+      "loss": 1.1455,
+      "step": 32
+    },
+    {
+      "epoch": 0.06883365200764818,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0002918918918918919,
+      "loss": 1.1358,
+      "step": 36
+    },
+    {
+      "epoch": 0.07648183556405354,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00029999949274434724,
+      "loss": 1.1201,
+      "step": 40
+    },
+    {
+      "epoch": 0.0841300191204589,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0002999972382816974,
+      "loss": 1.0549,
+      "step": 44
+    },
+    {
+      "epoch": 0.09177820267686425,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0002999931802773903,
+      "loss": 1.0946,
+      "step": 48
+    },
+    {
+      "epoch": 0.0994263862332696,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00029998731878021884,
+      "loss": 1.0841,
+      "step": 52
+    },
+    {
+      "epoch": 0.10707456978967496,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.00029997965386066057,
+      "loss": 1.0904,
+      "step": 56
+    },
+    {
+      "epoch": 0.1147227533460803,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0002999701856108772,
+      "loss": 1.0673,
+      "step": 60
+    },
+    {
+      "epoch": 0.12237093690248566,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00029995891414471334,
+      "loss": 1.0211,
+      "step": 64
+    },
+    {
+      "epoch": 0.13001912045889102,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0002999458395976953,
+      "loss": 1.0497,
+      "step": 68
+    },
+    {
+      "epoch": 0.13766730401529637,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0002999309621270293,
+      "loss": 1.0711,
+      "step": 72
+    },
+    {
+      "epoch": 0.14531548757170173,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.00029991428191159935,
+      "loss": 1.047,
+      "step": 76
+    },
+    {
+      "epoch": 0.15296367112810708,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.00029989579915196574,
+      "loss": 1.0442,
+      "step": 80
+    },
+    {
+      "epoch": 0.16061185468451242,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.000299875514070362,
+      "loss": 1.0197,
+      "step": 84
+    },
+    {
+      "epoch": 0.1682600382409178,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.00029985342691069255,
+      "loss": 1.0185,
+      "step": 88
+    },
+    {
+      "epoch": 0.17590822179732313,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0002998295379385297,
+      "loss": 1.0168,
+      "step": 92
+    },
+    {
+      "epoch": 0.1835564053537285,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00029980384744111047,
+      "loss": 1.0211,
+      "step": 96
+    },
+    {
+      "epoch": 0.19120458891013384,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0002997763557273331,
+      "loss": 1.0178,
+      "step": 100
+    },
+    {
+      "epoch": 0.1988527724665392,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0002997470631277533,
+      "loss": 0.9871,
+      "step": 104
+    },
+    {
+      "epoch": 0.20650095602294455,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0002997159699945804,
+      "loss": 1.0197,
+      "step": 108
+    },
+    {
+      "epoch": 0.21414913957934992,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002996830767016731,
+      "loss": 1.0,
+      "step": 112
+    },
+    {
+      "epoch": 0.22179732313575526,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0002996483836445347,
+      "loss": 0.9885,
+      "step": 116
+    },
+    {
+      "epoch": 0.2294455066921606,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00029961189124030885,
+      "loss": 0.9664,
+      "step": 120
+    },
+    {
+      "epoch": 0.23709369024856597,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00029957359992777404,
+      "loss": 0.9831,
+      "step": 124
+    },
+    {
+      "epoch": 0.2447418738049713,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00029953351016733854,
+      "loss": 0.9718,
+      "step": 128
+    },
+    {
+      "epoch": 0.25239005736137665,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.000299491622441035,
+      "loss": 0.9592,
+      "step": 132
+    },
+    {
+      "epoch": 0.26003824091778205,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00029944793725251436,
+      "loss": 0.9514,
+      "step": 136
+    },
+    {
+      "epoch": 0.2676864244741874,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.00029940245512704005,
+      "loss": 0.9435,
+      "step": 140
+    },
+    {
+      "epoch": 0.27533460803059273,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0002993551766114815,
+      "loss": 0.9545,
+      "step": 144
+    },
+    {
+      "epoch": 0.2829827915869981,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00029930610227430767,
+      "loss": 0.9378,
+      "step": 148
+    },
+    {
+      "epoch": 0.29063097514340347,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0002992552327055802,
+      "loss": 0.9085,
+      "step": 152
+    },
+    {
+      "epoch": 0.2982791586998088,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002992025685169461,
+      "loss": 0.9482,
+      "step": 156
+    },
+    {
+      "epoch": 0.30592734225621415,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.00029914811034163096,
+      "loss": 0.949,
+      "step": 160
+    },
+    {
+      "epoch": 0.3135755258126195,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00029909185883443063,
+      "loss": 0.9796,
+      "step": 164
+    },
+    {
+      "epoch": 0.32122370936902483,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.0002990338146717039,
+      "loss": 0.9351,
+      "step": 168
+    },
+    {
+      "epoch": 0.32887189292543023,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0002989739785513639,
+      "loss": 0.9166,
+      "step": 172
+    },
+    {
+      "epoch": 0.3365200764818356,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0002989123511928703,
+      "loss": 0.9204,
+      "step": 176
+    },
+    {
+      "epoch": 0.3441682600382409,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.00029884893333722,
+      "loss": 0.9356,
+      "step": 180
+    },
+    {
+      "epoch": 0.35181644359464626,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0002987837257469387,
+      "loss": 0.9342,
+      "step": 184
+    },
+    {
+      "epoch": 0.35946462715105165,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00029871672920607153,
+      "loss": 0.9026,
+      "step": 188
+    },
+    {
+      "epoch": 0.367112810707457,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0002986479445201737,
+      "loss": 0.8983,
+      "step": 192
+    },
+    {
+      "epoch": 0.37476099426386233,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002985773725163008,
+      "loss": 0.922,
+      "step": 196
+    },
+    {
+      "epoch": 0.3824091778202677,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002985050140429986,
+      "loss": 0.9099,
+      "step": 200
+    },
+    {
+      "epoch": 0.390057361376673,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002984308699702935,
+      "loss": 0.8825,
+      "step": 204
+    },
+    {
+      "epoch": 0.3977055449330784,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0002983549411896812,
+      "loss": 0.893,
+      "step": 208
+    },
+    {
+      "epoch": 0.40535372848948376,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002982772286141167,
+      "loss": 0.9068,
+      "step": 212
+    },
+    {
+      "epoch": 0.4130019120458891,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.000298197733178003,
+      "loss": 0.8536,
+      "step": 216
+    },
+    {
+      "epoch": 0.42065009560229444,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.00029811645583717987,
+      "loss": 0.8843,
+      "step": 220
+    },
+    {
+      "epoch": 0.42829827915869984,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00029803339756891254,
+      "loss": 0.8627,
+      "step": 224
+    },
+    {
+      "epoch": 0.4359464627151052,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00029794855937187963,
+      "loss": 0.8572,
+      "step": 228
+    },
+    {
+      "epoch": 0.4435946462715105,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0002978619422661613,
+      "loss": 0.8255,
+      "step": 232
+    },
+    {
+      "epoch": 0.45124282982791586,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0002977735472932273,
+      "loss": 0.8274,
+      "step": 236
+    },
+    {
+      "epoch": 0.4588910133843212,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.00029768337551592394,
+      "loss": 0.8569,
+      "step": 240
+    },
+    {
+      "epoch": 0.4665391969407266,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.00029759142801846143,
+      "loss": 0.8791,
+      "step": 244
+    },
+    {
+      "epoch": 0.47418738049713194,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00029749770590640123,
+      "loss": 0.8215,
+      "step": 248
+    },
+    {
+      "epoch": 0.4818355640535373,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00029740221030664216,
+      "loss": 0.8546,
+      "step": 252
+    },
+    {
+      "epoch": 0.4894837476099426,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00029730494236740744,
+      "loss": 0.8518,
+      "step": 256
+    },
+    {
+      "epoch": 0.497131931166348,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002972059032582304,
+      "loss": 0.8115,
+      "step": 260
+    },
+    {
+      "epoch": 0.5047801147227533,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002971050941699407,
+      "loss": 0.8818,
+      "step": 264
+    },
+    {
+      "epoch": 0.5124282982791587,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00029700251631464993,
+      "loss": 0.8834,
+      "step": 268
+    },
+    {
+      "epoch": 0.5200764818355641,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.000296898170925737,
+      "loss": 0.8502,
+      "step": 272
+    },
+    {
+      "epoch": 0.5277246653919694,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0002967920592578335,
+      "loss": 0.854,
+      "step": 276
+    },
+    {
+      "epoch": 0.5353728489483748,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002966841825868082,
+      "loss": 0.8164,
+      "step": 280
+    },
+    {
+      "epoch": 0.5430210325047801,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.00029657454220975216,
+      "loss": 0.9058,
+      "step": 284
+    },
+    {
+      "epoch": 0.5506692160611855,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.00029646313944496297,
+      "loss": 0.8646,
+      "step": 288
+    },
+    {
+      "epoch": 0.5583173996175909,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.00029634997563192866,
+      "loss": 0.8536,
+      "step": 292
+    },
+    {
+      "epoch": 0.5659655831739961,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002962350521313122,
+      "loss": 0.8532,
+      "step": 296
+    },
+    {
+      "epoch": 0.5736137667304015,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0002961183703249342,
+      "loss": 0.8228,
+      "step": 300
+    },
+    {
+      "epoch": 0.5812619502868069,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002959999316157573,
+      "loss": 0.8088,
+      "step": 304
+    },
+    {
+      "epoch": 0.5889101338432122,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00029587973742786875,
+      "loss": 0.8291,
+      "step": 308
+    },
+    {
+      "epoch": 0.5965583173996176,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002957577892064632,
+      "loss": 0.8087,
+      "step": 312
+    },
+    {
+      "epoch": 0.6042065009560229,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.00029563408841782576,
+      "loss": 0.853,
+      "step": 316
+    },
+    {
+      "epoch": 0.6118546845124283,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00029550863654931385,
+      "loss": 0.847,
+      "step": 320
+    },
+    {
+      "epoch": 0.6195028680688337,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002953814351093398,
+      "loss": 0.8087,
+      "step": 324
+    },
+    {
+      "epoch": 0.627151051625239,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0002952524856273524,
+      "loss": 0.8514,
+      "step": 328
+    },
+    {
+      "epoch": 0.6347992351816444,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.00029512178965381854,
+      "loss": 0.8501,
+      "step": 332
+    },
+    {
+      "epoch": 0.6424474187380497,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.00029498934876020475,
+      "loss": 0.8029,
+      "step": 336
+    },
+    {
+      "epoch": 0.6500956022944551,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.00029485516453895826,
+      "loss": 0.8293,
+      "step": 340
+    },
+    {
+      "epoch": 0.6577437858508605,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002947192386034874,
+      "loss": 0.8695,
+      "step": 344
+    },
+    {
+      "epoch": 0.6653919694072657,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00029458157258814316,
+      "loss": 0.8249,
+      "step": 348
+    },
+    {
+      "epoch": 0.6730401529636711,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.00029444216814819834,
+      "loss": 0.8009,
+      "step": 352
+    },
+    {
+      "epoch": 0.6806883365200764,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.00029430102695982875,
+      "loss": 0.8642,
+      "step": 356
+    },
+    {
+      "epoch": 0.6883365200764818,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.00029415815072009237,
+      "loss": 0.8562,
+      "step": 360
+    },
+    {
+      "epoch": 0.6959847036328872,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00029401354114690905,
+      "loss": 0.8274,
+      "step": 364
+    },
+    {
+      "epoch": 0.7036328871892925,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0002938671999790402,
+      "loss": 0.8214,
+      "step": 368
+    },
+    {
+      "epoch": 0.7112810707456979,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00029371912897606736,
+      "loss": 0.8537,
+      "step": 372
+    },
+    {
+      "epoch": 0.7189292543021033,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.00029356932991837163,
+      "loss": 0.8378,
+      "step": 376
+    },
+    {
+      "epoch": 0.7265774378585086,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002934178046071116,
+      "loss": 0.8064,
+      "step": 380
+    },
+    {
+      "epoch": 0.734225621414914,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002932645548642024,
+      "loss": 0.8427,
+      "step": 384
+    },
+    {
+      "epoch": 0.7418738049713193,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002931095825322931,
+      "loss": 0.7602,
+      "step": 388
+    },
+    {
+      "epoch": 0.7495219885277247,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.00029295288947474513,
+      "loss": 0.881,
+      "step": 392
+    },
+    {
+      "epoch": 0.7571701720841301,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002927944775756095,
+      "loss": 0.865,
+      "step": 396
+    },
+    {
+      "epoch": 0.7648183556405354,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002926343487396044,
+      "loss": 0.839,
+      "step": 400
+    },
+    {
+      "epoch": 0.7724665391969407,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00029247250489209217,
+      "loss": 0.7939,
+      "step": 404
+    },
+    {
+      "epoch": 0.780114722753346,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00029230894797905595,
+      "loss": 0.7748,
+      "step": 408
+    },
+    {
+      "epoch": 0.7877629063097514,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00029214367996707676,
+      "loss": 0.7829,
+      "step": 412
+    },
+    {
+      "epoch": 0.7954110898661568,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00029197670284330954,
+      "loss": 0.7867,
+      "step": 416
+    },
+    {
+      "epoch": 0.8030592734225621,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.00029180801861545906,
+      "loss": 0.7971,
+      "step": 420
+    },
+    {
+      "epoch": 0.8107074569789675,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002916376293117564,
+      "loss": 0.8241,
+      "step": 424
+    },
+    {
+      "epoch": 0.8183556405353728,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00029146553698093387,
+      "loss": 0.8119,
+      "step": 428
+    },
+    {
+      "epoch": 0.8260038240917782,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00029129174369220087,
+      "loss": 0.8048,
+      "step": 432
+    },
+    {
+      "epoch": 0.8336520076481836,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.00029111625153521877,
+      "loss": 0.76,
+      "step": 436
+    },
+    {
+      "epoch": 0.8413001912045889,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00029093906262007583,
+      "loss": 0.7833,
+      "step": 440
+    },
+    {
+      "epoch": 0.8489483747609943,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.00029076017907726196,
+      "loss": 0.8027,
+      "step": 444
+    },
+    {
+      "epoch": 0.8565965583173997,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0002905796030576428,
+      "loss": 0.818,
+      "step": 448
+    },
+    {
+      "epoch": 0.864244741873805,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.00029039733673243416,
+      "loss": 0.8358,
+      "step": 452
+    },
+    {
+      "epoch": 0.8718929254302104,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0002902133822931759,
+      "loss": 0.7543,
+      "step": 456
+    },
+    {
+      "epoch": 0.8795411089866156,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00029002774195170525,
+      "loss": 0.7765,
+      "step": 460
+    },
+    {
+      "epoch": 0.887189292543021,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002898404179401306,
+      "loss": 0.8094,
+      "step": 464
+    },
+    {
+      "epoch": 0.8948374760994264,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0002896514125108045,
+      "loss": 0.7657,
+      "step": 468
+    },
+    {
+      "epoch": 0.9024856596558317,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002894607279362966,
+      "loss": 0.7774,
+      "step": 472
+    },
+    {
+      "epoch": 0.9101338432122371,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0002892683665093662,
+      "loss": 0.8148,
+      "step": 476
+    },
+    {
+      "epoch": 0.9177820267686424,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0002890743305429348,
+      "loss": 0.7882,
+      "step": 480
+    },
+    {
+      "epoch": 0.9254302103250478,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.0002888786223700585,
+      "loss": 0.7656,
+      "step": 484
+    },
+    {
+      "epoch": 0.9330783938814532,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.00028868124434389944,
+      "loss": 0.7802,
+      "step": 488
+    },
+    {
+      "epoch": 0.9407265774378585,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00028848219883769805,
+      "loss": 0.7773,
+      "step": 492
+    },
+    {
+      "epoch": 0.9483747609942639,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.000288281488244744,
+      "loss": 0.7803,
+      "step": 496
+    },
+    {
+      "epoch": 0.9560229445506692,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.000288079114978348,
+      "loss": 0.8056,
+      "step": 500
+    },
+    {
+      "epoch": 0.9636711281070746,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002878750814718121,
+      "loss": 0.8309,
+      "step": 504
+    },
+    {
+      "epoch": 0.97131931166348,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.00028766939017840114,
+      "loss": 0.7737,
+      "step": 508
+    },
+    {
+      "epoch": 0.9789674952198852,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00028746204357131273,
+      "loss": 0.8039,
+      "step": 512
+    },
+    {
+      "epoch": 0.9866156787762906,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0002872530441436477,
+      "loss": 0.7341,
+      "step": 516
+    },
+    {
+      "epoch": 0.994263862332696,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0002870423944083801,
+      "loss": 0.8122,
+      "step": 520
+    },
+    {
+      "epoch": 1.0019120458891013,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0002868300968983271,
+      "loss": 0.7403,
+      "step": 524
+    },
+    {
+      "epoch": 1.0095602294455066,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0002866161541661185,
+      "loss": 0.697,
+      "step": 528
+    },
+    {
+      "epoch": 1.0172084130019121,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0002864005687841656,
+      "loss": 0.7442,
+      "step": 532
+    },
+    {
+      "epoch": 1.0248565965583174,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002861833433446312,
+      "loss": 0.6853,
+      "step": 536
+    },
+    {
+      "epoch": 1.0325047801147227,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.00028596448045939735,
+      "loss": 0.7398,
+      "step": 540
+    },
+    {
+      "epoch": 1.0401529636711282,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002857439827600348,
+      "loss": 0.7912,
+      "step": 544
+    },
+    {
+      "epoch": 1.0478011472275335,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002855218528977709,
+      "loss": 0.7138,
+      "step": 548
+    },
+    {
+      "epoch": 1.0554493307839388,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00028529809354345794,
+      "loss": 0.726,
+      "step": 552
+    },
+    {
+      "epoch": 1.063097514340344,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002850727073875409,
+      "loss": 0.7058,
+      "step": 556
+    },
+    {
+      "epoch": 1.0707456978967496,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.00028484569714002517,
+      "loss": 0.7102,
+      "step": 560
+    },
+    {
+      "epoch": 1.0783938814531548,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002846170655304438,
+      "loss": 0.6534,
+      "step": 564
+    },
+    {
+      "epoch": 1.0860420650095601,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0002843868153078251,
+      "loss": 0.6918,
+      "step": 568
+    },
+    {
+      "epoch": 1.0936902485659656,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.000284154949240659,
+      "loss": 0.6673,
+      "step": 572
+    },
+    {
+      "epoch": 1.101338432122371,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002839214701168644,
+      "loss": 0.6722,
+      "step": 576
+    },
+    {
+      "epoch": 1.1089866156787762,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00028368638074375516,
+      "loss": 0.7141,
+      "step": 580
+    },
+    {
+      "epoch": 1.1166347992351817,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002834496839480063,
+      "loss": 0.665,
+      "step": 584
+    },
+    {
+      "epoch": 1.124282982791587,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00028321138257562066,
+      "loss": 0.6886,
+      "step": 588
+    },
+    {
+      "epoch": 1.1319311663479923,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00028297147949189386,
+      "loss": 0.6997,
+      "step": 592
+    },
+    {
+      "epoch": 1.1395793499043978,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00028272997758138044,
+      "loss": 0.7051,
+      "step": 596
+    },
+    {
+      "epoch": 1.147227533460803,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00028248687974785896,
+      "loss": 0.7188,
+      "step": 600
+    },
+    {
+      "epoch": 1.1548757170172084,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002822421889142969,
+      "loss": 0.6757,
+      "step": 604
+    },
+    {
+      "epoch": 1.1625239005736137,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00028199590802281595,
+      "loss": 0.7203,
+      "step": 608
+    },
+    {
+      "epoch": 1.1701720841300192,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00028174804003465616,
+      "loss": 0.6943,
+      "step": 612
+    },
+    {
+      "epoch": 1.1778202676864244,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0002814985879301408,
+      "loss": 0.7037,
+      "step": 616
+    },
+    {
+      "epoch": 1.1854684512428297,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002812475547086401,
+      "loss": 0.7215,
+      "step": 620
+    },
+    {
+      "epoch": 1.1931166347992352,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00028099494338853554,
+      "loss": 0.6863,
+      "step": 624
+    },
+    {
+      "epoch": 1.2007648183556405,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002807407570071832,
+      "loss": 0.7432,
+      "step": 628
+    },
+    {
+      "epoch": 1.2084130019120458,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00028048499862087757,
+      "loss": 0.7265,
+      "step": 632
+    },
+    {
+      "epoch": 1.2160611854684513,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00028022767130481466,
+      "loss": 0.6848,
+      "step": 636
+    },
+    {
+      "epoch": 1.2237093690248566,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0002799687781530549,
+      "loss": 0.717,
+      "step": 640
+    },
+    {
+      "epoch": 1.231357552581262,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00027970832227848627,
+      "loss": 0.7011,
+      "step": 644
+    },
+    {
+      "epoch": 1.2390057361376674,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002794463068127866,
+      "loss": 0.7696,
+      "step": 648
+    },
+    {
+      "epoch": 1.2466539196940727,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.00027918273490638574,
+      "loss": 0.6922,
+      "step": 652
+    },
+    {
+      "epoch": 1.254302103250478,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0002789176097284283,
+      "loss": 0.6521,
+      "step": 656
+    },
+    {
+      "epoch": 1.2619502868068833,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002786509344667349,
+      "loss": 0.6642,
+      "step": 660
+    },
+    {
+      "epoch": 1.2695984703632888,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002783827123277643,
+      "loss": 0.7773,
+      "step": 664
+    },
+    {
+      "epoch": 1.277246653919694,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.00027811294653657444,
+      "loss": 0.7314,
+      "step": 668
+    },
+    {
+      "epoch": 1.2848948374760996,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.000277841640336784,
+      "loss": 0.7461,
+      "step": 672
+    },
+    {
+      "epoch": 1.2925430210325048,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00027756879699053337,
+      "loss": 0.7426,
+      "step": 676
+    },
+    {
+      "epoch": 1.3001912045889101,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002772944197784451,
+      "loss": 0.6986,
+      "step": 680
+    },
+    {
+      "epoch": 1.3078393881453154,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0002770185119995849,
+      "loss": 0.7379,
+      "step": 684
+    },
+    {
+      "epoch": 1.3154875717017207,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0002767410769714216,
+      "loss": 0.7146,
+      "step": 688
+    },
+    {
+      "epoch": 1.3231357552581262,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002764621180297875,
+      "loss": 0.7061,
+      "step": 692
+    },
+    {
+      "epoch": 1.3307839388145315,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002761816385288382,
+      "loss": 0.6547,
+      "step": 696
+    },
+    {
+      "epoch": 1.338432122370937,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002758996418410122,
+      "loss": 0.7018,
+      "step": 700
+    },
+    {
+      "epoch": 1.3460803059273423,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002756161313569904,
+      "loss": 0.7062,
+      "step": 704
+    },
+    {
+      "epoch": 1.3537284894837476,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00027533111048565537,
+      "loss": 0.778,
+      "step": 708
+    },
+    {
+      "epoch": 1.3613766730401529,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00027504458265405034,
+      "loss": 0.6916,
+      "step": 712
+    },
+    {
+      "epoch": 1.3690248565965584,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00027475655130733786,
+      "loss": 0.6862,
+      "step": 716
+    },
+    {
+      "epoch": 1.3766730401529637,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00027446701990875864,
+      "loss": 0.7037,
+      "step": 720
+    },
+    {
+      "epoch": 1.384321223709369,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.00027417599193958964,
+      "loss": 0.6976,
+      "step": 724
+    },
+    {
+      "epoch": 1.3919694072657744,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00027388347089910253,
+      "loss": 0.7209,
+      "step": 728
+    },
+    {
+      "epoch": 1.3996175908221797,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002735894603045211,
+      "loss": 0.7009,
+      "step": 732
+    },
+    {
+      "epoch": 1.407265774378585,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002732939636909796,
+      "loss": 0.6583,
+      "step": 736
+    },
+    {
+      "epoch": 1.4149139579349903,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00027299698461147966,
+      "loss": 0.6999,
+      "step": 740
+    },
+    {
+      "epoch": 1.4225621414913958,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002726985266368481,
+      "loss": 0.7269,
+      "step": 744
+    },
+    {
+      "epoch": 1.430210325047801,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002723985933556936,
+      "loss": 0.6256,
+      "step": 748
+    },
+    {
+      "epoch": 1.4378585086042066,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00027209718837436353,
+      "loss": 0.7129,
+      "step": 752
+    },
+    {
+      "epoch": 1.445506692160612,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.000271794315316901,
+      "loss": 0.6623,
+      "step": 756
+    },
+    {
+      "epoch": 1.4531548757170172,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00027148997782500085,
+      "loss": 0.6869,
+      "step": 760
+    },
+    {
+      "epoch": 1.4608030592734225,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002711841795579661,
+      "loss": 0.7426,
+      "step": 764
+    },
+    {
+      "epoch": 1.468451242829828,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00027087692419266383,
+      "loss": 0.6731,
+      "step": 768
+    },
+    {
+      "epoch": 1.4760994263862333,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00027056821542348114,
+      "loss": 0.7591,
+      "step": 772
+    },
+    {
+      "epoch": 1.4837476099426385,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002702580569622805,
+      "loss": 0.7129,
+      "step": 776
+    },
+    {
+      "epoch": 1.491395793499044,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002699464525383552,
+      "loss": 0.7307,
+      "step": 780
+    },
+    {
+      "epoch": 1.4990439770554493,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002696334058983848,
+      "loss": 0.7317,
+      "step": 784
+    },
+    {
+      "epoch": 1.5066921606118546,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002693189208063894,
+      "loss": 0.6994,
+      "step": 788
+    },
+    {
+      "epoch": 1.51434034416826,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00026900300104368524,
+      "loss": 0.72,
+      "step": 792
+    },
+    {
+      "epoch": 1.5219885277246654,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002686856504088385,
+      "loss": 0.7112,
+      "step": 796
+    },
+    {
+      "epoch": 1.5296367112810707,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.00026836687271762015,
+      "loss": 0.6912,
+      "step": 800
+    },
+    {
+      "epoch": 1.5372848948374762,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0002680466718029596,
+      "loss": 0.6801,
+      "step": 804
+    },
+    {
+      "epoch": 1.5449330783938815,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00026772505151489897,
+      "loss": 0.7077,
+      "step": 808
+    },
+    {
+      "epoch": 1.5525812619502868,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00026740201572054685,
+      "loss": 0.6926,
+      "step": 812
+    },
+    {
+      "epoch": 1.560229445506692,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00026707756830403144,
+      "loss": 0.702,
+      "step": 816
+    },
+    {
+      "epoch": 1.5678776290630974,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00026675171316645403,
+      "loss": 0.7178,
+      "step": 820
+    },
+    {
+      "epoch": 1.5755258126195029,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00026642445422584224,
+      "loss": 0.6843,
+      "step": 824
+    },
+    {
+      "epoch": 1.5831739961759084,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002660957954171028,
+      "loss": 0.6722,
+      "step": 828
+    },
+    {
+      "epoch": 1.5908221797323137,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00026576574069197406,
+      "loss": 0.6518,
+      "step": 832
+    },
+    {
+      "epoch": 1.598470363288719,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00026543429401897875,
+      "loss": 0.6998,
+      "step": 836
+    },
+    {
+      "epoch": 1.6061185468451242,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002651014593833762,
+      "loss": 0.6966,
+      "step": 840
+    },
+    {
+      "epoch": 1.6137667304015295,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00026476724078711416,
+      "loss": 0.7054,
+      "step": 844
+    },
+    {
+      "epoch": 1.621414913957935,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00026443164224878115,
+      "loss": 0.6655,
+      "step": 848
+    },
+    {
+      "epoch": 1.6290630975143403,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002640946678035576,
+      "loss": 0.7098,
+      "step": 852
+    },
+    {
+      "epoch": 1.6367112810707458,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002637563215031679,
+      "loss": 0.7099,
+      "step": 856
+    },
+    {
+      "epoch": 1.644359464627151,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00026341660741583127,
+      "loss": 0.6704,
+      "step": 860
+    },
+    {
+      "epoch": 1.6520076481835564,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00026307552962621293,
+      "loss": 0.686,
+      "step": 864
+    },
+    {
+      "epoch": 1.6596558317399617,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00026273309223537507,
+      "loss": 0.69,
+      "step": 868
+    },
+    {
+      "epoch": 1.667304015296367,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002623892993607275,
+      "loss": 0.6957,
+      "step": 872
+    },
+    {
+      "epoch": 1.6749521988527725,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00026204415513597813,
+      "loss": 0.7071,
+      "step": 876
+    },
+    {
+      "epoch": 1.682600382409178,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002616976637110832,
+      "loss": 0.6313,
+      "step": 880
+    },
+    {
+      "epoch": 1.6902485659655833,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002613498292521977,
+      "loss": 0.6809,
+      "step": 884
+    },
+    {
+      "epoch": 1.6978967495219885,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00026100065594162475,
+      "loss": 0.6867,
+      "step": 888
+    },
+    {
+      "epoch": 1.7055449330783938,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00026065014797776575,
+      "loss": 0.7065,
+      "step": 892
+    },
+    {
+      "epoch": 1.7131931166347991,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002602983095750698,
+      "loss": 0.6938,
+      "step": 896
+    },
+    {
+      "epoch": 1.7208413001912046,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002599451449639828,
+      "loss": 0.7138,
+      "step": 900
+    },
+    {
+      "epoch": 1.72848948374761,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00025959065839089684,
+      "loss": 0.6976,
+      "step": 904
+    },
+    {
+      "epoch": 1.7361376673040154,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.00025923485411809917,
+      "loss": 0.6792,
+      "step": 908
+    },
+    {
+      "epoch": 1.7437858508604207,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00025887773642372064,
+      "loss": 0.7016,
+      "step": 912
+    },
+    {
+      "epoch": 1.751434034416826,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00025851930960168464,
+      "loss": 0.6845,
+      "step": 916
+    },
+    {
+      "epoch": 1.7590822179732313,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002581595779616552,
+      "loss": 0.6932,
+      "step": 920
+    },
+    {
+      "epoch": 1.7667304015296366,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002577985458289852,
+      "loss": 0.6911,
+      "step": 924
+    },
+    {
+      "epoch": 1.774378585086042,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00025743621754466457,
+      "loss": 0.6611,
+      "step": 928
+    },
+    {
+      "epoch": 1.7820267686424476,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002570725974652679,
+      "loss": 0.7158,
+      "step": 932
+    },
+    {
+      "epoch": 1.7896749521988529,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002567076899629021,
+      "loss": 0.7176,
+      "step": 936
+    },
+    {
+      "epoch": 1.7973231357552581,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002563414994251538,
+      "loss": 0.7001,
+      "step": 940
+    },
+    {
+      "epoch": 1.8049713193116634,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002559740302550366,
+      "loss": 0.7583,
+      "step": 944
+    },
+    {
+      "epoch": 1.8126195028680687,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002556052868709383,
+      "loss": 0.6374,
+      "step": 948
+    },
+    {
+      "epoch": 1.8202676864244742,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00025523527370656753,
+      "loss": 0.6771,
+      "step": 952
+    },
+    {
+      "epoch": 1.8279158699808795,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002548639952109006,
+      "loss": 0.6547,
+      "step": 956
+    },
+    {
+      "epoch": 1.835564053537285,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002544914558481279,
+      "loss": 0.7078,
+      "step": 960
+    },
+    {
+      "epoch": 1.8432122370936903,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.00025411766009760027,
+      "loss": 0.6913,
+      "step": 964
+    },
+    {
+      "epoch": 1.8508604206500956,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00025374261245377525,
+      "loss": 0.725,
+      "step": 968
+    },
+    {
+      "epoch": 1.8585086042065009,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002533663174261628,
+      "loss": 0.6871,
+      "step": 972
+    },
+    {
+      "epoch": 1.8661567877629062,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0002529887795392713,
+      "loss": 0.6748,
+      "step": 976
+    },
+    {
+      "epoch": 1.8738049713193117,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00025261000333255305,
+      "loss": 0.6224,
+      "step": 980
+    },
+    {
+      "epoch": 1.8814531548757172,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002522299933603497,
+      "loss": 0.6914,
+      "step": 984
+    },
+    {
+      "epoch": 1.8891013384321225,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002518487541918374,
+      "loss": 0.684,
+      "step": 988
+    },
+    {
+      "epoch": 1.8967495219885278,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.000251466290410972,
+      "loss": 0.7131,
+      "step": 992
+    },
+    {
+      "epoch": 1.904397705544933,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002510826066164341,
+      "loss": 0.6956,
+      "step": 996
+    },
+    {
+      "epoch": 1.9120458891013383,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00025069770742157317,
+      "loss": 0.6683,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9196940726577438,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00025031159745435267,
+      "loss": 0.6941,
+      "step": 1004
+    },
+    {
+      "epoch": 1.9273422562141491,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0002499242813572942,
+      "loss": 0.7458,
+      "step": 1008
+    },
+    {
+      "epoch": 1.9349904397705546,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002495357637874215,
+      "loss": 0.6724,
+      "step": 1012
+    },
+    {
+      "epoch": 1.94263862332696,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002491460494162048,
+      "loss": 0.6662,
+      "step": 1016
+    },
+    {
+      "epoch": 1.9502868068833652,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00024875514292950447,
+      "loss": 0.652,
+      "step": 1020
+    },
+    {
+      "epoch": 1.9579349904397705,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00024836304902751445,
+      "loss": 0.7223,
+      "step": 1024
+    },
+    {
+      "epoch": 1.9655831739961758,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002479697724247062,
+      "loss": 0.7065,
+      "step": 1028
+    },
+    {
+      "epoch": 1.9732313575525813,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002475753178497716,
+      "loss": 0.7307,
+      "step": 1032
+    },
+    {
+      "epoch": 1.9808795411089866,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00024717969004556646,
+      "loss": 0.7086,
+      "step": 1036
+    },
+    {
+      "epoch": 1.988527724665392,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002467828937690532,
+      "loss": 0.7051,
+      "step": 1040
+    },
+    {
+      "epoch": 1.9961759082217974,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002463849337912437,
+      "loss": 0.7084,
+      "step": 1044
+    },
+    {
+      "epoch": 2.0038240917782026,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00024598581489714206,
+      "loss": 0.5785,
+      "step": 1048
+    },
+    {
+      "epoch": 2.011472275334608,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002455855418856869,
+      "loss": 0.5793,
+      "step": 1052
+    },
+    {
+      "epoch": 2.019120458891013,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.000245184119569694,
+      "loss": 0.6212,
+      "step": 1056
+    },
+    {
+      "epoch": 2.026768642447419,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002447815527757979,
+      "loss": 0.569,
+      "step": 1060
+    },
+    {
+      "epoch": 2.0344168260038242,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002443778463443944,
+      "loss": 0.5868,
+      "step": 1064
+    },
+    {
+      "epoch": 2.0420650095602295,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0002439730051295818,
+      "loss": 0.6195,
+      "step": 1068
+    },
+    {
+      "epoch": 2.049713193116635,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002435670339991031,
+      "loss": 0.5886,
+      "step": 1072
+    },
+    {
+      "epoch": 2.05736137667304,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00024315993783428718,
+      "loss": 0.6097,
+      "step": 1076
+    },
+    {
+      "epoch": 2.0650095602294454,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00024275172152999006,
+      "loss": 0.5727,
+      "step": 1080
+    },
+    {
+      "epoch": 2.0726577437858507,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00024234238999453614,
+      "loss": 0.6085,
+      "step": 1084
+    },
+    {
+      "epoch": 2.0803059273422564,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00024193194814965934,
+      "loss": 0.6144,
+      "step": 1088
+    },
+    {
+      "epoch": 2.0879541108986617,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00024152040093044353,
+      "loss": 0.5868,
+      "step": 1092
+    },
+    {
+      "epoch": 2.095602294455067,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00024110775328526352,
+      "loss": 0.6278,
+      "step": 1096
+    },
+    {
+      "epoch": 2.1032504780114722,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00024069401017572543,
+      "loss": 0.5923,
+      "step": 1100
+    },
+    {
+      "epoch": 2.1108986615678775,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00024027917657660713,
+      "loss": 0.5759,
+      "step": 1104
+    },
+    {
+      "epoch": 2.118546845124283,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00023986325747579824,
+      "loss": 0.6138,
+      "step": 1108
+    },
+    {
+      "epoch": 2.126195028680688,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0002394462578742403,
+      "loss": 0.5786,
+      "step": 1112
+    },
+    {
+      "epoch": 2.133843212237094,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002390281827858668,
+      "loss": 0.64,
+      "step": 1116
+    },
+    {
+      "epoch": 2.141491395793499,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002386090372375424,
+      "loss": 0.6307,
+      "step": 1120
+    },
+    {
+      "epoch": 2.1491395793499044,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00023818882626900294,
+      "loss": 0.5641,
+      "step": 1124
+    },
+    {
+      "epoch": 2.1567877629063097,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00023776755493279473,
+      "loss": 0.623,
+      "step": 1128
+    },
+    {
+      "epoch": 2.164435946462715,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00023734522829421372,
+      "loss": 0.6022,
+      "step": 1132
+    },
+    {
+      "epoch": 2.1720841300191203,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00023692185143124464,
+      "loss": 0.6121,
+      "step": 1136
+    },
+    {
+      "epoch": 2.179732313575526,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00023649742943449996,
+      "loss": 0.5878,
+      "step": 1140
+    },
+    {
+      "epoch": 2.1873804971319313,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00023607196740715858,
+      "loss": 0.6143,
+      "step": 1144
+    },
+    {
+      "epoch": 2.1950286806883366,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00023564547046490468,
+      "loss": 0.5655,
+      "step": 1148
+    },
+    {
+      "epoch": 2.202676864244742,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00023521794373586603,
+      "loss": 0.5685,
+      "step": 1152
+    },
+    {
+      "epoch": 2.210325047801147,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00023478939236055228,
+      "loss": 0.5845,
+      "step": 1156
+    },
+    {
+      "epoch": 2.2179732313575524,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00023435982149179346,
+      "loss": 0.6108,
+      "step": 1160
+    },
+    {
+      "epoch": 2.2256214149139577,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002339292362946777,
+      "loss": 0.6221,
+      "step": 1164
+    },
+    {
+      "epoch": 2.2332695984703634,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0002334976419464892,
+      "loss": 0.5739,
+      "step": 1168
+    },
+    {
+      "epoch": 2.2409177820267687,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00023306504363664613,
+      "loss": 0.5928,
+      "step": 1172
+    },
+    {
+      "epoch": 2.248565965583174,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00023263144656663801,
+      "loss": 0.5422,
+      "step": 1176
+    },
+    {
+      "epoch": 2.2562141491395793,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00023219685594996347,
+      "loss": 0.5815,
+      "step": 1180
+    },
+    {
+      "epoch": 2.2638623326959846,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00023176127701206713,
+      "loss": 0.5786,
+      "step": 1184
+    },
+    {
+      "epoch": 2.27151051625239,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00023132471499027717,
+      "loss": 0.5634,
+      "step": 1188
+    },
+    {
+      "epoch": 2.2791586998087956,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002308871751337422,
+      "loss": 0.5969,
+      "step": 1192
+    },
+    {
+      "epoch": 2.286806883365201,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00023044866270336822,
+      "loss": 0.587,
+      "step": 1196
+    },
+    {
+      "epoch": 2.294455066921606,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00023000918297175506,
+      "loss": 0.6312,
+      "step": 1200
+    },
+    {
+      "epoch": 2.3021032504780115,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00022956874122313347,
+      "loss": 0.6268,
+      "step": 1204
+    },
+    {
+      "epoch": 2.3097514340344167,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00022912734275330117,
+      "loss": 0.586,
+      "step": 1208
+    },
+    {
+      "epoch": 2.317399617590822,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00022868499286955943,
+      "loss": 0.571,
+      "step": 1212
+    },
+    {
+      "epoch": 2.3250478011472273,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00022824169689064915,
+      "loss": 0.6062,
+      "step": 1216
+    },
+    {
+      "epoch": 2.332695984703633,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00022779746014668683,
+      "loss": 0.5991,
+      "step": 1220
+    },
+    {
+      "epoch": 2.3403441682600383,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00022735228797910066,
+      "loss": 0.6193,
+      "step": 1224
+    },
+    {
+      "epoch": 2.3479923518164436,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002269061857405662,
+      "loss": 0.5719,
+      "step": 1228
+    },
+    {
+      "epoch": 2.355640535372849,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00022645915879494202,
+      "loss": 0.6171,
+      "step": 1232
+    },
+    {
+      "epoch": 2.363288718929254,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00022601121251720514,
+      "loss": 0.6213,
+      "step": 1236
+    },
+    {
+      "epoch": 2.3709369024856595,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002255623522933866,
+      "loss": 0.5933,
+      "step": 1240
+    },
+    {
+      "epoch": 2.378585086042065,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00022511258352050649,
+      "loss": 0.5598,
+      "step": 1244
+    },
+    {
+      "epoch": 2.3862332695984705,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00022466191160650916,
+      "loss": 0.6435,
+      "step": 1248
+    },
+    {
+      "epoch": 2.3938814531548758,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00022421034197019822,
+      "loss": 0.5553,
+      "step": 1252
+    },
+    {
+      "epoch": 2.401529636711281,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00022375788004117128,
+      "loss": 0.6193,
+      "step": 1256
+    },
+    {
+      "epoch": 2.4091778202676863,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00022330453125975474,
+      "loss": 0.6117,
+      "step": 1260
+    },
+    {
+      "epoch": 2.4168260038240916,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0002228503010769384,
+      "loss": 0.5761,
+      "step": 1264
+    },
+    {
+      "epoch": 2.424474187380497,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002223951949543098,
+      "loss": 0.6343,
+      "step": 1268
+    },
+    {
+      "epoch": 2.4321223709369026,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00022193921836398875,
+      "loss": 0.5763,
+      "step": 1272
+    },
+    {
+      "epoch": 2.439770554493308,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00022148237678856138,
+      "loss": 0.5807,
+      "step": 1276
+    },
+    {
+      "epoch": 2.447418738049713,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002210246757210142,
+      "loss": 0.62,
+      "step": 1280
+    },
+    {
+      "epoch": 2.4550669216061185,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00022056612066466817,
+      "loss": 0.6255,
+      "step": 1284
+    },
+    {
+      "epoch": 2.462715105162524,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00022010671713311238,
+      "loss": 0.6786,
+      "step": 1288
+    },
+    {
+      "epoch": 2.470363288718929,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.000219646470650138,
+      "loss": 0.6173,
+      "step": 1292
+    },
+    {
+      "epoch": 2.478011472275335,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00021918538674967156,
+      "loss": 0.6113,
+      "step": 1296
+    },
+    {
+      "epoch": 2.48565965583174,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0002187234709757087,
+      "loss": 0.5949,
+      "step": 1300
+    },
+    {
+      "epoch": 2.4933078393881454,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00021826072888224716,
+      "loss": 0.6248,
+      "step": 1304
+    },
+    {
+      "epoch": 2.5009560229445507,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00021779716603322034,
+      "loss": 0.5849,
+      "step": 1308
+    },
+    {
+      "epoch": 2.508604206500956,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002173327880024303,
+      "loss": 0.5947,
+      "step": 1312
+    },
+    {
+      "epoch": 2.5162523900573612,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00021686760037348065,
+      "loss": 0.5689,
+      "step": 1316
+    },
+    {
+      "epoch": 2.5239005736137665,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00021640160873970954,
+      "loss": 0.6302,
+      "step": 1320
+    },
+    {
+      "epoch": 2.5315487571701722,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00021593481870412217,
+      "loss": 0.6117,
+      "step": 1324
+    },
+    {
+      "epoch": 2.5391969407265775,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002154672358793238,
+      "loss": 0.6024,
+      "step": 1328
+    },
+    {
+      "epoch": 2.546845124282983,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00021499886588745195,
+      "loss": 0.5417,
+      "step": 1332
+    },
+    {
+      "epoch": 2.554493307839388,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00021452971436010886,
+      "loss": 0.5975,
+      "step": 1336
+    },
+    {
+      "epoch": 2.5621414913957934,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00021405978693829397,
+      "loss": 0.5997,
+      "step": 1340
+    },
+    {
+      "epoch": 2.569789674952199,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00021358908927233576,
+      "loss": 0.6047,
+      "step": 1344
+    },
+    {
+      "epoch": 2.5774378585086044,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00021311762702182414,
+      "loss": 0.6135,
+      "step": 1348
+    },
+    {
+      "epoch": 2.5850860420650097,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00021264540585554215,
+      "loss": 0.6251,
+      "step": 1352
+    },
+    {
+      "epoch": 2.592734225621415,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00021217243145139802,
+      "loss": 0.6308,
+      "step": 1356
+    },
+    {
+      "epoch": 2.6003824091778203,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002116987094963567,
+      "loss": 0.5828,
+      "step": 1360
+    },
+    {
+      "epoch": 2.6080305927342256,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00021122424568637157,
+      "loss": 0.6057,
+      "step": 1364
+    },
+    {
+      "epoch": 2.615678776290631,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00021074904572631606,
+      "loss": 0.6435,
+      "step": 1368
+    },
+    {
+      "epoch": 2.623326959847036,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00021027311532991475,
+      "loss": 0.6201,
+      "step": 1372
+    },
+    {
+      "epoch": 2.6309751434034414,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00020979646021967503,
+      "loss": 0.6192,
+      "step": 1376
+    },
+    {
+      "epoch": 2.638623326959847,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00020931908612681805,
+      "loss": 0.6072,
+      "step": 1380
+    },
+    {
+      "epoch": 2.6462715105162524,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00020884099879120993,
+      "loss": 0.5332,
+      "step": 1384
+    },
+    {
+      "epoch": 2.6539196940726577,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00020836220396129265,
+      "loss": 0.5923,
+      "step": 1388
+    },
+    {
+      "epoch": 2.661567877629063,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00020788270739401505,
+      "loss": 0.6293,
+      "step": 1392
+    },
+    {
+      "epoch": 2.6692160611854687,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00020740251485476345,
+      "loss": 0.5851,
+      "step": 1396
+    },
+    {
+      "epoch": 2.676864244741874,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00020692163211729253,
+      "loss": 0.6088,
+      "step": 1400
+    },
+    {
+      "epoch": 2.6845124282982793,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0002064400649636557,
+      "loss": 0.6033,
+      "step": 1404
+    },
+    {
+      "epoch": 2.6921606118546846,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002059578191841357,
+      "loss": 0.5948,
+      "step": 1408
+    },
+    {
+      "epoch": 2.69980879541109,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00020547490057717499,
+      "loss": 0.6287,
+      "step": 1412
+    },
+    {
+      "epoch": 2.707456978967495,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00020499131494930602,
+      "loss": 0.5736,
+      "step": 1416
+    },
+    {
+      "epoch": 2.7151051625239004,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002045070681150813,
+      "loss": 0.6496,
+      "step": 1420
+    },
+    {
+      "epoch": 2.7227533460803057,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00020402216589700362,
+      "loss": 0.5993,
+      "step": 1424
+    },
+    {
+      "epoch": 2.730401529636711,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00020353661412545598,
+      "loss": 0.596,
+      "step": 1428
+    },
+    {
+      "epoch": 2.7380497131931167,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00020305041863863152,
+      "loss": 0.639,
+      "step": 1432
+    },
+    {
+      "epoch": 2.745697896749522,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00020256358528246334,
+      "loss": 0.5703,
+      "step": 1436
+    },
+    {
+      "epoch": 2.7533460803059273,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00020207611991055407,
+      "loss": 0.5838,
+      "step": 1440
+    },
+    {
+      "epoch": 2.7609942638623326,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002015880283841057,
+      "loss": 0.5845,
+      "step": 1444
+    },
+    {
+      "epoch": 2.768642447418738,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00020109931657184894,
+      "loss": 0.6169,
+      "step": 1448
+    },
+    {
+      "epoch": 2.7762906309751436,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002006099903499727,
+      "loss": 0.6026,
+      "step": 1452
+    },
+    {
+      "epoch": 2.783938814531549,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00020012005560205356,
+      "loss": 0.6278,
+      "step": 1456
+    },
+    {
+      "epoch": 2.791586998087954,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0001996295182189847,
+      "loss": 0.6273,
+      "step": 1460
+    },
+    {
+      "epoch": 2.7992351816443595,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00019913838409890548,
+      "loss": 0.6084,
+      "step": 1464
+    },
+    {
+      "epoch": 2.8068833652007648,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00019864665914713024,
+      "loss": 0.6295,
+      "step": 1468
+    },
+    {
+      "epoch": 2.81453154875717,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0001981543492760774,
+      "loss": 0.5889,
+      "step": 1472
+    },
+    {
+      "epoch": 2.8221797323135753,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00019766146040519836,
+      "loss": 0.6064,
+      "step": 1476
+    },
+    {
+      "epoch": 2.8298279158699806,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00019716799846090634,
+      "loss": 0.6269,
+      "step": 1480
+    },
+    {
+      "epoch": 2.8374760994263863,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00019667396937650506,
+      "loss": 0.5742,
+      "step": 1484
+    },
+    {
+      "epoch": 2.8451242829827916,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001961793790921174,
+      "loss": 0.5701,
+      "step": 1488
+    },
+    {
+      "epoch": 2.852772466539197,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00019568423355461402,
+      "loss": 0.5973,
+      "step": 1492
+    },
+    {
+      "epoch": 2.860420650095602,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00019518853871754204,
+      "loss": 0.609,
+      "step": 1496
+    },
+    {
+      "epoch": 2.8680688336520075,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00019469230054105295,
+      "loss": 0.5944,
+      "step": 1500
+    },
+    {
+      "epoch": 2.875717017208413,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0001941955249918315,
+      "loss": 0.5914,
+      "step": 1504
+    },
+    {
+      "epoch": 2.8833652007648185,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00019369821804302365,
+      "loss": 0.6191,
+      "step": 1508
+    },
+    {
+      "epoch": 2.891013384321224,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00019320038567416484,
+      "loss": 0.6409,
+      "step": 1512
+    },
+    {
+      "epoch": 2.898661567877629,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00019270203387110798,
+      "loss": 0.5779,
+      "step": 1516
+    },
+    {
+      "epoch": 2.9063097514340344,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00019220316862595167,
+      "loss": 0.5956,
+      "step": 1520
+    },
+    {
+      "epoch": 2.9139579349904396,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00019170379593696802,
+      "loss": 0.5916,
+      "step": 1524
+    },
+    {
+      "epoch": 2.921606118546845,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00019120392180853058,
+      "loss": 0.6069,
+      "step": 1528
+    },
+    {
+      "epoch": 2.92925430210325,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0001907035522510421,
+      "loss": 0.6029,
+      "step": 1532
+    },
+    {
+      "epoch": 2.936902485659656,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00019020269328086226,
+      "loss": 0.5706,
+      "step": 1536
+    },
+    {
+      "epoch": 2.9445506692160612,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0001897013509202354,
+      "loss": 0.6024,
+      "step": 1540
+    },
+    {
+      "epoch": 2.9521988527724665,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00018919953119721808,
+      "loss": 0.6326,
+      "step": 1544
+    },
+    {
+      "epoch": 2.959847036328872,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0001886972401456065,
+      "loss": 0.5744,
+      "step": 1548
+    },
+    {
+      "epoch": 2.967495219885277,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00018819448380486413,
+      "loss": 0.5679,
+      "step": 1552
+    },
+    {
+      "epoch": 2.975143403441683,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00018769126822004898,
+      "loss": 0.5992,
+      "step": 1556
+    },
+    {
+      "epoch": 2.982791586998088,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00018718759944174086,
+      "loss": 0.5981,
+      "step": 1560
+    },
+    {
+      "epoch": 2.9904397705544934,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0001866834835259688,
+      "loss": 0.6188,
+      "step": 1564
+    },
+    {
+      "epoch": 2.9980879541108987,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001861789265341381,
+      "loss": 0.617,
+      "step": 1568
+    },
+    {
+      "epoch": 3.005736137667304,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00018567393453295742,
+      "loss": 0.4644,
+      "step": 1572
+    },
+    {
+      "epoch": 3.0133843212237093,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00018516851359436602,
+      "loss": 0.4965,
+      "step": 1576
+    },
+    {
+      "epoch": 3.0210325047801145,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00018466266979546057,
+      "loss": 0.501,
+      "step": 1580
+    },
+    {
+      "epoch": 3.0286806883365203,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0001841564092184221,
+      "loss": 0.4787,
+      "step": 1584
+    },
+    {
+      "epoch": 3.0363288718929256,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00018364973795044294,
+      "loss": 0.5116,
+      "step": 1588
+    },
+    {
+      "epoch": 3.043977055449331,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00018314266208365357,
+      "loss": 0.5309,
+      "step": 1592
+    },
+    {
+      "epoch": 3.051625239005736,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00018263518771504924,
+      "loss": 0.4979,
+      "step": 1596
+    },
+    {
+      "epoch": 3.0592734225621414,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00018212732094641666,
+      "loss": 0.4647,
+      "step": 1600
+    },
+    {
+      "epoch": 3.0669216061185467,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00018161906788426076,
+      "loss": 0.5367,
+      "step": 1604
+    },
+    {
+      "epoch": 3.0745697896749524,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00018111043463973122,
+      "loss": 0.5095,
+      "step": 1608
+    },
+    {
+      "epoch": 3.0822179732313577,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00018060142732854894,
+      "loss": 0.4615,
+      "step": 1612
+    },
+    {
+      "epoch": 3.089866156787763,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00018009205207093252,
+      "loss": 0.5105,
+      "step": 1616
+    },
+    {
+      "epoch": 3.0975143403441683,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00017958231499152463,
+      "loss": 0.5326,
+      "step": 1620
+    },
+    {
+      "epoch": 3.1051625239005736,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0001790722222193186,
+      "loss": 0.5383,
+      "step": 1624
+    },
+    {
+      "epoch": 3.112810707456979,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00017856177988758438,
+      "loss": 0.5192,
+      "step": 1628
+    },
+    {
+      "epoch": 3.120458891013384,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00017805099413379508,
+      "loss": 0.5029,
+      "step": 1632
+    },
+    {
+      "epoch": 3.12810707456979,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00017753987109955297,
+      "loss": 0.4896,
+      "step": 1636
+    },
+    {
+      "epoch": 3.135755258126195,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00017702841693051577,
+      "loss": 0.5254,
+      "step": 1640
+    },
+    {
+      "epoch": 3.1434034416826004,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001765166377763227,
+      "loss": 0.4829,
+      "step": 1644
+    },
+    {
+      "epoch": 3.1510516252390057,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00017600453979052055,
+      "loss": 0.5461,
+      "step": 1648
+    },
+    {
+      "epoch": 3.158699808795411,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0001754921291304897,
+      "loss": 0.5415,
+      "step": 1652
+    },
+    {
+      "epoch": 3.1663479923518163,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00017497941195737004,
+      "loss": 0.5501,
+      "step": 1656
+    },
+    {
+      "epoch": 3.173996175908222,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00017446639443598696,
+      "loss": 0.4964,
+      "step": 1660
+    },
+    {
+      "epoch": 3.1816443594646273,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00017395308273477714,
+      "loss": 0.4938,
+      "step": 1664
+    },
+    {
+      "epoch": 3.1892925430210326,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00017343948302571446,
+      "loss": 0.5409,
+      "step": 1668
+    },
+    {
+      "epoch": 3.196940726577438,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00017292560148423578,
+      "loss": 0.4844,
+      "step": 1672
+    },
+    {
+      "epoch": 3.204588910133843,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00017241144428916655,
+      "loss": 0.539,
+      "step": 1676
+    },
+    {
+      "epoch": 3.2122370936902485,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00017189701762264687,
+      "loss": 0.4974,
+      "step": 1680
+    },
+    {
+      "epoch": 3.2198852772466537,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001713823276700567,
+      "loss": 0.5443,
+      "step": 1684
+    },
+    {
+      "epoch": 3.2275334608030595,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00017086738061994176,
+      "loss": 0.4936,
+      "step": 1688
+    },
+    {
+      "epoch": 3.2351816443594648,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00017035218266393918,
+      "loss": 0.5027,
+      "step": 1692
+    },
+    {
+      "epoch": 3.24282982791587,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00016983673999670273,
+      "loss": 0.5352,
+      "step": 1696
+    },
+    {
+      "epoch": 3.2504780114722753,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0001693210588158287,
+      "loss": 0.5147,
+      "step": 1700
+    },
+    {
+      "epoch": 3.2581261950286806,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00016880514532178123,
+      "loss": 0.5013,
+      "step": 1704
+    },
+    {
+      "epoch": 3.265774378585086,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00016828900571781767,
+      "loss": 0.5408,
+      "step": 1708
+    },
+    {
+      "epoch": 3.2734225621414916,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00016777264620991414,
+      "loss": 0.4758,
+      "step": 1712
+    },
+    {
+      "epoch": 3.281070745697897,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00016725607300669087,
+      "loss": 0.5154,
+      "step": 1716
+    },
+    {
+      "epoch": 3.288718929254302,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0001667392923193375,
+      "loss": 0.4882,
+      "step": 1720
+    },
+    {
+      "epoch": 3.2963671128107075,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00016622231036153836,
+      "loss": 0.5423,
+      "step": 1724
+    },
+    {
+      "epoch": 3.3040152963671128,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001657051333493978,
+      "loss": 0.509,
+      "step": 1728
+    },
+    {
+      "epoch": 3.311663479923518,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00016518776750136578,
+      "loss": 0.5447,
+      "step": 1732
+    },
+    {
+      "epoch": 3.3193116634799233,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00016467021903816237,
+      "loss": 0.5048,
+      "step": 1736
+    },
+    {
+      "epoch": 3.3269598470363286,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00016415249418270364,
+      "loss": 0.5183,
+      "step": 1740
+    },
+    {
+      "epoch": 3.3346080305927344,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00016363459916002643,
+      "loss": 0.4915,
+      "step": 1744
+    },
+    {
+      "epoch": 3.3422562141491396,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00016311654019721377,
+      "loss": 0.5016,
+      "step": 1748
+    },
+    {
+      "epoch": 3.349904397705545,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00016259832352331978,
+      "loss": 0.5276,
+      "step": 1752
+    },
+    {
+      "epoch": 3.35755258126195,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001620799553692949,
+      "loss": 0.5436,
+      "step": 1756
+    },
+    {
+      "epoch": 3.3652007648183555,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00016156144196791103,
+      "loss": 0.5152,
+      "step": 1760
+    },
+    {
+      "epoch": 3.3728489483747612,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0001610427895536863,
+      "loss": 0.4845,
+      "step": 1764
+    },
+    {
+      "epoch": 3.3804971319311665,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00016052400436281046,
+      "loss": 0.51,
+      "step": 1768
+    },
+    {
+      "epoch": 3.388145315487572,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00016000509263306976,
+      "loss": 0.5163,
+      "step": 1772
+    },
+    {
+      "epoch": 3.395793499043977,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001594860606037719,
+      "loss": 0.559,
+      "step": 1776
+    },
+    {
+      "epoch": 3.4034416826003824,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001589669145156709,
+      "loss": 0.523,
+      "step": 1780
+    },
+    {
+      "epoch": 3.4110898661567877,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00015844766061089241,
+      "loss": 0.4994,
+      "step": 1784
+    },
+    {
+      "epoch": 3.418738049713193,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00015792830513285838,
+      "loss": 0.5259,
+      "step": 1788
+    },
+    {
+      "epoch": 3.4263862332695982,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.000157408854326212,
+      "loss": 0.4734,
+      "step": 1792
+    },
+    {
+      "epoch": 3.434034416826004,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00015688931443674276,
+      "loss": 0.5163,
+      "step": 1796
+    },
+    {
+      "epoch": 3.4416826003824093,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001563696917113112,
+      "loss": 0.514,
+      "step": 1800
+    },
+    {
+      "epoch": 3.4493307839388145,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00015584999239777393,
+      "loss": 0.5691,
+      "step": 1804
+    },
+    {
+      "epoch": 3.45697896749522,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001553302227449084,
+      "loss": 0.5365,
+      "step": 1808
+    },
+    {
+      "epoch": 3.464627151051625,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0001548103890023378,
+      "loss": 0.5111,
+      "step": 1812
+    },
+    {
+      "epoch": 3.472275334608031,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00015429049742045591,
+      "loss": 0.5272,
+      "step": 1816
+    },
+    {
+      "epoch": 3.479923518164436,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.000153770554250352,
+      "loss": 0.4746,
+      "step": 1820
+    },
+    {
+      "epoch": 3.4875717017208414,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00015325056574373564,
+      "loss": 0.5091,
+      "step": 1824
+    },
+    {
+      "epoch": 3.4952198852772467,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00015273053815286153,
+      "loss": 0.5043,
+      "step": 1828
+    },
+    {
+      "epoch": 3.502868068833652,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00015221047773045424,
+      "loss": 0.5157,
+      "step": 1832
+    },
+    {
+      "epoch": 3.5105162523900573,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00015169039072963312,
+      "loss": 0.525,
+      "step": 1836
+    },
+    {
+      "epoch": 3.5181644359464626,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00015117028340383713,
+      "loss": 0.536,
+      "step": 1840
+    },
+    {
+      "epoch": 3.525812619502868,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00015065016200674963,
+      "loss": 0.556,
+      "step": 1844
+    },
+    {
+      "epoch": 3.5334608030592736,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00015013003279222312,
+      "loss": 0.5199,
+      "step": 1848
+    },
+    {
+      "epoch": 3.541108986615679,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0001496099020142041,
+      "loss": 0.5381,
+      "step": 1852
+    },
+    {
+      "epoch": 3.548757170172084,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00014908977592665787,
+      "loss": 0.5092,
+      "step": 1856
+    },
+    {
+      "epoch": 3.5564053537284894,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00014856966078349339,
+      "loss": 0.5101,
+      "step": 1860
+    },
+    {
+      "epoch": 3.5640535372848947,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00014804956283848793,
+      "loss": 0.5093,
+      "step": 1864
+    },
+    {
+      "epoch": 3.5717017208413004,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00014752948834521206,
+      "loss": 0.499,
+      "step": 1868
+    },
+    {
+      "epoch": 3.5793499043977057,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00014700944355695432,
+      "loss": 0.4342,
+      "step": 1872
+    },
+    {
+      "epoch": 3.586998087954111,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00014648943472664612,
+      "loss": 0.541,
+      "step": 1876
+    },
+    {
+      "epoch": 3.5946462715105163,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00014596946810678646,
+      "loss": 0.5089,
+      "step": 1880
+    },
+    {
+      "epoch": 3.6022944550669216,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00014544954994936689,
+      "loss": 0.4995,
+      "step": 1884
+    },
+    {
+      "epoch": 3.609942638623327,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0001449296865057962,
+      "loss": 0.5299,
+      "step": 1888
+    },
+    {
+      "epoch": 3.617590822179732,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00014440988402682526,
+      "loss": 0.5933,
+      "step": 1892
+    },
+    {
+      "epoch": 3.6252390057361374,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00014389014876247205,
+      "loss": 0.5045,
+      "step": 1896
+    },
+    {
+      "epoch": 3.632887189292543,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00014337048696194625,
+      "loss": 0.4814,
+      "step": 1900
+    },
+    {
+      "epoch": 3.6405353728489485,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00014285090487357427,
+      "loss": 0.5416,
+      "step": 1904
+    },
+    {
+      "epoch": 3.6481835564053537,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0001423314087447241,
+      "loss": 0.5236,
+      "step": 1908
+    },
+    {
+      "epoch": 3.655831739961759,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00014181200482173015,
+      "loss": 0.5281,
+      "step": 1912
+    },
+    {
+      "epoch": 3.6634799235181643,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00014129269934981802,
+      "loss": 0.5446,
+      "step": 1916
+    },
+    {
+      "epoch": 3.67112810707457,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00014077349857302983,
+      "loss": 0.4949,
+      "step": 1920
+    },
+    {
+      "epoch": 3.6787762906309753,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00014025440873414863,
+      "loss": 0.4875,
+      "step": 1924
+    },
+    {
+      "epoch": 3.6864244741873806,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001397354360746237,
+      "loss": 0.528,
+      "step": 1928
+    },
+    {
+      "epoch": 3.694072657743786,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0001392165868344953,
+      "loss": 0.4827,
+      "step": 1932
+    },
+    {
+      "epoch": 3.701720841300191,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001386978672523198,
+      "loss": 0.5073,
+      "step": 1936
+    },
+    {
+      "epoch": 3.7093690248565965,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001381792835650945,
+      "loss": 0.5036,
+      "step": 1940
+    },
+    {
+      "epoch": 3.7170172084130018,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00013766084200818272,
+      "loss": 0.5396,
+      "step": 1944
+    },
+    {
+      "epoch": 3.724665391969407,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001371425488152389,
+      "loss": 0.4815,
+      "step": 1948
+    },
+    {
+      "epoch": 3.7323135755258128,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001366244102181335,
+      "loss": 0.5306,
+      "step": 1952
+    },
+    {
+      "epoch": 3.739961759082218,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00013610643244687826,
+      "loss": 0.5419,
+      "step": 1956
+    },
+    {
+      "epoch": 3.7476099426386233,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00013558862172955105,
+      "loss": 0.5204,
+      "step": 1960
+    },
+    {
+      "epoch": 3.7552581261950286,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00013507098429222115,
+      "loss": 0.4982,
+      "step": 1964
+    },
+    {
+      "epoch": 3.762906309751434,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00013455352635887438,
+      "loss": 0.4667,
+      "step": 1968
+    },
+    {
+      "epoch": 3.7705544933078396,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00013403625415133824,
+      "loss": 0.4302,
+      "step": 1972
+    },
+    {
+      "epoch": 3.778202676864245,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00013351917388920704,
+      "loss": 0.4545,
+      "step": 1976
+    },
+    {
+      "epoch": 3.78585086042065,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00013300229178976722,
+      "loss": 0.4953,
+      "step": 1980
+    },
+    {
+      "epoch": 3.7934990439770555,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001324856140679225,
+      "loss": 0.4966,
+      "step": 1984
+    },
+    {
+      "epoch": 3.801147227533461,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001319691469361193,
+      "loss": 0.5236,
+      "step": 1988
+    },
+    {
+      "epoch": 3.808795411089866,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00013145289660427173,
+      "loss": 0.5244,
+      "step": 1992
+    },
+    {
+      "epoch": 3.8164435946462714,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00013093686927968738,
+      "loss": 0.4982,
+      "step": 1996
+    },
+    {
+      "epoch": 3.8240917782026767,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00013042107116699228,
+      "loss": 0.4899,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8317399617590824,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00012990550846805654,
+      "loss": 0.5296,
+      "step": 2004
+    },
+    {
+      "epoch": 3.8393881453154877,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0001293901873819196,
+      "loss": 0.5331,
+      "step": 2008
+    },
+    {
+      "epoch": 3.847036328871893,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00012887511410471589,
+      "loss": 0.501,
+      "step": 2012
+    },
+    {
+      "epoch": 3.8546845124282982,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00012836029482960018,
+      "loss": 0.5254,
+      "step": 2016
+    },
+    {
+      "epoch": 3.8623326959847035,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00012784573574667316,
+      "loss": 0.5009,
+      "step": 2020
+    },
+    {
+      "epoch": 3.8699808795411093,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00012733144304290697,
+      "loss": 0.5107,
+      "step": 2024
+    },
+    {
+      "epoch": 3.8776290630975145,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0001268174229020709,
+      "loss": 0.5025,
+      "step": 2028
+    },
+    {
+      "epoch": 3.88527724665392,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0001263036815046571,
+      "loss": 0.5239,
+      "step": 2032
+    },
+    {
+      "epoch": 3.892925430210325,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00012579022502780596,
+      "loss": 0.5112,
+      "step": 2036
+    },
+    {
+      "epoch": 3.9005736137667304,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00012527705964523209,
+      "loss": 0.5182,
+      "step": 2040
+    },
+    {
+      "epoch": 3.9082217973231357,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00012476419152715007,
+      "loss": 0.5505,
+      "step": 2044
+    },
+    {
+      "epoch": 3.915869980879541,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00012425162684020024,
+      "loss": 0.4957,
+      "step": 2048
+    },
+    {
+      "epoch": 3.9235181644359463,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0001237393717473745,
+      "loss": 0.5132,
+      "step": 2052
+    },
+    {
+      "epoch": 3.9311663479923515,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0001232274324079422,
+      "loss": 0.516,
+      "step": 2056
+    },
+    {
+      "epoch": 3.9388145315487573,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00012271581497737619,
+      "loss": 0.5156,
+      "step": 2060
+    },
+    {
+      "epoch": 3.9464627151051626,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00012220452560727875,
+      "loss": 0.4621,
+      "step": 2064
+    },
+    {
+      "epoch": 3.954110898661568,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00012169357044530758,
+      "loss": 0.5206,
+      "step": 2068
+    },
+    {
+      "epoch": 3.961759082217973,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0001211829556351019,
+      "loss": 0.4511,
+      "step": 2072
+    },
+    {
+      "epoch": 3.969407265774379,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00012067268731620861,
+      "loss": 0.5047,
+      "step": 2076
+    },
+    {
+      "epoch": 3.977055449330784,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00012016277162400848,
+      "loss": 0.5295,
+      "step": 2080
+    },
+    {
+      "epoch": 3.9847036328871894,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00011965321468964237,
+      "loss": 0.5204,
+      "step": 2084
+    },
+    {
+      "epoch": 3.9923518164435947,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00011914402263993745,
+      "loss": 0.5064,
+      "step": 2088
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00011863520159733357,
+      "loss": 0.53,
+      "step": 2092
+    },
+    {
+      "epoch": 4.007648183556405,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00011812675767980972,
+      "loss": 0.4532,
+      "step": 2096
+    },
+    {
+      "epoch": 4.015296367112811,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00011761869700081036,
+      "loss": 0.4748,
+      "step": 2100
+    },
+    {
+      "epoch": 4.022944550669216,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00011711102566917194,
+      "loss": 0.4188,
+      "step": 2104
+    },
+    {
+      "epoch": 4.030592734225621,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00011660374978904947,
+      "loss": 0.4466,
+      "step": 2108
+    },
+    {
+      "epoch": 4.038240917782026,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00011609687545984315,
+      "loss": 0.4112,
+      "step": 2112
+    },
+    {
+      "epoch": 4.045889101338432,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00011559040877612497,
+      "loss": 0.4566,
+      "step": 2116
+    },
+    {
+      "epoch": 4.053537284894838,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00011508435582756545,
+      "loss": 0.4413,
+      "step": 2120
+    },
+    {
+      "epoch": 4.061185468451243,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00011457872269886043,
+      "loss": 0.4435,
+      "step": 2124
+    },
+    {
+      "epoch": 4.0688336520076485,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011407351546965796,
+      "loss": 0.4568,
+      "step": 2128
+    },
+    {
+      "epoch": 4.076481835564054,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00011356874021448506,
+      "loss": 0.4247,
+      "step": 2132
+    },
+    {
+      "epoch": 4.084130019120459,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00011306440300267482,
+      "loss": 0.3762,
+      "step": 2136
+    },
+    {
+      "epoch": 4.091778202676864,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00011256050989829337,
+      "loss": 0.4713,
+      "step": 2140
+    },
+    {
+      "epoch": 4.09942638623327,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00011205706696006698,
+      "loss": 0.4178,
+      "step": 2144
+    },
+    {
+      "epoch": 4.107074569789675,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00011155408024130921,
+      "loss": 0.4266,
+      "step": 2148
+    },
+    {
+      "epoch": 4.11472275334608,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00011105155578984795,
+      "loss": 0.4242,
+      "step": 2152
+    },
+    {
+      "epoch": 4.1223709369024855,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00011054949964795307,
+      "loss": 0.464,
+      "step": 2156
+    },
+    {
+      "epoch": 4.130019120458891,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00011004791785226347,
+      "loss": 0.4216,
+      "step": 2160
+    },
+    {
+      "epoch": 4.137667304015296,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00010954681643371462,
+      "loss": 0.4222,
+      "step": 2164
+    },
+    {
+      "epoch": 4.145315487571701,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00010904620141746601,
+      "loss": 0.4321,
+      "step": 2168
+    },
+    {
+      "epoch": 4.1529636711281075,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001085460788228287,
+      "loss": 0.441,
+      "step": 2172
+    },
+    {
+      "epoch": 4.160611854684513,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00010804645466319292,
+      "loss": 0.4468,
+      "step": 2176
+    },
+    {
+      "epoch": 4.168260038240918,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001075473349459559,
+      "loss": 0.3948,
+      "step": 2180
+    },
+    {
+      "epoch": 4.175908221797323,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00010704872567244948,
+      "loss": 0.4233,
+      "step": 2184
+    },
+    {
+      "epoch": 4.183556405353729,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00010655063283786795,
+      "loss": 0.4227,
+      "step": 2188
+    },
+    {
+      "epoch": 4.191204588910134,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00010605306243119617,
+      "loss": 0.4242,
+      "step": 2192
+    },
+    {
+      "epoch": 4.198852772466539,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00010555602043513724,
+      "loss": 0.4428,
+      "step": 2196
+    },
+    {
+      "epoch": 4.2065009560229445,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00010505951282604088,
+      "loss": 0.4132,
+      "step": 2200
+    },
+    {
+      "epoch": 4.21414913957935,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00010456354557383139,
+      "loss": 0.4046,
+      "step": 2204
+    },
+    {
+      "epoch": 4.221797323135755,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00010406812464193584,
+      "loss": 0.4527,
+      "step": 2208
+    },
+    {
+      "epoch": 4.22944550669216,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00010357325598721255,
+      "loss": 0.4107,
+      "step": 2212
+    },
+    {
+      "epoch": 4.237093690248566,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00010307894555987927,
+      "loss": 0.4375,
+      "step": 2216
+    },
+    {
+      "epoch": 4.244741873804971,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00010258519930344179,
+      "loss": 0.4328,
+      "step": 2220
+    },
+    {
+      "epoch": 4.252390057361376,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001020920231546223,
+      "loss": 0.4388,
+      "step": 2224
+    },
+    {
+      "epoch": 4.260038240917782,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00010159942304328819,
+      "loss": 0.4443,
+      "step": 2228
+    },
+    {
+      "epoch": 4.267686424474188,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00010110740489238066,
+      "loss": 0.4446,
+      "step": 2232
+    },
+    {
+      "epoch": 4.275334608030593,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00010061597461784346,
+      "loss": 0.4269,
+      "step": 2236
+    },
+    {
+      "epoch": 4.282982791586998,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00010012513812855191,
+      "loss": 0.4326,
+      "step": 2240
+    },
+    {
+      "epoch": 4.2906309751434035,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.963490132624169e-05,
+      "loss": 0.4342,
+      "step": 2244
+    },
+    {
+      "epoch": 4.298279158699809,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.914527010543795e-05,
+      "loss": 0.4157,
+      "step": 2248
+    },
+    {
+      "epoch": 4.305927342256214,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.865625035338447e-05,
+      "loss": 0.4237,
+      "step": 2252
+    },
+    {
+      "epoch": 4.313575525812619,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.816784794997275e-05,
+      "loss": 0.4033,
+      "step": 2256
+    },
+    {
+      "epoch": 4.321223709369025,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.76800687676715e-05,
+      "loss": 0.4436,
+      "step": 2260
+    },
+    {
+      "epoch": 4.32887189292543,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.719291867145583e-05,
+      "loss": 0.4384,
+      "step": 2264
+    },
+    {
+      "epoch": 4.336520076481835,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.670640351873688e-05,
+      "loss": 0.4512,
+      "step": 2268
+    },
+    {
+      "epoch": 4.3441682600382405,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.62205291592913e-05,
+      "loss": 0.4259,
+      "step": 2272
+    },
+    {
+      "epoch": 4.351816443594647,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.573530143519098e-05,
+      "loss": 0.4178,
+      "step": 2276
+    },
+    {
+      "epoch": 4.359464627151052,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.525072618073277e-05,
+      "loss": 0.3608,
+      "step": 2280
+    },
+    {
+      "epoch": 4.367112810707457,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.476680922236831e-05,
+      "loss": 0.4489,
+      "step": 2284
+    },
+    {
+      "epoch": 4.374760994263863,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.428355637863402e-05,
+      "loss": 0.4346,
+      "step": 2288
+    },
+    {
+      "epoch": 4.382409177820268,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.380097346008112e-05,
+      "loss": 0.4542,
+      "step": 2292
+    },
+    {
+      "epoch": 4.390057361376673,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.331906626920576e-05,
+      "loss": 0.4395,
+      "step": 2296
+    },
+    {
+      "epoch": 4.397705544933078,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.283784060037921e-05,
+      "loss": 0.4858,
+      "step": 2300
+    },
+    {
+      "epoch": 4.405353728489484,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.235730223977837e-05,
+      "loss": 0.4148,
+      "step": 2304
+    },
+    {
+      "epoch": 4.413001912045889,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.187745696531584e-05,
+      "loss": 0.4579,
+      "step": 2308
+    },
+    {
+      "epoch": 4.420650095602294,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.139831054657081e-05,
+      "loss": 0.457,
+      "step": 2312
+    },
+    {
+      "epoch": 4.4282982791587,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.091986874471956e-05,
+      "loss": 0.4257,
+      "step": 2316
+    },
+    {
+      "epoch": 4.435946462715105,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.044213731246614e-05,
+      "loss": 0.4287,
+      "step": 2320
+    },
+    {
+      "epoch": 4.44359464627151,
+      "grad_norm": 0.3359375,
+      "learning_rate": 8.99651219939732e-05,
+      "loss": 0.4482,
+      "step": 2324
+    },
+    {
+      "epoch": 4.451242829827915,
+      "grad_norm": 0.345703125,
+      "learning_rate": 8.948882852479305e-05,
+      "loss": 0.4772,
+      "step": 2328
+    },
+    {
+      "epoch": 4.458891013384322,
+      "grad_norm": 0.33203125,
+      "learning_rate": 8.901326263179851e-05,
+      "loss": 0.4067,
+      "step": 2332
+    },
+    {
+      "epoch": 4.466539196940727,
+      "grad_norm": 0.373046875,
+      "learning_rate": 8.85384300331142e-05,
+      "loss": 0.4399,
+      "step": 2336
+    },
+    {
+      "epoch": 4.474187380497132,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.80643364380477e-05,
+      "loss": 0.4726,
+      "step": 2340
+    },
+    {
+      "epoch": 4.4818355640535374,
+      "grad_norm": 0.349609375,
+      "learning_rate": 8.759098754702099e-05,
+      "loss": 0.4514,
+      "step": 2344
+    },
+    {
+      "epoch": 4.489483747609943,
+      "grad_norm": 0.36328125,
+      "learning_rate": 8.711838905150179e-05,
+      "loss": 0.4502,
+      "step": 2348
+    },
+    {
+      "epoch": 4.497131931166348,
+      "grad_norm": 0.328125,
+      "learning_rate": 8.664654663393516e-05,
+      "loss": 0.4366,
+      "step": 2352
+    },
+    {
+      "epoch": 4.504780114722753,
+      "grad_norm": 0.341796875,
+      "learning_rate": 8.617546596767534e-05,
+      "loss": 0.437,
+      "step": 2356
+    },
+    {
+      "epoch": 4.512428298279159,
+      "grad_norm": 0.337890625,
+      "learning_rate": 8.570515271691723e-05,
+      "loss": 0.4313,
+      "step": 2360
+    },
+    {
+      "epoch": 4.520076481835564,
+      "grad_norm": 0.34765625,
+      "learning_rate": 8.523561253662864e-05,
+      "loss": 0.447,
+      "step": 2364
+    },
+    {
+      "epoch": 4.527724665391969,
+      "grad_norm": 0.357421875,
+      "learning_rate": 8.476685107248197e-05,
+      "loss": 0.4488,
+      "step": 2368
+    },
+    {
+      "epoch": 4.5353728489483744,
+      "grad_norm": 0.33203125,
+      "learning_rate": 8.429887396078655e-05,
+      "loss": 0.469,
+      "step": 2372
+    },
+    {
+      "epoch": 4.54302103250478,
+      "grad_norm": 0.3515625,
+      "learning_rate": 8.38316868284207e-05,
+      "loss": 0.4886,
+      "step": 2376
+    },
+    {
+      "epoch": 4.550669216061186,
+      "grad_norm": 0.341796875,
+      "learning_rate": 8.336529529276421e-05,
+      "loss": 0.433,
+      "step": 2380
+    },
+    {
+      "epoch": 4.558317399617591,
+      "grad_norm": 0.33984375,
+      "learning_rate": 8.289970496163085e-05,
+      "loss": 0.4029,
+      "step": 2384
+    },
+    {
+      "epoch": 4.5659655831739965,
+      "grad_norm": 0.322265625,
+      "learning_rate": 8.243492143320058e-05,
+      "loss": 0.4198,
+      "step": 2388
+    },
+    {
+      "epoch": 4.573613766730402,
+      "grad_norm": 0.341796875,
+      "learning_rate": 8.197095029595276e-05,
+      "loss": 0.4377,
+      "step": 2392
+    },
+    {
+      "epoch": 4.581261950286807,
+      "grad_norm": 0.359375,
+      "learning_rate": 8.150779712859854e-05,
+      "loss": 0.4263,
+      "step": 2396
+    },
+    {
+      "epoch": 4.588910133843212,
+      "grad_norm": 0.349609375,
+      "learning_rate": 8.104546750001402e-05,
+      "loss": 0.4784,
+      "step": 2400
+    },
+    {
+      "epoch": 4.596558317399618,
+      "grad_norm": 0.337890625,
+      "learning_rate": 8.05839669691732e-05,
+      "loss": 0.4549,
+      "step": 2404
+    },
+    {
+      "epoch": 4.604206500956023,
+      "grad_norm": 0.33203125,
+      "learning_rate": 8.01233010850811e-05,
+      "loss": 0.4519,
+      "step": 2408
+    },
+    {
+      "epoch": 4.611854684512428,
+      "grad_norm": 0.37109375,
+      "learning_rate": 7.966347538670712e-05,
+      "loss": 0.4242,
+      "step": 2412
+    },
+    {
+      "epoch": 4.6195028680688335,
+      "grad_norm": 0.333984375,
+      "learning_rate": 7.92044954029184e-05,
+      "loss": 0.4647,
+      "step": 2416
+    },
+    {
+      "epoch": 4.627151051625239,
+      "grad_norm": 0.341796875,
+      "learning_rate": 7.874636665241335e-05,
+      "loss": 0.46,
+      "step": 2420
+    },
+    {
+      "epoch": 4.634799235181644,
+      "grad_norm": 0.34375,
+      "learning_rate": 7.828909464365531e-05,
+      "loss": 0.4169,
+      "step": 2424
+    },
+    {
+      "epoch": 4.642447418738049,
+      "grad_norm": 0.349609375,
+      "learning_rate": 7.783268487480626e-05,
+      "loss": 0.4251,
+      "step": 2428
+    },
+    {
+      "epoch": 4.650095602294455,
+      "grad_norm": 0.337890625,
+      "learning_rate": 7.73771428336608e-05,
+      "loss": 0.4232,
+      "step": 2432
+    },
+    {
+      "epoch": 4.657743785850861,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.692247399758008e-05,
+      "loss": 0.4836,
+      "step": 2436
+    },
+    {
+      "epoch": 4.665391969407266,
+      "grad_norm": 0.341796875,
+      "learning_rate": 7.6468683833426e-05,
+      "loss": 0.3917,
+      "step": 2440
+    },
+    {
+      "epoch": 4.673040152963671,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.601577779749545e-05,
+      "loss": 0.4153,
+      "step": 2444
+    },
+    {
+      "epoch": 4.680688336520077,
+      "grad_norm": 0.33984375,
+      "learning_rate": 7.55637613354547e-05,
+      "loss": 0.4103,
+      "step": 2448
+    },
+    {
+      "epoch": 4.688336520076482,
+      "grad_norm": 0.34765625,
+      "learning_rate": 7.511263988227397e-05,
+      "loss": 0.4858,
+      "step": 2452
+    },
+    {
+      "epoch": 4.695984703632887,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.466241886216198e-05,
+      "loss": 0.468,
+      "step": 2456
+    },
+    {
+      "epoch": 4.7036328871892925,
+      "grad_norm": 0.34765625,
+      "learning_rate": 7.421310368850085e-05,
+      "loss": 0.4817,
+      "step": 2460
+    },
+    {
+      "epoch": 4.711281070745698,
+      "grad_norm": 0.35546875,
+      "learning_rate": 7.376469976378094e-05,
+      "loss": 0.3923,
+      "step": 2464
+    },
+    {
+      "epoch": 4.718929254302103,
+      "grad_norm": 0.337890625,
+      "learning_rate": 7.33172124795359e-05,
+      "loss": 0.3954,
+      "step": 2468
+    },
+    {
+      "epoch": 4.726577437858508,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.287064721627782e-05,
+      "loss": 0.4606,
+      "step": 2472
+    },
+    {
+      "epoch": 4.734225621414914,
+      "grad_norm": 0.357421875,
+      "learning_rate": 7.242500934343262e-05,
+      "loss": 0.4253,
+      "step": 2476
+    },
+    {
+      "epoch": 4.741873804971319,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.19803042192754e-05,
+      "loss": 0.4431,
+      "step": 2480
+    },
+    {
+      "epoch": 4.749521988527725,
+      "grad_norm": 0.322265625,
+      "learning_rate": 7.153653719086604e-05,
+      "loss": 0.4097,
+      "step": 2484
+    },
+    {
+      "epoch": 4.75717017208413,
+      "grad_norm": 0.322265625,
+      "learning_rate": 7.109371359398493e-05,
+      "loss": 0.4208,
+      "step": 2488
+    },
+    {
+      "epoch": 4.764818355640536,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.06518387530688e-05,
+      "loss": 0.4396,
+      "step": 2492
+    },
+    {
+      "epoch": 4.772466539196941,
+      "grad_norm": 0.328125,
+      "learning_rate": 7.021091798114667e-05,
+      "loss": 0.4212,
+      "step": 2496
+    },
+    {
+      "epoch": 4.780114722753346,
+      "grad_norm": 0.33203125,
+      "learning_rate": 6.977095657977603e-05,
+      "loss": 0.4742,
+      "step": 2500
+    },
+    {
+      "epoch": 4.7877629063097515,
+      "grad_norm": 0.341796875,
+      "learning_rate": 6.933195983897905e-05,
+      "loss": 0.4198,
+      "step": 2504
+    },
+    {
+      "epoch": 4.795411089866157,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.889393303717898e-05,
+      "loss": 0.4401,
+      "step": 2508
+    },
+    {
+      "epoch": 4.803059273422562,
+      "grad_norm": 0.34765625,
+      "learning_rate": 6.845688144113663e-05,
+      "loss": 0.4273,
+      "step": 2512
+    },
+    {
+      "epoch": 4.810707456978967,
+      "grad_norm": 0.322265625,
+      "learning_rate": 6.802081030588722e-05,
+      "loss": 0.4589,
+      "step": 2516
+    },
+    {
+      "epoch": 4.818355640535373,
+      "grad_norm": 0.34765625,
+      "learning_rate": 6.758572487467698e-05,
+      "loss": 0.4835,
+      "step": 2520
+    },
+    {
+      "epoch": 4.826003824091778,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.715163037890021e-05,
+      "loss": 0.4493,
+      "step": 2524
+    },
+    {
+      "epoch": 4.833652007648183,
+      "grad_norm": 0.3515625,
+      "learning_rate": 6.671853203803641e-05,
+      "loss": 0.4032,
+      "step": 2528
+    },
+    {
+      "epoch": 4.8413001912045885,
+      "grad_norm": 0.34765625,
+      "learning_rate": 6.628643505958742e-05,
+      "loss": 0.4498,
+      "step": 2532
+    },
+    {
+      "epoch": 4.848948374760994,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.585534463901493e-05,
+      "loss": 0.4255,
+      "step": 2536
+    },
+    {
+      "epoch": 4.8565965583174,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.542526595967795e-05,
+      "loss": 0.4285,
+      "step": 2540
+    },
+    {
+      "epoch": 4.864244741873805,
+      "grad_norm": 0.337890625,
+      "learning_rate": 6.499620419277036e-05,
+      "loss": 0.4507,
+      "step": 2544
+    },
+    {
+      "epoch": 4.871892925430211,
+      "grad_norm": 0.333984375,
+      "learning_rate": 6.456816449725892e-05,
+      "loss": 0.4196,
+      "step": 2548
+    },
+    {
+      "epoch": 4.879541108986616,
+      "grad_norm": 0.345703125,
+      "learning_rate": 6.414115201982134e-05,
+      "loss": 0.4061,
+      "step": 2552
+    },
+    {
+      "epoch": 4.887189292543021,
+      "grad_norm": 0.33984375,
+      "learning_rate": 6.371517189478403e-05,
+      "loss": 0.4199,
+      "step": 2556
+    },
+    {
+      "epoch": 4.894837476099426,
+      "grad_norm": 0.3515625,
+      "learning_rate": 6.329022924406061e-05,
+      "loss": 0.452,
+      "step": 2560
+    },
+    {
+      "epoch": 4.902485659655832,
+      "grad_norm": 0.345703125,
+      "learning_rate": 6.286632917709031e-05,
+      "loss": 0.4516,
+      "step": 2564
+    },
+    {
+      "epoch": 4.910133843212237,
+      "grad_norm": 0.33984375,
+      "learning_rate": 6.244347679077651e-05,
+      "loss": 0.4622,
+      "step": 2568
+    },
+    {
+      "epoch": 4.917782026768642,
+      "grad_norm": 0.333984375,
+      "learning_rate": 6.202167716942543e-05,
+      "loss": 0.4387,
+      "step": 2572
+    },
+    {
+      "epoch": 4.925430210325048,
+      "grad_norm": 0.35546875,
+      "learning_rate": 6.160093538468505e-05,
+      "loss": 0.4064,
+      "step": 2576
+    },
+    {
+      "epoch": 4.933078393881453,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.118125649548405e-05,
+      "loss": 0.4728,
+      "step": 2580
+    },
+    {
+      "epoch": 4.940726577437858,
+      "grad_norm": 0.35546875,
+      "learning_rate": 6.076264554797112e-05,
+      "loss": 0.4555,
+      "step": 2584
+    },
+    {
+      "epoch": 4.948374760994264,
+      "grad_norm": 0.337890625,
+      "learning_rate": 6.0345107575454105e-05,
+      "loss": 0.427,
+      "step": 2588
+    },
+    {
+      "epoch": 4.95602294455067,
+      "grad_norm": 0.341796875,
+      "learning_rate": 5.992864759833963e-05,
+      "loss": 0.3841,
+      "step": 2592
+    },
+    {
+      "epoch": 4.963671128107075,
+      "grad_norm": 0.330078125,
+      "learning_rate": 5.9513270624072655e-05,
+      "loss": 0.4289,
+      "step": 2596
+    },
+    {
+      "epoch": 4.97131931166348,
+      "grad_norm": 0.353515625,
+      "learning_rate": 5.9098981647076345e-05,
+      "loss": 0.4538,
+      "step": 2600
+    },
+    {
+      "epoch": 4.9789674952198855,
+      "grad_norm": 0.33203125,
+      "learning_rate": 5.8685785648691894e-05,
+      "loss": 0.3954,
+      "step": 2604
+    },
+    {
+      "epoch": 4.986615678776291,
+      "grad_norm": 0.34765625,
+      "learning_rate": 5.8273687597118765e-05,
+      "loss": 0.4614,
+      "step": 2608
+    },
+    {
+      "epoch": 4.994263862332696,
+      "grad_norm": 0.3359375,
+      "learning_rate": 5.786269244735488e-05,
+      "loss": 0.4429,
+      "step": 2612
+    },
+    {
+      "epoch": 5.001912045889101,
+      "grad_norm": 0.30859375,
+      "learning_rate": 5.7452805141137034e-05,
+      "loss": 0.3411,
+      "step": 2616
+    },
+    {
+      "epoch": 5.009560229445507,
+      "grad_norm": 0.318359375,
+      "learning_rate": 5.704403060688158e-05,
+      "loss": 0.3973,
+      "step": 2620
+    },
+    {
+      "epoch": 5.017208413001912,
+      "grad_norm": 0.314453125,
+      "learning_rate": 5.663637375962489e-05,
+      "loss": 0.4319,
+      "step": 2624
+    },
+    {
+      "epoch": 5.024856596558317,
+      "grad_norm": 0.3203125,
+      "learning_rate": 5.6229839500964635e-05,
+      "loss": 0.3768,
+      "step": 2628
+    },
+    {
+      "epoch": 5.0325047801147225,
+      "grad_norm": 0.314453125,
+      "learning_rate": 5.582443271900063e-05,
+      "loss": 0.3683,
+      "step": 2632
+    },
+    {
+      "epoch": 5.040152963671128,
+      "grad_norm": 0.35546875,
+      "learning_rate": 5.542015828827609e-05,
+      "loss": 0.4145,
+      "step": 2636
+    },
+    {
+      "epoch": 5.047801147227533,
+      "grad_norm": 0.35546875,
+      "learning_rate": 5.5017021069719014e-05,
+      "loss": 0.4116,
+      "step": 2640
+    },
+    {
+      "epoch": 5.055449330783939,
+      "grad_norm": 0.3359375,
+      "learning_rate": 5.4615025910583756e-05,
+      "loss": 0.4069,
+      "step": 2644
+    },
+    {
+      "epoch": 5.0630975143403445,
+      "grad_norm": 0.35546875,
+      "learning_rate": 5.421417764439276e-05,
+      "loss": 0.375,
+      "step": 2648
+    },
+    {
+      "epoch": 5.07074569789675,
+      "grad_norm": 0.333984375,
+      "learning_rate": 5.3814481090878374e-05,
+      "loss": 0.3576,
+      "step": 2652
+    },
+    {
+      "epoch": 5.078393881453155,
+      "grad_norm": 0.337890625,
+      "learning_rate": 5.3415941055924974e-05,
+      "loss": 0.4027,
+      "step": 2656
+    },
+    {
+      "epoch": 5.08604206500956,
+      "grad_norm": 0.328125,
+      "learning_rate": 5.301856233151123e-05,
+      "loss": 0.3981,
+      "step": 2660
+    },
+    {
+      "epoch": 5.093690248565966,
+      "grad_norm": 0.318359375,
+      "learning_rate": 5.262234969565226e-05,
+      "loss": 0.3928,
+      "step": 2664
+    },
+    {
+      "epoch": 5.101338432122371,
+      "grad_norm": 0.33203125,
+      "learning_rate": 5.222730791234246e-05,
+      "loss": 0.3813,
+      "step": 2668
+    },
+    {
+      "epoch": 5.108986615678776,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.183344173149798e-05,
+      "loss": 0.4151,
+      "step": 2672
+    },
+    {
+      "epoch": 5.1166347992351815,
+      "grad_norm": 0.35546875,
+      "learning_rate": 5.14407558888998e-05,
+      "loss": 0.4051,
+      "step": 2676
+    },
+    {
+      "epoch": 5.124282982791587,
+      "grad_norm": 0.330078125,
+      "learning_rate": 5.104925510613668e-05,
+      "loss": 0.3973,
+      "step": 2680
+    },
+    {
+      "epoch": 5.131931166347992,
+      "grad_norm": 0.3203125,
+      "learning_rate": 5.0658944090548436e-05,
+      "loss": 0.3958,
+      "step": 2684
+    },
+    {
+      "epoch": 5.139579349904397,
+      "grad_norm": 0.34765625,
+      "learning_rate": 5.0269827535169306e-05,
+      "loss": 0.3731,
+      "step": 2688
+    },
+    {
+      "epoch": 5.147227533460803,
+      "grad_norm": 0.33984375,
+      "learning_rate": 4.988191011867153e-05,
+      "loss": 0.4028,
+      "step": 2692
+    },
+    {
+      "epoch": 5.154875717017209,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.9495196505309196e-05,
+      "loss": 0.4201,
+      "step": 2696
+    },
+    {
+      "epoch": 5.162523900573614,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.9109691344861886e-05,
+      "loss": 0.3815,
+      "step": 2700
+    },
+    {
+      "epoch": 5.170172084130019,
+      "grad_norm": 0.326171875,
+      "learning_rate": 4.8725399272579075e-05,
+      "loss": 0.4043,
+      "step": 2704
+    },
+    {
+      "epoch": 5.177820267686425,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.8342324909124256e-05,
+      "loss": 0.3828,
+      "step": 2708
+    },
+    {
+      "epoch": 5.18546845124283,
+      "grad_norm": 0.34765625,
+      "learning_rate": 4.7960472860519365e-05,
+      "loss": 0.4199,
+      "step": 2712
+    },
+    {
+      "epoch": 5.193116634799235,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.757984771808947e-05,
+      "loss": 0.4071,
+      "step": 2716
+    },
+    {
+      "epoch": 5.2007648183556405,
+      "grad_norm": 0.34765625,
+      "learning_rate": 4.72004540584075e-05,
+      "loss": 0.4288,
+      "step": 2720
+    },
+    {
+      "epoch": 5.208413001912046,
+      "grad_norm": 0.353515625,
+      "learning_rate": 4.682229644323922e-05,
+      "loss": 0.3935,
+      "step": 2724
+    },
+    {
+      "epoch": 5.216061185468451,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.6445379419488436e-05,
+      "loss": 0.3801,
+      "step": 2728
+    },
+    {
+      "epoch": 5.223709369024856,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.606970751914229e-05,
+      "loss": 0.4181,
+      "step": 2732
+    },
+    {
+      "epoch": 5.231357552581262,
+      "grad_norm": 0.341796875,
+      "learning_rate": 4.569528525921672e-05,
+      "loss": 0.3816,
+      "step": 2736
+    },
+    {
+      "epoch": 5.239005736137667,
+      "grad_norm": 0.349609375,
+      "learning_rate": 4.532211714170229e-05,
+      "loss": 0.4331,
+      "step": 2740
+    },
+    {
+      "epoch": 5.246653919694072,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.495020765350988e-05,
+      "loss": 0.4331,
+      "step": 2744
+    },
+    {
+      "epoch": 5.254302103250478,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.4579561266416855e-05,
+      "loss": 0.3889,
+      "step": 2748
+    },
+    {
+      "epoch": 5.261950286806884,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.421018243701327e-05,
+      "loss": 0.3804,
+      "step": 2752
+    },
+    {
+      "epoch": 5.269598470363289,
+      "grad_norm": 0.55859375,
+      "learning_rate": 4.384207560664825e-05,
+      "loss": 0.4386,
+      "step": 2756
+    },
+    {
+      "epoch": 5.277246653919694,
+      "grad_norm": 0.34375,
+      "learning_rate": 4.347524520137667e-05,
+      "loss": 0.3547,
+      "step": 2760
+    },
+    {
+      "epoch": 5.2848948374761,
+      "grad_norm": 0.337890625,
+      "learning_rate": 4.310969563190578e-05,
+      "loss": 0.3719,
+      "step": 2764
+    },
+    {
+      "epoch": 5.292543021032505,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.274543129354245e-05,
+      "loss": 0.3611,
+      "step": 2768
+    },
+    {
+      "epoch": 5.30019120458891,
+      "grad_norm": 0.3125,
+      "learning_rate": 4.2382456566139985e-05,
+      "loss": 0.3173,
+      "step": 2772
+    },
+    {
+      "epoch": 5.307839388145315,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.202077581404574e-05,
+      "loss": 0.3848,
+      "step": 2776
+    },
+    {
+      "epoch": 5.315487571701721,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.166039338604838e-05,
+      "loss": 0.3921,
+      "step": 2780
+    },
+    {
+      "epoch": 5.323135755258126,
+      "grad_norm": 0.318359375,
+      "learning_rate": 4.130131361532586e-05,
+      "loss": 0.4049,
+      "step": 2784
+    },
+    {
+      "epoch": 5.330783938814531,
+      "grad_norm": 0.35546875,
+      "learning_rate": 4.094354081939317e-05,
+      "loss": 0.3866,
+      "step": 2788
+    },
+    {
+      "epoch": 5.338432122370937,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.058707930005048e-05,
+      "loss": 0.3727,
+      "step": 2792
+    },
+    {
+      "epoch": 5.346080305927342,
+      "grad_norm": 0.341796875,
+      "learning_rate": 4.023193334333132e-05,
+      "loss": 0.3957,
+      "step": 2796
+    },
+    {
+      "epoch": 5.353728489483748,
+      "grad_norm": 0.341796875,
+      "learning_rate": 3.9878107219451206e-05,
+      "loss": 0.344,
+      "step": 2800
+    },
+    {
+      "epoch": 5.361376673040153,
+      "grad_norm": 0.375,
+      "learning_rate": 3.9525605182756134e-05,
+      "loss": 0.4048,
+      "step": 2804
+    },
+    {
+      "epoch": 5.369024856596559,
+      "grad_norm": 0.349609375,
+      "learning_rate": 3.917443147167152e-05,
+      "loss": 0.4343,
+      "step": 2808
+    },
+    {
+      "epoch": 5.376673040152964,
+      "grad_norm": 0.345703125,
+      "learning_rate": 3.882459030865124e-05,
+      "loss": 0.3908,
+      "step": 2812
+    },
+    {
+      "epoch": 5.384321223709369,
+      "grad_norm": 0.34765625,
+      "learning_rate": 3.8476085900126776e-05,
+      "loss": 0.3491,
+      "step": 2816
+    },
+    {
+      "epoch": 5.3919694072657744,
+      "grad_norm": 0.3515625,
+      "learning_rate": 3.8128922436456766e-05,
+      "loss": 0.4092,
+      "step": 2820
+    },
+    {
+      "epoch": 5.39961759082218,
+      "grad_norm": 0.375,
+      "learning_rate": 3.7783104091876524e-05,
+      "loss": 0.3766,
+      "step": 2824
+    },
+    {
+      "epoch": 5.407265774378585,
+      "grad_norm": 0.34375,
+      "learning_rate": 3.743863502444783e-05,
+      "loss": 0.3835,
+      "step": 2828
+    },
+    {
+      "epoch": 5.41491395793499,
+      "grad_norm": 0.33203125,
+      "learning_rate": 3.709551937600909e-05,
+      "loss": 0.3733,
+      "step": 2832
+    },
+    {
+      "epoch": 5.422562141491396,
+      "grad_norm": 0.34375,
+      "learning_rate": 3.675376127212532e-05,
+      "loss": 0.4127,
+      "step": 2836
+    },
+    {
+      "epoch": 5.430210325047801,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.64133648220387e-05,
+      "loss": 0.4091,
+      "step": 2840
+    },
+    {
+      "epoch": 5.437858508604206,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.607433411861912e-05,
+      "loss": 0.4612,
+      "step": 2844
+    },
+    {
+      "epoch": 5.4455066921606115,
+      "grad_norm": 0.330078125,
+      "learning_rate": 3.5736673238314914e-05,
+      "loss": 0.3668,
+      "step": 2848
+    },
+    {
+      "epoch": 5.453154875717018,
+      "grad_norm": 0.3359375,
+      "learning_rate": 3.5400386241103946e-05,
+      "loss": 0.3645,
+      "step": 2852
+    },
+    {
+      "epoch": 5.460803059273423,
+      "grad_norm": 0.33984375,
+      "learning_rate": 3.506547717044472e-05,
+      "loss": 0.348,
+      "step": 2856
+    },
+    {
+      "epoch": 5.468451242829828,
+      "grad_norm": 0.33984375,
+      "learning_rate": 3.473195005322776e-05,
+      "loss": 0.3812,
+      "step": 2860
+    },
+    {
+      "epoch": 5.4760994263862335,
+      "grad_norm": 0.333984375,
+      "learning_rate": 3.439980889972723e-05,
+      "loss": 0.3705,
+      "step": 2864
+    },
+    {
+      "epoch": 5.483747609942639,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.406905770355274e-05,
+      "loss": 0.3687,
+      "step": 2868
+    },
+    {
+      "epoch": 5.491395793499044,
+      "grad_norm": 0.32421875,
+      "learning_rate": 3.373970044160121e-05,
+      "loss": 0.4348,
+      "step": 2872
+    },
+    {
+      "epoch": 5.499043977055449,
+      "grad_norm": 0.36328125,
+      "learning_rate": 3.341174107400916e-05,
+      "loss": 0.4039,
+      "step": 2876
+    },
+    {
+      "epoch": 5.506692160611855,
+      "grad_norm": 0.3359375,
+      "learning_rate": 3.30851835441051e-05,
+      "loss": 0.4072,
+      "step": 2880
+    },
+    {
+      "epoch": 5.51434034416826,
+      "grad_norm": 0.353515625,
+      "learning_rate": 3.276003177836203e-05,
+      "loss": 0.3796,
+      "step": 2884
+    },
+    {
+      "epoch": 5.521988527724665,
+      "grad_norm": 0.345703125,
+      "learning_rate": 3.2436289686350285e-05,
+      "loss": 0.3647,
+      "step": 2888
+    },
+    {
+      "epoch": 5.5296367112810705,
+      "grad_norm": 0.341796875,
+      "learning_rate": 3.211396116069055e-05,
+      "loss": 0.3895,
+      "step": 2892
+    },
+    {
+      "epoch": 5.537284894837476,
+      "grad_norm": 0.34375,
+      "learning_rate": 3.179305007700697e-05,
+      "loss": 0.3689,
+      "step": 2896
+    },
+    {
+      "epoch": 5.544933078393882,
+      "grad_norm": 0.349609375,
+      "learning_rate": 3.147356029388067e-05,
+      "loss": 0.3782,
+      "step": 2900
+    },
+    {
+      "epoch": 5.552581261950287,
+      "grad_norm": 0.349609375,
+      "learning_rate": 3.115549565280325e-05,
+      "loss": 0.3985,
+      "step": 2904
+    },
+    {
+      "epoch": 5.5602294455066925,
+      "grad_norm": 0.341796875,
+      "learning_rate": 3.083885997813066e-05,
+      "loss": 0.4289,
+      "step": 2908
+    },
+    {
+      "epoch": 5.567877629063098,
+      "grad_norm": 0.33984375,
+      "learning_rate": 3.052365707703718e-05,
+      "loss": 0.436,
+      "step": 2912
+    },
+    {
+      "epoch": 5.575525812619503,
+      "grad_norm": 0.341796875,
+      "learning_rate": 3.0209890739469693e-05,
+      "loss": 0.387,
+      "step": 2916
+    },
+    {
+      "epoch": 5.583173996175908,
+      "grad_norm": 0.326171875,
+      "learning_rate": 2.989756473810203e-05,
+      "loss": 0.4034,
+      "step": 2920
+    },
+    {
+      "epoch": 5.590822179732314,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.9586682828289738e-05,
+      "loss": 0.4206,
+      "step": 2924
+    },
+    {
+      "epoch": 5.598470363288719,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.9277248748024763e-05,
+      "loss": 0.4529,
+      "step": 2928
+    },
+    {
+      "epoch": 5.606118546845124,
+      "grad_norm": 0.330078125,
+      "learning_rate": 2.8969266217890648e-05,
+      "loss": 0.3527,
+      "step": 2932
+    },
+    {
+      "epoch": 5.6137667304015295,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.866273894101776e-05,
+      "loss": 0.389,
+      "step": 2936
+    },
+    {
+      "epoch": 5.621414913957935,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.835767060303865e-05,
+      "loss": 0.4017,
+      "step": 2940
+    },
+    {
+      "epoch": 5.62906309751434,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.8054064872043917e-05,
+      "loss": 0.3973,
+      "step": 2944
+    },
+    {
+      "epoch": 5.636711281070745,
+      "grad_norm": 0.375,
+      "learning_rate": 2.7751925398537993e-05,
+      "loss": 0.4281,
+      "step": 2948
+    },
+    {
+      "epoch": 5.644359464627151,
+      "grad_norm": 0.330078125,
+      "learning_rate": 2.745125581539523e-05,
+      "loss": 0.4071,
+      "step": 2952
+    },
+    {
+      "epoch": 5.652007648183556,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.7152059737816395e-05,
+      "loss": 0.3866,
+      "step": 2956
+    },
+    {
+      "epoch": 5.659655831739962,
+      "grad_norm": 0.32421875,
+      "learning_rate": 2.6854340763284954e-05,
+      "loss": 0.4029,
+      "step": 2960
+    },
+    {
+      "epoch": 5.667304015296367,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.6558102471523975e-05,
+      "loss": 0.4207,
+      "step": 2964
+    },
+    {
+      "epoch": 5.674952198852773,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.6263348424453012e-05,
+      "loss": 0.3769,
+      "step": 2968
+    },
+    {
+      "epoch": 5.682600382409178,
+      "grad_norm": 0.337890625,
+      "learning_rate": 2.597008216614534e-05,
+      "loss": 0.3527,
+      "step": 2972
+    },
+    {
+      "epoch": 5.690248565965583,
+      "grad_norm": 0.345703125,
+      "learning_rate": 2.5678307222785315e-05,
+      "loss": 0.4091,
+      "step": 2976
+    },
+    {
+      "epoch": 5.6978967495219885,
+      "grad_norm": 0.328125,
+      "learning_rate": 2.5388027102625945e-05,
+      "loss": 0.4175,
+      "step": 2980
+    },
+    {
+      "epoch": 5.705544933078394,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.5099245295946764e-05,
+      "loss": 0.3557,
+      "step": 2984
+    },
+    {
+      "epoch": 5.713193116634799,
+      "grad_norm": 0.328125,
+      "learning_rate": 2.4811965275011825e-05,
+      "loss": 0.4021,
+      "step": 2988
+    },
+    {
+      "epoch": 5.720841300191204,
+      "grad_norm": 0.33203125,
+      "learning_rate": 2.4526190494027953e-05,
+      "loss": 0.3868,
+      "step": 2992
+    },
+    {
+      "epoch": 5.72848948374761,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.4241924389103227e-05,
+      "loss": 0.4399,
+      "step": 2996
+    },
+    {
+      "epoch": 5.736137667304015,
+      "grad_norm": 0.3359375,
+      "learning_rate": 2.395917037820566e-05,
+      "loss": 0.3907,
+      "step": 3000
+    },
+    {
+      "epoch": 5.743785850860421,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.3677931861122084e-05,
+      "loss": 0.4195,
+      "step": 3004
+    },
+    {
+      "epoch": 5.751434034416826,
+      "grad_norm": 0.3359375,
+      "learning_rate": 2.339821221941731e-05,
+      "loss": 0.3867,
+      "step": 3008
+    },
+    {
+      "epoch": 5.759082217973232,
+      "grad_norm": 0.3359375,
+      "learning_rate": 2.312001481639348e-05,
+      "loss": 0.3583,
+      "step": 3012
+    },
+    {
+      "epoch": 5.766730401529637,
+      "grad_norm": 0.3359375,
+      "learning_rate": 2.2843342997049445e-05,
+      "loss": 0.3527,
+      "step": 3016
+    },
+    {
+      "epoch": 5.774378585086042,
+      "grad_norm": 0.30859375,
+      "learning_rate": 2.2568200088040867e-05,
+      "loss": 0.3393,
+      "step": 3020
+    },
+    {
+      "epoch": 5.782026768642448,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.2294589397639978e-05,
+      "loss": 0.4225,
+      "step": 3024
+    },
+    {
+      "epoch": 5.789674952198853,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.2022514215695842e-05,
+      "loss": 0.4191,
+      "step": 3028
+    },
+    {
+      "epoch": 5.797323135755258,
+      "grad_norm": 0.341796875,
+      "learning_rate": 2.175197781359485e-05,
+      "loss": 0.3792,
+      "step": 3032
+    },
+    {
+      "epoch": 5.804971319311663,
+      "grad_norm": 0.337890625,
+      "learning_rate": 2.1482983444221402e-05,
+      "loss": 0.3942,
+      "step": 3036
+    },
+    {
+      "epoch": 5.812619502868069,
+      "grad_norm": 0.326171875,
+      "learning_rate": 2.1215534341918707e-05,
+      "loss": 0.3753,
+      "step": 3040
+    },
+    {
+      "epoch": 5.820267686424474,
+      "grad_norm": 0.33984375,
+      "learning_rate": 2.0949633722449915e-05,
+      "loss": 0.4234,
+      "step": 3044
+    },
+    {
+      "epoch": 5.827915869980879,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.0685284782959566e-05,
+      "loss": 0.3925,
+      "step": 3048
+    },
+    {
+      "epoch": 5.835564053537285,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.0422490701934996e-05,
+      "loss": 0.412,
+      "step": 3052
+    },
+    {
+      "epoch": 5.84321223709369,
+      "grad_norm": 0.328125,
+      "learning_rate": 2.0161254639168183e-05,
+      "loss": 0.3981,
+      "step": 3056
+    },
+    {
+      "epoch": 5.850860420650095,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.9901579735717743e-05,
+      "loss": 0.4204,
+      "step": 3060
+    },
+    {
+      "epoch": 5.858508604206501,
+      "grad_norm": 0.333984375,
+      "learning_rate": 1.964346911387127e-05,
+      "loss": 0.3878,
+      "step": 3064
+    },
+    {
+      "epoch": 5.866156787762907,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.9386925877107585e-05,
+      "loss": 0.4047,
+      "step": 3068
+    },
+    {
+      "epoch": 5.873804971319312,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.913195311005959e-05,
+      "loss": 0.3899,
+      "step": 3072
+    },
+    {
+      "epoch": 5.881453154875717,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.8878553878477105e-05,
+      "loss": 0.4179,
+      "step": 3076
+    },
+    {
+      "epoch": 5.8891013384321225,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.8626731229190016e-05,
+      "loss": 0.403,
+      "step": 3080
+    },
+    {
+      "epoch": 5.896749521988528,
+      "grad_norm": 0.31640625,
+      "learning_rate": 1.8376488190071666e-05,
+      "loss": 0.3925,
+      "step": 3084
+    },
+    {
+      "epoch": 5.904397705544933,
+      "grad_norm": 0.3359375,
+      "learning_rate": 1.8127827770002423e-05,
+      "loss": 0.3647,
+      "step": 3088
+    },
+    {
+      "epoch": 5.912045889101338,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.7880752958833543e-05,
+      "loss": 0.4301,
+      "step": 3092
+    },
+    {
+      "epoch": 5.919694072657744,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.7635266727351092e-05,
+      "loss": 0.3988,
+      "step": 3096
+    },
+    {
+      "epoch": 5.927342256214149,
+      "grad_norm": 0.34375,
+      "learning_rate": 1.73913720272404e-05,
+      "loss": 0.4028,
+      "step": 3100
+    },
+    {
+      "epoch": 5.934990439770554,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.714907179105049e-05,
+      "loss": 0.3756,
+      "step": 3104
+    },
+    {
+      "epoch": 5.9426386233269595,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.6908368932158777e-05,
+      "loss": 0.4023,
+      "step": 3108
+    },
+    {
+      "epoch": 5.950286806883366,
+      "grad_norm": 0.345703125,
+      "learning_rate": 1.6669266344736104e-05,
+      "loss": 0.3784,
+      "step": 3112
+    },
+    {
+      "epoch": 5.957934990439771,
+      "grad_norm": 0.34375,
+      "learning_rate": 1.6431766903711914e-05,
+      "loss": 0.3622,
+      "step": 3116
+    },
+    {
+      "epoch": 5.965583173996176,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.6195873464739702e-05,
+      "loss": 0.416,
+      "step": 3120
+    },
+    {
+      "epoch": 5.9732313575525815,
+      "grad_norm": 0.3359375,
+      "learning_rate": 1.5961588864162627e-05,
+      "loss": 0.4191,
+      "step": 3124
+    },
+    {
+      "epoch": 5.980879541108987,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.5728915918979477e-05,
+      "loss": 0.3683,
+      "step": 3128
+    },
+    {
+      "epoch": 5.988527724665392,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.5497857426810756e-05,
+      "loss": 0.365,
+      "step": 3132
+    },
+    {
+      "epoch": 5.996175908221797,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.5268416165865055e-05,
+      "loss": 0.401,
+      "step": 3136
+    },
+    {
+      "epoch": 6.003824091778203,
+      "grad_norm": 0.33203125,
+      "learning_rate": 1.5040594894905628e-05,
+      "loss": 0.3805,
+      "step": 3140
+    },
+    {
+      "epoch": 6.011472275334608,
+      "grad_norm": 0.34765625,
+      "learning_rate": 1.481439635321729e-05,
+      "loss": 0.3635,
+      "step": 3144
+    },
+    {
+      "epoch": 6.019120458891013,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.458982326057338e-05,
+      "loss": 0.37,
+      "step": 3148
+    },
+    {
+      "epoch": 6.0267686424474185,
+      "grad_norm": 0.34765625,
+      "learning_rate": 1.436687831720314e-05,
+      "loss": 0.3981,
+      "step": 3152
+    },
+    {
+      "epoch": 6.034416826003824,
+      "grad_norm": 0.345703125,
+      "learning_rate": 1.4145564203759219e-05,
+      "loss": 0.4484,
+      "step": 3156
+    },
+    {
+      "epoch": 6.042065009560229,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.3925883581285401e-05,
+      "loss": 0.3988,
+      "step": 3160
+    },
+    {
+      "epoch": 6.049713193116634,
+      "grad_norm": 0.357421875,
+      "learning_rate": 1.3707839091184702e-05,
+      "loss": 0.3422,
+      "step": 3164
+    },
+    {
+      "epoch": 6.0573613766730405,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.349143335518752e-05,
+      "loss": 0.3994,
+      "step": 3168
+    },
+    {
+      "epoch": 6.065009560229446,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.3276668975320165e-05,
+      "loss": 0.4038,
+      "step": 3172
+    },
+    {
+      "epoch": 6.072657743785851,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.3063548533873536e-05,
+      "loss": 0.4065,
+      "step": 3176
+    },
+    {
+      "epoch": 6.080305927342256,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.2852074593372142e-05,
+      "loss": 0.3902,
+      "step": 3180
+    },
+    {
+      "epoch": 6.087954110898662,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.2642249696543178e-05,
+      "loss": 0.388,
+      "step": 3184
+    },
+    {
+      "epoch": 6.095602294455067,
+      "grad_norm": 0.34375,
+      "learning_rate": 1.243407636628605e-05,
+      "loss": 0.3979,
+      "step": 3188
+    },
+    {
+      "epoch": 6.103250478011472,
+      "grad_norm": 0.3359375,
+      "learning_rate": 1.2227557105642e-05,
+      "loss": 0.404,
+      "step": 3192
+    },
+    {
+      "epoch": 6.1108986615678775,
+      "grad_norm": 0.34765625,
+      "learning_rate": 1.2022694397763993e-05,
+      "loss": 0.4059,
+      "step": 3196
+    },
+    {
+      "epoch": 6.118546845124283,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.1819490705886914e-05,
+      "loss": 0.3766,
+      "step": 3200
+    },
+    {
+      "epoch": 6.126195028680688,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.16179484732979e-05,
+      "loss": 0.3629,
+      "step": 3204
+    },
+    {
+      "epoch": 6.133843212237093,
+      "grad_norm": 0.375,
+      "learning_rate": 1.1418070123306989e-05,
+      "loss": 0.3733,
+      "step": 3208
+    },
+    {
+      "epoch": 6.141491395793499,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.1219858059217951e-05,
+      "loss": 0.4169,
+      "step": 3212
+    },
+    {
+      "epoch": 6.149139579349905,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.1023314664299455e-05,
+      "loss": 0.4131,
+      "step": 3216
+    },
+    {
+      "epoch": 6.15678776290631,
+      "grad_norm": 0.318359375,
+      "learning_rate": 1.0828442301756312e-05,
+      "loss": 0.3505,
+      "step": 3220
+    },
+    {
+      "epoch": 6.164435946462715,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.0635243314701163e-05,
+      "loss": 0.3978,
+      "step": 3224
+    },
+    {
+      "epoch": 6.172084130019121,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.0443720026126273e-05,
+      "loss": 0.3982,
+      "step": 3228
+    },
+    {
+      "epoch": 6.179732313575526,
+      "grad_norm": 0.357421875,
+      "learning_rate": 1.025387473887554e-05,
+      "loss": 0.3935,
+      "step": 3232
+    },
+    {
+      "epoch": 6.187380497131931,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.0065709735616917e-05,
+      "loss": 0.3709,
+      "step": 3236
+    },
+    {
+      "epoch": 6.195028680688337,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.87922727881484e-06,
+      "loss": 0.3922,
+      "step": 3240
+    },
+    {
+      "epoch": 6.202676864244742,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.694429610703153e-06,
+      "loss": 0.3719,
+      "step": 3244
+    },
+    {
+      "epoch": 6.210325047801147,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.511318953258013e-06,
+      "loss": 0.2995,
+      "step": 3248
+    },
+    {
+      "epoch": 6.217973231357552,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.329897508171296e-06,
+      "loss": 0.3932,
+      "step": 3252
+    },
+    {
+      "epoch": 6.225621414913958,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.150167456824065e-06,
+      "loss": 0.4016,
+      "step": 3256
+    },
+    {
+      "epoch": 6.233269598470363,
+      "grad_norm": 0.33984375,
+      "learning_rate": 8.972130960260326e-06,
+      "loss": 0.3622,
+      "step": 3260
+    },
+    {
+      "epoch": 6.240917782026768,
+      "grad_norm": 0.337890625,
+      "learning_rate": 8.795790159161098e-06,
+      "loss": 0.4053,
+      "step": 3264
+    },
+    {
+      "epoch": 6.248565965583174,
+      "grad_norm": 0.3203125,
+      "learning_rate": 8.621147173818587e-06,
+      "loss": 0.3517,
+      "step": 3268
+    },
+    {
+      "epoch": 6.25621414913958,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.448204104110818e-06,
+      "loss": 0.3448,
+      "step": 3272
+    },
+    {
+      "epoch": 6.263862332695985,
+      "grad_norm": 0.349609375,
+      "learning_rate": 8.276963029476275e-06,
+      "loss": 0.3758,
+      "step": 3276
+    },
+    {
+      "epoch": 6.27151051625239,
+      "grad_norm": 0.37109375,
+      "learning_rate": 8.107426008888934e-06,
+      "loss": 0.3911,
+      "step": 3280
+    },
+    {
+      "epoch": 6.279158699808796,
+      "grad_norm": 0.322265625,
+      "learning_rate": 7.93959508083351e-06,
+      "loss": 0.3491,
+      "step": 3284
+    },
+    {
+      "epoch": 6.286806883365201,
+      "grad_norm": 0.34375,
+      "learning_rate": 7.773472263280977e-06,
+      "loss": 0.3954,
+      "step": 3288
+    },
+    {
+      "epoch": 6.294455066921606,
+      "grad_norm": 0.34375,
+      "learning_rate": 7.609059553664254e-06,
+      "loss": 0.4018,
+      "step": 3292
+    },
+    {
+      "epoch": 6.3021032504780115,
+      "grad_norm": 0.330078125,
+      "learning_rate": 7.446358928854207e-06,
+      "loss": 0.3823,
+      "step": 3296
+    },
+    {
+      "epoch": 6.309751434034417,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.2853723451358705e-06,
+      "loss": 0.4097,
+      "step": 3300
+    },
+    {
+      "epoch": 6.317399617590822,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.126101738184964e-06,
+      "loss": 0.3676,
+      "step": 3304
+    },
+    {
+      "epoch": 6.325047801147227,
+      "grad_norm": 0.337890625,
+      "learning_rate": 6.9685490230445615e-06,
+      "loss": 0.3901,
+      "step": 3308
+    },
+    {
+      "epoch": 6.332695984703633,
+      "grad_norm": 0.32421875,
+      "learning_rate": 6.812716094102128e-06,
+      "loss": 0.3652,
+      "step": 3312
+    },
+    {
+      "epoch": 6.340344168260038,
+      "grad_norm": 0.35546875,
+      "learning_rate": 6.658604825066683e-06,
+      "loss": 0.3631,
+      "step": 3316
+    },
+    {
+      "epoch": 6.347992351816444,
+      "grad_norm": 0.341796875,
+      "learning_rate": 6.50621706894629e-06,
+      "loss": 0.4053,
+      "step": 3320
+    },
+    {
+      "epoch": 6.355640535372849,
+      "grad_norm": 0.33203125,
+      "learning_rate": 6.355554658025791e-06,
+      "loss": 0.3451,
+      "step": 3324
+    },
+    {
+      "epoch": 6.363288718929255,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.206619403844804e-06,
+      "loss": 0.4109,
+      "step": 3328
+    },
+    {
+      "epoch": 6.37093690248566,
+      "grad_norm": 0.318359375,
+      "learning_rate": 6.059413097175808e-06,
+      "loss": 0.3897,
+      "step": 3332
+    },
+    {
+      "epoch": 6.378585086042065,
+      "grad_norm": 0.341796875,
+      "learning_rate": 5.913937508002797e-06,
+      "loss": 0.3573,
+      "step": 3336
+    },
+    {
+      "epoch": 6.3862332695984705,
+      "grad_norm": 0.341796875,
+      "learning_rate": 5.770194385499877e-06,
+      "loss": 0.3928,
+      "step": 3340
+    },
+    {
+      "epoch": 6.393881453154876,
+      "grad_norm": 0.330078125,
+      "learning_rate": 5.628185458010248e-06,
+      "loss": 0.3662,
+      "step": 3344
+    },
+    {
+      "epoch": 6.401529636711281,
+      "grad_norm": 0.349609375,
+      "learning_rate": 5.487912433025493e-06,
+      "loss": 0.3974,
+      "step": 3348
+    },
+    {
+      "epoch": 6.409177820267686,
+      "grad_norm": 0.32421875,
+      "learning_rate": 5.349376997164923e-06,
+      "loss": 0.3477,
+      "step": 3352
+    },
+    {
+      "epoch": 6.416826003824092,
+      "grad_norm": 0.330078125,
+      "learning_rate": 5.212580816155426e-06,
+      "loss": 0.3992,
+      "step": 3356
+    },
+    {
+      "epoch": 6.424474187380497,
+      "grad_norm": 0.345703125,
+      "learning_rate": 5.077525534811339e-06,
+      "loss": 0.3881,
+      "step": 3360
+    },
+    {
+      "epoch": 6.432122370936902,
+      "grad_norm": 0.322265625,
+      "learning_rate": 4.9442127770147385e-06,
+      "loss": 0.3767,
+      "step": 3364
+    },
+    {
+      "epoch": 6.4397705544933075,
+      "grad_norm": 0.34765625,
+      "learning_rate": 4.812644145695915e-06,
+      "loss": 0.3939,
+      "step": 3368
+    },
+    {
+      "epoch": 6.447418738049713,
+      "grad_norm": 0.3359375,
+      "learning_rate": 4.682821222813998e-06,
+      "loss": 0.3322,
+      "step": 3372
+    },
+    {
+      "epoch": 6.455066921606119,
+      "grad_norm": 0.34375,
+      "learning_rate": 4.554745569338092e-06,
+      "loss": 0.3414,
+      "step": 3376
+    },
+    {
+      "epoch": 6.462715105162524,
+      "grad_norm": 0.35546875,
+      "learning_rate": 4.428418725228372e-06,
+      "loss": 0.3982,
+      "step": 3380
+    },
+    {
+      "epoch": 6.4703632887189295,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.303842209417652e-06,
+      "loss": 0.3522,
+      "step": 3384
+    },
+    {
+      "epoch": 6.478011472275335,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.181017519793079e-06,
+      "loss": 0.4167,
+      "step": 3388
+    },
+    {
+      "epoch": 6.48565965583174,
+      "grad_norm": 0.349609375,
+      "learning_rate": 4.059946133178132e-06,
+      "loss": 0.3875,
+      "step": 3392
+    },
+    {
+      "epoch": 6.493307839388145,
+      "grad_norm": 0.33203125,
+      "learning_rate": 3.94062950531489e-06,
+      "loss": 0.3662,
+      "step": 3396
+    },
+    {
+      "epoch": 6.500956022944551,
+      "grad_norm": 0.34375,
+      "learning_rate": 3.823069070846474e-06,
+      "loss": 0.3625,
+      "step": 3400
+    },
+    {
+      "epoch": 6.508604206500956,
+      "grad_norm": 0.318359375,
+      "learning_rate": 3.707266243299861e-06,
+      "loss": 0.3445,
+      "step": 3404
+    },
+    {
+      "epoch": 6.516252390057361,
+      "grad_norm": 0.353515625,
+      "learning_rate": 3.5932224150688526e-06,
+      "loss": 0.3705,
+      "step": 3408
+    },
+    {
+      "epoch": 6.5239005736137665,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.4809389573973e-06,
+      "loss": 0.396,
+      "step": 3412
+    },
+    {
+      "epoch": 6.531548757170172,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.3704172203627035e-06,
+      "loss": 0.3689,
+      "step": 3416
+    },
+    {
+      "epoch": 6.539196940726577,
+      "grad_norm": 0.33203125,
+      "learning_rate": 3.2616585328599065e-06,
+      "loss": 0.4253,
+      "step": 3420
+    },
+    {
+      "epoch": 6.546845124282983,
+      "grad_norm": 0.3359375,
+      "learning_rate": 3.154664202585128e-06,
+      "loss": 0.387,
+      "step": 3424
+    },
+    {
+      "epoch": 6.5544933078393885,
+      "grad_norm": 0.353515625,
+      "learning_rate": 3.049435516020271e-06,
+      "loss": 0.3585,
+      "step": 3428
+    },
+    {
+      "epoch": 6.562141491395794,
+      "grad_norm": 0.34765625,
+      "learning_rate": 2.94597373841744e-06,
+      "loss": 0.3912,
+      "step": 3432
+    },
+    {
+      "epoch": 6.569789674952199,
+      "grad_norm": 0.337890625,
+      "learning_rate": 2.844280113783698e-06,
+      "loss": 0.3863,
+      "step": 3436
+    },
+    {
+      "epoch": 6.577437858508604,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.7443558648661656e-06,
+      "loss": 0.4214,
+      "step": 3440
+    },
+    {
+      "epoch": 6.58508604206501,
+      "grad_norm": 0.33203125,
+      "learning_rate": 2.646202193137248e-06,
+      "loss": 0.429,
+      "step": 3444
+    },
+    {
+      "epoch": 6.592734225621415,
+      "grad_norm": 0.330078125,
+      "learning_rate": 2.549820278780246e-06,
+      "loss": 0.3523,
+      "step": 3448
+    },
+    {
+      "epoch": 6.60038240917782,
+      "grad_norm": 0.33984375,
+      "learning_rate": 2.455211280675168e-06,
+      "loss": 0.4001,
+      "step": 3452
+    },
+    {
+      "epoch": 6.6080305927342256,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.3623763363847246e-06,
+      "loss": 0.3819,
+      "step": 3456
+    },
+    {
+      "epoch": 6.615678776290631,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.271316562140757e-06,
+      "loss": 0.3632,
+      "step": 3460
+    },
+    {
+      "epoch": 6.623326959847036,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.182033052830695e-06,
+      "loss": 0.3842,
+      "step": 3464
+    },
+    {
+      "epoch": 6.630975143403441,
+      "grad_norm": 0.34375,
+      "learning_rate": 2.094526881984521e-06,
+      "loss": 0.3844,
+      "step": 3468
+    },
+    {
+      "epoch": 6.638623326959847,
+      "grad_norm": 0.328125,
+      "learning_rate": 2.0087991017617598e-06,
+      "loss": 0.3697,
+      "step": 3472
+    },
+    {
+      "epoch": 6.646271510516252,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.924850742938894e-06,
+      "loss": 0.3886,
+      "step": 3476
+    },
+    {
+      "epoch": 6.653919694072657,
+      "grad_norm": 0.349609375,
+      "learning_rate": 1.8426828148969008e-06,
+      "loss": 0.363,
+      "step": 3480
+    },
+    {
+      "epoch": 6.661567877629063,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.7622963056091843e-06,
+      "loss": 0.3671,
+      "step": 3484
+    },
+    {
+      "epoch": 6.669216061185469,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.6836921816296644e-06,
+      "loss": 0.4043,
+      "step": 3488
+    },
+    {
+      "epoch": 6.676864244741874,
+      "grad_norm": 0.33203125,
+      "learning_rate": 1.6068713880811546e-06,
+      "loss": 0.398,
+      "step": 3492
+    },
+    {
+      "epoch": 6.684512428298279,
+      "grad_norm": 0.330078125,
+      "learning_rate": 1.531834848643987e-06,
+      "loss": 0.3847,
+      "step": 3496
+    },
+    {
+      "epoch": 6.692160611854685,
+      "grad_norm": 0.34765625,
+      "learning_rate": 1.4585834655449547e-06,
+      "loss": 0.4041,
+      "step": 3500
+    }
+  ],
+  "logging_steps": 4,
+  "max_steps": 3661,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 7,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.2872303824598467e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}