diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14908 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.678410117434508,
+  "eval_steps": 500,
+  "global_step": 8500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0036133694670280035,
+      "grad_norm": 0.294921875,
+      "learning_rate": 1.3483146067415729e-05,
+      "loss": 1.3019,
+      "step": 4
+    },
+    {
+      "epoch": 0.007226738934056007,
+      "grad_norm": 0.2890625,
+      "learning_rate": 2.6966292134831458e-05,
+      "loss": 1.3449,
+      "step": 8
+    },
+    {
+      "epoch": 0.01084010840108401,
+      "grad_norm": 0.23046875,
+      "learning_rate": 4.0449438202247185e-05,
+      "loss": 1.2588,
+      "step": 12
+    },
+    {
+      "epoch": 0.014453477868112014,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 5.3932584269662916e-05,
+      "loss": 1.2878,
+      "step": 16
+    },
+    {
+      "epoch": 0.018066847335140017,
+      "grad_norm": 0.240234375,
+      "learning_rate": 6.741573033707865e-05,
+      "loss": 1.2136,
+      "step": 20
+    },
+    {
+      "epoch": 0.02168021680216802,
+      "grad_norm": 0.212890625,
+      "learning_rate": 8.089887640449437e-05,
+      "loss": 1.2275,
+      "step": 24
+    },
+    {
+      "epoch": 0.025293586269196026,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 9.438202247191011e-05,
+      "loss": 1.2667,
+      "step": 28
+    },
+    {
+      "epoch": 0.028906955736224028,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.00010786516853932583,
+      "loss": 1.1905,
+      "step": 32
+    },
+    {
+      "epoch": 0.032520325203252036,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00012134831460674156,
+      "loss": 1.2345,
+      "step": 36
+    },
+    {
+      "epoch": 0.036133694670280034,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0001348314606741573,
+      "loss": 1.2052,
+      "step": 40
+    },
+    {
+      "epoch": 0.03974706413730804,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00014831460674157303,
+      "loss": 1.146,
+      "step": 44
+    },
+    {
+      "epoch": 0.04336043360433604,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00016179775280898874,
+      "loss": 1.1099,
+      "step": 48
+    },
+    {
+      "epoch": 0.04697380307136405,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.00017528089887640448,
+      "loss": 1.1519,
+      "step": 52
+    },
+    {
+      "epoch": 0.05058717253839205,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.00018876404494382021,
+      "loss": 1.1701,
+      "step": 56
+    },
+    {
+      "epoch": 0.05420054200542006,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.00020224719101123593,
+      "loss": 1.1546,
+      "step": 60
+    },
+    {
+      "epoch": 0.057813911472448055,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.00021573033707865166,
+      "loss": 1.1587,
+      "step": 64
+    },
+    {
+      "epoch": 0.06142728093947606,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0002292134831460674,
+      "loss": 1.1414,
+      "step": 68
+    },
+    {
+      "epoch": 0.06504065040650407,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0002426966292134831,
+      "loss": 1.1381,
+      "step": 72
+    },
+    {
+      "epoch": 0.06865401987353206,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0002561797752808988,
+      "loss": 1.121,
+      "step": 76
+    },
+    {
+      "epoch": 0.07226738934056007,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0002696629213483146,
+      "loss": 1.053,
+      "step": 80
+    },
+    {
+      "epoch": 0.07588075880758807,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0002831460674157303,
+      "loss": 1.1037,
+      "step": 84
+    },
+    {
+      "epoch": 0.07949412827461608,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.00029662921348314606,
+      "loss": 1.127,
+      "step": 88
+    },
+    {
+      "epoch": 0.08310749774164408,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0002999999133234806,
+      "loss": 1.0572,
+      "step": 92
+    },
+    {
+      "epoch": 0.08672086720867209,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.00029999952809470757,
+      "loss": 1.1133,
+      "step": 96
+    },
+    {
+      "epoch": 0.09033423667570009,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.00029999883468374714,
+      "loss": 1.1132,
+      "step": 100
+    },
+    {
+      "epoch": 0.0939476061427281,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.000299997833092024,
+      "loss": 1.0945,
+      "step": 104
+    },
+    {
+      "epoch": 0.0975609756097561,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00029999652332159593,
+      "loss": 1.1265,
+      "step": 108
+    },
+    {
+      "epoch": 0.1011743450767841,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00029999490537515396,
+      "loss": 1.0732,
+      "step": 112
+    },
+    {
+      "epoch": 0.10478771454381211,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.00029999297925602226,
+      "loss": 1.1023,
+      "step": 116
+    },
+    {
+      "epoch": 0.10840108401084012,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.00029999074496815805,
+      "loss": 1.0428,
+      "step": 120
+    },
+    {
+      "epoch": 0.1120144534778681,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.00029998820251615196,
+      "loss": 1.0927,
+      "step": 124
+    },
+    {
+      "epoch": 0.11562782294489611,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.00029998535190522746,
+      "loss": 1.0348,
+      "step": 128
+    },
+    {
+      "epoch": 0.11924119241192412,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.00029998219314124135,
+      "loss": 1.0602,
+      "step": 132
+    },
+    {
+      "epoch": 0.12285456187895212,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00029997872623068345,
+      "loss": 1.0567,
+      "step": 136
+    },
+    {
+      "epoch": 0.12646793134598014,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00029997495118067676,
+      "loss": 1.0755,
+      "step": 140
+    },
+    {
+      "epoch": 0.13008130081300814,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.00029997086799897726,
+      "loss": 1.0485,
+      "step": 144
+    },
+    {
+      "epoch": 0.13369467028003612,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0002999664766939741,
+      "loss": 1.0222,
+      "step": 148
+    },
+    {
+      "epoch": 0.13730803974706413,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.00029996177727468945,
+      "loss": 1.0111,
+      "step": 152
+    },
+    {
+      "epoch": 0.14092140921409213,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.00029995676975077857,
+      "loss": 1.0359,
+      "step": 156
+    },
+    {
+      "epoch": 0.14453477868112014,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0002999514541325296,
+      "loss": 1.005,
+      "step": 160
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.00029994583043086386,
+      "loss": 1.0534,
+      "step": 164
+    },
+    {
+      "epoch": 0.15176151761517614,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.00029993989865733555,
+      "loss": 1.0187,
+      "step": 168
+    },
+    {
+      "epoch": 0.15537488708220415,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0002999336588241317,
+      "loss": 1.0295,
+      "step": 172
+    },
+    {
+      "epoch": 0.15898825654923215,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00029992711094407247,
+      "loss": 1.043,
+      "step": 176
+    },
+    {
+      "epoch": 0.16260162601626016,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0002999202550306109,
+      "loss": 1.035,
+      "step": 180
+    },
+    {
+      "epoch": 0.16621499548328816,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0002999130910978327,
+      "loss": 0.9977,
+      "step": 184
+    },
+    {
+      "epoch": 0.16982836495031617,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0002999056191604567,
+      "loss": 0.9963,
+      "step": 188
+    },
+    {
+      "epoch": 0.17344173441734417,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0002998978392338342,
+      "loss": 1.0241,
+      "step": 192
+    },
+    {
+      "epoch": 0.17705510388437218,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0002998897513339497,
+      "loss": 1.0046,
+      "step": 196
+    },
+    {
+      "epoch": 0.18066847335140018,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0002998813554774202,
+      "loss": 1.0136,
+      "step": 200
+    },
+    {
+      "epoch": 0.1842818428184282,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0002998726516814953,
+      "loss": 0.9941,
+      "step": 204
+    },
+    {
+      "epoch": 0.1878952122854562,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.00029986363996405754,
+      "loss": 0.9964,
+      "step": 208
+    },
+    {
+      "epoch": 0.1915085817524842,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.00029985432034362197,
+      "loss": 0.9962,
+      "step": 212
+    },
+    {
+      "epoch": 0.1951219512195122,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002998446928393362,
+      "loss": 1.0432,
+      "step": 216
+    },
+    {
+      "epoch": 0.1987353206865402,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0002998347574709805,
+      "loss": 0.9932,
+      "step": 220
+    },
+    {
+      "epoch": 0.2023486901535682,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0002998245142589677,
+      "loss": 0.9796,
+      "step": 224
+    },
+    {
+      "epoch": 0.20596205962059622,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00029981396322434285,
+      "loss": 0.9699,
+      "step": 228
+    },
+    {
+      "epoch": 0.20957542908762422,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0002998031043887838,
+      "loss": 0.9858,
+      "step": 232
+    },
+    {
+      "epoch": 0.21318879855465223,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.00029979193777460053,
+      "loss": 0.968,
+      "step": 236
+    },
+    {
+      "epoch": 0.21680216802168023,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0002997804634047355,
+      "loss": 0.9406,
+      "step": 240
+    },
+    {
+      "epoch": 0.2204155374887082,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002997686813027634,
+      "loss": 0.9795,
+      "step": 244
+    },
+    {
+      "epoch": 0.2240289069557362,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002997565914928912,
+      "loss": 0.9785,
+      "step": 248
+    },
+    {
+      "epoch": 0.22764227642276422,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.00029974419399995814,
+      "loss": 0.9495,
+      "step": 252
+    },
+    {
+      "epoch": 0.23125564588979222,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0002997314888494354,
+      "loss": 0.9771,
+      "step": 256
+    },
+    {
+      "epoch": 0.23486901535682023,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00029971847606742645,
+      "loss": 0.9933,
+      "step": 260
+    },
+    {
+      "epoch": 0.23848238482384823,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002997051556806669,
+      "loss": 0.9808,
+      "step": 264
+    },
+    {
+      "epoch": 0.24209575429087624,
+      "grad_norm": 0.25,
+      "learning_rate": 0.000299691527716524,
+      "loss": 0.9665,
+      "step": 268
+    },
+    {
+      "epoch": 0.24570912375790424,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00029967759220299734,
+      "loss": 0.9489,
+      "step": 272
+    },
+    {
+      "epoch": 0.24932249322493225,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00029966334916871806,
+      "loss": 0.924,
+      "step": 276
+    },
+    {
+      "epoch": 0.2529358626919603,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.00029964879864294926,
+      "loss": 0.9947,
+      "step": 280
+    },
+    {
+      "epoch": 0.2565492321589883,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002996339406555858,
+      "loss": 0.9652,
+      "step": 284
+    },
+    {
+      "epoch": 0.2601626016260163,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0002996187752371543,
+      "loss": 0.9315,
+      "step": 288
+    },
+    {
+      "epoch": 0.26377597109304424,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002996033024188128,
+      "loss": 0.972,
+      "step": 292
+    },
+    {
+      "epoch": 0.26738934056007224,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00029958752223235114,
+      "loss": 0.9273,
+      "step": 296
+    },
+    {
+      "epoch": 0.27100271002710025,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00029957143471019053,
+      "loss": 0.9286,
+      "step": 300
+    },
+    {
+      "epoch": 0.27461607949412825,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002995550398853837,
+      "loss": 0.949,
+      "step": 304
+    },
+    {
+      "epoch": 0.27822944896115626,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0002995383377916147,
+      "loss": 0.9447,
+      "step": 308
+    },
+    {
+      "epoch": 0.28184281842818426,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002995213284631989,
+      "loss": 0.9322,
+      "step": 312
+    },
+    {
+      "epoch": 0.28545618789521227,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002995040119350828,
+      "loss": 0.9584,
+      "step": 316
+    },
+    {
+      "epoch": 0.28906955736224027,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0002994863882428442,
+      "loss": 0.9464,
+      "step": 320
+    },
+    {
+      "epoch": 0.2926829268292683,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.000299468457422692,
+      "loss": 0.8938,
+      "step": 324
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00029945021951146596,
+      "loss": 0.9366,
+      "step": 328
+    },
+    {
+      "epoch": 0.2999096657633243,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00029943167454663693,
+      "loss": 0.9055,
+      "step": 332
+    },
+    {
+      "epoch": 0.3035230352303523,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0002994128225663065,
+      "loss": 0.9295,
+      "step": 336
+    },
+    {
+      "epoch": 0.3071364046973803,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0002993936636092071,
+      "loss": 0.9431,
+      "step": 340
+    },
+    {
+      "epoch": 0.3107497741644083,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00029937419771470185,
+      "loss": 0.9077,
+      "step": 344
+    },
+    {
+      "epoch": 0.3143631436314363,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002993544249227845,
+      "loss": 0.8974,
+      "step": 348
+    },
+    {
+      "epoch": 0.3179765130984643,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00029933434527407924,
+      "loss": 0.9345,
+      "step": 352
+    },
+    {
+      "epoch": 0.3215898825654923,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.000299313958809841,
+      "loss": 0.9295,
+      "step": 356
+    },
+    {
+      "epoch": 0.3252032520325203,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0002992932655719547,
+      "loss": 0.9544,
+      "step": 360
+    },
+    {
+      "epoch": 0.3288166214995483,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002992722656029357,
+      "loss": 0.904,
+      "step": 364
+    },
+    {
+      "epoch": 0.3324299909665763,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00029925095894592975,
+      "loss": 0.8913,
+      "step": 368
+    },
+    {
+      "epoch": 0.33604336043360433,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.00029922934564471243,
+      "loss": 0.9099,
+      "step": 372
+    },
+    {
+      "epoch": 0.33965672990063234,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002992074257436894,
+      "loss": 0.9094,
+      "step": 376
+    },
+    {
+      "epoch": 0.34327009936766034,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002991851992878964,
+      "loss": 0.9316,
+      "step": 380
+    },
+    {
+      "epoch": 0.34688346883468835,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00029916266632299884,
+      "loss": 0.8649,
+      "step": 384
+    },
+    {
+      "epoch": 0.35049683830171635,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002991398268952919,
+      "loss": 0.933,
+      "step": 388
+    },
+    {
+      "epoch": 0.35411020776874436,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00029911668105170043,
+      "loss": 0.9069,
+      "step": 392
+    },
+    {
+      "epoch": 0.35772357723577236,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00029909322883977884,
+      "loss": 0.8697,
+      "step": 396
+    },
+    {
+      "epoch": 0.36133694670280037,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.00029906947030771097,
+      "loss": 0.875,
+      "step": 400
+    },
+    {
+      "epoch": 0.36495031616982837,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.00029904540550431004,
+      "loss": 0.9392,
+      "step": 404
+    },
+    {
+      "epoch": 0.3685636856368564,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.00029902103447901844,
+      "loss": 0.8518,
+      "step": 408
+    },
+    {
+      "epoch": 0.3721770551038844,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.00029899635728190783,
+      "loss": 0.9312,
+      "step": 412
+    },
+    {
+      "epoch": 0.3757904245709124,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002989713739636788,
+      "loss": 0.9395,
+      "step": 416
+    },
+    {
+      "epoch": 0.3794037940379404,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.000298946084575661,
+      "loss": 0.9087,
+      "step": 420
+    },
+    {
+      "epoch": 0.3830171635049684,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00029892048916981283,
+      "loss": 0.8754,
+      "step": 424
+    },
+    {
+      "epoch": 0.3866305329719964,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00029889458779872133,
+      "loss": 0.8802,
+      "step": 428
+    },
+    {
+      "epoch": 0.3902439024390244,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00029886838051560247,
+      "loss": 0.925,
+      "step": 432
+    },
+    {
+      "epoch": 0.3938572719060524,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0002988418673743004,
+      "loss": 0.9213,
+      "step": 436
+    },
+    {
+      "epoch": 0.3974706413730804,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00029881504842928783,
+      "loss": 0.8926,
+      "step": 440
+    },
+    {
+      "epoch": 0.4010840108401084,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00029878792373566577,
+      "loss": 0.8691,
+      "step": 444
+    },
+    {
+      "epoch": 0.4046973803071364,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0002987604933491634,
+      "loss": 0.8952,
+      "step": 448
+    },
+    {
+      "epoch": 0.4083107497741644,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002987327573261379,
+      "loss": 0.8955,
+      "step": 452
+    },
+    {
+      "epoch": 0.41192411924119243,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0002987047157235745,
+      "loss": 0.8932,
+      "step": 456
+    },
+    {
+      "epoch": 0.41553748870822044,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002986763685990861,
+      "loss": 0.8703,
+      "step": 460
+    },
+    {
+      "epoch": 0.41915085817524844,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002986477160109134,
+      "loss": 0.8803,
+      "step": 464
+    },
+    {
+      "epoch": 0.42276422764227645,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0002986187580179248,
+      "loss": 0.9077,
+      "step": 468
+    },
+    {
+      "epoch": 0.42637759710930445,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00029858949467961605,
+      "loss": 0.8915,
+      "step": 472
+    },
+    {
+      "epoch": 0.42999096657633246,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002985599260561102,
+      "loss": 0.826,
+      "step": 476
+    },
+    {
+      "epoch": 0.43360433604336046,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002985300522081577,
+      "loss": 0.8754,
+      "step": 480
+    },
+    {
+      "epoch": 0.4372177055103884,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002984998731971359,
+      "loss": 0.8972,
+      "step": 484
+    },
+    {
+      "epoch": 0.4408310749774164,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00029846938908504927,
+      "loss": 0.8418,
+      "step": 488
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002984385999345291,
+      "loss": 0.8761,
+      "step": 492
+    },
+    {
+      "epoch": 0.4480578139114724,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.00029840750580883323,
+      "loss": 0.8642,
+      "step": 496
+    },
+    {
+      "epoch": 0.45167118337850043,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002983761067718464,
+      "loss": 0.8763,
+      "step": 500
+    },
+    {
+      "epoch": 0.45528455284552843,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002983444028880796,
+      "loss": 0.8694,
+      "step": 504
+    },
+    {
+      "epoch": 0.45889792231255644,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0002983123942226701,
+      "loss": 0.8567,
+      "step": 508
+    },
+    {
+      "epoch": 0.46251129177958444,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00029828008084138155,
+      "loss": 0.8337,
+      "step": 512
+    },
+    {
+      "epoch": 0.46612466124661245,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0002982474628106035,
+      "loss": 0.9067,
+      "step": 516
+    },
+    {
+      "epoch": 0.46973803071364045,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00029821454019735147,
+      "loss": 0.8444,
+      "step": 520
+    },
+    {
+      "epoch": 0.47335140018066846,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00029818131306926676,
+      "loss": 0.8703,
+      "step": 524
+    },
+    {
+      "epoch": 0.47696476964769646,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002981477814946163,
+      "loss": 0.8683,
+      "step": 528
+    },
+    {
+      "epoch": 0.48057813911472447,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00029811394554229254,
+      "loss": 0.8359,
+      "step": 532
+    },
+    {
+      "epoch": 0.48419150858175247,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002980798052818133,
+      "loss": 0.8693,
+      "step": 536
+    },
+    {
+      "epoch": 0.4878048780487805,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0002980453607833216,
+      "loss": 0.8598,
+      "step": 540
+    },
+    {
+      "epoch": 0.4914182475158085,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00029801061211758554,
+      "loss": 0.8279,
+      "step": 544
+    },
+    {
+      "epoch": 0.4950316169828365,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00029797555935599806,
+      "loss": 0.8298,
+      "step": 548
+    },
+    {
+      "epoch": 0.4986449864498645,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00029794020257057713,
+      "loss": 0.8548,
+      "step": 552
+    },
+    {
+      "epoch": 0.5022583559168925,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00029790454183396505,
+      "loss": 0.896,
+      "step": 556
+    },
+    {
+      "epoch": 0.5058717253839206,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0002978685772194288,
+      "loss": 0.8971,
+      "step": 560
+    },
+    {
+      "epoch": 0.5094850948509485,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002978323088008596,
+      "loss": 0.8629,
+      "step": 564
+    },
+    {
+      "epoch": 0.5130984643179766,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002977957366527729,
+      "loss": 0.8855,
+      "step": 568
+    },
+    {
+      "epoch": 0.5167118337850045,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.00029775886085030816,
+      "loss": 0.8323,
+      "step": 572
+    },
+    {
+      "epoch": 0.5203252032520326,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002977216814692288,
+      "loss": 0.8541,
+      "step": 576
+    },
+    {
+      "epoch": 0.5239385727190605,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002976841985859218,
+      "loss": 0.8808,
+      "step": 580
+    },
+    {
+      "epoch": 0.5275519421860885,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00029764641227739776,
+      "loss": 0.835,
+      "step": 584
+    },
+    {
+      "epoch": 0.5311653116531165,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002976083226212908,
+      "loss": 0.8281,
+      "step": 588
+    },
+    {
+      "epoch": 0.5347786811201445,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00029756992969585814,
+      "loss": 0.8838,
+      "step": 592
+    },
+    {
+      "epoch": 0.5383920505871725,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0002975312335799801,
+      "loss": 0.8677,
+      "step": 596
+    },
+    {
+      "epoch": 0.5420054200542005,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.00029749223435315993,
+      "loss": 0.8747,
+      "step": 600
+    },
+    {
+      "epoch": 0.5456187895212286,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0002974529320955238,
+      "loss": 0.8742,
+      "step": 604
+    },
+    {
+      "epoch": 0.5492321589882565,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002974133268878202,
+      "loss": 0.8547,
+      "step": 608
+    },
+    {
+      "epoch": 0.5528455284552846,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002973734188114203,
+      "loss": 0.8832,
+      "step": 612
+    },
+    {
+      "epoch": 0.5564588979223125,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0002973332079483173,
+      "loss": 0.8457,
+      "step": 616
+    },
+    {
+      "epoch": 0.5600722673893406,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00029729269438112665,
+      "loss": 0.8019,
+      "step": 620
+    },
+    {
+      "epoch": 0.5636856368563685,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002972518781930857,
+      "loss": 0.8371,
+      "step": 624
+    },
+    {
+      "epoch": 0.5672990063233966,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00029721075946805345,
+      "loss": 0.8103,
+      "step": 628
+    },
+    {
+      "epoch": 0.5709123757904245,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002971693382905107,
+      "loss": 0.8604,
+      "step": 632
+    },
+    {
+      "epoch": 0.5745257452574526,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002971276147455594,
+      "loss": 0.8282,
+      "step": 636
+    },
+    {
+      "epoch": 0.5781391147244805,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00029708558891892284,
+      "loss": 0.8315,
+      "step": 640
+    },
+    {
+      "epoch": 0.5817524841915086,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0002970432608969454,
+      "loss": 0.856,
+      "step": 644
+    },
+    {
+      "epoch": 0.5853658536585366,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002970006307665923,
+      "loss": 0.8755,
+      "step": 648
+    },
+    {
+      "epoch": 0.5889792231255646,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00029695769861544944,
+      "loss": 0.8204,
+      "step": 652
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002969144645317233,
+      "loss": 0.8371,
+      "step": 656
+    },
+    {
+      "epoch": 0.5962059620596206,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00029687092860424067,
+      "loss": 0.8409,
+      "step": 660
+    },
+    {
+      "epoch": 0.5998193315266486,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002968270909224484,
+      "loss": 0.8478,
+      "step": 664
+    },
+    {
+      "epoch": 0.6034327009936766,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0002967829515764134,
+      "loss": 0.8357,
+      "step": 668
+    },
+    {
+      "epoch": 0.6070460704607046,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00029673851065682244,
+      "loss": 0.775,
+      "step": 672
+    },
+    {
+      "epoch": 0.6106594399277326,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00029669376825498177,
+      "loss": 0.8624,
+      "step": 676
+    },
+    {
+      "epoch": 0.6142728093947606,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00029664872446281704,
+      "loss": 0.8696,
+      "step": 680
+    },
+    {
+      "epoch": 0.6178861788617886,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002966033793728732,
+      "loss": 0.8846,
+      "step": 684
+    },
+    {
+      "epoch": 0.6214995483288166,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002965577330783142,
+      "loss": 0.8533,
+      "step": 688
+    },
+    {
+      "epoch": 0.6251129177958447,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00029651178567292294,
+      "loss": 0.8841,
+      "step": 692
+    },
+    {
+      "epoch": 0.6287262872628726,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002964655372511007,
+      "loss": 0.8371,
+      "step": 696
+    },
+    {
+      "epoch": 0.6323396567299007,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002964189879078674,
+      "loss": 0.8183,
+      "step": 700
+    },
+    {
+      "epoch": 0.6359530261969286,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0002963721377388613,
+      "loss": 0.7912,
+      "step": 704
+    },
+    {
+      "epoch": 0.6395663956639567,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002963249868403385,
+      "loss": 0.8244,
+      "step": 708
+    },
+    {
+      "epoch": 0.6431797651309846,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00029627753530917325,
+      "loss": 0.852,
+      "step": 712
+    },
+    {
+      "epoch": 0.6467931345980127,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002962297832428571,
+      "loss": 0.8214,
+      "step": 716
+    },
+    {
+      "epoch": 0.6504065040650406,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00029618173073949936,
+      "loss": 0.8618,
+      "step": 720
+    },
+    {
+      "epoch": 0.6540198735320687,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0002961333778978266,
+      "loss": 0.9167,
+      "step": 724
+    },
+    {
+      "epoch": 0.6576332429990966,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002960847248171822,
+      "loss": 0.8186,
+      "step": 728
+    },
+    {
+      "epoch": 0.6612466124661247,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.00029603577159752664,
+      "loss": 0.8127,
+      "step": 732
+    },
+    {
+      "epoch": 0.6648599819331527,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00029598651833943705,
+      "loss": 0.8128,
+      "step": 736
+    },
+    {
+      "epoch": 0.6684733514001807,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002959369651441068,
+      "loss": 0.8373,
+      "step": 740
+    },
+    {
+      "epoch": 0.6720867208672087,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002958871121133456,
+      "loss": 0.7941,
+      "step": 744
+    },
+    {
+      "epoch": 0.6757000903342367,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0002958369593495794,
+      "loss": 0.8336,
+      "step": 748
+    },
+    {
+      "epoch": 0.6793134598012647,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00029578650695584956,
+      "loss": 0.8253,
+      "step": 752
+    },
+    {
+      "epoch": 0.6829268292682927,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002957357550358134,
+      "loss": 0.7859,
+      "step": 756
+    },
+    {
+      "epoch": 0.6865401987353207,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00029568470369374357,
+      "loss": 0.8203,
+      "step": 760
+    },
+    {
+      "epoch": 0.6901535682023487,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002956333530345276,
+      "loss": 0.8076,
+      "step": 764
+    },
+    {
+      "epoch": 0.6937669376693767,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002955817031636684,
+      "loss": 0.7564,
+      "step": 768
+    },
+    {
+      "epoch": 0.6973803071364046,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00029552975418728337,
+      "loss": 0.8361,
+      "step": 772
+    },
+    {
+      "epoch": 0.7009936766034327,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002954775062121045,
+      "loss": 0.8071,
+      "step": 776
+    },
+    {
+      "epoch": 0.7046070460704607,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00029542495934547813,
+      "loss": 0.8453,
+      "step": 780
+    },
+    {
+      "epoch": 0.7082204155374887,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002953721136953646,
+      "loss": 0.8342,
+      "step": 784
+    },
+    {
+      "epoch": 0.7118337850045167,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00029531896937033827,
+      "loss": 0.8209,
+      "step": 788
+    },
+    {
+      "epoch": 0.7154471544715447,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00029526552647958694,
+      "loss": 0.8183,
+      "step": 792
+    },
+    {
+      "epoch": 0.7190605239385727,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.000295211785132912,
+      "loss": 0.8761,
+      "step": 796
+    },
+    {
+      "epoch": 0.7226738934056007,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00029515774544072796,
+      "loss": 0.85,
+      "step": 800
+    },
+    {
+      "epoch": 0.7262872628726287,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002951034075140623,
+      "loss": 0.8058,
+      "step": 804
+    },
+    {
+      "epoch": 0.7299006323396567,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00029504877146455523,
+      "loss": 0.8008,
+      "step": 808
+    },
+    {
+      "epoch": 0.7335140018066847,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00029499383740445943,
+      "loss": 0.8391,
+      "step": 812
+    },
+    {
+      "epoch": 0.7371273712737128,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00029493860544664005,
+      "loss": 0.816,
+      "step": 816
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002948830757045741,
+      "loss": 0.8301,
+      "step": 820
+    },
+    {
+      "epoch": 0.7443541102077688,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00029482724829235033,
+      "loss": 0.8092,
+      "step": 824
+    },
+    {
+      "epoch": 0.7479674796747967,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002947711233246694,
+      "loss": 0.8431,
+      "step": 828
+    },
+    {
+      "epoch": 0.7515808491418248,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.00029471470091684297,
+      "loss": 0.8175,
+      "step": 832
+    },
+    {
+      "epoch": 0.7551942186088527,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.000294657981184794,
+      "loss": 0.7798,
+      "step": 836
+    },
+    {
+      "epoch": 0.7588075880758808,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00029460096424505626,
+      "loss": 0.8456,
+      "step": 840
+    },
+    {
+      "epoch": 0.7624209575429087,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002945436502147741,
+      "loss": 0.8109,
+      "step": 844
+    },
+    {
+      "epoch": 0.7660343270099368,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002944860392117023,
+      "loss": 0.8604,
+      "step": 848
+    },
+    {
+      "epoch": 0.7696476964769647,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00029442813135420595,
+      "loss": 0.8226,
+      "step": 852
+    },
+    {
+      "epoch": 0.7732610659439928,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00029436992676125975,
+      "loss": 0.8219,
+      "step": 856
+    },
+    {
+      "epoch": 0.7768744354110207,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0002943114255524482,
+      "loss": 0.8126,
+      "step": 860
+    },
+    {
+      "epoch": 0.7804878048780488,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00029425262784796535,
+      "loss": 0.7987,
+      "step": 864
+    },
+    {
+      "epoch": 0.7841011743450768,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00029419353376861415,
+      "loss": 0.8254,
+      "step": 868
+    },
+    {
+      "epoch": 0.7877145438121048,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0002941341434358066,
+      "loss": 0.8161,
+      "step": 872
+    },
+    {
+      "epoch": 0.7913279132791328,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002940744569715634,
+      "loss": 0.7733,
+      "step": 876
+    },
+    {
+      "epoch": 0.7949412827461608,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002940144744985137,
+      "loss": 0.8475,
+      "step": 880
+    },
+    {
+      "epoch": 0.7985546522131888,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002939541961398947,
+      "loss": 0.8765,
+      "step": 884
+    },
+    {
+      "epoch": 0.8021680216802168,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0002938936220195515,
+      "loss": 0.847,
+      "step": 888
+    },
+    {
+      "epoch": 0.8057813911472448,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002938327522619371,
+      "loss": 0.8139,
+      "step": 892
+    },
+    {
+      "epoch": 0.8093947606142728,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0002937715869921115,
+      "loss": 0.8334,
+      "step": 896
+    },
+    {
+      "epoch": 0.8130081300813008,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00029371012633574225,
+      "loss": 0.8142,
+      "step": 900
+    },
+    {
+      "epoch": 0.8166214995483289,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002936483704191035,
+      "loss": 0.8294,
+      "step": 904
+    },
+    {
+      "epoch": 0.8202348690153568,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00029358631936907625,
+      "loss": 0.8177,
+      "step": 908
+    },
+    {
+      "epoch": 0.8238482384823849,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00029352397331314765,
+      "loss": 0.8487,
+      "step": 912
+    },
+    {
+      "epoch": 0.8274616079494128,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0002934613323794111,
+      "loss": 0.7998,
+      "step": 916
+    },
+    {
+      "epoch": 0.8310749774164409,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00029339839669656584,
+      "loss": 0.8358,
+      "step": 920
+    },
+    {
+      "epoch": 0.8346883468834688,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002933351663939166,
+      "loss": 0.8398,
+      "step": 924
+    },
+    {
+      "epoch": 0.8383017163504969,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0002932716416013735,
+      "loss": 0.8188,
+      "step": 928
+    },
+    {
+      "epoch": 0.8419150858175248,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00029320782244945167,
+      "loss": 0.841,
+      "step": 932
+    },
+    {
+      "epoch": 0.8455284552845529,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00029314370906927107,
+      "loss": 0.8309,
+      "step": 936
+    },
+    {
+      "epoch": 0.8491418247515808,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.000293079301592556,
+      "loss": 0.8081,
+      "step": 940
+    },
+    {
+      "epoch": 0.8527551942186089,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002930146001516353,
+      "loss": 0.8106,
+      "step": 944
+    },
+    {
+      "epoch": 0.8563685636856369,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002929496048794415,
+      "loss": 0.8113,
+      "step": 948
+    },
+    {
+      "epoch": 0.8599819331526649,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00029288431590951094,
+      "loss": 0.8077,
+      "step": 952
+    },
+    {
+      "epoch": 0.8635953026196929,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00029281873337598334,
+      "loss": 0.8226,
+      "step": 956
+    },
+    {
+      "epoch": 0.8672086720867209,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002927528574136016,
+      "loss": 0.7979,
+      "step": 960
+    },
+    {
+      "epoch": 0.8708220415537489,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002926866881577115,
+      "loss": 0.7547,
+      "step": 964
+    },
+    {
+      "epoch": 0.8744354110207768,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002926202257442613,
+      "loss": 0.7724,
+      "step": 968
+    },
+    {
+      "epoch": 0.8780487804878049,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00029255347030980166,
+      "loss": 0.8242,
+      "step": 972
+    },
+    {
+      "epoch": 0.8816621499548328,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002924864219914853,
+      "loss": 0.7917,
+      "step": 976
+    },
+    {
+      "epoch": 0.8852755194218609,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002924190809270666,
+      "loss": 0.8284,
+      "step": 980
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00029235144725490146,
+      "loss": 0.7862,
+      "step": 984
+    },
+    {
+      "epoch": 0.8925022583559169,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002922835211139469,
+      "loss": 0.8303,
+      "step": 988
+    },
+    {
+      "epoch": 0.8961156278229448,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00029221530264376093,
+      "loss": 0.7848,
+      "step": 992
+    },
+    {
+      "epoch": 0.8997289972899729,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002921467919845021,
+      "loss": 0.7719,
+      "step": 996
+    },
+    {
+      "epoch": 0.9033423667570009,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00029207798927692916,
+      "loss": 0.8385,
+      "step": 1000
+    },
+    {
+      "epoch": 0.9069557362240289,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00029200889466240114,
+      "loss": 0.808,
+      "step": 1004
+    },
+    {
+      "epoch": 0.9105691056910569,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002919395082828767,
+      "loss": 0.7802,
+      "step": 1008
+    },
+    {
+      "epoch": 0.9141824751580849,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00029186983028091384,
+      "loss": 0.8207,
+      "step": 1012
+    },
+    {
+      "epoch": 0.9177958446251129,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00029179986079966985,
+      "loss": 0.8426,
+      "step": 1016
+    },
+    {
+      "epoch": 0.9214092140921409,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00029172959998290083,
+      "loss": 0.794,
+      "step": 1020
+    },
+    {
+      "epoch": 0.9250225835591689,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00029165904797496145,
+      "loss": 0.8051,
+      "step": 1024
+    },
+    {
+      "epoch": 0.928635953026197,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002915882049208047,
+      "loss": 0.834,
+      "step": 1028
+    },
+    {
+      "epoch": 0.9322493224932249,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00029151707096598134,
+      "loss": 0.8063,
+      "step": 1032
+    },
+    {
+      "epoch": 0.935862691960253,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002914456462566401,
+      "loss": 0.8119,
+      "step": 1036
+    },
+    {
+      "epoch": 0.9394760614272809,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00029137393093952685,
+      "loss": 0.7962,
+      "step": 1040
+    },
+    {
+      "epoch": 0.943089430894309,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002913019251619846,
+      "loss": 0.8037,
+      "step": 1044
+    },
+    {
+      "epoch": 0.9467028003613369,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002912296290719532,
+      "loss": 0.803,
+      "step": 1048
+    },
+    {
+      "epoch": 0.950316169828365,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002911570428179689,
+      "loss": 0.8887,
+      "step": 1052
+    },
+    {
+      "epoch": 0.9539295392953929,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00029108416654916405,
+      "loss": 0.772,
+      "step": 1056
+    },
+    {
+      "epoch": 0.957542908762421,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002910110004152669,
+      "loss": 0.7948,
+      "step": 1060
+    },
+    {
+      "epoch": 0.9611562782294489,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002909375445666013,
+      "loss": 0.7825,
+      "step": 1064
+    },
+    {
+      "epoch": 0.964769647696477,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00029086379915408626,
+      "loss": 0.8405,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9683830171635049,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00029078976432923576,
+      "loss": 0.7632,
+      "step": 1072
+    },
+    {
+      "epoch": 0.971996386630533,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0002907154402441583,
+      "loss": 0.8101,
+      "step": 1076
+    },
+    {
+      "epoch": 0.975609756097561,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00029064082705155685,
+      "loss": 0.7764,
+      "step": 1080
+    },
+    {
+      "epoch": 0.979223125564589,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002905659249047283,
+      "loss": 0.7553,
+      "step": 1084
+    },
+    {
+      "epoch": 0.982836495031617,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00029049073395756304,
+      "loss": 0.7814,
+      "step": 1088
+    },
+    {
+      "epoch": 0.986449864498645,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00029041525436454503,
+      "loss": 0.7669,
+      "step": 1092
+    },
+    {
+      "epoch": 0.990063233965673,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0002903394862807512,
+      "loss": 0.8187,
+      "step": 1096
+    },
+    {
+      "epoch": 0.993676603432701,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00029026342986185127,
+      "loss": 0.7845,
+      "step": 1100
+    },
+    {
+      "epoch": 0.997289972899729,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00029018708526410715,
+      "loss": 0.7785,
+      "step": 1104
+    },
+    {
+      "epoch": 1.000903342366757,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00029011045264437305,
+      "loss": 0.7601,
+      "step": 1108
+    },
+    {
+      "epoch": 1.004516711833785,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002900335321600949,
+      "loss": 0.7283,
+      "step": 1112
+    },
+    {
+      "epoch": 1.008130081300813,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002899563239693099,
+      "loss": 0.7177,
+      "step": 1116
+    },
+    {
+      "epoch": 1.0117434507678411,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002898788282306466,
+      "loss": 0.7408,
+      "step": 1120
+    },
+    {
+      "epoch": 1.015356820234869,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00028980104510332414,
+      "loss": 0.7268,
+      "step": 1124
+    },
+    {
+      "epoch": 1.018970189701897,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002897229747471521,
+      "loss": 0.7105,
+      "step": 1128
+    },
+    {
+      "epoch": 1.022583559168925,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002896446173225304,
+      "loss": 0.7627,
+      "step": 1132
+    },
+    {
+      "epoch": 1.0261969286359531,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002895659729904487,
+      "loss": 0.7362,
+      "step": 1136
+    },
+    {
+      "epoch": 1.029810298102981,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00028948704191248585,
+      "loss": 0.7749,
+      "step": 1140
+    },
+    {
+      "epoch": 1.033423667570009,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00028940782425081017,
+      "loss": 0.7324,
+      "step": 1144
+    },
+    {
+      "epoch": 1.037037037037037,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00028932832016817864,
+      "loss": 0.7389,
+      "step": 1148
+    },
+    {
+      "epoch": 1.040650406504065,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002892485298279367,
+      "loss": 0.7309,
+      "step": 1152
+    },
+    {
+      "epoch": 1.044263775971093,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.000289168453394018,
+      "loss": 0.7174,
+      "step": 1156
+    },
+    {
+      "epoch": 1.047877145438121,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002890880910309438,
+      "loss": 0.7076,
+      "step": 1160
+    },
+    {
+      "epoch": 1.051490514905149,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00028900744290382313,
+      "loss": 0.7059,
+      "step": 1164
+    },
+    {
+      "epoch": 1.055103884372177,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002889265091783517,
+      "loss": 0.7081,
+      "step": 1168
+    },
+    {
+      "epoch": 1.0587172538392051,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002888452900208125,
+      "loss": 0.7433,
+      "step": 1172
+    },
+    {
+      "epoch": 1.062330623306233,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00028876378559807464,
+      "loss": 0.6917,
+      "step": 1176
+    },
+    {
+      "epoch": 1.065943992773261,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00028868199607759323,
+      "loss": 0.6994,
+      "step": 1180
+    },
+    {
+      "epoch": 1.069557362240289,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00028859992162740954,
+      "loss": 0.7196,
+      "step": 1184
+    },
+    {
+      "epoch": 1.0731707317073171,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00028851756241614975,
+      "loss": 0.7514,
+      "step": 1188
+    },
+    {
+      "epoch": 1.076784101174345,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002884349186130255,
+      "loss": 0.7266,
+      "step": 1192
+    },
+    {
+      "epoch": 1.080397470641373,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00028835199038783295,
+      "loss": 0.6985,
+      "step": 1196
+    },
+    {
+      "epoch": 1.084010840108401,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00028826877791095256,
+      "loss": 0.7516,
+      "step": 1200
+    },
+    {
+      "epoch": 1.0876242095754292,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.000288185281353349,
+      "loss": 0.677,
+      "step": 1204
+    },
+    {
+      "epoch": 1.091237579042457,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00028810150088657047,
+      "loss": 0.6898,
+      "step": 1208
+    },
+    {
+      "epoch": 1.094850948509485,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00028801743668274845,
+      "loss": 0.7318,
+      "step": 1212
+    },
+    {
+      "epoch": 1.098464317976513,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002879330889145974,
+      "loss": 0.7356,
+      "step": 1216
+    },
+    {
+      "epoch": 1.1020776874435412,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002878484577554144,
+      "loss": 0.7169,
+      "step": 1220
+    },
+    {
+      "epoch": 1.1056910569105691,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002877635433790789,
+      "loss": 0.7866,
+      "step": 1224
+    },
+    {
+      "epoch": 1.109304426377597,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002876783459600519,
+      "loss": 0.7008,
+      "step": 1228
+    },
+    {
+      "epoch": 1.112917795844625,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00028759286567337633,
+      "loss": 0.6943,
+      "step": 1232
+    },
+    {
+      "epoch": 1.1165311653116532,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00028750710269467595,
+      "loss": 0.6751,
+      "step": 1236
+    },
+    {
+      "epoch": 1.1201445347786811,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002874210572001555,
+      "loss": 0.7603,
+      "step": 1240
+    },
+    {
+      "epoch": 1.123757904245709,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00028733472936660014,
+      "loss": 0.7299,
+      "step": 1244
+    },
+    {
+      "epoch": 1.127371273712737,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002872481193713751,
+      "loss": 0.7056,
+      "step": 1248
+    },
+    {
+      "epoch": 1.1309846431797652,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00028716122739242533,
+      "loss": 0.7833,
+      "step": 1252
+    },
+    {
+      "epoch": 1.1345980126467932,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00028707405360827506,
+      "loss": 0.7327,
+      "step": 1256
+    },
+    {
+      "epoch": 1.1382113821138211,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00028698659819802766,
+      "loss": 0.6559,
+      "step": 1260
+    },
+    {
+      "epoch": 1.141824751580849,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000286898861341365,
+      "loss": 0.7023,
+      "step": 1264
+    },
+    {
+      "epoch": 1.1454381210478772,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002868108432185472,
+      "loss": 0.7236,
+      "step": 1268
+    },
+    {
+      "epoch": 1.1490514905149052,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002867225440104123,
+      "loss": 0.7181,
+      "step": 1272
+    },
+    {
+      "epoch": 1.1526648599819331,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002866339638983758,
+      "loss": 0.7769,
+      "step": 1276
+    },
+    {
+      "epoch": 1.156278229448961,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00028654510306443034,
+      "loss": 0.7196,
+      "step": 1280
+    },
+    {
+      "epoch": 1.1598915989159893,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002864559616911454,
+      "loss": 0.7283,
+      "step": 1284
+    },
+    {
+      "epoch": 1.1635049683830172,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00028636653996166677,
+      "loss": 0.7252,
+      "step": 1288
+    },
+    {
+      "epoch": 1.1671183378500452,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002862768380597162,
+      "loss": 0.7131,
+      "step": 1292
+    },
+    {
+      "epoch": 1.170731707317073,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002861868561695912,
+      "loss": 0.7274,
+      "step": 1296
+    },
+    {
+      "epoch": 1.174345076784101,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002860965944761644,
+      "loss": 0.7314,
+      "step": 1300
+    },
+    {
+      "epoch": 1.1779584462511292,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00028600605316488336,
+      "loss": 0.7095,
+      "step": 1304
+    },
+    {
+      "epoch": 1.1815718157181572,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002859152324217701,
+      "loss": 0.6816,
+      "step": 1308
+    },
+    {
+      "epoch": 1.1851851851851851,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00028582413243342095,
+      "loss": 0.7008,
+      "step": 1312
+    },
+    {
+      "epoch": 1.1887985546522133,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0002857327533870055,
+      "loss": 0.7118,
+      "step": 1316
+    },
+    {
+      "epoch": 1.1924119241192412,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002856410954702672,
+      "loss": 0.7472,
+      "step": 1320
+    },
+    {
+      "epoch": 1.1960252935862692,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0002855491588715222,
+      "loss": 0.7236,
+      "step": 1324
+    },
+    {
+      "epoch": 1.1996386630532971,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0002854569437796591,
+      "loss": 0.7762,
+      "step": 1328
+    },
+    {
+      "epoch": 1.203252032520325,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002853644503841389,
+      "loss": 0.7694,
+      "step": 1332
+    },
+    {
+      "epoch": 1.2068654019873533,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00028527167887499444,
+      "loss": 0.7559,
+      "step": 1336
+    },
+    {
+      "epoch": 1.2104787714543812,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00028517862944282964,
+      "loss": 0.7358,
+      "step": 1340
+    },
+    {
+      "epoch": 1.2140921409214092,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002850853022788197,
+      "loss": 0.7076,
+      "step": 1344
+    },
+    {
+      "epoch": 1.2177055103884373,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00028499169757471035,
+      "loss": 0.7354,
+      "step": 1348
+    },
+    {
+      "epoch": 1.2213188798554653,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00028489781552281756,
+      "loss": 0.7191,
+      "step": 1352
+    },
+    {
+      "epoch": 1.2249322493224932,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00028480365631602706,
+      "loss": 0.7447,
+      "step": 1356
+    },
+    {
+      "epoch": 1.2285456187895212,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002847092201477941,
+      "loss": 0.7253,
+      "step": 1360
+    },
+    {
+      "epoch": 1.2321589882565491,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00028461450721214293,
+      "loss": 0.7286,
+      "step": 1364
+    },
+    {
+      "epoch": 1.2357723577235773,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002845195177036664,
+      "loss": 0.7333,
+      "step": 1368
+    },
+    {
+      "epoch": 1.2393857271906052,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002844242518175256,
+      "loss": 0.7299,
+      "step": 1372
+    },
+    {
+      "epoch": 1.2429990966576332,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002843287097494496,
+      "loss": 0.755,
+      "step": 1376
+    },
+    {
+      "epoch": 1.2466124661246614,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00028423289169573465,
+      "loss": 0.6887,
+      "step": 1380
+    },
+    {
+      "epoch": 1.2502258355916893,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00028413679785324413,
+      "loss": 0.7668,
+      "step": 1384
+    },
+    {
+      "epoch": 1.2538392050587173,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00028404042841940813,
+      "loss": 0.7157,
+      "step": 1388
+    },
+    {
+      "epoch": 1.2574525745257452,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00028394378359222294,
+      "loss": 0.7247,
+      "step": 1392
+    },
+    {
+      "epoch": 1.2610659439927732,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002838468635702505,
+      "loss": 0.6949,
+      "step": 1396
+    },
+    {
+      "epoch": 1.2646793134598013,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002837496685526183,
+      "loss": 0.7799,
+      "step": 1400
+    },
+    {
+      "epoch": 1.2682926829268293,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00028365219873901885,
+      "loss": 0.6787,
+      "step": 1404
+    },
+    {
+      "epoch": 1.2719060523938572,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00028355445432970915,
+      "loss": 0.7171,
+      "step": 1408
+    },
+    {
+      "epoch": 1.2755194218608854,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00028345643552551037,
+      "loss": 0.722,
+      "step": 1412
+    },
+    {
+      "epoch": 1.2791327913279134,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002833581425278075,
+      "loss": 0.7501,
+      "step": 1416
+    },
+    {
+      "epoch": 1.2827461607949413,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00028325957553854885,
+      "loss": 0.7811,
+      "step": 1420
+    },
+    {
+      "epoch": 1.2863595302619693,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00028316073476024567,
+      "loss": 0.7448,
+      "step": 1424
+    },
+    {
+      "epoch": 1.2899728997289972,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00028306162039597166,
+      "loss": 0.7354,
+      "step": 1428
+    },
+    {
+      "epoch": 1.2935862691960254,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00028296223264936277,
+      "loss": 0.6888,
+      "step": 1432
+    },
+    {
+      "epoch": 1.2971996386630533,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00028286257172461637,
+      "loss": 0.7354,
+      "step": 1436
+    },
+    {
+      "epoch": 1.3008130081300813,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002827626378264914,
+      "loss": 0.7097,
+      "step": 1440
+    },
+    {
+      "epoch": 1.3044263775971092,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00028266243116030753,
+      "loss": 0.7167,
+      "step": 1444
+    },
+    {
+      "epoch": 1.3080397470641372,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00028256195193194464,
+      "loss": 0.7424,
+      "step": 1448
+    },
+    {
+      "epoch": 1.3116531165311653,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002824612003478428,
+      "loss": 0.7203,
+      "step": 1452
+    },
+    {
+      "epoch": 1.3152664859981933,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00028236017661500176,
+      "loss": 0.7034,
+      "step": 1456
+    },
+    {
+      "epoch": 1.3188798554652212,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0002822588809409801,
+      "loss": 0.7605,
+      "step": 1460
+    },
+    {
+      "epoch": 1.3224932249322494,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002821573135338954,
+      "loss": 0.7303,
+      "step": 1464
+    },
+    {
+      "epoch": 1.3261065943992774,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00028205547460242336,
+      "loss": 0.7412,
+      "step": 1468
+    },
+    {
+      "epoch": 1.3297199638663053,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00028195336435579764,
+      "loss": 0.7397,
+      "step": 1472
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002818509830038093,
+      "loss": 0.7335,
+      "step": 1476
+    },
+    {
+      "epoch": 1.3369467028003612,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002817483307568064,
+      "loss": 0.7167,
+      "step": 1480
+    },
+    {
+      "epoch": 1.3405600722673894,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00028164540782569343,
+      "loss": 0.7075,
+      "step": 1484
+    },
+    {
+      "epoch": 1.3441734417344173,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00028154221442193135,
+      "loss": 0.729,
+      "step": 1488
+    },
+    {
+      "epoch": 1.3477868112014453,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00028143875075753655,
+      "loss": 0.695,
+      "step": 1492
+    },
+    {
+      "epoch": 1.3514001806684734,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002813350170450807,
+      "loss": 0.748,
+      "step": 1496
+    },
+    {
+      "epoch": 1.3550135501355014,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00028123101349769033,
+      "loss": 0.741,
+      "step": 1500
+    },
+    {
+      "epoch": 1.3586269196025293,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002811267403290465,
+      "loss": 0.7294,
+      "step": 1504
+    },
+    {
+      "epoch": 1.3622402890695573,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00028102219775338406,
+      "loss": 0.743,
+      "step": 1508
+    },
+    {
+      "epoch": 1.3658536585365852,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00028091738598549145,
+      "loss": 0.7018,
+      "step": 1512
+    },
+    {
+      "epoch": 1.3694670280036134,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00028081230524071,
+      "loss": 0.7036,
+      "step": 1516
+    },
+    {
+      "epoch": 1.3730803974706414,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00028070695573493394,
+      "loss": 0.7261,
+      "step": 1520
+    },
+    {
+      "epoch": 1.3766937669376693,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00028060133768460956,
+      "loss": 0.7342,
+      "step": 1524
+    },
+    {
+      "epoch": 1.3803071364046975,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002804954513067349,
+      "loss": 0.7539,
+      "step": 1528
+    },
+    {
+      "epoch": 1.3839205058717254,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002803892968188592,
+      "loss": 0.7647,
+      "step": 1532
+    },
+    {
+      "epoch": 1.3875338753387534,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002802828744390826,
+      "loss": 0.704,
+      "step": 1536
+    },
+    {
+      "epoch": 1.3911472448057813,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00028017618438605583,
+      "loss": 0.7364,
+      "step": 1540
+    },
+    {
+      "epoch": 1.3947606142728093,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002800692268789793,
+      "loss": 0.6884,
+      "step": 1544
+    },
+    {
+      "epoch": 1.3983739837398375,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.000279962002137603,
+      "loss": 0.7081,
+      "step": 1548
+    },
+    {
+      "epoch": 1.4019873532068654,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.000279854510382226,
+      "loss": 0.7041,
+      "step": 1552
+    },
+    {
+      "epoch": 1.4056007226738934,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0002797467518336961,
+      "loss": 0.6753,
+      "step": 1556
+    },
+    {
+      "epoch": 1.4092140921409215,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00027963872671340887,
+      "loss": 0.7222,
+      "step": 1560
+    },
+    {
+      "epoch": 1.4128274616079495,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000279530435243308,
+      "loss": 0.7318,
+      "step": 1564
+    },
+    {
+      "epoch": 1.4164408310749774,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00027942187764588405,
+      "loss": 0.7012,
+      "step": 1568
+    },
+    {
+      "epoch": 1.4200542005420054,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002793130541441746,
+      "loss": 0.7813,
+      "step": 1572
+    },
+    {
+      "epoch": 1.4236675700090333,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002792039649617634,
+      "loss": 0.696,
+      "step": 1576
+    },
+    {
+      "epoch": 1.4272809394760615,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00027909461032278013,
+      "loss": 0.7043,
+      "step": 1580
+    },
+    {
+      "epoch": 1.4308943089430894,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002789849904518999,
+      "loss": 0.7521,
+      "step": 1584
+    },
+    {
+      "epoch": 1.4345076784101174,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00027887510557434255,
+      "loss": 0.7258,
+      "step": 1588
+    },
+    {
+      "epoch": 1.4381210478771456,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002787649559158727,
+      "loss": 0.7201,
+      "step": 1592
+    },
+    {
+      "epoch": 1.4417344173441735,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002786545417027987,
+      "loss": 0.7215,
+      "step": 1596
+    },
+    {
+      "epoch": 1.4453477868112015,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002785438631619726,
+      "loss": 0.6907,
+      "step": 1600
+    },
+    {
+      "epoch": 1.4489611562782294,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00027843292052078943,
+      "loss": 0.7227,
+      "step": 1604
+    },
+    {
+      "epoch": 1.4525745257452574,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.000278321714007187,
+      "loss": 0.6769,
+      "step": 1608
+    },
+    {
+      "epoch": 1.4561878952122855,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.000278210243849645,
+      "loss": 0.7191,
+      "step": 1612
+    },
+    {
+      "epoch": 1.4598012646793135,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.000278098510277185,
+      "loss": 0.6908,
+      "step": 1616
+    },
+    {
+      "epoch": 1.4634146341463414,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002779865135193697,
+      "loss": 0.7349,
+      "step": 1620
+    },
+    {
+      "epoch": 1.4670280036133696,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00027787425380630253,
+      "loss": 0.6604,
+      "step": 1624
+    },
+    {
+      "epoch": 1.4706413730803976,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00027776173136862713,
+      "loss": 0.6614,
+      "step": 1628
+    },
+    {
+      "epoch": 1.4742547425474255,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00027764894643752706,
+      "loss": 0.7078,
+      "step": 1632
+    },
+    {
+      "epoch": 1.4778681120144534,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.000277535899244725,
+      "loss": 0.7213,
+      "step": 1636
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00027742259002248263,
+      "loss": 0.7703,
+      "step": 1640
+    },
+    {
+      "epoch": 1.4850948509485096,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00027730901900359985,
+      "loss": 0.7415,
+      "step": 1644
+    },
+    {
+      "epoch": 1.4887082204155375,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002771951864214145,
+      "loss": 0.7169,
+      "step": 1648
+    },
+    {
+      "epoch": 1.4923215898825655,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00027708109250980183,
+      "loss": 0.6944,
+      "step": 1652
+    },
+    {
+      "epoch": 1.4959349593495934,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.000276966737503174,
+      "loss": 0.6798,
+      "step": 1656
+    },
+    {
+      "epoch": 1.4995483288166214,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00027685212163647955,
+      "loss": 0.7072,
+      "step": 1660
+    },
+    {
+      "epoch": 1.5031616982836495,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00027673724514520306,
+      "loss": 0.7185,
+      "step": 1664
+    },
+    {
+      "epoch": 1.5067750677506775,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002766221082653645,
+      "loss": 0.707,
+      "step": 1668
+    },
+    {
+      "epoch": 1.5103884372177054,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00027650671123351884,
+      "loss": 0.7042,
+      "step": 1672
+    },
+    {
+      "epoch": 1.5140018066847336,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00027639105428675556,
+      "loss": 0.7221,
+      "step": 1676
+    },
+    {
+      "epoch": 1.5176151761517616,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002762751376626982,
+      "loss": 0.7181,
+      "step": 1680
+    },
+    {
+      "epoch": 1.5212285456187895,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00027615896159950367,
+      "loss": 0.7511,
+      "step": 1684
+    },
+    {
+      "epoch": 1.5248419150858177,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002760425263358621,
+      "loss": 0.6876,
+      "step": 1688
+    },
+    {
+      "epoch": 1.5284552845528454,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.000275925832110996,
+      "loss": 0.7231,
+      "step": 1692
+    },
+    {
+      "epoch": 1.5320686540198736,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00027580887916466007,
+      "loss": 0.7025,
+      "step": 1696
+    },
+    {
+      "epoch": 1.5356820234869015,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00027569166773714043,
+      "loss": 0.6513,
+      "step": 1700
+    },
+    {
+      "epoch": 1.5392953929539295,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00027557419806925436,
+      "loss": 0.7532,
+      "step": 1704
+    },
+    {
+      "epoch": 1.5429087624209576,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002754564704023497,
+      "loss": 0.7641,
+      "step": 1708
+    },
+    {
+      "epoch": 1.5465221318879856,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00027533848497830434,
+      "loss": 0.7356,
+      "step": 1712
+    },
+    {
+      "epoch": 1.5501355013550135,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00027522024203952575,
+      "loss": 0.7366,
+      "step": 1716
+    },
+    {
+      "epoch": 1.5537488708220417,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00027510174182895046,
+      "loss": 0.6814,
+      "step": 1720
+    },
+    {
+      "epoch": 1.5573622402890694,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00027498298459004356,
+      "loss": 0.6878,
+      "step": 1724
+    },
+    {
+      "epoch": 1.5609756097560976,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00027486397056679835,
+      "loss": 0.7133,
+      "step": 1728
+    },
+    {
+      "epoch": 1.5645889792231256,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00027474470000373553,
+      "loss": 0.7425,
+      "step": 1732
+    },
+    {
+      "epoch": 1.5682023486901535,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.000274625173145903,
+      "loss": 0.724,
+      "step": 1736
+    },
+    {
+      "epoch": 1.5718157181571817,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00027450539023887515,
+      "loss": 0.7249,
+      "step": 1740
+    },
+    {
+      "epoch": 1.5754290876242094,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00027438535152875254,
+      "loss": 0.7067,
+      "step": 1744
+    },
+    {
+      "epoch": 1.5790424570912376,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002742650572621612,
+      "loss": 0.7092,
+      "step": 1748
+    },
+    {
+      "epoch": 1.5826558265582655,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002741445076862522,
+      "loss": 0.6639,
+      "step": 1752
+    },
+    {
+      "epoch": 1.5862691960252935,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00027402370304870126,
+      "loss": 0.6936,
+      "step": 1756
+    },
+    {
+      "epoch": 1.5898825654923217,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00027390264359770797,
+      "loss": 0.721,
+      "step": 1760
+    },
+    {
+      "epoch": 1.5934959349593496,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00027378132958199577,
+      "loss": 0.7253,
+      "step": 1764
+    },
+    {
+      "epoch": 1.5971093044263776,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00027365976125081064,
+      "loss": 0.7269,
+      "step": 1768
+    },
+    {
+      "epoch": 1.6007226738934057,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00027353793885392155,
+      "loss": 0.7555,
+      "step": 1772
+    },
+    {
+      "epoch": 1.6043360433604335,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00027341586264161905,
+      "loss": 0.7226,
+      "step": 1776
+    },
+    {
+      "epoch": 1.6079494128274616,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00027329353286471554,
+      "loss": 0.7104,
+      "step": 1780
+    },
+    {
+      "epoch": 1.6115627822944896,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002731709497745441,
+      "loss": 0.7279,
+      "step": 1784
+    },
+    {
+      "epoch": 1.6151761517615175,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0002730481136229583,
+      "loss": 0.7066,
+      "step": 1788
+    },
+    {
+      "epoch": 1.6187895212285457,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00027292502466233184,
+      "loss": 0.7169,
+      "step": 1792
+    },
+    {
+      "epoch": 1.6224028906955736,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002728016831455575,
+      "loss": 0.6813,
+      "step": 1796
+    },
+    {
+      "epoch": 1.6260162601626016,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002726780893260473,
+      "loss": 0.6798,
+      "step": 1800
+    },
+    {
+      "epoch": 1.6296296296296298,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002725542434577314,
+      "loss": 0.6663,
+      "step": 1804
+    },
+    {
+      "epoch": 1.6332429990966575,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0002724301457950578,
+      "loss": 0.6847,
+      "step": 1808
+    },
+    {
+      "epoch": 1.6368563685636857,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002723057965929921,
+      "loss": 0.7253,
+      "step": 1812
+    },
+    {
+      "epoch": 1.6404697380307136,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002721811961070163,
+      "loss": 0.7399,
+      "step": 1816
+    },
+    {
+      "epoch": 1.6440831074977416,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00027205634459312884,
+      "loss": 0.7394,
+      "step": 1820
+    },
+    {
+      "epoch": 1.6476964769647697,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00027193124230784414,
+      "loss": 0.7702,
+      "step": 1824
+    },
+    {
+      "epoch": 1.6513098464317977,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00027180588950819157,
+      "loss": 0.6959,
+      "step": 1828
+    },
+    {
+      "epoch": 1.6549232158988256,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00027168028645171516,
+      "loss": 0.7626,
+      "step": 1832
+    },
+    {
+      "epoch": 1.6585365853658538,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00027155443339647335,
+      "loss": 0.7254,
+      "step": 1836
+    },
+    {
+      "epoch": 1.6621499548328815,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00027142833060103807,
+      "loss": 0.728,
+      "step": 1840
+    },
+    {
+      "epoch": 1.6657633242999097,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002713019783244944,
+      "loss": 0.717,
+      "step": 1844
+    },
+    {
+      "epoch": 1.6693766937669376,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002711753768264398,
+      "loss": 0.729,
+      "step": 1848
+    },
+    {
+      "epoch": 1.6729900632339656,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002710485263669841,
+      "loss": 0.7737,
+      "step": 1852
+    },
+    {
+      "epoch": 1.6766034327009938,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0002709214272067484,
+      "loss": 0.7194,
+      "step": 1856
+    },
+    {
+      "epoch": 1.6802168021680217,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00027079407960686487,
+      "loss": 0.7001,
+      "step": 1860
+    },
+    {
+      "epoch": 1.6838301716350497,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00027066648382897604,
+      "loss": 0.7283,
+      "step": 1864
+    },
+    {
+      "epoch": 1.6874435411020778,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0002705386401352344,
+      "loss": 0.6987,
+      "step": 1868
+    },
+    {
+      "epoch": 1.6910569105691056,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00027041054878830176,
+      "loss": 0.7261,
+      "step": 1872
+    },
+    {
+      "epoch": 1.6946702800361337,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002702822100513487,
+      "loss": 0.7475,
+      "step": 1876
+    },
+    {
+      "epoch": 1.6982836495031617,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00027015362418805424,
+      "loss": 0.7149,
+      "step": 1880
+    },
+    {
+      "epoch": 1.7018970189701896,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00027002479146260504,
+      "loss": 0.7151,
+      "step": 1884
+    },
+    {
+      "epoch": 1.7055103884372178,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002698957121396948,
+      "loss": 0.778,
+      "step": 1888
+    },
+    {
+      "epoch": 1.7091237579042458,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00026976638648452413,
+      "loss": 0.7065,
+      "step": 1892
+    },
+    {
+      "epoch": 1.7127371273712737,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00026963681476279956,
+      "loss": 0.6981,
+      "step": 1896
+    },
+    {
+      "epoch": 1.7163504968383019,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00026950699724073333,
+      "loss": 0.6818,
+      "step": 1900
+    },
+    {
+      "epoch": 1.7199638663053296,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00026937693418504246,
+      "loss": 0.7459,
+      "step": 1904
+    },
+    {
+      "epoch": 1.7235772357723578,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002692466258629486,
+      "loss": 0.7163,
+      "step": 1908
+    },
+    {
+      "epoch": 1.7271906052393857,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002691160725421774,
+      "loss": 0.7107,
+      "step": 1912
+    },
+    {
+      "epoch": 1.7308039747064137,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00026898527449095765,
+      "loss": 0.7582,
+      "step": 1916
+    },
+    {
+      "epoch": 1.7344173441734418,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0002688542319780211,
+      "loss": 0.7266,
+      "step": 1920
+    },
+    {
+      "epoch": 1.7380307136404698,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002687229452726017,
+      "loss": 0.7625,
+      "step": 1924
+    },
+    {
+      "epoch": 1.7416440831074977,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00026859141464443515,
+      "loss": 0.7086,
+      "step": 1928
+    },
+    {
+      "epoch": 1.7452574525745257,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00026845964036375825,
+      "loss": 0.7283,
+      "step": 1932
+    },
+    {
+      "epoch": 1.7488708220415536,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00026832762270130844,
+      "loss": 0.7321,
+      "step": 1936
+    },
+    {
+      "epoch": 1.7524841915085818,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002681953619283232,
+      "loss": 0.6968,
+      "step": 1940
+    },
+    {
+      "epoch": 1.7560975609756098,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00026806285831653943,
+      "loss": 0.7761,
+      "step": 1944
+    },
+    {
+      "epoch": 1.7597109304426377,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00026793011213819304,
+      "loss": 0.7347,
+      "step": 1948
+    },
+    {
+      "epoch": 1.7633242999096659,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002677971236660183,
+      "loss": 0.7694,
+      "step": 1952
+    },
+    {
+      "epoch": 1.7669376693766936,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002676638931732472,
+      "loss": 0.7586,
+      "step": 1956
+    },
+    {
+      "epoch": 1.7705510388437218,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002675304209336091,
+      "loss": 0.7129,
+      "step": 1960
+    },
+    {
+      "epoch": 1.7741644083107497,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00026739670722132993,
+      "loss": 0.7161,
+      "step": 1964
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00026726275231113173,
+      "loss": 0.7078,
+      "step": 1968
+    },
+    {
+      "epoch": 1.7813911472448059,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002671285564782323,
+      "loss": 0.7319,
+      "step": 1972
+    },
+    {
+      "epoch": 1.7850045167118338,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002669941199983441,
+      "loss": 0.7176,
+      "step": 1976
+    },
+    {
+      "epoch": 1.7886178861788617,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002668594431476743,
+      "loss": 0.757,
+      "step": 1980
+    },
+    {
+      "epoch": 1.79223125564589,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002667245262029238,
+      "loss": 0.7165,
+      "step": 1984
+    },
+    {
+      "epoch": 1.7958446251129176,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002665893694412868,
+      "loss": 0.7306,
+      "step": 1988
+    },
+    {
+      "epoch": 1.7994579945799458,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0002664539731404502,
+      "loss": 0.6908,
+      "step": 1992
+    },
+    {
+      "epoch": 1.8030713640469738,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00026631833757859304,
+      "loss": 0.6951,
+      "step": 1996
+    },
+    {
+      "epoch": 1.8066847335140017,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.000266182463034386,
+      "loss": 0.6831,
+      "step": 2000
+    },
+    {
+      "epoch": 1.8102981029810299,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00026604634978699075,
+      "loss": 0.7225,
+      "step": 2004
+    },
+    {
+      "epoch": 1.8139114724480578,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002659099981160592,
+      "loss": 0.7541,
+      "step": 2008
+    },
+    {
+      "epoch": 1.8175248419150858,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002657734083017335,
+      "loss": 0.6766,
+      "step": 2012
+    },
+    {
+      "epoch": 1.821138211382114,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00026563658062464464,
+      "loss": 0.7134,
+      "step": 2016
+    },
+    {
+      "epoch": 1.8247515808491417,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00026549951536591264,
+      "loss": 0.7577,
+      "step": 2020
+    },
+    {
+      "epoch": 1.8283649503161699,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002653622128071455,
+      "loss": 0.6919,
+      "step": 2024
+    },
+    {
+      "epoch": 1.8319783197831978,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00026522467323043884,
+      "loss": 0.7221,
+      "step": 2028
+    },
+    {
+      "epoch": 1.8355916892502258,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00026508689691837517,
+      "loss": 0.6636,
+      "step": 2032
+    },
+    {
+      "epoch": 1.839205058717254,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00026494888415402336,
+      "loss": 0.7046,
+      "step": 2036
+    },
+    {
+      "epoch": 1.8428184281842819,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002648106352209382,
+      "loss": 0.7064,
+      "step": 2040
+    },
+    {
+      "epoch": 1.8464317976513098,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00026467215040315963,
+      "loss": 0.6979,
+      "step": 2044
+    },
+    {
+      "epoch": 1.850045167118338,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002645334299852122,
+      "loss": 0.7182,
+      "step": 2048
+    },
+    {
+      "epoch": 1.8536585365853657,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002643944742521046,
+      "loss": 0.7247,
+      "step": 2052
+    },
+    {
+      "epoch": 1.857271906052394,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.000264255283489329,
+      "loss": 0.6865,
+      "step": 2056
+    },
+    {
+      "epoch": 1.8608852755194218,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002641158579828602,
+      "loss": 0.6904,
+      "step": 2060
+    },
+    {
+      "epoch": 1.8644986449864498,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002639761980191557,
+      "loss": 0.6798,
+      "step": 2064
+    },
+    {
+      "epoch": 1.868112014453478,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00026383630388515433,
+      "loss": 0.6785,
+      "step": 2068
+    },
+    {
+      "epoch": 1.871725383920506,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00026369617586827627,
+      "loss": 0.7089,
+      "step": 2072
+    },
+    {
+      "epoch": 1.8753387533875339,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002635558142564221,
+      "loss": 0.729,
+      "step": 2076
+    },
+    {
+      "epoch": 1.878952122854562,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002634152193379725,
+      "loss": 0.7456,
+      "step": 2080
+    },
+    {
+      "epoch": 1.8825654923215898,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00026327439140178726,
+      "loss": 0.6812,
+      "step": 2084
+    },
+    {
+      "epoch": 1.886178861788618,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00026313333073720507,
+      "loss": 0.6964,
+      "step": 2088
+    },
+    {
+      "epoch": 1.8897922312556459,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002629920376340427,
+      "loss": 0.7094,
+      "step": 2092
+    },
+    {
+      "epoch": 1.8934056007226738,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00026285051238259465,
+      "loss": 0.7041,
+      "step": 2096
+    },
+    {
+      "epoch": 1.897018970189702,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002627087552736321,
+      "loss": 0.6752,
+      "step": 2100
+    },
+    {
+      "epoch": 1.90063233965673,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00026256676659840275,
+      "loss": 0.72,
+      "step": 2104
+    },
+    {
+      "epoch": 1.904245709123758,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00026242454664863015,
+      "loss": 0.7093,
+      "step": 2108
+    },
+    {
+      "epoch": 1.907859078590786,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002622820957165128,
+      "loss": 0.6896,
+      "step": 2112
+    },
+    {
+      "epoch": 1.9114724480578138,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.000262139414094724,
+      "loss": 0.7046,
+      "step": 2116
+    },
+    {
+      "epoch": 1.915085817524842,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00026199650207641085,
+      "loss": 0.729,
+      "step": 2120
+    },
+    {
+      "epoch": 1.91869918699187,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00026185335995519384,
+      "loss": 0.7202,
+      "step": 2124
+    },
+    {
+      "epoch": 1.9223125564588979,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00026170998802516624,
+      "loss": 0.6912,
+      "step": 2128
+    },
+    {
+      "epoch": 1.925925925925926,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002615663865808935,
+      "loss": 0.7145,
+      "step": 2132
+    },
+    {
+      "epoch": 1.9295392953929538,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00026142255591741267,
+      "loss": 0.722,
+      "step": 2136
+    },
+    {
+      "epoch": 1.933152664859982,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00026127849633023157,
+      "loss": 0.6763,
+      "step": 2140
+    },
+    {
+      "epoch": 1.9367660343270099,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002611342081153284,
+      "loss": 0.7073,
+      "step": 2144
+    },
+    {
+      "epoch": 1.9403794037940378,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002609896915691513,
+      "loss": 0.6981,
+      "step": 2148
+    },
+    {
+      "epoch": 1.943992773261066,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00026084494698861723,
+      "loss": 0.7239,
+      "step": 2152
+    },
+    {
+      "epoch": 1.947606142728094,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.000260699974671112,
+      "loss": 0.7198,
+      "step": 2156
+    },
+    {
+      "epoch": 1.951219512195122,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002605547749144889,
+      "loss": 0.6963,
+      "step": 2160
+    },
+    {
+      "epoch": 1.95483288166215,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00026040934801706874,
+      "loss": 0.7097,
+      "step": 2164
+    },
+    {
+      "epoch": 1.9584462511291778,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00026026369427763923,
+      "loss": 0.7057,
+      "step": 2168
+    },
+    {
+      "epoch": 1.962059620596206,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00026011781399545365,
+      "loss": 0.6887,
+      "step": 2172
+    },
+    {
+      "epoch": 1.965672990063234,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00025997170747023106,
+      "loss": 0.6859,
+      "step": 2176
+    },
+    {
+      "epoch": 1.9692863595302619,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00025982537500215524,
+      "loss": 0.6541,
+      "step": 2180
+    },
+    {
+      "epoch": 1.97289972899729,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00025967881689187424,
+      "loss": 0.6997,
+      "step": 2184
+    },
+    {
+      "epoch": 1.976513098464318,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00025953203344049965,
+      "loss": 0.6942,
+      "step": 2188
+    },
+    {
+      "epoch": 1.980126467931346,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00025938502494960607,
+      "loss": 0.77,
+      "step": 2192
+    },
+    {
+      "epoch": 1.9837398373983741,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00025923779172123035,
+      "loss": 0.6931,
+      "step": 2196
+    },
+    {
+      "epoch": 1.9873532068654018,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00025909033405787133,
+      "loss": 0.6612,
+      "step": 2200
+    },
+    {
+      "epoch": 1.99096657633243,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002589426522624886,
+      "loss": 0.691,
+      "step": 2204
+    },
+    {
+      "epoch": 1.994579945799458,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002587947466385026,
+      "loss": 0.7166,
+      "step": 2208
+    },
+    {
+      "epoch": 1.998193315266486,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002586466174897934,
+      "loss": 0.6862,
+      "step": 2212
+    },
+    {
+      "epoch": 2.001806684733514,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002584982651207005,
+      "loss": 0.6855,
+      "step": 2216
+    },
+    {
+      "epoch": 2.005420054200542,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00025834968983602175,
+      "loss": 0.606,
+      "step": 2220
+    },
+    {
+      "epoch": 2.00903342366757,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00025820089194101317,
+      "loss": 0.6066,
+      "step": 2224
+    },
+    {
+      "epoch": 2.012646793134598,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002580518717413882,
+      "loss": 0.647,
+      "step": 2228
+    },
+    {
+      "epoch": 2.016260162601626,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00025790262954331696,
+      "loss": 0.6474,
+      "step": 2232
+    },
+    {
+      "epoch": 2.019873532068654,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00025775316565342565,
+      "loss": 0.64,
+      "step": 2236
+    },
+    {
+      "epoch": 2.0234869015356822,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0002576034803787959,
+      "loss": 0.651,
+      "step": 2240
+    },
+    {
+      "epoch": 2.02710027100271,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00025745357402696424,
+      "loss": 0.6223,
+      "step": 2244
+    },
+    {
+      "epoch": 2.030713640469738,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002573034469059215,
+      "loss": 0.6823,
+      "step": 2248
+    },
+    {
+      "epoch": 2.034327009936766,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00025715309932411196,
+      "loss": 0.6411,
+      "step": 2252
+    },
+    {
+      "epoch": 2.037940379403794,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00025700253159043296,
+      "loss": 0.6004,
+      "step": 2256
+    },
+    {
+      "epoch": 2.041553748870822,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.000256851744014234,
+      "loss": 0.6005,
+      "step": 2260
+    },
+    {
+      "epoch": 2.04516711833785,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002567007369053164,
+      "loss": 0.6616,
+      "step": 2264
+    },
+    {
+      "epoch": 2.048780487804878,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00025654951057393254,
+      "loss": 0.5758,
+      "step": 2268
+    },
+    {
+      "epoch": 2.0523938572719063,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002563980653307851,
+      "loss": 0.6309,
+      "step": 2272
+    },
+    {
+      "epoch": 2.056007226738934,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00025624640148702647,
+      "loss": 0.6173,
+      "step": 2276
+    },
+    {
+      "epoch": 2.059620596205962,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00025609451935425844,
+      "loss": 0.6838,
+      "step": 2280
+    },
+    {
+      "epoch": 2.06323396567299,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00025594241924453096,
+      "loss": 0.6917,
+      "step": 2284
+    },
+    {
+      "epoch": 2.066847335140018,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0002557901014703421,
+      "loss": 0.6417,
+      "step": 2288
+    },
+    {
+      "epoch": 2.0704607046070462,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002556375663446369,
+      "loss": 0.6295,
+      "step": 2292
+    },
+    {
+      "epoch": 2.074074074074074,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00025548481418080713,
+      "loss": 0.6559,
+      "step": 2296
+    },
+    {
+      "epoch": 2.077687443541102,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002553318452926904,
+      "loss": 0.6531,
+      "step": 2300
+    },
+    {
+      "epoch": 2.08130081300813,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00025517865999456963,
+      "loss": 0.6311,
+      "step": 2304
+    },
+    {
+      "epoch": 2.084914182475158,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0002550252586011723,
+      "loss": 0.6364,
+      "step": 2308
+    },
+    {
+      "epoch": 2.088527551942186,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00025487164142767,
+      "loss": 0.6485,
+      "step": 2312
+    },
+    {
+      "epoch": 2.092140921409214,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00025471780878967744,
+      "loss": 0.6544,
+      "step": 2316
+    },
+    {
+      "epoch": 2.095754290876242,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00025456376100325224,
+      "loss": 0.6185,
+      "step": 2320
+    },
+    {
+      "epoch": 2.0993676603432703,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00025440949838489394,
+      "loss": 0.6092,
+      "step": 2324
+    },
+    {
+      "epoch": 2.102981029810298,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0002542550212515435,
+      "loss": 0.5907,
+      "step": 2328
+    },
+    {
+      "epoch": 2.106594399277326,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002541003299205825,
+      "loss": 0.6177,
+      "step": 2332
+    },
+    {
+      "epoch": 2.110207768744354,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0002539454247098328,
+      "loss": 0.6671,
+      "step": 2336
+    },
+    {
+      "epoch": 2.113821138211382,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00025379030593755545,
+      "loss": 0.5961,
+      "step": 2340
+    },
+    {
+      "epoch": 2.1174345076784102,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002536349739224505,
+      "loss": 0.6423,
+      "step": 2344
+    },
+    {
+      "epoch": 2.121047877145438,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00025347942898365603,
+      "loss": 0.5715,
+      "step": 2348
+    },
+    {
+      "epoch": 2.124661246612466,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002533236714407475,
+      "loss": 0.623,
+      "step": 2352
+    },
+    {
+      "epoch": 2.1282746160794943,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0002531677016137374,
+      "loss": 0.6196,
+      "step": 2356
+    },
+    {
+      "epoch": 2.131887985546522,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.000253011519823074,
+      "loss": 0.6454,
+      "step": 2360
+    },
+    {
+      "epoch": 2.13550135501355,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00025285512638964145,
+      "loss": 0.6169,
+      "step": 2364
+    },
+    {
+      "epoch": 2.139114724480578,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002526985216347585,
+      "loss": 0.6842,
+      "step": 2368
+    },
+    {
+      "epoch": 2.142728093947606,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0002525417058801781,
+      "loss": 0.6178,
+      "step": 2372
+    },
+    {
+      "epoch": 2.1463414634146343,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0002523846794480869,
+      "loss": 0.6206,
+      "step": 2376
+    },
+    {
+      "epoch": 2.149954832881662,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0002522274426611041,
+      "loss": 0.6268,
+      "step": 2380
+    },
+    {
+      "epoch": 2.15356820234869,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00025206999584228125,
+      "loss": 0.6667,
+      "step": 2384
+    },
+    {
+      "epoch": 2.1571815718157183,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00025191233931510143,
+      "loss": 0.646,
+      "step": 2388
+    },
+    {
+      "epoch": 2.160794941282746,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00025175447340347856,
+      "loss": 0.6293,
+      "step": 2392
+    },
+    {
+      "epoch": 2.1644083107497742,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002515963984317567,
+      "loss": 0.6158,
+      "step": 2396
+    },
+    {
+      "epoch": 2.168021680216802,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00025143811472470943,
+      "loss": 0.6386,
+      "step": 2400
+    },
+    {
+      "epoch": 2.17163504968383,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00025127962260753934,
+      "loss": 0.6441,
+      "step": 2404
+    },
+    {
+      "epoch": 2.1752484191508583,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002511209224058771,
+      "loss": 0.6115,
+      "step": 2408
+    },
+    {
+      "epoch": 2.178861788617886,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002509620144457808,
+      "loss": 0.6351,
+      "step": 2412
+    },
+    {
+      "epoch": 2.182475158084914,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0002508028990537356,
+      "loss": 0.6807,
+      "step": 2416
+    },
+    {
+      "epoch": 2.1860885275519424,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002506435765566527,
+      "loss": 0.6026,
+      "step": 2420
+    },
+    {
+      "epoch": 2.18970189701897,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0002504840472818687,
+      "loss": 0.6566,
+      "step": 2424
+    },
+    {
+      "epoch": 2.1933152664859983,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002503243115571454,
+      "loss": 0.6153,
+      "step": 2428
+    },
+    {
+      "epoch": 2.196928635953026,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00025016436971066837,
+      "loss": 0.632,
+      "step": 2432
+    },
+    {
+      "epoch": 2.200542005420054,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00025000422207104684,
+      "loss": 0.6327,
+      "step": 2436
+    },
+    {
+      "epoch": 2.2041553748870824,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002498438689673129,
+      "loss": 0.6611,
+      "step": 2440
+    },
+    {
+      "epoch": 2.20776874435411,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0002496833107289207,
+      "loss": 0.6004,
+      "step": 2444
+    },
+    {
+      "epoch": 2.2113821138211383,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00024952254768574584,
+      "loss": 0.581,
+      "step": 2448
+    },
+    {
+      "epoch": 2.2149954832881664,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002493615801680848,
+      "loss": 0.6382,
+      "step": 2452
+    },
+    {
+      "epoch": 2.218608852755194,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0002492004085066541,
+      "loss": 0.6183,
+      "step": 2456
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0002490390330325896,
+      "loss": 0.6295,
+      "step": 2460
+    },
+    {
+      "epoch": 2.22583559168925,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00024887745407744605,
+      "loss": 0.6535,
+      "step": 2464
+    },
+    {
+      "epoch": 2.229448961156278,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00024871567197319616,
+      "loss": 0.6478,
+      "step": 2468
+    },
+    {
+      "epoch": 2.2330623306233064,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002485536870522301,
+      "loss": 0.6159,
+      "step": 2472
+    },
+    {
+      "epoch": 2.236675700090334,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002483914996473547,
+      "loss": 0.6679,
+      "step": 2476
+    },
+    {
+      "epoch": 2.2402890695573623,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00024822911009179276,
+      "loss": 0.6093,
+      "step": 2480
+    },
+    {
+      "epoch": 2.2439024390243905,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0002480665187191825,
+      "loss": 0.6462,
+      "step": 2484
+    },
+    {
+      "epoch": 2.247515808491418,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00024790372586357666,
+      "loss": 0.6456,
+      "step": 2488
+    },
+    {
+      "epoch": 2.2511291779584464,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002477407318594421,
+      "loss": 0.6216,
+      "step": 2492
+    },
+    {
+      "epoch": 2.254742547425474,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002475775370416589,
+      "loss": 0.6621,
+      "step": 2496
+    },
+    {
+      "epoch": 2.2583559168925023,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0002474141417455195,
+      "loss": 0.6094,
+      "step": 2500
+    },
+    {
+      "epoch": 2.2619692863595304,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00024725054630672866,
+      "loss": 0.6318,
+      "step": 2504
+    },
+    {
+      "epoch": 2.265582655826558,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002470867510614019,
+      "loss": 0.6301,
+      "step": 2508
+    },
+    {
+      "epoch": 2.2691960252935863,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00024692275634606564,
+      "loss": 0.6492,
+      "step": 2512
+    },
+    {
+      "epoch": 2.272809394760614,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0002467585624976558,
+      "loss": 0.6677,
+      "step": 2516
+    },
+    {
+      "epoch": 2.2764227642276422,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00024659416985351763,
+      "loss": 0.6629,
+      "step": 2520
+    },
+    {
+      "epoch": 2.2800361336946704,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002464295787514047,
+      "loss": 0.6153,
+      "step": 2524
+    },
+    {
+      "epoch": 2.283649503161698,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00024626478952947847,
+      "loss": 0.5951,
+      "step": 2528
+    },
+    {
+      "epoch": 2.2872628726287263,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00024609980252630735,
+      "loss": 0.6298,
+      "step": 2532
+    },
+    {
+      "epoch": 2.2908762420957545,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.000245934618080866,
+      "loss": 0.6122,
+      "step": 2536
+    },
+    {
+      "epoch": 2.294489611562782,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002457692365325349,
+      "loss": 0.6431,
+      "step": 2540
+    },
+    {
+      "epoch": 2.2981029810298104,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00024560365822109953,
+      "loss": 0.6434,
+      "step": 2544
+    },
+    {
+      "epoch": 2.3017163504968385,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0002454378834867494,
+      "loss": 0.6324,
+      "step": 2548
+    },
+    {
+      "epoch": 2.3053297199638663,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0002452719126700778,
+      "loss": 0.6332,
+      "step": 2552
+    },
+    {
+      "epoch": 2.3089430894308944,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0002451057461120808,
+      "loss": 0.676,
+      "step": 2556
+    },
+    {
+      "epoch": 2.312556458897922,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00024493938415415657,
+      "loss": 0.6499,
+      "step": 2560
+    },
+    {
+      "epoch": 2.3161698283649503,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00024477282713810484,
+      "loss": 0.5823,
+      "step": 2564
+    },
+    {
+      "epoch": 2.3197831978319785,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00024460607540612603,
+      "loss": 0.6632,
+      "step": 2568
+    },
+    {
+      "epoch": 2.3233965672990062,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0002444391293008207,
+      "loss": 0.5896,
+      "step": 2572
+    },
+    {
+      "epoch": 2.3270099367660344,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002442719891651886,
+      "loss": 0.6416,
+      "step": 2576
+    },
+    {
+      "epoch": 2.330623306233062,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00024410465534262834,
+      "loss": 0.6521,
+      "step": 2580
+    },
+    {
+      "epoch": 2.3342366757000903,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0002439371281769363,
+      "loss": 0.5804,
+      "step": 2584
+    },
+    {
+      "epoch": 2.3378500451671185,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00024376940801230614,
+      "loss": 0.6091,
+      "step": 2588
+    },
+    {
+      "epoch": 2.341463414634146,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00024360149519332808,
+      "loss": 0.6804,
+      "step": 2592
+    },
+    {
+      "epoch": 2.3450767841011744,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00024343339006498813,
+      "loss": 0.6016,
+      "step": 2596
+    },
+    {
+      "epoch": 2.348690153568202,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00024326509297266746,
+      "loss": 0.6019,
+      "step": 2600
+    },
+    {
+      "epoch": 2.3523035230352303,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00024309660426214154,
+      "loss": 0.6194,
+      "step": 2604
+    },
+    {
+      "epoch": 2.3559168925022584,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00024292792427957962,
+      "loss": 0.6322,
+      "step": 2608
+    },
+    {
+      "epoch": 2.359530261969286,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00024275905337154386,
+      "loss": 0.6148,
+      "step": 2612
+    },
+    {
+      "epoch": 2.3631436314363143,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00024258999188498878,
+      "loss": 0.6429,
+      "step": 2616
+    },
+    {
+      "epoch": 2.3667570009033425,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00024242074016726035,
+      "loss": 0.6343,
+      "step": 2620
+    },
+    {
+      "epoch": 2.3703703703703702,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00024225129856609545,
+      "loss": 0.6325,
+      "step": 2624
+    },
+    {
+      "epoch": 2.3739837398373984,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00024208166742962107,
+      "loss": 0.6194,
+      "step": 2628
+    },
+    {
+      "epoch": 2.3775971093044266,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00024191184710635358,
+      "loss": 0.6261,
+      "step": 2632
+    },
+    {
+      "epoch": 2.3812104787714543,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0002417418379451981,
+      "loss": 0.5721,
+      "step": 2636
+    },
+    {
+      "epoch": 2.3848238482384825,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00024157164029544765,
+      "loss": 0.6571,
+      "step": 2640
+    },
+    {
+      "epoch": 2.38843721770551,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0002414012545067826,
+      "loss": 0.6082,
+      "step": 2644
+    },
+    {
+      "epoch": 2.3920505871725384,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0002412306809292698,
+      "loss": 0.5889,
+      "step": 2648
+    },
+    {
+      "epoch": 2.3956639566395665,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00024105991991336197,
+      "loss": 0.6266,
+      "step": 2652
+    },
+    {
+      "epoch": 2.3992773261065943,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00024088897180989683,
+      "loss": 0.6239,
+      "step": 2656
+    },
+    {
+      "epoch": 2.4028906955736224,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00024071783697009666,
+      "loss": 0.6489,
+      "step": 2660
+    },
+    {
+      "epoch": 2.40650406504065,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00024054651574556717,
+      "loss": 0.654,
+      "step": 2664
+    },
+    {
+      "epoch": 2.4101174345076783,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00024037500848829725,
+      "loss": 0.5871,
+      "step": 2668
+    },
+    {
+      "epoch": 2.4137308039747065,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0002402033155506578,
+      "loss": 0.6203,
+      "step": 2672
+    },
+    {
+      "epoch": 2.4173441734417342,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00024003143728540136,
+      "loss": 0.6587,
+      "step": 2676
+    },
+    {
+      "epoch": 2.4209575429087624,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00023985937404566115,
+      "loss": 0.6234,
+      "step": 2680
+    },
+    {
+      "epoch": 2.4245709123757906,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00023968712618495044,
+      "loss": 0.6166,
+      "step": 2684
+    },
+    {
+      "epoch": 2.4281842818428183,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002395146940571618,
+      "loss": 0.6067,
+      "step": 2688
+    },
+    {
+      "epoch": 2.4317976513098465,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00023934207801656651,
+      "loss": 0.6243,
+      "step": 2692
+    },
+    {
+      "epoch": 2.4354110207768747,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00023916927841781356,
+      "loss": 0.6855,
+      "step": 2696
+    },
+    {
+      "epoch": 2.4390243902439024,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00023899629561592903,
+      "loss": 0.6323,
+      "step": 2700
+    },
+    {
+      "epoch": 2.4426377597109306,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00023882312996631566,
+      "loss": 0.6511,
+      "step": 2704
+    },
+    {
+      "epoch": 2.4462511291779583,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00023864978182475154,
+      "loss": 0.6638,
+      "step": 2708
+    },
+    {
+      "epoch": 2.4498644986449865,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00023847625154738999,
+      "loss": 0.6051,
+      "step": 2712
+    },
+    {
+      "epoch": 2.4534778681120146,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00023830253949075827,
+      "loss": 0.6598,
+      "step": 2716
+    },
+    {
+      "epoch": 2.4570912375790424,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00023812864601175735,
+      "loss": 0.618,
+      "step": 2720
+    },
+    {
+      "epoch": 2.4607046070460705,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00023795457146766078,
+      "loss": 0.6816,
+      "step": 2724
+    },
+    {
+      "epoch": 2.4643179765130983,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0002377803162161142,
+      "loss": 0.6624,
+      "step": 2728
+    },
+    {
+      "epoch": 2.4679313459801264,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002376058806151345,
+      "loss": 0.6298,
+      "step": 2732
+    },
+    {
+      "epoch": 2.4715447154471546,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00023743126502310914,
+      "loss": 0.5964,
+      "step": 2736
+    },
+    {
+      "epoch": 2.4751580849141823,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00023725646979879528,
+      "loss": 0.6039,
+      "step": 2740
+    },
+    {
+      "epoch": 2.4787714543812105,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00023708149530131923,
+      "loss": 0.6232,
+      "step": 2744
+    },
+    {
+      "epoch": 2.4823848238482387,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00023690634189017567,
+      "loss": 0.6234,
+      "step": 2748
+    },
+    {
+      "epoch": 2.4859981933152664,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00023673100992522674,
+      "loss": 0.5961,
+      "step": 2752
+    },
+    {
+      "epoch": 2.4896115627822946,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00023655549976670152,
+      "loss": 0.6586,
+      "step": 2756
+    },
+    {
+      "epoch": 2.4932249322493227,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0002363798117751952,
+      "loss": 0.6162,
+      "step": 2760
+    },
+    {
+      "epoch": 2.4968383017163505,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00023620394631166828,
+      "loss": 0.6455,
+      "step": 2764
+    },
+    {
+      "epoch": 2.5004516711833786,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00023602790373744594,
+      "loss": 0.6159,
+      "step": 2768
+    },
+    {
+      "epoch": 2.5040650406504064,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002358516844142172,
+      "loss": 0.6267,
+      "step": 2772
+    },
+    {
+      "epoch": 2.5076784101174345,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00023567528870403425,
+      "loss": 0.6341,
+      "step": 2776
+    },
+    {
+      "epoch": 2.5112917795844627,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00023549871696931167,
+      "loss": 0.6187,
+      "step": 2780
+    },
+    {
+      "epoch": 2.5149051490514904,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00023532196957282568,
+      "loss": 0.6156,
+      "step": 2784
+    },
+    {
+      "epoch": 2.5185185185185186,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00023514504687771346,
+      "loss": 0.5952,
+      "step": 2788
+    },
+    {
+      "epoch": 2.5221318879855463,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00023496794924747228,
+      "loss": 0.6097,
+      "step": 2792
+    },
+    {
+      "epoch": 2.5257452574525745,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0002347906770459588,
+      "loss": 0.6406,
+      "step": 2796
+    },
+    {
+      "epoch": 2.5293586269196027,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0002346132306373885,
+      "loss": 0.6756,
+      "step": 2800
+    },
+    {
+      "epoch": 2.5329719963866304,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0002344356103863346,
+      "loss": 0.6112,
+      "step": 2804
+    },
+    {
+      "epoch": 2.5365853658536586,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00023425781665772758,
+      "loss": 0.6363,
+      "step": 2808
+    },
+    {
+      "epoch": 2.5401987353206863,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00023407984981685436,
+      "loss": 0.6156,
+      "step": 2812
+    },
+    {
+      "epoch": 2.5438121047877145,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00023390171022935746,
+      "loss": 0.6203,
+      "step": 2816
+    },
+    {
+      "epoch": 2.5474254742547426,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00023372339826123432,
+      "loss": 0.5995,
+      "step": 2820
+    },
+    {
+      "epoch": 2.551038843721771,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00023354491427883664,
+      "loss": 0.6362,
+      "step": 2824
+    },
+    {
+      "epoch": 2.5546522131887985,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00023336625864886943,
+      "loss": 0.6234,
+      "step": 2828
+    },
+    {
+      "epoch": 2.5582655826558267,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002331874317383904,
+      "loss": 0.6176,
+      "step": 2832
+    },
+    {
+      "epoch": 2.5618789521228544,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00023300843391480916,
+      "loss": 0.5973,
+      "step": 2836
+    },
+    {
+      "epoch": 2.5654923215898826,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002328292655458865,
+      "loss": 0.6319,
+      "step": 2840
+    },
+    {
+      "epoch": 2.569105691056911,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00023264992699973357,
+      "loss": 0.6267,
+      "step": 2844
+    },
+    {
+      "epoch": 2.5727190605239385,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0002324704186448111,
+      "loss": 0.6169,
+      "step": 2848
+    },
+    {
+      "epoch": 2.5763324299909667,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00023229074084992884,
+      "loss": 0.6681,
+      "step": 2852
+    },
+    {
+      "epoch": 2.5799457994579944,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00023211089398424457,
+      "loss": 0.6289,
+      "step": 2856
+    },
+    {
+      "epoch": 2.5835591689250226,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00023193087841726347,
+      "loss": 0.66,
+      "step": 2860
+    },
+    {
+      "epoch": 2.5871725383920507,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00023175069451883727,
+      "loss": 0.6165,
+      "step": 2864
+    },
+    {
+      "epoch": 2.5907859078590785,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00023157034265916364,
+      "loss": 0.6014,
+      "step": 2868
+    },
+    {
+      "epoch": 2.5943992773261066,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0002313898232087852,
+      "loss": 0.5906,
+      "step": 2872
+    },
+    {
+      "epoch": 2.5980126467931344,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0002312091365385891,
+      "loss": 0.6539,
+      "step": 2876
+    },
+    {
+      "epoch": 2.6016260162601625,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00023102828301980582,
+      "loss": 0.6245,
+      "step": 2880
+    },
+    {
+      "epoch": 2.6052393857271907,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00023084726302400883,
+      "loss": 0.6089,
+      "step": 2884
+    },
+    {
+      "epoch": 2.6088527551942184,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00023066607692311347,
+      "loss": 0.6076,
+      "step": 2888
+    },
+    {
+      "epoch": 2.6124661246612466,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00023048472508937652,
+      "loss": 0.6506,
+      "step": 2892
+    },
+    {
+      "epoch": 2.6160794941282743,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0002303032078953951,
+      "loss": 0.6154,
+      "step": 2896
+    },
+    {
+      "epoch": 2.6196928635953025,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00023012152571410627,
+      "loss": 0.6112,
+      "step": 2900
+    },
+    {
+      "epoch": 2.6233062330623307,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00022993967891878585,
+      "loss": 0.6254,
+      "step": 2904
+    },
+    {
+      "epoch": 2.626919602529359,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.000229757667883048,
+      "loss": 0.6436,
+      "step": 2908
+    },
+    {
+      "epoch": 2.6305329719963866,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00022957549298084433,
+      "loss": 0.6586,
+      "step": 2912
+    },
+    {
+      "epoch": 2.6341463414634148,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.000229393154586463,
+      "loss": 0.6565,
+      "step": 2916
+    },
+    {
+      "epoch": 2.6377597109304425,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00022921065307452825,
+      "loss": 0.6168,
+      "step": 2920
+    },
+    {
+      "epoch": 2.6413730803974707,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0002290279888199993,
+      "loss": 0.6327,
+      "step": 2924
+    },
+    {
+      "epoch": 2.644986449864499,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002288451621981698,
+      "loss": 0.5869,
+      "step": 2928
+    },
+    {
+      "epoch": 2.6485998193315266,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00022866217358466704,
+      "loss": 0.6394,
+      "step": 2932
+    },
+    {
+      "epoch": 2.6522131887985547,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00022847902335545097,
+      "loss": 0.6287,
+      "step": 2936
+    },
+    {
+      "epoch": 2.6558265582655824,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00022829571188681382,
+      "loss": 0.6535,
+      "step": 2940
+    },
+    {
+      "epoch": 2.6594399277326106,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00022811223955537886,
+      "loss": 0.676,
+      "step": 2944
+    },
+    {
+      "epoch": 2.663053297199639,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00022792860673810005,
+      "loss": 0.6782,
+      "step": 2948
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00022774481381226095,
+      "loss": 0.6651,
+      "step": 2952
+    },
+    {
+      "epoch": 2.6702800361336947,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00022756086115547416,
+      "loss": 0.624,
+      "step": 2956
+    },
+    {
+      "epoch": 2.6738934056007224,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00022737674914568039,
+      "loss": 0.6237,
+      "step": 2960
+    },
+    {
+      "epoch": 2.6775067750677506,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00022719247816114783,
+      "loss": 0.6197,
+      "step": 2964
+    },
+    {
+      "epoch": 2.6811201445347788,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0002270080485804711,
+      "loss": 0.6159,
+      "step": 2968
+    },
+    {
+      "epoch": 2.684733514001807,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.000226823460782571,
+      "loss": 0.6186,
+      "step": 2972
+    },
+    {
+      "epoch": 2.6883468834688347,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00022663871514669304,
+      "loss": 0.6349,
+      "step": 2976
+    },
+    {
+      "epoch": 2.691960252935863,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00022645381205240722,
+      "loss": 0.6122,
+      "step": 2980
+    },
+    {
+      "epoch": 2.6955736224028906,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00022626875187960703,
+      "loss": 0.6423,
+      "step": 2984
+    },
+    {
+      "epoch": 2.6991869918699187,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00022608353500850863,
+      "loss": 0.6259,
+      "step": 2988
+    },
+    {
+      "epoch": 2.702800361336947,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00022589816181965022,
+      "loss": 0.6267,
+      "step": 2992
+    },
+    {
+      "epoch": 2.7064137308039746,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.000225712632693891,
+      "loss": 0.6657,
+      "step": 2996
+    },
+    {
+      "epoch": 2.710027100271003,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00022552694801241066,
+      "loss": 0.6439,
+      "step": 3000
+    },
+    {
+      "epoch": 2.7136404697380305,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00022534110815670855,
+      "loss": 0.6932,
+      "step": 3004
+    },
+    {
+      "epoch": 2.7172538392050587,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002251551135086027,
+      "loss": 0.6355,
+      "step": 3008
+    },
+    {
+      "epoch": 2.720867208672087,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00022496896445022924,
+      "loss": 0.6464,
+      "step": 3012
+    },
+    {
+      "epoch": 2.7244805781391146,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00022478266136404153,
+      "loss": 0.6115,
+      "step": 3016
+    },
+    {
+      "epoch": 2.7280939476061428,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0002245962046328094,
+      "loss": 0.633,
+      "step": 3020
+    },
+    {
+      "epoch": 2.7317073170731705,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00022440959463961832,
+      "loss": 0.6263,
+      "step": 3024
+    },
+    {
+      "epoch": 2.7353206865401987,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00022422283176786871,
+      "loss": 0.6686,
+      "step": 3028
+    },
+    {
+      "epoch": 2.738934056007227,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00022403591640127505,
+      "loss": 0.6192,
+      "step": 3032
+    },
+    {
+      "epoch": 2.742547425474255,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00022384884892386509,
+      "loss": 0.6031,
+      "step": 3036
+    },
+    {
+      "epoch": 2.7461607949412827,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00022366162971997915,
+      "loss": 0.645,
+      "step": 3040
+    },
+    {
+      "epoch": 2.749774164408311,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00022347425917426927,
+      "loss": 0.6432,
+      "step": 3044
+    },
+    {
+      "epoch": 2.7533875338753386,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00022328673767169841,
+      "loss": 0.626,
+      "step": 3048
+    },
+    {
+      "epoch": 2.757000903342367,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00022309906559753965,
+      "loss": 0.6708,
+      "step": 3052
+    },
+    {
+      "epoch": 2.760614272809395,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002229112433373756,
+      "loss": 0.6306,
+      "step": 3056
+    },
+    {
+      "epoch": 2.7642276422764227,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00022272327127709717,
+      "loss": 0.6488,
+      "step": 3060
+    },
+    {
+      "epoch": 2.767841011743451,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00022253514980290326,
+      "loss": 0.6636,
+      "step": 3064
+    },
+    {
+      "epoch": 2.7714543812104786,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002223468793012997,
+      "loss": 0.6511,
+      "step": 3068
+    },
+    {
+      "epoch": 2.7750677506775068,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00022215846015909835,
+      "loss": 0.6433,
+      "step": 3072
+    },
+    {
+      "epoch": 2.778681120144535,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0002219698927634167,
+      "loss": 0.6107,
+      "step": 3076
+    },
+    {
+      "epoch": 2.7822944896115627,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00022178117750167663,
+      "loss": 0.6756,
+      "step": 3080
+    },
+    {
+      "epoch": 2.785907859078591,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00022159231476160402,
+      "loss": 0.6706,
+      "step": 3084
+    },
+    {
+      "epoch": 2.7895212285456186,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00022140330493122752,
+      "loss": 0.6185,
+      "step": 3088
+    },
+    {
+      "epoch": 2.7931345980126467,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00022121414839887813,
+      "loss": 0.6646,
+      "step": 3092
+    },
+    {
+      "epoch": 2.796747967479675,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00022102484555318834,
+      "loss": 0.6591,
+      "step": 3096
+    },
+    {
+      "epoch": 2.8003613369467026,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00022083539678309098,
+      "loss": 0.6143,
+      "step": 3100
+    },
+    {
+      "epoch": 2.803974706413731,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00022064580247781903,
+      "loss": 0.6596,
+      "step": 3104
+    },
+    {
+      "epoch": 2.8075880758807585,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00022045606302690412,
+      "loss": 0.6317,
+      "step": 3108
+    },
+    {
+      "epoch": 2.8112014453477867,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00022026617882017638,
+      "loss": 0.6856,
+      "step": 3112
+    },
+    {
+      "epoch": 2.814814814814815,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0002200761502477632,
+      "loss": 0.6679,
+      "step": 3116
+    },
+    {
+      "epoch": 2.818428184281843,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0002198859777000886,
+      "loss": 0.6058,
+      "step": 3120
+    },
+    {
+      "epoch": 2.822041553748871,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00021969566156787248,
+      "loss": 0.6112,
+      "step": 3124
+    },
+    {
+      "epoch": 2.825654923215899,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00021950520224212967,
+      "loss": 0.6011,
+      "step": 3128
+    },
+    {
+      "epoch": 2.8292682926829267,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00021931460011416913,
+      "loss": 0.6622,
+      "step": 3132
+    },
+    {
+      "epoch": 2.832881662149955,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0002191238555755934,
+      "loss": 0.6222,
+      "step": 3136
+    },
+    {
+      "epoch": 2.836495031616983,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00021893296901829743,
+      "loss": 0.608,
+      "step": 3140
+    },
+    {
+      "epoch": 2.8401084010840107,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0002187419408344681,
+      "loss": 0.6183,
+      "step": 3144
+    },
+    {
+      "epoch": 2.843721770551039,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00021855077141658317,
+      "loss": 0.6442,
+      "step": 3148
+    },
+    {
+      "epoch": 2.8473351400180666,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00021835946115741057,
+      "loss": 0.6435,
+      "step": 3152
+    },
+    {
+      "epoch": 2.850948509485095,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00021816801045000767,
+      "loss": 0.643,
+      "step": 3156
+    },
+    {
+      "epoch": 2.854561878952123,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00021797641968772038,
+      "loss": 0.6622,
+      "step": 3160
+    },
+    {
+      "epoch": 2.8581752484191507,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00021778468926418227,
+      "loss": 0.6305,
+      "step": 3164
+    },
+    {
+      "epoch": 2.861788617886179,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.000217592819573314,
+      "loss": 0.6354,
+      "step": 3168
+    },
+    {
+      "epoch": 2.8654019873532066,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00021740081100932225,
+      "loss": 0.6947,
+      "step": 3172
+    },
+    {
+      "epoch": 2.869015356820235,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00021720866396669904,
+      "loss": 0.6697,
+      "step": 3176
+    },
+    {
+      "epoch": 2.872628726287263,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00021701637884022094,
+      "loss": 0.5763,
+      "step": 3180
+    },
+    {
+      "epoch": 2.876242095754291,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00021682395602494817,
+      "loss": 0.6741,
+      "step": 3184
+    },
+    {
+      "epoch": 2.879855465221319,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0002166313959162239,
+      "loss": 0.6487,
+      "step": 3188
+    },
+    {
+      "epoch": 2.883468834688347,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00021643869890967336,
+      "loss": 0.6081,
+      "step": 3192
+    },
+    {
+      "epoch": 2.8870822041553748,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00021624586540120296,
+      "loss": 0.6229,
+      "step": 3196
+    },
+    {
+      "epoch": 2.890695573622403,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002160528957869997,
+      "loss": 0.6686,
+      "step": 3200
+    },
+    {
+      "epoch": 2.894308943089431,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00021585979046353008,
+      "loss": 0.638,
+      "step": 3204
+    },
+    {
+      "epoch": 2.897922312556459,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00021566654982753955,
+      "loss": 0.5879,
+      "step": 3208
+    },
+    {
+      "epoch": 2.901535682023487,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00021547317427605148,
+      "loss": 0.5981,
+      "step": 3212
+    },
+    {
+      "epoch": 2.9051490514905147,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00021527966420636644,
+      "loss": 0.6293,
+      "step": 3216
+    },
+    {
+      "epoch": 2.908762420957543,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00021508602001606145,
+      "loss": 0.6284,
+      "step": 3220
+    },
+    {
+      "epoch": 2.912375790424571,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00021489224210298894,
+      "loss": 0.5764,
+      "step": 3224
+    },
+    {
+      "epoch": 2.915989159891599,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00021469833086527624,
+      "loss": 0.6103,
+      "step": 3228
+    },
+    {
+      "epoch": 2.919602529358627,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00021450428670132448,
+      "loss": 0.6215,
+      "step": 3232
+    },
+    {
+      "epoch": 2.9232158988256547,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.000214310110009808,
+      "loss": 0.6189,
+      "step": 3236
+    },
+    {
+      "epoch": 2.926829268292683,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00021411580118967335,
+      "loss": 0.6523,
+      "step": 3240
+    },
+    {
+      "epoch": 2.930442637759711,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002139213606401386,
+      "loss": 0.651,
+      "step": 3244
+    },
+    {
+      "epoch": 2.934056007226739,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00021372678876069236,
+      "loss": 0.6704,
+      "step": 3248
+    },
+    {
+      "epoch": 2.937669376693767,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00021353208595109317,
+      "loss": 0.6155,
+      "step": 3252
+    },
+    {
+      "epoch": 2.941282746160795,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00021333725261136856,
+      "loss": 0.6136,
+      "step": 3256
+    },
+    {
+      "epoch": 2.944896115627823,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00021314228914181418,
+      "loss": 0.6262,
+      "step": 3260
+    },
+    {
+      "epoch": 2.948509485094851,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00021294719594299315,
+      "loss": 0.6668,
+      "step": 3264
+    },
+    {
+      "epoch": 2.952122854561879,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.000212751973415735,
+      "loss": 0.6747,
+      "step": 3268
+    },
+    {
+      "epoch": 2.955736224028907,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00021255662196113502,
+      "loss": 0.6239,
+      "step": 3272
+    },
+    {
+      "epoch": 2.959349593495935,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00021236114198055342,
+      "loss": 0.6522,
+      "step": 3276
+    },
+    {
+      "epoch": 2.962962962962963,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00021216553387561445,
+      "loss": 0.6569,
+      "step": 3280
+    },
+    {
+      "epoch": 2.966576332429991,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00021196979804820562,
+      "loss": 0.6878,
+      "step": 3284
+    },
+    {
+      "epoch": 2.970189701897019,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00021177393490047678,
+      "loss": 0.6494,
+      "step": 3288
+    },
+    {
+      "epoch": 2.973803071364047,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00021157794483483947,
+      "loss": 0.6326,
+      "step": 3292
+    },
+    {
+      "epoch": 2.977416440831075,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00021138182825396594,
+      "loss": 0.6859,
+      "step": 3296
+    },
+    {
+      "epoch": 2.9810298102981028,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00021118558556078827,
+      "loss": 0.6028,
+      "step": 3300
+    },
+    {
+      "epoch": 2.984643179765131,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00021098921715849786,
+      "loss": 0.6063,
+      "step": 3304
+    },
+    {
+      "epoch": 2.988256549232159,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00021079272345054422,
+      "loss": 0.6191,
+      "step": 3308
+    },
+    {
+      "epoch": 2.991869918699187,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00021059610484063437,
+      "loss": 0.5924,
+      "step": 3312
+    },
+    {
+      "epoch": 2.995483288166215,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00021039936173273196,
+      "loss": 0.6085,
+      "step": 3316
+    },
+    {
+      "epoch": 2.9990966576332427,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00021020249453105637,
+      "loss": 0.5925,
+      "step": 3320
+    },
+    {
+      "epoch": 3.002710027100271,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00021000550364008198,
+      "loss": 0.5536,
+      "step": 3324
+    },
+    {
+      "epoch": 3.006323396567299,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00020980838946453727,
+      "loss": 0.5074,
+      "step": 3328
+    },
+    {
+      "epoch": 3.009936766034327,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00020961115240940404,
+      "loss": 0.5476,
+      "step": 3332
+    },
+    {
+      "epoch": 3.013550135501355,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00020941379287991657,
+      "loss": 0.5756,
+      "step": 3336
+    },
+    {
+      "epoch": 3.017163504968383,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00020921631128156075,
+      "loss": 0.5602,
+      "step": 3340
+    },
+    {
+      "epoch": 3.020776874435411,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00020901870802007323,
+      "loss": 0.5553,
+      "step": 3344
+    },
+    {
+      "epoch": 3.024390243902439,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00020882098350144066,
+      "loss": 0.5324,
+      "step": 3348
+    },
+    {
+      "epoch": 3.028003613369467,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00020862313813189886,
+      "loss": 0.5725,
+      "step": 3352
+    },
+    {
+      "epoch": 3.031616982836495,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0002084251723179318,
+      "loss": 0.5595,
+      "step": 3356
+    },
+    {
+      "epoch": 3.035230352303523,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00020822708646627117,
+      "loss": 0.5841,
+      "step": 3360
+    },
+    {
+      "epoch": 3.038843721770551,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00020802888098389493,
+      "loss": 0.5697,
+      "step": 3364
+    },
+    {
+      "epoch": 3.042457091237579,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0002078305562780272,
+      "loss": 0.5432,
+      "step": 3368
+    },
+    {
+      "epoch": 3.046070460704607,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00020763211275613678,
+      "loss": 0.5347,
+      "step": 3372
+    },
+    {
+      "epoch": 3.049683830171635,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00020743355082593662,
+      "loss": 0.5611,
+      "step": 3376
+    },
+    {
+      "epoch": 3.053297199638663,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00020723487089538317,
+      "loss": 0.5526,
+      "step": 3380
+    },
+    {
+      "epoch": 3.0569105691056913,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.000207036073372675,
+      "loss": 0.5453,
+      "step": 3384
+    },
+    {
+      "epoch": 3.060523938572719,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00020683715866625253,
+      "loss": 0.5596,
+      "step": 3388
+    },
+    {
+      "epoch": 3.064137308039747,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00020663812718479675,
+      "loss": 0.5554,
+      "step": 3392
+    },
+    {
+      "epoch": 3.067750677506775,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00020643897933722875,
+      "loss": 0.5474,
+      "step": 3396
+    },
+    {
+      "epoch": 3.071364046973803,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002062397155327086,
+      "loss": 0.5386,
+      "step": 3400
+    },
+    {
+      "epoch": 3.0749774164408312,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00020604033618063454,
+      "loss": 0.538,
+      "step": 3404
+    },
+    {
+      "epoch": 3.078590785907859,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0002058408416906424,
+      "loss": 0.5471,
+      "step": 3408
+    },
+    {
+      "epoch": 3.082204155374887,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00020564123247260432,
+      "loss": 0.573,
+      "step": 3412
+    },
+    {
+      "epoch": 3.0858175248419153,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00020544150893662846,
+      "loss": 0.5385,
+      "step": 3416
+    },
+    {
+      "epoch": 3.089430894308943,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00020524167149305758,
+      "loss": 0.5666,
+      "step": 3420
+    },
+    {
+      "epoch": 3.093044263775971,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0002050417205524686,
+      "loss": 0.5123,
+      "step": 3424
+    },
+    {
+      "epoch": 3.096657633242999,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00020484165652567157,
+      "loss": 0.5652,
+      "step": 3428
+    },
+    {
+      "epoch": 3.100271002710027,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00020464147982370896,
+      "loss": 0.5819,
+      "step": 3432
+    },
+    {
+      "epoch": 3.1038843721770553,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00020444119085785467,
+      "loss": 0.5436,
+      "step": 3436
+    },
+    {
+      "epoch": 3.107497741644083,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00020424079003961328,
+      "loss": 0.5348,
+      "step": 3440
+    },
+    {
+      "epoch": 3.111111111111111,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0002040402777807192,
+      "loss": 0.5701,
+      "step": 3444
+    },
+    {
+      "epoch": 3.114724480578139,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00020383965449313567,
+      "loss": 0.5627,
+      "step": 3448
+    },
+    {
+      "epoch": 3.118337850045167,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00020363892058905422,
+      "loss": 0.5389,
+      "step": 3452
+    },
+    {
+      "epoch": 3.1219512195121952,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00020343807648089357,
+      "loss": 0.5448,
+      "step": 3456
+    },
+    {
+      "epoch": 3.125564588979223,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00020323712258129884,
+      "loss": 0.5465,
+      "step": 3460
+    },
+    {
+      "epoch": 3.129177958446251,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00020303605930314077,
+      "loss": 0.5307,
+      "step": 3464
+    },
+    {
+      "epoch": 3.1327913279132793,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00020283488705951475,
+      "loss": 0.5384,
+      "step": 3468
+    },
+    {
+      "epoch": 3.136404697380307,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0002026336062637402,
+      "loss": 0.5594,
+      "step": 3472
+    },
+    {
+      "epoch": 3.140018066847335,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00020243221732935932,
+      "loss": 0.529,
+      "step": 3476
+    },
+    {
+      "epoch": 3.1436314363143634,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00020223072067013673,
+      "loss": 0.5833,
+      "step": 3480
+    },
+    {
+      "epoch": 3.147244805781391,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00020202911670005823,
+      "loss": 0.5319,
+      "step": 3484
+    },
+    {
+      "epoch": 3.1508581752484193,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00020182740583333012,
+      "loss": 0.5311,
+      "step": 3488
+    },
+    {
+      "epoch": 3.154471544715447,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00020162558848437842,
+      "loss": 0.5411,
+      "step": 3492
+    },
+    {
+      "epoch": 3.158084914182475,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0002014236650678478,
+      "loss": 0.5302,
+      "step": 3496
+    },
+    {
+      "epoch": 3.1616982836495033,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00020122163599860086,
+      "loss": 0.5373,
+      "step": 3500
+    },
+    {
+      "epoch": 3.165311653116531,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00020101950169171737,
+      "loss": 0.5507,
+      "step": 3504
+    },
+    {
+      "epoch": 3.1689250225835592,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0002008172625624932,
+      "loss": 0.5737,
+      "step": 3508
+    },
+    {
+      "epoch": 3.172538392050587,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00020061491902643972,
+      "loss": 0.5877,
+      "step": 3512
+    },
+    {
+      "epoch": 3.176151761517615,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00020041247149928267,
+      "loss": 0.5461,
+      "step": 3516
+    },
+    {
+      "epoch": 3.1797651309846433,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00020020992039696152,
+      "loss": 0.5582,
+      "step": 3520
+    },
+    {
+      "epoch": 3.183378500451671,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00020000726613562852,
+      "loss": 0.5615,
+      "step": 3524
+    },
+    {
+      "epoch": 3.186991869918699,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00019980450913164785,
+      "loss": 0.5438,
+      "step": 3528
+    },
+    {
+      "epoch": 3.1906052393857274,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019960164980159484,
+      "loss": 0.5615,
+      "step": 3532
+    },
+    {
+      "epoch": 3.194218608852755,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.000199398688562255,
+      "loss": 0.5521,
+      "step": 3536
+    },
+    {
+      "epoch": 3.1978319783197833,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0001991956258306232,
+      "loss": 0.5773,
+      "step": 3540
+    },
+    {
+      "epoch": 3.201445347786811,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00019899246202390294,
+      "loss": 0.5693,
+      "step": 3544
+    },
+    {
+      "epoch": 3.205058717253839,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0001987891975595052,
+      "loss": 0.5533,
+      "step": 3548
+    },
+    {
+      "epoch": 3.2086720867208673,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.000198585832855048,
+      "loss": 0.5302,
+      "step": 3552
+    },
+    {
+      "epoch": 3.212285456187895,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001983823683283551,
+      "loss": 0.5504,
+      "step": 3556
+    },
+    {
+      "epoch": 3.2158988256549232,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00019817880439745546,
+      "loss": 0.5243,
+      "step": 3560
+    },
+    {
+      "epoch": 3.2195121951219514,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00019797514148058224,
+      "loss": 0.5859,
+      "step": 3564
+    },
+    {
+      "epoch": 3.223125564588979,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.000197771379996172,
+      "loss": 0.588,
+      "step": 3568
+    },
+    {
+      "epoch": 3.2267389340560073,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019756752036286375,
+      "loss": 0.5883,
+      "step": 3572
+    },
+    {
+      "epoch": 3.230352303523035,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0001973635629994982,
+      "loss": 0.514,
+      "step": 3576
+    },
+    {
+      "epoch": 3.233965672990063,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001971595083251169,
+      "loss": 0.5427,
+      "step": 3580
+    },
+    {
+      "epoch": 3.2375790424570914,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019695535675896115,
+      "loss": 0.513,
+      "step": 3584
+    },
+    {
+      "epoch": 3.241192411924119,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001967511087204716,
+      "loss": 0.5919,
+      "step": 3588
+    },
+    {
+      "epoch": 3.2448057813911473,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001965467646292868,
+      "loss": 0.5145,
+      "step": 3592
+    },
+    {
+      "epoch": 3.2484191508581755,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019634232490524297,
+      "loss": 0.5296,
+      "step": 3596
+    },
+    {
+      "epoch": 3.252032520325203,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019613778996837252,
+      "loss": 0.5701,
+      "step": 3600
+    },
+    {
+      "epoch": 3.2556458897922314,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019593316023890358,
+      "loss": 0.5738,
+      "step": 3604
+    },
+    {
+      "epoch": 3.259259259259259,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019572843613725917,
+      "loss": 0.5653,
+      "step": 3608
+    },
+    {
+      "epoch": 3.2628726287262872,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00019552361808405596,
+      "loss": 0.5736,
+      "step": 3612
+    },
+    {
+      "epoch": 3.2664859981933154,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001953187065001039,
+      "loss": 0.584,
+      "step": 3616
+    },
+    {
+      "epoch": 3.270099367660343,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019511370180640486,
+      "loss": 0.5677,
+      "step": 3620
+    },
+    {
+      "epoch": 3.2737127371273713,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00019490860442415226,
+      "loss": 0.5222,
+      "step": 3624
+    },
+    {
+      "epoch": 3.2773261065943995,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001947034147747297,
+      "loss": 0.5507,
+      "step": 3628
+    },
+    {
+      "epoch": 3.280939476061427,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00019449813327971057,
+      "loss": 0.5827,
+      "step": 3632
+    },
+    {
+      "epoch": 3.2845528455284554,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001942927603608568,
+      "loss": 0.5695,
+      "step": 3636
+    },
+    {
+      "epoch": 3.288166214995483,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019408729644011826,
+      "loss": 0.5417,
+      "step": 3640
+    },
+    {
+      "epoch": 3.2917795844625113,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00019388174193963174,
+      "loss": 0.5685,
+      "step": 3644
+    },
+    {
+      "epoch": 3.2953929539295395,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019367609728172016,
+      "loss": 0.6039,
+      "step": 3648
+    },
+    {
+      "epoch": 3.299006323396567,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001934703628888916,
+      "loss": 0.5437,
+      "step": 3652
+    },
+    {
+      "epoch": 3.3026196928635954,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001932645391838386,
+      "loss": 0.521,
+      "step": 3656
+    },
+    {
+      "epoch": 3.306233062330623,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0001930586265894371,
+      "loss": 0.563,
+      "step": 3660
+    },
+    {
+      "epoch": 3.3098464317976513,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001928526255287458,
+      "loss": 0.5217,
+      "step": 3664
+    },
+    {
+      "epoch": 3.3134598012646794,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.000192646536425005,
+      "loss": 0.5684,
+      "step": 3668
+    },
+    {
+      "epoch": 3.317073170731707,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019244035970163604,
+      "loss": 0.5654,
+      "step": 3672
+    },
+    {
+      "epoch": 3.3206865401987353,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019223409578224022,
+      "loss": 0.5945,
+      "step": 3676
+    },
+    {
+      "epoch": 3.3242999096657635,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019202774509059789,
+      "loss": 0.5358,
+      "step": 3680
+    },
+    {
+      "epoch": 3.3279132791327912,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019182130805066785,
+      "loss": 0.5368,
+      "step": 3684
+    },
+    {
+      "epoch": 3.3315266485998194,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019161478508658615,
+      "loss": 0.5286,
+      "step": 3688
+    },
+    {
+      "epoch": 3.3351400180668476,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019140817662266556,
+      "loss": 0.5547,
+      "step": 3692
+    },
+    {
+      "epoch": 3.3387533875338753,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0001912014830833943,
+      "loss": 0.5808,
+      "step": 3696
+    },
+    {
+      "epoch": 3.3423667570009035,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019099470489343558,
+      "loss": 0.5341,
+      "step": 3700
+    },
+    {
+      "epoch": 3.345980126467931,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019078784247762633,
+      "loss": 0.6313,
+      "step": 3704
+    },
+    {
+      "epoch": 3.3495934959349594,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019058089626097668,
+      "loss": 0.6229,
+      "step": 3708
+    },
+    {
+      "epoch": 3.3532068654019875,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001903738666686689,
+      "loss": 0.5368,
+      "step": 3712
+    },
+    {
+      "epoch": 3.3568202348690153,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019016675412605652,
+      "loss": 0.5718,
+      "step": 3716
+    },
+    {
+      "epoch": 3.3604336043360434,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0001899595590586636,
+      "loss": 0.5737,
+      "step": 3720
+    },
+    {
+      "epoch": 3.364046973803071,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00018975228189218354,
+      "loss": 0.5556,
+      "step": 3724
+    },
+    {
+      "epoch": 3.3676603432700993,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00018954492305247865,
+      "loss": 0.5544,
+      "step": 3728
+    },
+    {
+      "epoch": 3.3712737127371275,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001893374829655789,
+      "loss": 0.5282,
+      "step": 3732
+    },
+    {
+      "epoch": 3.3748870822041552,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018912996205768125,
+      "loss": 0.5734,
+      "step": 3736
+    },
+    {
+      "epoch": 3.3785004516711834,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00018892236075514869,
+      "loss": 0.6076,
+      "step": 3740
+    },
+    {
+      "epoch": 3.3821138211382116,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018871467948450936,
+      "loss": 0.5469,
+      "step": 3744
+    },
+    {
+      "epoch": 3.3857271906052393,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00018850691867245572,
+      "loss": 0.5731,
+      "step": 3748
+    },
+    {
+      "epoch": 3.3893405600722675,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018829907874584376,
+      "loss": 0.5417,
+      "step": 3752
+    },
+    {
+      "epoch": 3.392953929539295,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018809116013169177,
+      "loss": 0.5449,
+      "step": 3756
+    },
+    {
+      "epoch": 3.3965672990063234,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018788316325717993,
+      "loss": 0.537,
+      "step": 3760
+    },
+    {
+      "epoch": 3.4001806684733515,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00018767508854964908,
+      "loss": 0.5881,
+      "step": 3764
+    },
+    {
+      "epoch": 3.4037940379403793,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00018746693643660007,
+      "loss": 0.5387,
+      "step": 3768
+    },
+    {
+      "epoch": 3.4074074074074074,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001872587073456927,
+      "loss": 0.5376,
+      "step": 3772
+    },
+    {
+      "epoch": 3.4110207768744356,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00018705040170474495,
+      "loss": 0.56,
+      "step": 3776
+    },
+    {
+      "epoch": 3.4146341463414633,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00018684201994173216,
+      "loss": 0.6074,
+      "step": 3780
+    },
+    {
+      "epoch": 3.4182475158084915,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018663356248478592,
+      "loss": 0.5337,
+      "step": 3784
+    },
+    {
+      "epoch": 3.4218608852755192,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0001864250297621935,
+      "loss": 0.5546,
+      "step": 3788
+    },
+    {
+      "epoch": 3.4254742547425474,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00018621642220239653,
+      "loss": 0.5278,
+      "step": 3792
+    },
+    {
+      "epoch": 3.4290876242095756,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001860077402339908,
+      "loss": 0.5097,
+      "step": 3796
+    },
+    {
+      "epoch": 3.4327009936766033,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018579898428572466,
+      "loss": 0.5837,
+      "step": 3800
+    },
+    {
+      "epoch": 3.4363143631436315,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018559015478649858,
+      "loss": 0.5655,
+      "step": 3804
+    },
+    {
+      "epoch": 3.439927732610659,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00018538125216536412,
+      "loss": 0.5507,
+      "step": 3808
+    },
+    {
+      "epoch": 3.4435411020776874,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00018517227685152304,
+      "loss": 0.5943,
+      "step": 3812
+    },
+    {
+      "epoch": 3.4471544715447155,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00018496322927432653,
+      "loss": 0.5815,
+      "step": 3816
+    },
+    {
+      "epoch": 3.4507678410117433,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00018475410986327416,
+      "loss": 0.5206,
+      "step": 3820
+    },
+    {
+      "epoch": 3.4543812104787714,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018454491904801315,
+      "loss": 0.5722,
+      "step": 3824
+    },
+    {
+      "epoch": 3.4579945799457996,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00018433565725833753,
+      "loss": 0.5577,
+      "step": 3828
+    },
+    {
+      "epoch": 3.4616079494128273,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00018412632492418682,
+      "loss": 0.5678,
+      "step": 3832
+    },
+    {
+      "epoch": 3.4652213188798555,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00018391692247564582,
+      "loss": 0.5566,
+      "step": 3836
+    },
+    {
+      "epoch": 3.4688346883468837,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018370745034294316,
+      "loss": 0.5937,
+      "step": 3840
+    },
+    {
+      "epoch": 3.4724480578139114,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00018349790895645082,
+      "loss": 0.5795,
+      "step": 3844
+    },
+    {
+      "epoch": 3.4760614272809396,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001832882987466829,
+      "loss": 0.5501,
+      "step": 3848
+    },
+    {
+      "epoch": 3.4796747967479673,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.000183078620144295,
+      "loss": 0.5379,
+      "step": 3852
+    },
+    {
+      "epoch": 3.4832881662149955,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00018286887358008323,
+      "loss": 0.5591,
+      "step": 3856
+    },
+    {
+      "epoch": 3.4869015356820237,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001826590594849833,
+      "loss": 0.5886,
+      "step": 3860
+    },
+    {
+      "epoch": 3.4905149051490514,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00018244917829006964,
+      "loss": 0.5821,
+      "step": 3864
+    },
+    {
+      "epoch": 3.4941282746160796,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001822392304265546,
+      "loss": 0.5541,
+      "step": 3868
+    },
+    {
+      "epoch": 3.4977416440831073,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0001820292163257875,
+      "loss": 0.5662,
+      "step": 3872
+    },
+    {
+      "epoch": 3.5013550135501355,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018181913641925377,
+      "loss": 0.5446,
+      "step": 3876
+    },
+    {
+      "epoch": 3.5049683830171636,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018160899113857392,
+      "loss": 0.5613,
+      "step": 3880
+    },
+    {
+      "epoch": 3.5085817524841914,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018139878091550293,
+      "loss": 0.5821,
+      "step": 3884
+    },
+    {
+      "epoch": 3.5121951219512195,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00018118850618192904,
+      "loss": 0.5521,
+      "step": 3888
+    },
+    {
+      "epoch": 3.5158084914182473,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001809781673698732,
+      "loss": 0.5405,
+      "step": 3892
+    },
+    {
+      "epoch": 3.5194218608852754,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018076776491148794,
+      "loss": 0.5545,
+      "step": 3896
+    },
+    {
+      "epoch": 3.5230352303523036,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001805572992390565,
+      "loss": 0.5163,
+      "step": 3900
+    },
+    {
+      "epoch": 3.5266485998193318,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018034677078499216,
+      "loss": 0.5589,
+      "step": 3904
+    },
+    {
+      "epoch": 3.5302619692863595,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001801361799818369,
+      "loss": 0.6153,
+      "step": 3908
+    },
+    {
+      "epoch": 3.5338753387533877,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001799255272622611,
+      "loss": 0.5782,
+      "step": 3912
+    },
+    {
+      "epoch": 3.5374887082204154,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00017971481305906222,
+      "loss": 0.5632,
+      "step": 3916
+    },
+    {
+      "epoch": 3.5411020776874436,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00017950403780516398,
+      "loss": 0.557,
+      "step": 3920
+    },
+    {
+      "epoch": 3.5447154471544717,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00017929320193361574,
+      "loss": 0.5593,
+      "step": 3924
+    },
+    {
+      "epoch": 3.5483288166214995,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017908230587759108,
+      "loss": 0.5818,
+      "step": 3928
+    },
+    {
+      "epoch": 3.5519421860885276,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00017887135007038762,
+      "loss": 0.5731,
+      "step": 3932
+    },
+    {
+      "epoch": 3.5555555555555554,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00017866033494542535,
+      "loss": 0.5172,
+      "step": 3936
+    },
+    {
+      "epoch": 3.5591689250225835,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00017844926093624644,
+      "loss": 0.5789,
+      "step": 3940
+    },
+    {
+      "epoch": 3.5627822944896117,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00017823812847651382,
+      "loss": 0.5574,
+      "step": 3944
+    },
+    {
+      "epoch": 3.5663956639566394,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001780269380000107,
+      "loss": 0.5353,
+      "step": 3948
+    },
+    {
+      "epoch": 3.5700090334236676,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00017781568994063937,
+      "loss": 0.5513,
+      "step": 3952
+    },
+    {
+      "epoch": 3.5736224028906953,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0001776043847324205,
+      "loss": 0.557,
+      "step": 3956
+    },
+    {
+      "epoch": 3.5772357723577235,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.000177393022809492,
+      "loss": 0.5666,
+      "step": 3960
+    },
+    {
+      "epoch": 3.5808491418247517,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00017718160460610858,
+      "loss": 0.579,
+      "step": 3964
+    },
+    {
+      "epoch": 3.58446251129178,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00017697013055664035,
+      "loss": 0.57,
+      "step": 3968
+    },
+    {
+      "epoch": 3.5880758807588076,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00017675860109557225,
+      "loss": 0.5668,
+      "step": 3972
+    },
+    {
+      "epoch": 3.5916892502258357,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00017654701665750306,
+      "loss": 0.5952,
+      "step": 3976
+    },
+    {
+      "epoch": 3.5953026196928635,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00017633537767714462,
+      "loss": 0.5357,
+      "step": 3980
+    },
+    {
+      "epoch": 3.5989159891598916,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00017612368458932052,
+      "loss": 0.5801,
+      "step": 3984
+    },
+    {
+      "epoch": 3.60252935862692,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0001759119378289659,
+      "loss": 0.5527,
+      "step": 3988
+    },
+    {
+      "epoch": 3.6061427280939475,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00017570013783112592,
+      "loss": 0.5276,
+      "step": 3992
+    },
+    {
+      "epoch": 3.6097560975609757,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00017548828503095518,
+      "loss": 0.5595,
+      "step": 3996
+    },
+    {
+      "epoch": 3.6133694670280034,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00017527637986371685,
+      "loss": 0.5526,
+      "step": 4000
+    },
+    {
+      "epoch": 3.6169828364950316,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00017506442276478152,
+      "loss": 0.5703,
+      "step": 4004
+    },
+    {
+      "epoch": 3.6205962059620598,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00017485241416962668,
+      "loss": 0.5836,
+      "step": 4008
+    },
+    {
+      "epoch": 3.6242095754290875,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001746403545138354,
+      "loss": 0.5687,
+      "step": 4012
+    },
+    {
+      "epoch": 3.6278229448961157,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00017442824423309592,
+      "loss": 0.575,
+      "step": 4016
+    },
+    {
+      "epoch": 3.6314363143631434,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001742160837632002,
+      "loss": 0.5646,
+      "step": 4020
+    },
+    {
+      "epoch": 3.6350496838301716,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00017400387354004357,
+      "loss": 0.5465,
+      "step": 4024
+    },
+    {
+      "epoch": 3.6386630532971997,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0001737916139996234,
+      "loss": 0.5504,
+      "step": 4028
+    },
+    {
+      "epoch": 3.642276422764228,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017357930557803853,
+      "loss": 0.5749,
+      "step": 4032
+    },
+    {
+      "epoch": 3.6458897922312556,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017336694871148808,
+      "loss": 0.5385,
+      "step": 4036
+    },
+    {
+      "epoch": 3.649503161698284,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00017315454383627082,
+      "loss": 0.5664,
+      "step": 4040
+    },
+    {
+      "epoch": 3.6531165311653115,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001729420913887841,
+      "loss": 0.5726,
+      "step": 4044
+    },
+    {
+      "epoch": 3.6567299006323397,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001727295918055231,
+      "loss": 0.5914,
+      "step": 4048
+    },
+    {
+      "epoch": 3.660343270099368,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00017251704552307965,
+      "loss": 0.5505,
+      "step": 4052
+    },
+    {
+      "epoch": 3.6639566395663956,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00017230445297814175,
+      "loss": 0.5353,
+      "step": 4056
+    },
+    {
+      "epoch": 3.667570009033424,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00017209181460749228,
+      "loss": 0.5662,
+      "step": 4060
+    },
+    {
+      "epoch": 3.6711833785004515,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017187913084800837,
+      "loss": 0.5594,
+      "step": 4064
+    },
+    {
+      "epoch": 3.6747967479674797,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0001716664021366603,
+      "loss": 0.532,
+      "step": 4068
+    },
+    {
+      "epoch": 3.678410117434508,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00017145362891051084,
+      "loss": 0.5822,
+      "step": 4072
+    },
+    {
+      "epoch": 3.6820234869015356,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0001712408116067142,
+      "loss": 0.5387,
+      "step": 4076
+    },
+    {
+      "epoch": 3.6856368563685638,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00017102795066251502,
+      "loss": 0.5367,
+      "step": 4080
+    },
+    {
+      "epoch": 3.6892502258355915,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00017081504651524773,
+      "loss": 0.5445,
+      "step": 4084
+    },
+    {
+      "epoch": 3.6928635953026197,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00017060209960233544,
+      "loss": 0.5815,
+      "step": 4088
+    },
+    {
+      "epoch": 3.696476964769648,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001703891103612892,
+      "loss": 0.5598,
+      "step": 4092
+    },
+    {
+      "epoch": 3.7000903342366755,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00017017607922970693,
+      "loss": 0.5526,
+      "step": 4096
+    },
+    {
+      "epoch": 3.7037037037037037,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00016996300664527274,
+      "loss": 0.5901,
+      "step": 4100
+    },
+    {
+      "epoch": 3.7073170731707314,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016974989304575583,
+      "loss": 0.5884,
+      "step": 4104
+    },
+    {
+      "epoch": 3.7109304426377596,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00016953673886900967,
+      "loss": 0.6022,
+      "step": 4108
+    },
+    {
+      "epoch": 3.714543812104788,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00016932354455297113,
+      "loss": 0.5683,
+      "step": 4112
+    },
+    {
+      "epoch": 3.718157181571816,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00016911031053565943,
+      "loss": 0.5815,
+      "step": 4116
+    },
+    {
+      "epoch": 3.7217705510388437,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00016889703725517553,
+      "loss": 0.5582,
+      "step": 4120
+    },
+    {
+      "epoch": 3.725383920505872,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00016868372514970104,
+      "loss": 0.5383,
+      "step": 4124
+    },
+    {
+      "epoch": 3.7289972899728996,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016847037465749713,
+      "loss": 0.5236,
+      "step": 4128
+    },
+    {
+      "epoch": 3.7326106594399278,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00016825698621690419,
+      "loss": 0.5677,
+      "step": 4132
+    },
+    {
+      "epoch": 3.736224028906956,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00016804356026634017,
+      "loss": 0.5743,
+      "step": 4136
+    },
+    {
+      "epoch": 3.7398373983739837,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00016783009724430042,
+      "loss": 0.5377,
+      "step": 4140
+    },
+    {
+      "epoch": 3.743450767841012,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00016761659758935624,
+      "loss": 0.5916,
+      "step": 4144
+    },
+    {
+      "epoch": 3.7470641373080396,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00016740306174015435,
+      "loss": 0.5635,
+      "step": 4148
+    },
+    {
+      "epoch": 3.7506775067750677,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00016718949013541578,
+      "loss": 0.5483,
+      "step": 4152
+    },
+    {
+      "epoch": 3.754290876242096,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00016697588321393494,
+      "loss": 0.5791,
+      "step": 4156
+    },
+    {
+      "epoch": 3.7579042457091236,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00016676224141457893,
+      "loss": 0.6157,
+      "step": 4160
+    },
+    {
+      "epoch": 3.761517615176152,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001665485651762864,
+      "loss": 0.5631,
+      "step": 4164
+    },
+    {
+      "epoch": 3.7651309846431795,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00016633485493806676,
+      "loss": 0.5893,
+      "step": 4168
+    },
+    {
+      "epoch": 3.7687443541102077,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016612111113899943,
+      "loss": 0.5555,
+      "step": 4172
+    },
+    {
+      "epoch": 3.772357723577236,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00016590733421823254,
+      "loss": 0.5636,
+      "step": 4176
+    },
+    {
+      "epoch": 3.775971093044264,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001656935246149825,
+      "loss": 0.56,
+      "step": 4180
+    },
+    {
+      "epoch": 3.7795844625112918,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00016547968276853274,
+      "loss": 0.5431,
+      "step": 4184
+    },
+    {
+      "epoch": 3.78319783197832,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.000165265809118233,
+      "loss": 0.5402,
+      "step": 4188
+    },
+    {
+      "epoch": 3.7868112014453477,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00016505190410349817,
+      "loss": 0.5214,
+      "step": 4192
+    },
+    {
+      "epoch": 3.790424570912376,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0001648379681638079,
+      "loss": 0.5794,
+      "step": 4196
+    },
+    {
+      "epoch": 3.794037940379404,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00016462400173870514,
+      "loss": 0.579,
+      "step": 4200
+    },
+    {
+      "epoch": 3.7976513098464317,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00016441000526779556,
+      "loss": 0.5955,
+      "step": 4204
+    },
+    {
+      "epoch": 3.80126467931346,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00016419597919074656,
+      "loss": 0.5505,
+      "step": 4208
+    },
+    {
+      "epoch": 3.8048780487804876,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001639819239472864,
+      "loss": 0.5759,
+      "step": 4212
+    },
+    {
+      "epoch": 3.808491418247516,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001637678399772031,
+      "loss": 0.5646,
+      "step": 4216
+    },
+    {
+      "epoch": 3.812104787714544,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00016355372772034388,
+      "loss": 0.5929,
+      "step": 4220
+    },
+    {
+      "epoch": 3.8157181571815717,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00016333958761661398,
+      "loss": 0.5522,
+      "step": 4224
+    },
+    {
+      "epoch": 3.8193315266486,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00016312542010597593,
+      "loss": 0.5343,
+      "step": 4228
+    },
+    {
+      "epoch": 3.8229448961156276,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00016291122562844855,
+      "loss": 0.5524,
+      "step": 4232
+    },
+    {
+      "epoch": 3.8265582655826558,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00016269700462410603,
+      "loss": 0.539,
+      "step": 4236
+    },
+    {
+      "epoch": 3.830171635049684,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00016248275753307702,
+      "loss": 0.5513,
+      "step": 4240
+    },
+    {
+      "epoch": 3.8337850045167117,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00016226848479554382,
+      "loss": 0.5407,
+      "step": 4244
+    },
+    {
+      "epoch": 3.83739837398374,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001620541868517415,
+      "loss": 0.5228,
+      "step": 4248
+    },
+    {
+      "epoch": 3.8410117434507676,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00016183986414195677,
+      "loss": 0.5958,
+      "step": 4252
+    },
+    {
+      "epoch": 3.8446251129177957,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00016162551710652734,
+      "loss": 0.5598,
+      "step": 4256
+    },
+    {
+      "epoch": 3.848238482384824,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001614111461858408,
+      "loss": 0.5138,
+      "step": 4260
+    },
+    {
+      "epoch": 3.851851851851852,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00016119675182033391,
+      "loss": 0.5419,
+      "step": 4264
+    },
+    {
+      "epoch": 3.85546522131888,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001609823344504915,
+      "loss": 0.569,
+      "step": 4268
+    },
+    {
+      "epoch": 3.859078590785908,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00016076789451684574,
+      "loss": 0.549,
+      "step": 4272
+    },
+    {
+      "epoch": 3.8626919602529357,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00016055343245997518,
+      "loss": 0.5565,
+      "step": 4276
+    },
+    {
+      "epoch": 3.866305329719964,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00016033894872050368,
+      "loss": 0.5619,
+      "step": 4280
+    },
+    {
+      "epoch": 3.869918699186992,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00016012444373909987,
+      "loss": 0.5603,
+      "step": 4284
+    },
+    {
+      "epoch": 3.8735320686540198,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00015990991795647582,
+      "loss": 0.5487,
+      "step": 4288
+    },
+    {
+      "epoch": 3.877145438121048,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001596953718133864,
+      "loss": 0.5858,
+      "step": 4292
+    },
+    {
+      "epoch": 3.8807588075880757,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001594808057506284,
+      "loss": 0.5938,
+      "step": 4296
+    },
+    {
+      "epoch": 3.884372177055104,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001592662202090394,
+      "loss": 0.546,
+      "step": 4300
+    },
+    {
+      "epoch": 3.887985546522132,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00015905161562949712,
+      "loss": 0.5365,
+      "step": 4304
+    },
+    {
+      "epoch": 3.8915989159891597,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00015883699245291822,
+      "loss": 0.6163,
+      "step": 4308
+    },
+    {
+      "epoch": 3.895212285456188,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001586223511202579,
+      "loss": 0.5412,
+      "step": 4312
+    },
+    {
+      "epoch": 3.8988256549232156,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00015840769207250825,
+      "loss": 0.5812,
+      "step": 4316
+    },
+    {
+      "epoch": 3.902439024390244,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00015819301575069802,
+      "loss": 0.5649,
+      "step": 4320
+    },
+    {
+      "epoch": 3.906052393857272,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001579783225958914,
+      "loss": 0.5503,
+      "step": 4324
+    },
+    {
+      "epoch": 3.9096657633243,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00015776361304918715,
+      "loss": 0.5589,
+      "step": 4328
+    },
+    {
+      "epoch": 3.913279132791328,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00015754888755171773,
+      "loss": 0.6127,
+      "step": 4332
+    },
+    {
+      "epoch": 3.916892502258356,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00015733414654464835,
+      "loss": 0.6025,
+      "step": 4336
+    },
+    {
+      "epoch": 3.920505871725384,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00015711939046917606,
+      "loss": 0.5519,
+      "step": 4340
+    },
+    {
+      "epoch": 3.924119241192412,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00015690461976652884,
+      "loss": 0.613,
+      "step": 4344
+    },
+    {
+      "epoch": 3.92773261065944,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00015668983487796486,
+      "loss": 0.5976,
+      "step": 4348
+    },
+    {
+      "epoch": 3.931345980126468,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00015647503624477137,
+      "loss": 0.563,
+      "step": 4352
+    },
+    {
+      "epoch": 3.934959349593496,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00015626022430826375,
+      "loss": 0.5501,
+      "step": 4356
+    },
+    {
+      "epoch": 3.9385727190605238,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001560453995097849,
+      "loss": 0.5709,
+      "step": 4360
+    },
+    {
+      "epoch": 3.942186088527552,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00015583056229070405,
+      "loss": 0.5044,
+      "step": 4364
+    },
+    {
+      "epoch": 3.94579945799458,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00015561571309241585,
+      "loss": 0.5778,
+      "step": 4368
+    },
+    {
+      "epoch": 3.949412827461608,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00015540085235633973,
+      "loss": 0.5547,
+      "step": 4372
+    },
+    {
+      "epoch": 3.953026196928636,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00015518598052391874,
+      "loss": 0.5934,
+      "step": 4376
+    },
+    {
+      "epoch": 3.9566395663956637,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001549710980366188,
+      "loss": 0.5204,
+      "step": 4380
+    },
+    {
+      "epoch": 3.960252935862692,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00015475620533592758,
+      "loss": 0.5592,
+      "step": 4384
+    },
+    {
+      "epoch": 3.96386630532972,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001545413028633539,
+      "loss": 0.5684,
+      "step": 4388
+    },
+    {
+      "epoch": 3.9674796747967482,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00015432639106042646,
+      "loss": 0.5568,
+      "step": 4392
+    },
+    {
+      "epoch": 3.971093044263776,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00015411147036869334,
+      "loss": 0.5771,
+      "step": 4396
+    },
+    {
+      "epoch": 3.974706413730804,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001538965412297208,
+      "loss": 0.5472,
+      "step": 4400
+    },
+    {
+      "epoch": 3.978319783197832,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00015368160408509238,
+      "loss": 0.5926,
+      "step": 4404
+    },
+    {
+      "epoch": 3.98193315266486,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00015346665937640818,
+      "loss": 0.5737,
+      "step": 4408
+    },
+    {
+      "epoch": 3.985546522131888,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00015325170754528376,
+      "loss": 0.4884,
+      "step": 4412
+    },
+    {
+      "epoch": 3.989159891598916,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001530367490333494,
+      "loss": 0.5861,
+      "step": 4416
+    },
+    {
+      "epoch": 3.992773261065944,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.000152821784282249,
+      "loss": 0.5417,
+      "step": 4420
+    },
+    {
+      "epoch": 3.996386630532972,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00015260681373363932,
+      "loss": 0.5862,
+      "step": 4424
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001523918378291891,
+      "loss": 0.5813,
+      "step": 4428
+    },
+    {
+      "epoch": 4.003613369467028,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00015217685701057792,
+      "loss": 0.5454,
+      "step": 4432
+    },
+    {
+      "epoch": 4.007226738934056,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00015196187171949573,
+      "loss": 0.5102,
+      "step": 4436
+    },
+    {
+      "epoch": 4.010840108401084,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00015174688239764136,
+      "loss": 0.4544,
+      "step": 4440
+    },
+    {
+      "epoch": 4.014453477868112,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001515318894867221,
+      "loss": 0.4868,
+      "step": 4444
+    },
+    {
+      "epoch": 4.01806684733514,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0001513168934284526,
+      "loss": 0.486,
+      "step": 4448
+    },
+    {
+      "epoch": 4.021680216802168,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00015110189466455395,
+      "loss": 0.5037,
+      "step": 4452
+    },
+    {
+      "epoch": 4.025293586269196,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001508868936367528,
+      "loss": 0.4714,
+      "step": 4456
+    },
+    {
+      "epoch": 4.028906955736224,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0001506718907867804,
+      "loss": 0.4837,
+      "step": 4460
+    },
+    {
+      "epoch": 4.032520325203252,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001504568865563719,
+      "loss": 0.5177,
+      "step": 4464
+    },
+    {
+      "epoch": 4.03613369467028,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00015024188138726508,
+      "loss": 0.4873,
+      "step": 4468
+    },
+    {
+      "epoch": 4.039747064137308,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00015002687572119978,
+      "loss": 0.4903,
+      "step": 4472
+    },
+    {
+      "epoch": 4.043360433604336,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00014981186999991694,
+      "loss": 0.5045,
+      "step": 4476
+    },
+    {
+      "epoch": 4.0469738030713645,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00014959686466515729,
+      "loss": 0.508,
+      "step": 4480
+    },
+    {
+      "epoch": 4.050587172538392,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001493818601586612,
+      "loss": 0.4853,
+      "step": 4484
+    },
+    {
+      "epoch": 4.05420054200542,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00014916685692216696,
+      "loss": 0.4733,
+      "step": 4488
+    },
+    {
+      "epoch": 4.057813911472448,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00014895185539741054,
+      "loss": 0.4786,
+      "step": 4492
+    },
+    {
+      "epoch": 4.061427280939476,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00014873685602612415,
+      "loss": 0.4783,
+      "step": 4496
+    },
+    {
+      "epoch": 4.065040650406504,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00014852185925003577,
+      "loss": 0.4978,
+      "step": 4500
+    },
+    {
+      "epoch": 4.068654019873532,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.000148306865510868,
+      "loss": 0.5434,
+      "step": 4504
+    },
+    {
+      "epoch": 4.07226738934056,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.000148091875250337,
+      "loss": 0.4549,
+      "step": 4508
+    },
+    {
+      "epoch": 4.075880758807588,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00014787688891015218,
+      "loss": 0.4809,
+      "step": 4512
+    },
+    {
+      "epoch": 4.079494128274616,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00014766190693201443,
+      "loss": 0.4999,
+      "step": 4516
+    },
+    {
+      "epoch": 4.083107497741644,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001474469297576161,
+      "loss": 0.48,
+      "step": 4520
+    },
+    {
+      "epoch": 4.086720867208672,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00014723195782863937,
+      "loss": 0.491,
+      "step": 4524
+    },
+    {
+      "epoch": 4.0903342366757,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00014701699158675588,
+      "loss": 0.4985,
+      "step": 4528
+    },
+    {
+      "epoch": 4.093947606142728,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001468020314736254,
+      "loss": 0.466,
+      "step": 4532
+    },
+    {
+      "epoch": 4.097560975609756,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00014658707793089505,
+      "loss": 0.5192,
+      "step": 4536
+    },
+    {
+      "epoch": 4.101174345076784,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00014637213140019877,
+      "loss": 0.4989,
+      "step": 4540
+    },
+    {
+      "epoch": 4.1047877145438125,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00014615719232315575,
+      "loss": 0.4902,
+      "step": 4544
+    },
+    {
+      "epoch": 4.10840108401084,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00014594226114137014,
+      "loss": 0.5096,
+      "step": 4548
+    },
+    {
+      "epoch": 4.112014453477868,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00014572733829642962,
+      "loss": 0.4781,
+      "step": 4552
+    },
+    {
+      "epoch": 4.115627822944896,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00014551242422990496,
+      "loss": 0.4662,
+      "step": 4556
+    },
+    {
+      "epoch": 4.119241192411924,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00014529751938334876,
+      "loss": 0.4736,
+      "step": 4560
+    },
+    {
+      "epoch": 4.1228545618789525,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00014508262419829463,
+      "loss": 0.4831,
+      "step": 4564
+    },
+    {
+      "epoch": 4.12646793134598,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00014486773911625655,
+      "loss": 0.4733,
+      "step": 4568
+    },
+    {
+      "epoch": 4.130081300813008,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001446528645787275,
+      "loss": 0.5184,
+      "step": 4572
+    },
+    {
+      "epoch": 4.133694670280036,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.000144438001027179,
+      "loss": 0.5103,
+      "step": 4576
+    },
+    {
+      "epoch": 4.137308039747064,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001442231489030599,
+      "loss": 0.5411,
+      "step": 4580
+    },
+    {
+      "epoch": 4.1409214092140925,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00014400830864779539,
+      "loss": 0.4805,
+      "step": 4584
+    },
+    {
+      "epoch": 4.14453477868112,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00014379348070278668,
+      "loss": 0.5105,
+      "step": 4588
+    },
+    {
+      "epoch": 4.148148148148148,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001435786655094093,
+      "loss": 0.4928,
+      "step": 4592
+    },
+    {
+      "epoch": 4.151761517615176,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00014336386350901289,
+      "loss": 0.4993,
+      "step": 4596
+    },
+    {
+      "epoch": 4.155374887082204,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001431490751429197,
+      "loss": 0.4553,
+      "step": 4600
+    },
+    {
+      "epoch": 4.158988256549232,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00014293430085242424,
+      "loss": 0.4709,
+      "step": 4604
+    },
+    {
+      "epoch": 4.16260162601626,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00014271954107879196,
+      "loss": 0.5244,
+      "step": 4608
+    },
+    {
+      "epoch": 4.166214995483288,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00014250479626325832,
+      "loss": 0.4795,
+      "step": 4612
+    },
+    {
+      "epoch": 4.169828364950316,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0001422900668470284,
+      "loss": 0.5038,
+      "step": 4616
+    },
+    {
+      "epoch": 4.173441734417344,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001420753532712754,
+      "loss": 0.4542,
+      "step": 4620
+    },
+    {
+      "epoch": 4.177055103884372,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00014186065597714004,
+      "loss": 0.5415,
+      "step": 4624
+    },
+    {
+      "epoch": 4.180668473351401,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00014164597540572958,
+      "loss": 0.4913,
+      "step": 4628
+    },
+    {
+      "epoch": 4.184281842818428,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00014143131199811695,
+      "loss": 0.507,
+      "step": 4632
+    },
+    {
+      "epoch": 4.187895212285456,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00014121666619533986,
+      "loss": 0.4978,
+      "step": 4636
+    },
+    {
+      "epoch": 4.191508581752484,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00014100203843839956,
+      "loss": 0.4906,
+      "step": 4640
+    },
+    {
+      "epoch": 4.195121951219512,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00014078742916826068,
+      "loss": 0.4673,
+      "step": 4644
+    },
+    {
+      "epoch": 4.1987353206865405,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00014057283882584951,
+      "loss": 0.5296,
+      "step": 4648
+    },
+    {
+      "epoch": 4.202348690153568,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00014035826785205365,
+      "loss": 0.4449,
+      "step": 4652
+    },
+    {
+      "epoch": 4.205962059620596,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00014014371668772085,
+      "loss": 0.4812,
+      "step": 4656
+    },
+    {
+      "epoch": 4.209575429087624,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.000139929185773658,
+      "loss": 0.506,
+      "step": 4660
+    },
+    {
+      "epoch": 4.213188798554652,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00013971467555063064,
+      "loss": 0.4995,
+      "step": 4664
+    },
+    {
+      "epoch": 4.2168021680216805,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001395001864593617,
+      "loss": 0.4825,
+      "step": 4668
+    },
+    {
+      "epoch": 4.220415537488708,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00013928571894053065,
+      "loss": 0.5069,
+      "step": 4672
+    },
+    {
+      "epoch": 4.224028906955736,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00013907127343477265,
+      "loss": 0.5236,
+      "step": 4676
+    },
+    {
+      "epoch": 4.227642276422764,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00013885685038267774,
+      "loss": 0.4994,
+      "step": 4680
+    },
+    {
+      "epoch": 4.231255645889792,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00013864245022478968,
+      "loss": 0.4814,
+      "step": 4684
+    },
+    {
+      "epoch": 4.2348690153568205,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00013842807340160514,
+      "loss": 0.4836,
+      "step": 4688
+    },
+    {
+      "epoch": 4.238482384823849,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00013821372035357316,
+      "loss": 0.4885,
+      "step": 4692
+    },
+    {
+      "epoch": 4.242095754290876,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00013799939152109363,
+      "loss": 0.4962,
+      "step": 4696
+    },
+    {
+      "epoch": 4.245709123757904,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00013778508734451682,
+      "loss": 0.4996,
+      "step": 4700
+    },
+    {
+      "epoch": 4.249322493224932,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00013757080826414227,
+      "loss": 0.4894,
+      "step": 4704
+    },
+    {
+      "epoch": 4.2529358626919604,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00013735655472021807,
+      "loss": 0.4748,
+      "step": 4708
+    },
+    {
+      "epoch": 4.256549232158989,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00013714232715293978,
+      "loss": 0.4613,
+      "step": 4712
+    },
+    {
+      "epoch": 4.260162601626016,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00013692812600244944,
+      "loss": 0.532,
+      "step": 4716
+    },
+    {
+      "epoch": 4.263775971093044,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00013671395170883514,
+      "loss": 0.54,
+      "step": 4720
+    },
+    {
+      "epoch": 4.267389340560072,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00013649980471212953,
+      "loss": 0.4708,
+      "step": 4724
+    },
+    {
+      "epoch": 4.2710027100271,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00013628568545230927,
+      "loss": 0.4768,
+      "step": 4728
+    },
+    {
+      "epoch": 4.274616079494129,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001360715943692941,
+      "loss": 0.5263,
+      "step": 4732
+    },
+    {
+      "epoch": 4.278229448961156,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0001358575319029456,
+      "loss": 0.4815,
+      "step": 4736
+    },
+    {
+      "epoch": 4.281842818428184,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00013564349849306694,
+      "loss": 0.5291,
+      "step": 4740
+    },
+    {
+      "epoch": 4.285456187895212,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00013542949457940126,
+      "loss": 0.4996,
+      "step": 4744
+    },
+    {
+      "epoch": 4.28906955736224,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00013521552060163138,
+      "loss": 0.5285,
+      "step": 4748
+    },
+    {
+      "epoch": 4.2926829268292686,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00013500157699937836,
+      "loss": 0.4899,
+      "step": 4752
+    },
+    {
+      "epoch": 4.296296296296296,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.000134787664212201,
+      "loss": 0.5122,
+      "step": 4756
+    },
+    {
+      "epoch": 4.299909665763324,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00013457378267959483,
+      "loss": 0.5029,
+      "step": 4760
+    },
+    {
+      "epoch": 4.303523035230352,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00013435993284099091,
+      "loss": 0.4732,
+      "step": 4764
+    },
+    {
+      "epoch": 4.30713640469738,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00013414611513575553,
+      "loss": 0.5427,
+      "step": 4768
+    },
+    {
+      "epoch": 4.3107497741644085,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001339323300031887,
+      "loss": 0.504,
+      "step": 4772
+    },
+    {
+      "epoch": 4.314363143631437,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00013371857788252366,
+      "loss": 0.4896,
+      "step": 4776
+    },
+    {
+      "epoch": 4.317976513098464,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0001335048592129257,
+      "loss": 0.4816,
+      "step": 4780
+    },
+    {
+      "epoch": 4.321589882565492,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00013329117443349153,
+      "loss": 0.4952,
+      "step": 4784
+    },
+    {
+      "epoch": 4.32520325203252,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00013307752398324807,
+      "loss": 0.452,
+      "step": 4788
+    },
+    {
+      "epoch": 4.3288166214995485,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001328639083011518,
+      "loss": 0.5189,
+      "step": 4792
+    },
+    {
+      "epoch": 4.332429990966577,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00013265032782608786,
+      "loss": 0.4767,
+      "step": 4796
+    },
+    {
+      "epoch": 4.336043360433604,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001324367829968688,
+      "loss": 0.4891,
+      "step": 4800
+    },
+    {
+      "epoch": 4.339656729900632,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001322232742522342,
+      "loss": 0.4857,
+      "step": 4804
+    },
+    {
+      "epoch": 4.34327009936766,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00013200980203084936,
+      "loss": 0.5197,
+      "step": 4808
+    },
+    {
+      "epoch": 4.3468834688346885,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00013179636677130446,
+      "loss": 0.4855,
+      "step": 4812
+    },
+    {
+      "epoch": 4.350496838301717,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.000131582968912114,
+      "loss": 0.5269,
+      "step": 4816
+    },
+    {
+      "epoch": 4.354110207768745,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00013136960889171536,
+      "loss": 0.483,
+      "step": 4820
+    },
+    {
+      "epoch": 4.357723577235772,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00013115628714846842,
+      "loss": 0.4785,
+      "step": 4824
+    },
+    {
+      "epoch": 4.3613369467028,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00013094300412065424,
+      "loss": 0.4587,
+      "step": 4828
+    },
+    {
+      "epoch": 4.364950316169828,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0001307297602464744,
+      "loss": 0.5134,
+      "step": 4832
+    },
+    {
+      "epoch": 4.368563685636857,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00013051655596405006,
+      "loss": 0.5088,
+      "step": 4836
+    },
+    {
+      "epoch": 4.372177055103885,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00013030339171142094,
+      "loss": 0.5073,
+      "step": 4840
+    },
+    {
+      "epoch": 4.375790424570912,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001300902679265447,
+      "loss": 0.5079,
+      "step": 4844
+    },
+    {
+      "epoch": 4.37940379403794,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00012987718504729563,
+      "loss": 0.5293,
+      "step": 4848
+    },
+    {
+      "epoch": 4.383017163504968,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001296641435114642,
+      "loss": 0.4757,
+      "step": 4852
+    },
+    {
+      "epoch": 4.386630532971997,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00012945114375675573,
+      "loss": 0.4939,
+      "step": 4856
+    },
+    {
+      "epoch": 4.390243902439025,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001292381862207899,
+      "loss": 0.493,
+      "step": 4860
+    },
+    {
+      "epoch": 4.393857271906052,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001290252713410995,
+      "loss": 0.5238,
+      "step": 4864
+    },
+    {
+      "epoch": 4.39747064137308,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0001288123995551297,
+      "loss": 0.5016,
+      "step": 4868
+    },
+    {
+      "epoch": 4.401084010840108,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001285995713002372,
+      "loss": 0.4808,
+      "step": 4872
+    },
+    {
+      "epoch": 4.4046973803071365,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00012838678701368916,
+      "loss": 0.4892,
+      "step": 4876
+    },
+    {
+      "epoch": 4.408310749774165,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001281740471326626,
+      "loss": 0.4923,
+      "step": 4880
+    },
+    {
+      "epoch": 4.411924119241192,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001279613520942431,
+      "loss": 0.4856,
+      "step": 4884
+    },
+    {
+      "epoch": 4.41553748870822,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00012774870233542407,
+      "loss": 0.5078,
+      "step": 4888
+    },
+    {
+      "epoch": 4.419150858175248,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00012753609829310616,
+      "loss": 0.4861,
+      "step": 4892
+    },
+    {
+      "epoch": 4.4227642276422765,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001273235404040959,
+      "loss": 0.5044,
+      "step": 4896
+    },
+    {
+      "epoch": 4.426377597109305,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.000127111029105105,
+      "loss": 0.5363,
+      "step": 4900
+    },
+    {
+      "epoch": 4.429990966576333,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001268985648327495,
+      "loss": 0.4659,
+      "step": 4904
+    },
+    {
+      "epoch": 4.43360433604336,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0001266861480235488,
+      "loss": 0.5122,
+      "step": 4908
+    },
+    {
+      "epoch": 4.437217705510388,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001264737791139248,
+      "loss": 0.5029,
+      "step": 4912
+    },
+    {
+      "epoch": 4.4408310749774165,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00012626145854020092,
+      "loss": 0.4561,
+      "step": 4916
+    },
+    {
+      "epoch": 4.444444444444445,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0001260491867386014,
+      "loss": 0.504,
+      "step": 4920
+    },
+    {
+      "epoch": 4.448057813911473,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00012583696414525012,
+      "loss": 0.5015,
+      "step": 4924
+    },
+    {
+      "epoch": 4.4516711833785,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00012562479119617,
+      "loss": 0.4928,
+      "step": 4928
+    },
+    {
+      "epoch": 4.455284552845528,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00012541266832728187,
+      "loss": 0.4672,
+      "step": 4932
+    },
+    {
+      "epoch": 4.458897922312556,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00012520059597440373,
+      "loss": 0.4683,
+      "step": 4936
+    },
+    {
+      "epoch": 4.462511291779585,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00012498857457324974,
+      "loss": 0.4727,
+      "step": 4940
+    },
+    {
+      "epoch": 4.466124661246613,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0001247766045594294,
+      "loss": 0.5256,
+      "step": 4944
+    },
+    {
+      "epoch": 4.46973803071364,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00012456468636844664,
+      "loss": 0.485,
+      "step": 4948
+    },
+    {
+      "epoch": 4.473351400180668,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00012435282043569883,
+      "loss": 0.483,
+      "step": 4952
+    },
+    {
+      "epoch": 4.476964769647696,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001241410071964762,
+      "loss": 0.4951,
+      "step": 4956
+    },
+    {
+      "epoch": 4.480578139114725,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001239292470859605,
+      "loss": 0.4883,
+      "step": 4960
+    },
+    {
+      "epoch": 4.484191508581753,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00012371754053922427,
+      "loss": 0.4958,
+      "step": 4964
+    },
+    {
+      "epoch": 4.487804878048781,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012350588799123032,
+      "loss": 0.488,
+      "step": 4968
+    },
+    {
+      "epoch": 4.491418247515808,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001232942898768302,
+      "loss": 0.5583,
+      "step": 4972
+    },
+    {
+      "epoch": 4.495031616982836,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012308274663076378,
+      "loss": 0.4998,
+      "step": 4976
+    },
+    {
+      "epoch": 4.4986449864498645,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012287125868765814,
+      "loss": 0.4579,
+      "step": 4980
+    },
+    {
+      "epoch": 4.502258355916893,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001226598264820268,
+      "loss": 0.5235,
+      "step": 4984
+    },
+    {
+      "epoch": 4.505871725383921,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001224484504482687,
+      "loss": 0.5149,
+      "step": 4988
+    },
+    {
+      "epoch": 4.509485094850948,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012223713102066738,
+      "loss": 0.5022,
+      "step": 4992
+    },
+    {
+      "epoch": 4.513098464317976,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00012202586863339012,
+      "loss": 0.5239,
+      "step": 4996
+    },
+    {
+      "epoch": 4.5167118337850045,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00012181466372048696,
+      "loss": 0.4847,
+      "step": 5000
+    },
+    {
+      "epoch": 4.520325203252033,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012160351671588992,
+      "loss": 0.4884,
+      "step": 5004
+    },
+    {
+      "epoch": 4.523938572719061,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00012139242805341194,
+      "loss": 0.5007,
+      "step": 5008
+    },
+    {
+      "epoch": 4.527551942186088,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00012118139816674621,
+      "loss": 0.502,
+      "step": 5012
+    },
+    {
+      "epoch": 4.531165311653116,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00012097042748946512,
+      "loss": 0.488,
+      "step": 5016
+    },
+    {
+      "epoch": 4.5347786811201445,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00012075951645501932,
+      "loss": 0.5187,
+      "step": 5020
+    },
+    {
+      "epoch": 4.538392050587173,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00012054866549673708,
+      "loss": 0.5105,
+      "step": 5024
+    },
+    {
+      "epoch": 4.542005420054201,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00012033787504782311,
+      "loss": 0.5026,
+      "step": 5028
+    },
+    {
+      "epoch": 4.545618789521228,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00012012714554135788,
+      "loss": 0.5304,
+      "step": 5032
+    },
+    {
+      "epoch": 4.549232158988256,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.0001199164774102966,
+      "loss": 0.5012,
+      "step": 5036
+    },
+    {
+      "epoch": 4.5528455284552845,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00011970587108746841,
+      "loss": 0.5098,
+      "step": 5040
+    },
+    {
+      "epoch": 4.556458897922313,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00011949532700557547,
+      "loss": 0.4814,
+      "step": 5044
+    },
+    {
+      "epoch": 4.560072267389341,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.000119284845597192,
+      "loss": 0.5322,
+      "step": 5048
+    },
+    {
+      "epoch": 4.563685636856368,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00011907442729476354,
+      "loss": 0.4925,
+      "step": 5052
+    },
+    {
+      "epoch": 4.567299006323396,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00011886407253060592,
+      "loss": 0.4805,
+      "step": 5056
+    },
+    {
+      "epoch": 4.570912375790424,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00011865378173690446,
+      "loss": 0.4974,
+      "step": 5060
+    },
+    {
+      "epoch": 4.574525745257453,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00011844355534571302,
+      "loss": 0.5354,
+      "step": 5064
+    },
+    {
+      "epoch": 4.578139114724481,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00011823339378895312,
+      "loss": 0.4368,
+      "step": 5068
+    },
+    {
+      "epoch": 4.581752484191509,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00011802329749841316,
+      "loss": 0.4979,
+      "step": 5072
+    },
+    {
+      "epoch": 4.585365853658536,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00011781326690574733,
+      "loss": 0.5221,
+      "step": 5076
+    },
+    {
+      "epoch": 4.588979223125564,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.000117603302442475,
+      "loss": 0.4517,
+      "step": 5080
+    },
+    {
+      "epoch": 4.592592592592593,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00011739340453997946,
+      "loss": 0.5323,
+      "step": 5084
+    },
+    {
+      "epoch": 4.596205962059621,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00011718357362950747,
+      "loss": 0.4959,
+      "step": 5088
+    },
+    {
+      "epoch": 4.599819331526649,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.000116973810142168,
+      "loss": 0.4846,
+      "step": 5092
+    },
+    {
+      "epoch": 4.603432700993677,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00011676411450893152,
+      "loss": 0.5178,
+      "step": 5096
+    },
+    {
+      "epoch": 4.607046070460704,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00011655448716062916,
+      "loss": 0.4228,
+      "step": 5100
+    },
+    {
+      "epoch": 4.6106594399277325,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00011634492852795165,
+      "loss": 0.4944,
+      "step": 5104
+    },
+    {
+      "epoch": 4.614272809394761,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00011613543904144865,
+      "loss": 0.4711,
+      "step": 5108
+    },
+    {
+      "epoch": 4.617886178861789,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00011592601913152766,
+      "loss": 0.4716,
+      "step": 5112
+    },
+    {
+      "epoch": 4.621499548328817,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00011571666922845334,
+      "loss": 0.4811,
+      "step": 5116
+    },
+    {
+      "epoch": 4.625112917795844,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00011550738976234637,
+      "loss": 0.5702,
+      "step": 5120
+    },
+    {
+      "epoch": 4.6287262872628725,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00011529818116318281,
+      "loss": 0.5025,
+      "step": 5124
+    },
+    {
+      "epoch": 4.632339656729901,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00011508904386079317,
+      "loss": 0.4702,
+      "step": 5128
+    },
+    {
+      "epoch": 4.635953026196929,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00011487997828486129,
+      "loss": 0.5187,
+      "step": 5132
+    },
+    {
+      "epoch": 4.639566395663957,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00011467098486492389,
+      "loss": 0.5601,
+      "step": 5136
+    },
+    {
+      "epoch": 4.643179765130984,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00011446206403036927,
+      "loss": 0.5089,
+      "step": 5140
+    },
+    {
+      "epoch": 4.6467931345980125,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00011425321621043659,
+      "loss": 0.4851,
+      "step": 5144
+    },
+    {
+      "epoch": 4.650406504065041,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00011404444183421515,
+      "loss": 0.4992,
+      "step": 5148
+    },
+    {
+      "epoch": 4.654019873532069,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00011383574133064315,
+      "loss": 0.4739,
+      "step": 5152
+    },
+    {
+      "epoch": 4.657633242999097,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00011362711512850723,
+      "loss": 0.4987,
+      "step": 5156
+    },
+    {
+      "epoch": 4.661246612466124,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001134185636564412,
+      "loss": 0.5039,
+      "step": 5160
+    },
+    {
+      "epoch": 4.664859981933152,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00011321008734292544,
+      "loss": 0.4694,
+      "step": 5164
+    },
+    {
+      "epoch": 4.668473351400181,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00011300168661628587,
+      "loss": 0.4751,
+      "step": 5168
+    },
+    {
+      "epoch": 4.672086720867209,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00011279336190469305,
+      "loss": 0.5174,
+      "step": 5172
+    },
+    {
+      "epoch": 4.675700090334237,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00011258511363616151,
+      "loss": 0.4763,
+      "step": 5176
+    },
+    {
+      "epoch": 4.679313459801264,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001123769422385486,
+      "loss": 0.4803,
+      "step": 5180
+    },
+    {
+      "epoch": 4.682926829268292,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0001121688481395538,
+      "loss": 0.5152,
+      "step": 5184
+    },
+    {
+      "epoch": 4.686540198735321,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001119608317667177,
+      "loss": 0.4553,
+      "step": 5188
+    },
+    {
+      "epoch": 4.690153568202349,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00011175289354742134,
+      "loss": 0.4471,
+      "step": 5192
+    },
+    {
+      "epoch": 4.693766937669377,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00011154503390888503,
+      "loss": 0.4766,
+      "step": 5196
+    },
+    {
+      "epoch": 4.697380307136404,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00011133725327816771,
+      "loss": 0.4628,
+      "step": 5200
+    },
+    {
+      "epoch": 4.700993676603432,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00011112955208216601,
+      "loss": 0.5262,
+      "step": 5204
+    },
+    {
+      "epoch": 4.7046070460704605,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00011092193074761329,
+      "loss": 0.5055,
+      "step": 5208
+    },
+    {
+      "epoch": 4.708220415537489,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00011071438970107893,
+      "loss": 0.4822,
+      "step": 5212
+    },
+    {
+      "epoch": 4.711833785004517,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00011050692936896723,
+      "loss": 0.469,
+      "step": 5216
+    },
+    {
+      "epoch": 4.715447154471545,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00011029955017751677,
+      "loss": 0.4591,
+      "step": 5220
+    },
+    {
+      "epoch": 4.719060523938572,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00011009225255279938,
+      "loss": 0.5104,
+      "step": 5224
+    },
+    {
+      "epoch": 4.7226738934056005,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00010988503692071926,
+      "loss": 0.4972,
+      "step": 5228
+    },
+    {
+      "epoch": 4.726287262872629,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00010967790370701223,
+      "loss": 0.4893,
+      "step": 5232
+    },
+    {
+      "epoch": 4.729900632339657,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00010947085333724475,
+      "loss": 0.4707,
+      "step": 5236
+    },
+    {
+      "epoch": 4.733514001806685,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00010926388623681304,
+      "loss": 0.4775,
+      "step": 5240
+    },
+    {
+      "epoch": 4.737127371273713,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00010905700283094226,
+      "loss": 0.4952,
+      "step": 5244
+    },
+    {
+      "epoch": 4.7407407407407405,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00010885020354468563,
+      "loss": 0.5275,
+      "step": 5248
+    },
+    {
+      "epoch": 4.744354110207769,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00010864348880292354,
+      "loss": 0.5259,
+      "step": 5252
+    },
+    {
+      "epoch": 4.747967479674797,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00010843685903036262,
+      "loss": 0.4457,
+      "step": 5256
+    },
+    {
+      "epoch": 4.751580849141825,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00010823031465153505,
+      "loss": 0.4538,
+      "step": 5260
+    },
+    {
+      "epoch": 4.755194218608853,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00010802385609079739,
+      "loss": 0.5377,
+      "step": 5264
+    },
+    {
+      "epoch": 4.7588075880758804,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001078174837723301,
+      "loss": 0.5104,
+      "step": 5268
+    },
+    {
+      "epoch": 4.762420957542909,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00010761119812013625,
+      "loss": 0.5585,
+      "step": 5272
+    },
+    {
+      "epoch": 4.766034327009937,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00010740499955804091,
+      "loss": 0.4972,
+      "step": 5276
+    },
+    {
+      "epoch": 4.769647696476965,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00010719888850969031,
+      "loss": 0.5246,
+      "step": 5280
+    },
+    {
+      "epoch": 4.773261065943993,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00010699286539855076,
+      "loss": 0.514,
+      "step": 5284
+    },
+    {
+      "epoch": 4.77687443541102,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00010678693064790798,
+      "loss": 0.4796,
+      "step": 5288
+    },
+    {
+      "epoch": 4.780487804878049,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00010658108468086611,
+      "loss": 0.4739,
+      "step": 5292
+    },
+    {
+      "epoch": 4.784101174345077,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001063753279203468,
+      "loss": 0.5086,
+      "step": 5296
+    },
+    {
+      "epoch": 4.787714543812105,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00010616966078908861,
+      "loss": 0.5169,
+      "step": 5300
+    },
+    {
+      "epoch": 4.791327913279133,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00010596408370964579,
+      "loss": 0.5098,
+      "step": 5304
+    },
+    {
+      "epoch": 4.79494128274616,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00010575859710438765,
+      "loss": 0.4885,
+      "step": 5308
+    },
+    {
+      "epoch": 4.7985546522131886,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00010555320139549755,
+      "loss": 0.5206,
+      "step": 5312
+    },
+    {
+      "epoch": 4.802168021680217,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00010534789700497226,
+      "loss": 0.4649,
+      "step": 5316
+    },
+    {
+      "epoch": 4.805781391147245,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00010514268435462069,
+      "loss": 0.4921,
+      "step": 5320
+    },
+    {
+      "epoch": 4.809394760614273,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00010493756386606341,
+      "loss": 0.5084,
+      "step": 5324
+    },
+    {
+      "epoch": 4.8130081300813,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00010473253596073169,
+      "loss": 0.4661,
+      "step": 5328
+    },
+    {
+      "epoch": 4.8166214995483285,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00010452760105986643,
+      "loss": 0.4978,
+      "step": 5332
+    },
+    {
+      "epoch": 4.820234869015357,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0001043227595845176,
+      "loss": 0.5176,
+      "step": 5336
+    },
+    {
+      "epoch": 4.823848238482385,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00010411801195554308,
+      "loss": 0.514,
+      "step": 5340
+    },
+    {
+      "epoch": 4.827461607949413,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00010391335859360806,
+      "loss": 0.5451,
+      "step": 5344
+    },
+    {
+      "epoch": 4.831074977416441,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00010370879991918398,
+      "loss": 0.5037,
+      "step": 5348
+    },
+    {
+      "epoch": 4.8346883468834685,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00010350433635254775,
+      "loss": 0.4899,
+      "step": 5352
+    },
+    {
+      "epoch": 4.838301716350497,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00010329996831378091,
+      "loss": 0.5386,
+      "step": 5356
+    },
+    {
+      "epoch": 4.841915085817525,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00010309569622276866,
+      "loss": 0.5056,
+      "step": 5360
+    },
+    {
+      "epoch": 4.845528455284553,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00010289152049919922,
+      "loss": 0.4911,
+      "step": 5364
+    },
+    {
+      "epoch": 4.849141824751581,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00010268744156256264,
+      "loss": 0.4855,
+      "step": 5368
+    },
+    {
+      "epoch": 4.852755194218609,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00010248345983215015,
+      "loss": 0.513,
+      "step": 5372
+    },
+    {
+      "epoch": 4.856368563685637,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00010227957572705342,
+      "loss": 0.4928,
+      "step": 5376
+    },
+    {
+      "epoch": 4.859981933152665,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00010207578966616334,
+      "loss": 0.4715,
+      "step": 5380
+    },
+    {
+      "epoch": 4.863595302619693,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00010187210206816953,
+      "loss": 0.5328,
+      "step": 5384
+    },
+    {
+      "epoch": 4.867208672086721,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00010166851335155913,
+      "loss": 0.5068,
+      "step": 5388
+    },
+    {
+      "epoch": 4.870822041553749,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00010146502393461637,
+      "loss": 0.4827,
+      "step": 5392
+    },
+    {
+      "epoch": 4.874435411020777,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00010126163423542121,
+      "loss": 0.5177,
+      "step": 5396
+    },
+    {
+      "epoch": 4.878048780487805,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00010105834467184889,
+      "loss": 0.4967,
+      "step": 5400
+    },
+    {
+      "epoch": 4.881662149954833,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00010085515566156888,
+      "loss": 0.516,
+      "step": 5404
+    },
+    {
+      "epoch": 4.885275519421861,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.000100652067622044,
+      "loss": 0.4888,
+      "step": 5408
+    },
+    {
+      "epoch": 4.888888888888889,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00010044908097052976,
+      "loss": 0.5017,
+      "step": 5412
+    },
+    {
+      "epoch": 4.892502258355917,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00010024619612407317,
+      "loss": 0.4858,
+      "step": 5416
+    },
+    {
+      "epoch": 4.896115627822945,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00010004341349951228,
+      "loss": 0.4965,
+      "step": 5420
+    },
+    {
+      "epoch": 4.899728997289973,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.984073351347498e-05,
+      "loss": 0.4791,
+      "step": 5424
+    },
+    {
+      "epoch": 4.903342366757001,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.963815658237829e-05,
+      "loss": 0.5123,
+      "step": 5428
+    },
+    {
+      "epoch": 4.906955736224029,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.943568312242756e-05,
+      "loss": 0.5253,
+      "step": 5432
+    },
+    {
+      "epoch": 4.9105691056910565,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.923331354961551e-05,
+      "loss": 0.4842,
+      "step": 5436
+    },
+    {
+      "epoch": 4.914182475158085,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.903104827972147e-05,
+      "loss": 0.5174,
+      "step": 5440
+    },
+    {
+      "epoch": 4.917795844625113,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.882888772831041e-05,
+      "loss": 0.5506,
+      "step": 5444
+    },
+    {
+      "epoch": 4.921409214092141,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.862683231073215e-05,
+      "loss": 0.4965,
+      "step": 5448
+    },
+    {
+      "epoch": 4.925022583559169,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.842488244212057e-05,
+      "loss": 0.5504,
+      "step": 5452
+    },
+    {
+      "epoch": 4.9286359530261965,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.822303853739265e-05,
+      "loss": 0.4471,
+      "step": 5456
+    },
+    {
+      "epoch": 4.932249322493225,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.802130101124765e-05,
+      "loss": 0.4667,
+      "step": 5460
+    },
+    {
+      "epoch": 4.935862691960253,
+      "grad_norm": 0.376953125,
+      "learning_rate": 9.781967027816631e-05,
+      "loss": 0.4884,
+      "step": 5464
+    },
+    {
+      "epoch": 4.939476061427281,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.761814675240995e-05,
+      "loss": 0.4699,
+      "step": 5468
+    },
+    {
+      "epoch": 4.943089430894309,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.741673084801959e-05,
+      "loss": 0.4987,
+      "step": 5472
+    },
+    {
+      "epoch": 4.9467028003613365,
+      "grad_norm": 0.375,
+      "learning_rate": 9.721542297881511e-05,
+      "loss": 0.5054,
+      "step": 5476
+    },
+    {
+      "epoch": 4.950316169828365,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.701422355839458e-05,
+      "loss": 0.516,
+      "step": 5480
+    },
+    {
+      "epoch": 4.953929539295393,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.681313300013305e-05,
+      "loss": 0.5119,
+      "step": 5484
+    },
+    {
+      "epoch": 4.957542908762421,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.661215171718207e-05,
+      "loss": 0.4833,
+      "step": 5488
+    },
+    {
+      "epoch": 4.961156278229449,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.641128012246858e-05,
+      "loss": 0.5013,
+      "step": 5492
+    },
+    {
+      "epoch": 4.964769647696477,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.621051862869423e-05,
+      "loss": 0.4904,
+      "step": 5496
+    },
+    {
+      "epoch": 4.968383017163505,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.600986764833441e-05,
+      "loss": 0.4939,
+      "step": 5500
+    },
+    {
+      "epoch": 4.971996386630533,
+      "grad_norm": 0.376953125,
+      "learning_rate": 9.580932759363744e-05,
+      "loss": 0.4751,
+      "step": 5504
+    },
+    {
+      "epoch": 4.975609756097561,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.560889887662382e-05,
+      "loss": 0.5234,
+      "step": 5508
+    },
+    {
+      "epoch": 4.979223125564589,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.540858190908521e-05,
+      "loss": 0.5123,
+      "step": 5512
+    },
+    {
+      "epoch": 4.982836495031617,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.520837710258374e-05,
+      "loss": 0.4779,
+      "step": 5516
+    },
+    {
+      "epoch": 4.9864498644986455,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.500828486845106e-05,
+      "loss": 0.4851,
+      "step": 5520
+    },
+    {
+      "epoch": 4.990063233965673,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.480830561778751e-05,
+      "loss": 0.4492,
+      "step": 5524
+    },
+    {
+      "epoch": 4.993676603432701,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.460843976146142e-05,
+      "loss": 0.457,
+      "step": 5528
+    },
+    {
+      "epoch": 4.997289972899729,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.440868771010797e-05,
+      "loss": 0.4986,
+      "step": 5532
+    },
+    {
+      "epoch": 5.000903342366757,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.420904987412866e-05,
+      "loss": 0.5165,
+      "step": 5536
+    },
+    {
+      "epoch": 5.004516711833785,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.400952666369025e-05,
+      "loss": 0.4546,
+      "step": 5540
+    },
+    {
+      "epoch": 5.008130081300813,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.38101184887241e-05,
+      "loss": 0.5119,
+      "step": 5544
+    },
+    {
+      "epoch": 5.011743450767841,
+      "grad_norm": 0.365234375,
+      "learning_rate": 9.361082575892508e-05,
+      "loss": 0.4755,
+      "step": 5548
+    },
+    {
+      "epoch": 5.015356820234869,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.341164888375093e-05,
+      "loss": 0.458,
+      "step": 5552
+    },
+    {
+      "epoch": 5.018970189701897,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.321258827242145e-05,
+      "loss": 0.4431,
+      "step": 5556
+    },
+    {
+      "epoch": 5.022583559168925,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.301364433391738e-05,
+      "loss": 0.4425,
+      "step": 5560
+    },
+    {
+      "epoch": 5.026196928635953,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.281481747697994e-05,
+      "loss": 0.4147,
+      "step": 5564
+    },
+    {
+      "epoch": 5.029810298102981,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.261610811010969e-05,
+      "loss": 0.4241,
+      "step": 5568
+    },
+    {
+      "epoch": 5.033423667570009,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.241751664156588e-05,
+      "loss": 0.4338,
+      "step": 5572
+    },
+    {
+      "epoch": 5.037037037037037,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.221904347936543e-05,
+      "loss": 0.4574,
+      "step": 5576
+    },
+    {
+      "epoch": 5.040650406504065,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.202068903128222e-05,
+      "loss": 0.4555,
+      "step": 5580
+    },
+    {
+      "epoch": 5.044263775971093,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.182245370484633e-05,
+      "loss": 0.4773,
+      "step": 5584
+    },
+    {
+      "epoch": 5.047877145438121,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.162433790734292e-05,
+      "loss": 0.4972,
+      "step": 5588
+    },
+    {
+      "epoch": 5.051490514905149,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.142634204581177e-05,
+      "loss": 0.4566,
+      "step": 5592
+    },
+    {
+      "epoch": 5.055103884372177,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.122846652704608e-05,
+      "loss": 0.4461,
+      "step": 5596
+    },
+    {
+      "epoch": 5.058717253839205,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.103071175759188e-05,
+      "loss": 0.4297,
+      "step": 5600
+    },
+    {
+      "epoch": 5.062330623306233,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.083307814374715e-05,
+      "loss": 0.485,
+      "step": 5604
+    },
+    {
+      "epoch": 5.065943992773261,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.063556609156077e-05,
+      "loss": 0.4577,
+      "step": 5608
+    },
+    {
+      "epoch": 5.069557362240289,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.04381760068321e-05,
+      "loss": 0.4168,
+      "step": 5612
+    },
+    {
+      "epoch": 5.073170731707317,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.024090829510976e-05,
+      "loss": 0.4594,
+      "step": 5616
+    },
+    {
+      "epoch": 5.076784101174345,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.004376336169102e-05,
+      "loss": 0.4259,
+      "step": 5620
+    },
+    {
+      "epoch": 5.0803974706413735,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.984674161162085e-05,
+      "loss": 0.4593,
+      "step": 5624
+    },
+    {
+      "epoch": 5.084010840108401,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.964984344969111e-05,
+      "loss": 0.4226,
+      "step": 5628
+    },
+    {
+      "epoch": 5.087624209575429,
+      "grad_norm": 0.365234375,
+      "learning_rate": 8.945306928043988e-05,
+      "loss": 0.4343,
+      "step": 5632
+    },
+    {
+      "epoch": 5.091237579042457,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.925641950815026e-05,
+      "loss": 0.4335,
+      "step": 5636
+    },
+    {
+      "epoch": 5.094850948509485,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.905989453685003e-05,
+      "loss": 0.435,
+      "step": 5640
+    },
+    {
+      "epoch": 5.0984643179765134,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.88634947703103e-05,
+      "loss": 0.4573,
+      "step": 5644
+    },
+    {
+      "epoch": 5.102077687443541,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.866722061204522e-05,
+      "loss": 0.5077,
+      "step": 5648
+    },
+    {
+      "epoch": 5.105691056910569,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.847107246531064e-05,
+      "loss": 0.4731,
+      "step": 5652
+    },
+    {
+      "epoch": 5.109304426377597,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.827505073310353e-05,
+      "loss": 0.431,
+      "step": 5656
+    },
+    {
+      "epoch": 5.112917795844625,
+      "grad_norm": 0.375,
+      "learning_rate": 8.807915581816132e-05,
+      "loss": 0.4606,
+      "step": 5660
+    },
+    {
+      "epoch": 5.116531165311653,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.788338812296068e-05,
+      "loss": 0.4314,
+      "step": 5664
+    },
+    {
+      "epoch": 5.120144534778681,
+      "grad_norm": 0.39453125,
+      "learning_rate": 8.768774804971705e-05,
+      "loss": 0.4581,
+      "step": 5668
+    },
+    {
+      "epoch": 5.123757904245709,
+      "grad_norm": 0.37890625,
+      "learning_rate": 8.749223600038354e-05,
+      "loss": 0.4931,
+      "step": 5672
+    },
+    {
+      "epoch": 5.127371273712737,
+      "grad_norm": 0.375,
+      "learning_rate": 8.72968523766503e-05,
+      "loss": 0.4472,
+      "step": 5676
+    },
+    {
+      "epoch": 5.130984643179765,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.710159757994366e-05,
+      "loss": 0.4415,
+      "step": 5680
+    },
+    {
+      "epoch": 5.134598012646793,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.690647201142513e-05,
+      "loss": 0.4492,
+      "step": 5684
+    },
+    {
+      "epoch": 5.138211382113822,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.671147607199096e-05,
+      "loss": 0.4317,
+      "step": 5688
+    },
+    {
+      "epoch": 5.141824751580849,
+      "grad_norm": 0.37109375,
+      "learning_rate": 8.651661016227067e-05,
+      "loss": 0.4452,
+      "step": 5692
+    },
+    {
+      "epoch": 5.145438121047877,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.632187468262708e-05,
+      "loss": 0.4734,
+      "step": 5696
+    },
+    {
+      "epoch": 5.149051490514905,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.612727003315476e-05,
+      "loss": 0.4279,
+      "step": 5700
+    },
+    {
+      "epoch": 5.152664859981933,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.593279661367942e-05,
+      "loss": 0.4164,
+      "step": 5704
+    },
+    {
+      "epoch": 5.1562782294489615,
+      "grad_norm": 0.373046875,
+      "learning_rate": 8.573845482375754e-05,
+      "loss": 0.4438,
+      "step": 5708
+    },
+    {
+      "epoch": 5.159891598915989,
+      "grad_norm": 0.37890625,
+      "learning_rate": 8.554424506267461e-05,
+      "loss": 0.4514,
+      "step": 5712
+    },
+    {
+      "epoch": 5.163504968383017,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.535016772944534e-05,
+      "loss": 0.4454,
+      "step": 5716
+    },
+    {
+      "epoch": 5.167118337850045,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.515622322281203e-05,
+      "loss": 0.4425,
+      "step": 5720
+    },
+    {
+      "epoch": 5.170731707317073,
+      "grad_norm": 0.37109375,
+      "learning_rate": 8.496241194124436e-05,
+      "loss": 0.4466,
+      "step": 5724
+    },
+    {
+      "epoch": 5.1743450767841015,
+      "grad_norm": 0.369140625,
+      "learning_rate": 8.476873428293812e-05,
+      "loss": 0.428,
+      "step": 5728
+    },
+    {
+      "epoch": 5.177958446251129,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.457519064581444e-05,
+      "loss": 0.4494,
+      "step": 5732
+    },
+    {
+      "epoch": 5.181571815718157,
+      "grad_norm": 0.375,
+      "learning_rate": 8.438178142751939e-05,
+      "loss": 0.4485,
+      "step": 5736
+    },
+    {
+      "epoch": 5.185185185185185,
+      "grad_norm": 0.400390625,
+      "learning_rate": 8.418850702542265e-05,
+      "loss": 0.4973,
+      "step": 5740
+    },
+    {
+      "epoch": 5.188798554652213,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.399536783661713e-05,
+      "loss": 0.4924,
+      "step": 5744
+    },
+    {
+      "epoch": 5.1924119241192415,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.380236425791759e-05,
+      "loss": 0.4346,
+      "step": 5748
+    },
+    {
+      "epoch": 5.196025293586269,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.360949668586055e-05,
+      "loss": 0.4849,
+      "step": 5752
+    },
+    {
+      "epoch": 5.199638663053297,
+      "grad_norm": 0.369140625,
+      "learning_rate": 8.341676551670283e-05,
+      "loss": 0.4523,
+      "step": 5756
+    },
+    {
+      "epoch": 5.203252032520325,
+      "grad_norm": 0.400390625,
+      "learning_rate": 8.322417114642107e-05,
+      "loss": 0.4236,
+      "step": 5760
+    },
+    {
+      "epoch": 5.206865401987353,
+      "grad_norm": 0.373046875,
+      "learning_rate": 8.303171397071106e-05,
+      "loss": 0.4416,
+      "step": 5764
+    },
+    {
+      "epoch": 5.210478771454381,
+      "grad_norm": 0.359375,
+      "learning_rate": 8.283939438498627e-05,
+      "loss": 0.4522,
+      "step": 5768
+    },
+    {
+      "epoch": 5.21409214092141,
+      "grad_norm": 0.36328125,
+      "learning_rate": 8.264721278437795e-05,
+      "loss": 0.476,
+      "step": 5772
+    },
+    {
+      "epoch": 5.217705510388437,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.245516956373359e-05,
+      "loss": 0.4237,
+      "step": 5776
+    },
+    {
+      "epoch": 5.221318879855465,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.226326511761635e-05,
+      "loss": 0.4215,
+      "step": 5780
+    },
+    {
+      "epoch": 5.224932249322493,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.207149984030453e-05,
+      "loss": 0.4488,
+      "step": 5784
+    },
+    {
+      "epoch": 5.228545618789521,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.187987412579011e-05,
+      "loss": 0.4941,
+      "step": 5788
+    },
+    {
+      "epoch": 5.23215898825655,
+      "grad_norm": 0.375,
+      "learning_rate": 8.168838836777871e-05,
+      "loss": 0.4512,
+      "step": 5792
+    },
+    {
+      "epoch": 5.235772357723577,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.14970429596881e-05,
+      "loss": 0.4581,
+      "step": 5796
+    },
+    {
+      "epoch": 5.239385727190605,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.130583829464793e-05,
+      "loss": 0.4529,
+      "step": 5800
+    },
+    {
+      "epoch": 5.242999096657633,
+      "grad_norm": 0.373046875,
+      "learning_rate": 8.11147747654986e-05,
+      "loss": 0.4779,
+      "step": 5804
+    },
+    {
+      "epoch": 5.246612466124661,
+      "grad_norm": 0.37109375,
+      "learning_rate": 8.092385276479038e-05,
+      "loss": 0.3973,
+      "step": 5808
+    },
+    {
+      "epoch": 5.2502258355916895,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.073307268478303e-05,
+      "loss": 0.436,
+      "step": 5812
+    },
+    {
+      "epoch": 5.253839205058718,
+      "grad_norm": 0.37890625,
+      "learning_rate": 8.054243491744448e-05,
+      "loss": 0.4576,
+      "step": 5816
+    },
+    {
+      "epoch": 5.257452574525745,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.03519398544506e-05,
+      "loss": 0.4253,
+      "step": 5820
+    },
+    {
+      "epoch": 5.261065943992773,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.01615878871836e-05,
+      "loss": 0.4457,
+      "step": 5824
+    },
+    {
+      "epoch": 5.264679313459801,
+      "grad_norm": 0.390625,
+      "learning_rate": 7.997137940673212e-05,
+      "loss": 0.4777,
+      "step": 5828
+    },
+    {
+      "epoch": 5.2682926829268295,
+      "grad_norm": 0.39453125,
+      "learning_rate": 7.97813148038898e-05,
+      "loss": 0.4356,
+      "step": 5832
+    },
+    {
+      "epoch": 5.271906052393858,
+      "grad_norm": 0.37109375,
+      "learning_rate": 7.95913944691546e-05,
+      "loss": 0.4404,
+      "step": 5836
+    },
+    {
+      "epoch": 5.275519421860885,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.940161879272837e-05,
+      "loss": 0.4555,
+      "step": 5840
+    },
+    {
+      "epoch": 5.279132791327913,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.921198816451531e-05,
+      "loss": 0.4435,
+      "step": 5844
+    },
+    {
+      "epoch": 5.282746160794941,
+      "grad_norm": 0.375,
+      "learning_rate": 7.902250297412209e-05,
+      "loss": 0.4695,
+      "step": 5848
+    },
+    {
+      "epoch": 5.2863595302619695,
+      "grad_norm": 0.388671875,
+      "learning_rate": 7.883316361085618e-05,
+      "loss": 0.414,
+      "step": 5852
+    },
+    {
+      "epoch": 5.289972899728998,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.864397046372563e-05,
+      "loss": 0.4476,
+      "step": 5856
+    },
+    {
+      "epoch": 5.293586269196025,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.84549239214382e-05,
+      "loss": 0.4597,
+      "step": 5860
+    },
+    {
+      "epoch": 5.297199638663053,
+      "grad_norm": 0.359375,
+      "learning_rate": 7.826602437240003e-05,
+      "loss": 0.4677,
+      "step": 5864
+    },
+    {
+      "epoch": 5.300813008130081,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.80772722047157e-05,
+      "loss": 0.4376,
+      "step": 5868
+    },
+    {
+      "epoch": 5.304426377597109,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.788866780618672e-05,
+      "loss": 0.4448,
+      "step": 5872
+    },
+    {
+      "epoch": 5.308039747064138,
+      "grad_norm": 0.38671875,
+      "learning_rate": 7.770021156431115e-05,
+      "loss": 0.4528,
+      "step": 5876
+    },
+    {
+      "epoch": 5.311653116531165,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.751190386628261e-05,
+      "loss": 0.49,
+      "step": 5880
+    },
+    {
+      "epoch": 5.315266485998193,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.732374509898935e-05,
+      "loss": 0.4535,
+      "step": 5884
+    },
+    {
+      "epoch": 5.318879855465221,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.713573564901393e-05,
+      "loss": 0.4324,
+      "step": 5888
+    },
+    {
+      "epoch": 5.322493224932249,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.694787590263191e-05,
+      "loss": 0.4557,
+      "step": 5892
+    },
+    {
+      "epoch": 5.326106594399278,
+      "grad_norm": 0.34375,
+      "learning_rate": 7.676016624581143e-05,
+      "loss": 0.4346,
+      "step": 5896
+    },
+    {
+      "epoch": 5.329719963866305,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.657260706421217e-05,
+      "loss": 0.4433,
+      "step": 5900
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.638519874318461e-05,
+      "loss": 0.4558,
+      "step": 5904
+    },
+    {
+      "epoch": 5.336946702800361,
+      "grad_norm": 0.3828125,
+      "learning_rate": 7.619794166776943e-05,
+      "loss": 0.4721,
+      "step": 5908
+    },
+    {
+      "epoch": 5.340560072267389,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.60108362226963e-05,
+      "loss": 0.4502,
+      "step": 5912
+    },
+    {
+      "epoch": 5.3441734417344176,
+      "grad_norm": 0.359375,
+      "learning_rate": 7.582388279238381e-05,
+      "loss": 0.4422,
+      "step": 5916
+    },
+    {
+      "epoch": 5.347786811201446,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.563708176093765e-05,
+      "loss": 0.4733,
+      "step": 5920
+    },
+    {
+      "epoch": 5.351400180668473,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.545043351215092e-05,
+      "loss": 0.489,
+      "step": 5924
+    },
+    {
+      "epoch": 5.355013550135501,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.52639384295025e-05,
+      "loss": 0.4533,
+      "step": 5928
+    },
+    {
+      "epoch": 5.358626919602529,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.507759689615662e-05,
+      "loss": 0.4594,
+      "step": 5932
+    },
+    {
+      "epoch": 5.3622402890695575,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.489140929496229e-05,
+      "loss": 0.4292,
+      "step": 5936
+    },
+    {
+      "epoch": 5.365853658536586,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.470537600845182e-05,
+      "loss": 0.4571,
+      "step": 5940
+    },
+    {
+      "epoch": 5.369467028003613,
+      "grad_norm": 0.35546875,
+      "learning_rate": 7.45194974188409e-05,
+      "loss": 0.4081,
+      "step": 5944
+    },
+    {
+      "epoch": 5.373080397470641,
+      "grad_norm": 0.390625,
+      "learning_rate": 7.433377390802705e-05,
+      "loss": 0.4749,
+      "step": 5948
+    },
+    {
+      "epoch": 5.376693766937669,
+      "grad_norm": 0.396484375,
+      "learning_rate": 7.414820585758949e-05,
+      "loss": 0.4672,
+      "step": 5952
+    },
+    {
+      "epoch": 5.3803071364046975,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.396279364878781e-05,
+      "loss": 0.4913,
+      "step": 5956
+    },
+    {
+      "epoch": 5.383920505871726,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.377753766256135e-05,
+      "loss": 0.42,
+      "step": 5960
+    },
+    {
+      "epoch": 5.387533875338754,
+      "grad_norm": 0.3828125,
+      "learning_rate": 7.359243827952877e-05,
+      "loss": 0.4313,
+      "step": 5964
+    },
+    {
+      "epoch": 5.391147244805781,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.340749587998669e-05,
+      "loss": 0.4101,
+      "step": 5968
+    },
+    {
+      "epoch": 5.394760614272809,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.322271084390945e-05,
+      "loss": 0.4171,
+      "step": 5972
+    },
+    {
+      "epoch": 5.3983739837398375,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.303808355094791e-05,
+      "loss": 0.4458,
+      "step": 5976
+    },
+    {
+      "epoch": 5.401987353206866,
+      "grad_norm": 0.388671875,
+      "learning_rate": 7.285361438042887e-05,
+      "loss": 0.4423,
+      "step": 5980
+    },
+    {
+      "epoch": 5.405600722673894,
+      "grad_norm": 0.39453125,
+      "learning_rate": 7.266930371135426e-05,
+      "loss": 0.46,
+      "step": 5984
+    },
+    {
+      "epoch": 5.409214092140921,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.248515192240033e-05,
+      "loss": 0.4832,
+      "step": 5988
+    },
+    {
+      "epoch": 5.412827461607949,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.230115939191713e-05,
+      "loss": 0.4656,
+      "step": 5992
+    },
+    {
+      "epoch": 5.416440831074977,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.211732649792706e-05,
+      "loss": 0.4732,
+      "step": 5996
+    },
+    {
+      "epoch": 5.420054200542006,
+      "grad_norm": 0.34765625,
+      "learning_rate": 7.193365361812499e-05,
+      "loss": 0.4405,
+      "step": 6000
+    },
+    {
+      "epoch": 5.423667570009034,
+      "grad_norm": 0.37109375,
+      "learning_rate": 7.175014112987677e-05,
+      "loss": 0.4558,
+      "step": 6004
+    },
+    {
+      "epoch": 5.427280939476061,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.156678941021874e-05,
+      "loss": 0.4448,
+      "step": 6008
+    },
+    {
+      "epoch": 5.430894308943089,
+      "grad_norm": 0.375,
+      "learning_rate": 7.138359883585715e-05,
+      "loss": 0.4381,
+      "step": 6012
+    },
+    {
+      "epoch": 5.434507678410117,
+      "grad_norm": 0.37109375,
+      "learning_rate": 7.120056978316676e-05,
+      "loss": 0.4776,
+      "step": 6016
+    },
+    {
+      "epoch": 5.438121047877146,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.101770262819089e-05,
+      "loss": 0.4322,
+      "step": 6020
+    },
+    {
+      "epoch": 5.441734417344174,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.083499774663994e-05,
+      "loss": 0.4651,
+      "step": 6024
+    },
+    {
+      "epoch": 5.445347786811201,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.065245551389112e-05,
+      "loss": 0.4518,
+      "step": 6028
+    },
+    {
+      "epoch": 5.448961156278229,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.04700763049874e-05,
+      "loss": 0.4403,
+      "step": 6032
+    },
+    {
+      "epoch": 5.452574525745257,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.02878604946366e-05,
+      "loss": 0.4851,
+      "step": 6036
+    },
+    {
+      "epoch": 5.4561878952122855,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.010580845721119e-05,
+      "loss": 0.4262,
+      "step": 6040
+    },
+    {
+      "epoch": 5.459801264679314,
+      "grad_norm": 0.384765625,
+      "learning_rate": 6.992392056674687e-05,
+      "loss": 0.4951,
+      "step": 6044
+    },
+    {
+      "epoch": 5.463414634146342,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.974219719694229e-05,
+      "loss": 0.4697,
+      "step": 6048
+    },
+    {
+      "epoch": 5.467028003613369,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.956063872115796e-05,
+      "loss": 0.4828,
+      "step": 6052
+    },
+    {
+      "epoch": 5.470641373080397,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.937924551241562e-05,
+      "loss": 0.4649,
+      "step": 6056
+    },
+    {
+      "epoch": 5.4742547425474255,
+      "grad_norm": 0.353515625,
+      "learning_rate": 6.919801794339751e-05,
+      "loss": 0.4759,
+      "step": 6060
+    },
+    {
+      "epoch": 5.477868112014454,
+      "grad_norm": 0.39453125,
+      "learning_rate": 6.901695638644547e-05,
+      "loss": 0.4503,
+      "step": 6064
+    },
+    {
+      "epoch": 5.481481481481482,
+      "grad_norm": 0.365234375,
+      "learning_rate": 6.883606121356046e-05,
+      "loss": 0.4506,
+      "step": 6068
+    },
+    {
+      "epoch": 5.485094850948509,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.865533279640121e-05,
+      "loss": 0.4817,
+      "step": 6072
+    },
+    {
+      "epoch": 5.488708220415537,
+      "grad_norm": 0.365234375,
+      "learning_rate": 6.84747715062843e-05,
+      "loss": 0.4343,
+      "step": 6076
+    },
+    {
+      "epoch": 5.4923215898825655,
+      "grad_norm": 0.380859375,
+      "learning_rate": 6.829437771418257e-05,
+      "loss": 0.4502,
+      "step": 6080
+    },
+    {
+      "epoch": 5.495934959349594,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.811415179072501e-05,
+      "loss": 0.4064,
+      "step": 6084
+    },
+    {
+      "epoch": 5.499548328816622,
+      "grad_norm": 0.37109375,
+      "learning_rate": 6.793409410619556e-05,
+      "loss": 0.4308,
+      "step": 6088
+    },
+    {
+      "epoch": 5.50316169828365,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.775420503053239e-05,
+      "loss": 0.4258,
+      "step": 6092
+    },
+    {
+      "epoch": 5.506775067750677,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.757448493332752e-05,
+      "loss": 0.4541,
+      "step": 6096
+    },
+    {
+      "epoch": 5.510388437217705,
+      "grad_norm": 0.365234375,
+      "learning_rate": 6.739493418382555e-05,
+      "loss": 0.4554,
+      "step": 6100
+    },
+    {
+      "epoch": 5.514001806684734,
+      "grad_norm": 0.41796875,
+      "learning_rate": 6.721555315092341e-05,
+      "loss": 0.5069,
+      "step": 6104
+    },
+    {
+      "epoch": 5.517615176151762,
+      "grad_norm": 0.341796875,
+      "learning_rate": 6.703634220316914e-05,
+      "loss": 0.4701,
+      "step": 6108
+    },
+    {
+      "epoch": 5.52122854561879,
+      "grad_norm": 0.376953125,
+      "learning_rate": 6.685730170876122e-05,
+      "loss": 0.4212,
+      "step": 6112
+    },
+    {
+      "epoch": 5.524841915085817,
+      "grad_norm": 0.38671875,
+      "learning_rate": 6.667843203554827e-05,
+      "loss": 0.4479,
+      "step": 6116
+    },
+    {
+      "epoch": 5.528455284552845,
+      "grad_norm": 0.35546875,
+      "learning_rate": 6.649973355102765e-05,
+      "loss": 0.4272,
+      "step": 6120
+    },
+    {
+      "epoch": 5.532068654019874,
+      "grad_norm": 0.384765625,
+      "learning_rate": 6.632120662234521e-05,
+      "loss": 0.4513,
+      "step": 6124
+    },
+    {
+      "epoch": 5.535682023486902,
+      "grad_norm": 0.39453125,
+      "learning_rate": 6.614285161629417e-05,
+      "loss": 0.4464,
+      "step": 6128
+    },
+    {
+      "epoch": 5.53929539295393,
+      "grad_norm": 0.376953125,
+      "learning_rate": 6.596466889931462e-05,
+      "loss": 0.4526,
+      "step": 6132
+    },
+    {
+      "epoch": 5.542908762420957,
+      "grad_norm": 0.369140625,
+      "learning_rate": 6.578665883749262e-05,
+      "loss": 0.455,
+      "step": 6136
+    },
+    {
+      "epoch": 5.546522131887985,
+      "grad_norm": 0.359375,
+      "learning_rate": 6.560882179655946e-05,
+      "loss": 0.4683,
+      "step": 6140
+    },
+    {
+      "epoch": 5.5501355013550135,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.543115814189112e-05,
+      "loss": 0.4441,
+      "step": 6144
+    },
+    {
+      "epoch": 5.553748870822042,
+      "grad_norm": 0.3828125,
+      "learning_rate": 6.52536682385072e-05,
+      "loss": 0.445,
+      "step": 6148
+    },
+    {
+      "epoch": 5.55736224028907,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.507635245107036e-05,
+      "loss": 0.4254,
+      "step": 6152
+    },
+    {
+      "epoch": 5.560975609756097,
+      "grad_norm": 0.376953125,
+      "learning_rate": 6.489921114388547e-05,
+      "loss": 0.4385,
+      "step": 6156
+    },
+    {
+      "epoch": 5.564588979223125,
+      "grad_norm": 0.396484375,
+      "learning_rate": 6.472224468089909e-05,
+      "loss": 0.4733,
+      "step": 6160
+    },
+    {
+      "epoch": 5.5682023486901535,
+      "grad_norm": 0.376953125,
+      "learning_rate": 6.454545342569842e-05,
+      "loss": 0.4438,
+      "step": 6164
+    },
+    {
+      "epoch": 5.571815718157182,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.436883774151057e-05,
+      "loss": 0.4355,
+      "step": 6168
+    },
+    {
+      "epoch": 5.57542908762421,
+      "grad_norm": 0.396484375,
+      "learning_rate": 6.419239799120222e-05,
+      "loss": 0.486,
+      "step": 6172
+    },
+    {
+      "epoch": 5.579042457091237,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.401613453727834e-05,
+      "loss": 0.4683,
+      "step": 6176
+    },
+    {
+      "epoch": 5.582655826558265,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.384004774188184e-05,
+      "loss": 0.4502,
+      "step": 6180
+    },
+    {
+      "epoch": 5.5862691960252935,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.366413796679266e-05,
+      "loss": 0.4511,
+      "step": 6184
+    },
+    {
+      "epoch": 5.589882565492322,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.348840557342684e-05,
+      "loss": 0.4497,
+      "step": 6188
+    },
+    {
+      "epoch": 5.59349593495935,
+      "grad_norm": 0.369140625,
+      "learning_rate": 6.331285092283627e-05,
+      "loss": 0.4689,
+      "step": 6192
+    },
+    {
+      "epoch": 5.597109304426377,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.313747437570746e-05,
+      "loss": 0.4535,
+      "step": 6196
+    },
+    {
+      "epoch": 5.600722673893405,
+      "grad_norm": 0.376953125,
+      "learning_rate": 6.296227629236113e-05,
+      "loss": 0.4724,
+      "step": 6200
+    },
+    {
+      "epoch": 5.6043360433604335,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.278725703275124e-05,
+      "loss": 0.4687,
+      "step": 6204
+    },
+    {
+      "epoch": 5.607949412827462,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.261241695646437e-05,
+      "loss": 0.4474,
+      "step": 6208
+    },
+    {
+      "epoch": 5.61156278229449,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.243775642271895e-05,
+      "loss": 0.4471,
+      "step": 6212
+    },
+    {
+      "epoch": 5.615176151761518,
+      "grad_norm": 0.376953125,
+      "learning_rate": 6.226327579036451e-05,
+      "loss": 0.4865,
+      "step": 6216
+    },
+    {
+      "epoch": 5.618789521228545,
+      "grad_norm": 0.3828125,
+      "learning_rate": 6.208897541788109e-05,
+      "loss": 0.4483,
+      "step": 6220
+    },
+    {
+      "epoch": 5.622402890695573,
+      "grad_norm": 0.380859375,
+      "learning_rate": 6.191485566337825e-05,
+      "loss": 0.4523,
+      "step": 6224
+    },
+    {
+      "epoch": 5.626016260162602,
+      "grad_norm": 0.396484375,
+      "learning_rate": 6.174091688459448e-05,
+      "loss": 0.4679,
+      "step": 6228
+    },
+    {
+      "epoch": 5.62962962962963,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.156715943889641e-05,
+      "loss": 0.4603,
+      "step": 6232
+    },
+    {
+      "epoch": 5.633242999096658,
+      "grad_norm": 0.41796875,
+      "learning_rate": 6.139358368327826e-05,
+      "loss": 0.451,
+      "step": 6236
+    },
+    {
+      "epoch": 5.636856368563686,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.122018997436087e-05,
+      "loss": 0.4728,
+      "step": 6240
+    },
+    {
+      "epoch": 5.640469738030713,
+      "grad_norm": 0.396484375,
+      "learning_rate": 6.104697866839089e-05,
+      "loss": 0.4546,
+      "step": 6244
+    },
+    {
+      "epoch": 5.644083107497742,
+      "grad_norm": 0.42578125,
+      "learning_rate": 6.0873950121240505e-05,
+      "loss": 0.4323,
+      "step": 6248
+    },
+    {
+      "epoch": 5.64769647696477,
+      "grad_norm": 0.3828125,
+      "learning_rate": 6.070110468840617e-05,
+      "loss": 0.4531,
+      "step": 6252
+    },
+    {
+      "epoch": 5.651309846431798,
+      "grad_norm": 0.3828125,
+      "learning_rate": 6.052844272500836e-05,
+      "loss": 0.4546,
+      "step": 6256
+    },
+    {
+      "epoch": 5.654923215898826,
+      "grad_norm": 0.400390625,
+      "learning_rate": 6.0355964585790454e-05,
+      "loss": 0.449,
+      "step": 6260
+    },
+    {
+      "epoch": 5.658536585365853,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.018367062511799e-05,
+      "loss": 0.4686,
+      "step": 6264
+    },
+    {
+      "epoch": 5.6621499548328815,
+      "grad_norm": 0.375,
+      "learning_rate": 6.001156119697846e-05,
+      "loss": 0.4225,
+      "step": 6268
+    },
+    {
+      "epoch": 5.66576332429991,
+      "grad_norm": 0.380859375,
+      "learning_rate": 5.983963665497993e-05,
+      "loss": 0.4427,
+      "step": 6272
+    },
+    {
+      "epoch": 5.669376693766938,
+      "grad_norm": 0.39453125,
+      "learning_rate": 5.966789735235082e-05,
+      "loss": 0.4798,
+      "step": 6276
+    },
+    {
+      "epoch": 5.672990063233966,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.949634364193883e-05,
+      "loss": 0.4689,
+      "step": 6280
+    },
+    {
+      "epoch": 5.676603432700993,
+      "grad_norm": 0.376953125,
+      "learning_rate": 5.9324975876210376e-05,
+      "loss": 0.4478,
+      "step": 6284
+    },
+    {
+      "epoch": 5.6802168021680215,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.915379440724986e-05,
+      "loss": 0.5014,
+      "step": 6288
+    },
+    {
+      "epoch": 5.68383017163505,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.898279958675886e-05,
+      "loss": 0.4768,
+      "step": 6292
+    },
+    {
+      "epoch": 5.687443541102078,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.8811991766055636e-05,
+      "loss": 0.4688,
+      "step": 6296
+    },
+    {
+      "epoch": 5.691056910569106,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.864137129607409e-05,
+      "loss": 0.4298,
+      "step": 6300
+    },
+    {
+      "epoch": 5.694670280036133,
+      "grad_norm": 0.388671875,
+      "learning_rate": 5.847093852736326e-05,
+      "loss": 0.4591,
+      "step": 6304
+    },
+    {
+      "epoch": 5.6982836495031615,
+      "grad_norm": 0.357421875,
+      "learning_rate": 5.8300693810086485e-05,
+      "loss": 0.4219,
+      "step": 6308
+    },
+    {
+      "epoch": 5.70189701897019,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.813063749402089e-05,
+      "loss": 0.4736,
+      "step": 6312
+    },
+    {
+      "epoch": 5.705510388437218,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.796076992855645e-05,
+      "loss": 0.4435,
+      "step": 6316
+    },
+    {
+      "epoch": 5.709123757904246,
+      "grad_norm": 0.359375,
+      "learning_rate": 5.7791091462695134e-05,
+      "loss": 0.4688,
+      "step": 6320
+    },
+    {
+      "epoch": 5.712737127371273,
+      "grad_norm": 0.384765625,
+      "learning_rate": 5.7621602445050744e-05,
+      "loss": 0.4473,
+      "step": 6324
+    },
+    {
+      "epoch": 5.716350496838301,
+      "grad_norm": 0.39453125,
+      "learning_rate": 5.745230322384757e-05,
+      "loss": 0.4365,
+      "step": 6328
+    },
+    {
+      "epoch": 5.71996386630533,
+      "grad_norm": 0.384765625,
+      "learning_rate": 5.728319414692019e-05,
+      "loss": 0.45,
+      "step": 6332
+    },
+    {
+      "epoch": 5.723577235772358,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.7114275561712394e-05,
+      "loss": 0.4337,
+      "step": 6336
+    },
+    {
+      "epoch": 5.727190605239386,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.694554781527643e-05,
+      "loss": 0.4329,
+      "step": 6340
+    },
+    {
+      "epoch": 5.730803974706414,
+      "grad_norm": 0.361328125,
+      "learning_rate": 5.677701125427278e-05,
+      "loss": 0.459,
+      "step": 6344
+    },
+    {
+      "epoch": 5.734417344173441,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.66086662249689e-05,
+      "loss": 0.4699,
+      "step": 6348
+    },
+    {
+      "epoch": 5.73803071364047,
+      "grad_norm": 0.375,
+      "learning_rate": 5.6440513073238885e-05,
+      "loss": 0.4946,
+      "step": 6352
+    },
+    {
+      "epoch": 5.741644083107498,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.627255214456249e-05,
+      "loss": 0.4601,
+      "step": 6356
+    },
+    {
+      "epoch": 5.745257452574526,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.610478378402457e-05,
+      "loss": 0.4391,
+      "step": 6360
+    },
+    {
+      "epoch": 5.748870822041554,
+      "grad_norm": 0.380859375,
+      "learning_rate": 5.593720833631434e-05,
+      "loss": 0.4241,
+      "step": 6364
+    },
+    {
+      "epoch": 5.752484191508581,
+      "grad_norm": 0.375,
+      "learning_rate": 5.576982614572464e-05,
+      "loss": 0.4315,
+      "step": 6368
+    },
+    {
+      "epoch": 5.7560975609756095,
+      "grad_norm": 0.376953125,
+      "learning_rate": 5.560263755615137e-05,
+      "loss": 0.4534,
+      "step": 6372
+    },
+    {
+      "epoch": 5.759710930442638,
+      "grad_norm": 0.373046875,
+      "learning_rate": 5.5435642911092534e-05,
+      "loss": 0.4628,
+      "step": 6376
+    },
+    {
+      "epoch": 5.763324299909666,
+      "grad_norm": 0.388671875,
+      "learning_rate": 5.526884255364771e-05,
+      "loss": 0.4114,
+      "step": 6380
+    },
+    {
+      "epoch": 5.766937669376694,
+      "grad_norm": 0.421875,
+      "learning_rate": 5.5102236826517234e-05,
+      "loss": 0.495,
+      "step": 6384
+    },
+    {
+      "epoch": 5.770551038843722,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.493582607200176e-05,
+      "loss": 0.4702,
+      "step": 6388
+    },
+    {
+      "epoch": 5.7741644083107495,
+      "grad_norm": 0.38671875,
+      "learning_rate": 5.4769610632001164e-05,
+      "loss": 0.478,
+      "step": 6392
+    },
+    {
+      "epoch": 5.777777777777778,
+      "grad_norm": 0.375,
+      "learning_rate": 5.460359084801411e-05,
+      "loss": 0.4456,
+      "step": 6396
+    },
+    {
+      "epoch": 5.781391147244806,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.4437767061137285e-05,
+      "loss": 0.4366,
+      "step": 6400
+    },
+    {
+      "epoch": 5.785004516711834,
+      "grad_norm": 0.376953125,
+      "learning_rate": 5.42721396120646e-05,
+      "loss": 0.4836,
+      "step": 6404
+    },
+    {
+      "epoch": 5.788617886178862,
+      "grad_norm": 0.376953125,
+      "learning_rate": 5.41067088410868e-05,
+      "loss": 0.4784,
+      "step": 6408
+    },
+    {
+      "epoch": 5.7922312556458895,
+      "grad_norm": 0.357421875,
+      "learning_rate": 5.394147508809036e-05,
+      "loss": 0.4618,
+      "step": 6412
+    },
+    {
+      "epoch": 5.795844625112918,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.377643869255687e-05,
+      "loss": 0.4266,
+      "step": 6416
+    },
+    {
+      "epoch": 5.799457994579946,
+      "grad_norm": 0.380859375,
+      "learning_rate": 5.3611599993562726e-05,
+      "loss": 0.4439,
+      "step": 6420
+    },
+    {
+      "epoch": 5.803071364046974,
+      "grad_norm": 0.365234375,
+      "learning_rate": 5.344695932977791e-05,
+      "loss": 0.4716,
+      "step": 6424
+    },
+    {
+      "epoch": 5.806684733514002,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.328251703946573e-05,
+      "loss": 0.4725,
+      "step": 6428
+    },
+    {
+      "epoch": 5.8102981029810294,
+      "grad_norm": 0.3828125,
+      "learning_rate": 5.311827346048174e-05,
+      "loss": 0.4721,
+      "step": 6432
+    },
+    {
+      "epoch": 5.813911472448058,
+      "grad_norm": 0.38671875,
+      "learning_rate": 5.2954228930273306e-05,
+      "loss": 0.4933,
+      "step": 6436
+    },
+    {
+      "epoch": 5.817524841915086,
+      "grad_norm": 0.376953125,
+      "learning_rate": 5.279038378587884e-05,
+      "loss": 0.4656,
+      "step": 6440
+    },
+    {
+      "epoch": 5.821138211382114,
+      "grad_norm": 0.375,
+      "learning_rate": 5.262673836392707e-05,
+      "loss": 0.4642,
+      "step": 6444
+    },
+    {
+      "epoch": 5.824751580849142,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.2463293000636485e-05,
+      "loss": 0.4859,
+      "step": 6448
+    },
+    {
+      "epoch": 5.828364950316169,
+      "grad_norm": 0.373046875,
+      "learning_rate": 5.230004803181444e-05,
+      "loss": 0.477,
+      "step": 6452
+    },
+    {
+      "epoch": 5.831978319783198,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.213700379285659e-05,
+      "loss": 0.4692,
+      "step": 6456
+    },
+    {
+      "epoch": 5.835591689250226,
+      "grad_norm": 0.38671875,
+      "learning_rate": 5.197416061874615e-05,
+      "loss": 0.4841,
+      "step": 6460
+    },
+    {
+      "epoch": 5.839205058717254,
+      "grad_norm": 0.388671875,
+      "learning_rate": 5.181151884405336e-05,
+      "loss": 0.4608,
+      "step": 6464
+    },
+    {
+      "epoch": 5.842818428184282,
+      "grad_norm": 0.39453125,
+      "learning_rate": 5.1649078802934575e-05,
+      "loss": 0.4724,
+      "step": 6468
+    },
+    {
+      "epoch": 5.846431797651309,
+      "grad_norm": 0.361328125,
+      "learning_rate": 5.1486840829131664e-05,
+      "loss": 0.4707,
+      "step": 6472
+    },
+    {
+      "epoch": 5.8500451671183376,
+      "grad_norm": 0.37109375,
+      "learning_rate": 5.132480525597137e-05,
+      "loss": 0.4186,
+      "step": 6476
+    },
+    {
+      "epoch": 5.853658536585366,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.116297241636455e-05,
+      "loss": 0.4941,
+      "step": 6480
+    },
+    {
+      "epoch": 5.857271906052394,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.100134264280566e-05,
+      "loss": 0.4667,
+      "step": 6484
+    },
+    {
+      "epoch": 5.860885275519422,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.0839916267371863e-05,
+      "loss": 0.466,
+      "step": 6488
+    },
+    {
+      "epoch": 5.86449864498645,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.067869362172227e-05,
+      "loss": 0.4296,
+      "step": 6492
+    },
+    {
+      "epoch": 5.8681120144534775,
+      "grad_norm": 0.3984375,
+      "learning_rate": 5.051767503709776e-05,
+      "loss": 0.4593,
+      "step": 6496
+    },
+    {
+      "epoch": 5.871725383920506,
+      "grad_norm": 0.361328125,
+      "learning_rate": 5.035686084431961e-05,
+      "loss": 0.4522,
+      "step": 6500
+    },
+    {
+      "epoch": 5.875338753387534,
+      "grad_norm": 0.37109375,
+      "learning_rate": 5.0196251373789485e-05,
+      "loss": 0.4357,
+      "step": 6504
+    },
+    {
+      "epoch": 5.878952122854562,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.003584695548819e-05,
+      "loss": 0.467,
+      "step": 6508
+    },
+    {
+      "epoch": 5.88256549232159,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.987564791897532e-05,
+      "loss": 0.4651,
+      "step": 6512
+    },
+    {
+      "epoch": 5.886178861788618,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.971565459338854e-05,
+      "loss": 0.4648,
+      "step": 6516
+    },
+    {
+      "epoch": 5.889792231255646,
+      "grad_norm": 0.380859375,
+      "learning_rate": 4.9555867307442745e-05,
+      "loss": 0.4459,
+      "step": 6520
+    },
+    {
+      "epoch": 5.893405600722674,
+      "grad_norm": 0.380859375,
+      "learning_rate": 4.939628638942975e-05,
+      "loss": 0.4776,
+      "step": 6524
+    },
+    {
+      "epoch": 5.897018970189702,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.9236912167217154e-05,
+      "loss": 0.4891,
+      "step": 6528
+    },
+    {
+      "epoch": 5.90063233965673,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.9077744968247956e-05,
+      "loss": 0.4506,
+      "step": 6532
+    },
+    {
+      "epoch": 5.904245709123758,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.891878511953977e-05,
+      "loss": 0.4284,
+      "step": 6536
+    },
+    {
+      "epoch": 5.907859078590786,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.876003294768437e-05,
+      "loss": 0.4613,
+      "step": 6540
+    },
+    {
+      "epoch": 5.911472448057814,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.860148877884667e-05,
+      "loss": 0.4556,
+      "step": 6544
+    },
+    {
+      "epoch": 5.915085817524842,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.844315293876428e-05,
+      "loss": 0.4644,
+      "step": 6548
+    },
+    {
+      "epoch": 5.91869918699187,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.8285025752746794e-05,
+      "loss": 0.473,
+      "step": 6552
+    },
+    {
+      "epoch": 5.922312556458898,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.812710754567507e-05,
+      "loss": 0.4223,
+      "step": 6556
+    },
+    {
+      "epoch": 5.925925925925926,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.7969398642000745e-05,
+      "loss": 0.4756,
+      "step": 6560
+    },
+    {
+      "epoch": 5.929539295392954,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.781189936574534e-05,
+      "loss": 0.4198,
+      "step": 6564
+    },
+    {
+      "epoch": 5.933152664859982,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.7654610040499535e-05,
+      "loss": 0.4755,
+      "step": 6568
+    },
+    {
+      "epoch": 5.93676603432701,
+      "grad_norm": 0.400390625,
+      "learning_rate": 4.749753098942295e-05,
+      "loss": 0.4607,
+      "step": 6572
+    },
+    {
+      "epoch": 5.940379403794038,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.734066253524295e-05,
+      "loss": 0.4254,
+      "step": 6576
+    },
+    {
+      "epoch": 5.943992773261066,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.718400500025437e-05,
+      "loss": 0.4586,
+      "step": 6580
+    },
+    {
+      "epoch": 5.947606142728094,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.702755870631862e-05,
+      "loss": 0.4384,
+      "step": 6584
+    },
+    {
+      "epoch": 5.951219512195122,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.68713239748631e-05,
+      "loss": 0.3966,
+      "step": 6588
+    },
+    {
+      "epoch": 5.95483288166215,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.6715301126880574e-05,
+      "loss": 0.4167,
+      "step": 6592
+    },
+    {
+      "epoch": 5.958446251129178,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.6559490482928415e-05,
+      "loss": 0.4627,
+      "step": 6596
+    },
+    {
+      "epoch": 5.9620596205962055,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.640389236312816e-05,
+      "loss": 0.4451,
+      "step": 6600
+    },
+    {
+      "epoch": 5.965672990063234,
+      "grad_norm": 0.375,
+      "learning_rate": 4.6248507087164544e-05,
+      "loss": 0.4344,
+      "step": 6604
+    },
+    {
+      "epoch": 5.969286359530262,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.609333497428508e-05,
+      "loss": 0.4447,
+      "step": 6608
+    },
+    {
+      "epoch": 5.97289972899729,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.593837634329928e-05,
+      "loss": 0.4399,
+      "step": 6612
+    },
+    {
+      "epoch": 5.976513098464318,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.578363151257815e-05,
+      "loss": 0.459,
+      "step": 6616
+    },
+    {
+      "epoch": 5.9801264679313455,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.5629100800053315e-05,
+      "loss": 0.452,
+      "step": 6620
+    },
+    {
+      "epoch": 5.983739837398374,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.547478452321654e-05,
+      "loss": 0.4198,
+      "step": 6624
+    },
+    {
+      "epoch": 5.987353206865402,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.532068299911899e-05,
+      "loss": 0.4881,
+      "step": 6628
+    },
+    {
+      "epoch": 5.99096657633243,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.5166796544370584e-05,
+      "loss": 0.4498,
+      "step": 6632
+    },
+    {
+      "epoch": 5.994579945799458,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.501312547513953e-05,
+      "loss": 0.4678,
+      "step": 6636
+    },
+    {
+      "epoch": 5.998193315266486,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.485967010715131e-05,
+      "loss": 0.4275,
+      "step": 6640
+    },
+    {
+      "epoch": 6.001806684733514,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.4706430755688363e-05,
+      "loss": 0.4462,
+      "step": 6644
+    },
+    {
+      "epoch": 6.005420054200542,
+      "grad_norm": 0.353515625,
+      "learning_rate": 4.455340773558923e-05,
+      "loss": 0.4047,
+      "step": 6648
+    },
+    {
+      "epoch": 6.00903342366757,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.4400601361248e-05,
+      "loss": 0.4123,
+      "step": 6652
+    },
+    {
+      "epoch": 6.012646793134598,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.424801194661377e-05,
+      "loss": 0.4104,
+      "step": 6656
+    },
+    {
+      "epoch": 6.016260162601626,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.409563980518975e-05,
+      "loss": 0.4238,
+      "step": 6660
+    },
+    {
+      "epoch": 6.019873532068654,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.394348525003279e-05,
+      "loss": 0.4041,
+      "step": 6664
+    },
+    {
+      "epoch": 6.023486901535682,
+      "grad_norm": 0.361328125,
+      "learning_rate": 4.379154859375268e-05,
+      "loss": 0.431,
+      "step": 6668
+    },
+    {
+      "epoch": 6.02710027100271,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.363983014851151e-05,
+      "loss": 0.3976,
+      "step": 6672
+    },
+    {
+      "epoch": 6.030713640469738,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.3488330226023183e-05,
+      "loss": 0.4231,
+      "step": 6676
+    },
+    {
+      "epoch": 6.034327009936766,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.333704913755244e-05,
+      "loss": 0.4579,
+      "step": 6680
+    },
+    {
+      "epoch": 6.0379403794037945,
+      "grad_norm": 0.380859375,
+      "learning_rate": 4.318598719391455e-05,
+      "loss": 0.4168,
+      "step": 6684
+    },
+    {
+      "epoch": 6.041553748870822,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.303514470547442e-05,
+      "loss": 0.4644,
+      "step": 6688
+    },
+    {
+      "epoch": 6.04516711833785,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.288452198214625e-05,
+      "loss": 0.4276,
+      "step": 6692
+    },
+    {
+      "epoch": 6.048780487804878,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.273411933339255e-05,
+      "loss": 0.4901,
+      "step": 6696
+    },
+    {
+      "epoch": 6.052393857271906,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.258393706822376e-05,
+      "loss": 0.388,
+      "step": 6700
+    },
+    {
+      "epoch": 6.056007226738934,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.243397549519749e-05,
+      "loss": 0.4287,
+      "step": 6704
+    },
+    {
+      "epoch": 6.059620596205962,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.228423492241793e-05,
+      "loss": 0.4281,
+      "step": 6708
+    },
+    {
+      "epoch": 6.06323396567299,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.2134715657535315e-05,
+      "loss": 0.4374,
+      "step": 6712
+    },
+    {
+      "epoch": 6.066847335140018,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.198541800774502e-05,
+      "loss": 0.4366,
+      "step": 6716
+    },
+    {
+      "epoch": 6.070460704607046,
+      "grad_norm": 0.380859375,
+      "learning_rate": 4.183634227978724e-05,
+      "loss": 0.4594,
+      "step": 6720
+    },
+    {
+      "epoch": 6.074074074074074,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.168748877994613e-05,
+      "loss": 0.452,
+      "step": 6724
+    },
+    {
+      "epoch": 6.077687443541102,
+      "grad_norm": 0.3984375,
+      "learning_rate": 4.1538857814049266e-05,
+      "loss": 0.4504,
+      "step": 6728
+    },
+    {
+      "epoch": 6.08130081300813,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.139044968746713e-05,
+      "loss": 0.4236,
+      "step": 6732
+    },
+    {
+      "epoch": 6.084914182475158,
+      "grad_norm": 0.375,
+      "learning_rate": 4.124226470511228e-05,
+      "loss": 0.4384,
+      "step": 6736
+    },
+    {
+      "epoch": 6.088527551942186,
+      "grad_norm": 0.35546875,
+      "learning_rate": 4.109430317143876e-05,
+      "loss": 0.4464,
+      "step": 6740
+    },
+    {
+      "epoch": 6.092140921409214,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.0946565390441636e-05,
+      "loss": 0.4661,
+      "step": 6744
+    },
+    {
+      "epoch": 6.095754290876242,
+      "grad_norm": 0.361328125,
+      "learning_rate": 4.079905166565616e-05,
+      "loss": 0.4124,
+      "step": 6748
+    },
+    {
+      "epoch": 6.09936766034327,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.065176230015737e-05,
+      "loss": 0.3999,
+      "step": 6752
+    },
+    {
+      "epoch": 6.102981029810298,
+      "grad_norm": 0.41015625,
+      "learning_rate": 4.0504697596559244e-05,
+      "loss": 0.4859,
+      "step": 6756
+    },
+    {
+      "epoch": 6.106594399277326,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.035785785701422e-05,
+      "loss": 0.4529,
+      "step": 6760
+    },
+    {
+      "epoch": 6.110207768744354,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.021124338321249e-05,
+      "loss": 0.4509,
+      "step": 6764
+    },
+    {
+      "epoch": 6.1138211382113825,
+      "grad_norm": 0.400390625,
+      "learning_rate": 4.0064854476381525e-05,
+      "loss": 0.4201,
+      "step": 6768
+    },
+    {
+      "epoch": 6.11743450767841,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.991869143728526e-05,
+      "loss": 0.4354,
+      "step": 6772
+    },
+    {
+      "epoch": 6.121047877145438,
+      "grad_norm": 0.357421875,
+      "learning_rate": 3.977275456622363e-05,
+      "loss": 0.4321,
+      "step": 6776
+    },
+    {
+      "epoch": 6.124661246612466,
+      "grad_norm": 0.396484375,
+      "learning_rate": 3.962704416303184e-05,
+      "loss": 0.4153,
+      "step": 6780
+    },
+    {
+      "epoch": 6.128274616079494,
+      "grad_norm": 0.34375,
+      "learning_rate": 3.948156052707982e-05,
+      "loss": 0.4049,
+      "step": 6784
+    },
+    {
+      "epoch": 6.1318879855465225,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.933630395727167e-05,
+      "loss": 0.4081,
+      "step": 6788
+    },
+    {
+      "epoch": 6.13550135501355,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.91912747520449e-05,
+      "loss": 0.4504,
+      "step": 6792
+    },
+    {
+      "epoch": 6.139114724480578,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.9046473209369874e-05,
+      "loss": 0.4187,
+      "step": 6796
+    },
+    {
+      "epoch": 6.142728093947606,
+      "grad_norm": 0.38671875,
+      "learning_rate": 3.890189962674926e-05,
+      "loss": 0.4668,
+      "step": 6800
+    },
+    {
+      "epoch": 6.146341463414634,
+      "grad_norm": 0.33984375,
+      "learning_rate": 3.8757554301217266e-05,
+      "loss": 0.4381,
+      "step": 6804
+    },
+    {
+      "epoch": 6.1499548328816624,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.8613437529339335e-05,
+      "loss": 0.4509,
+      "step": 6808
+    },
+    {
+      "epoch": 6.15356820234869,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.8469549607211156e-05,
+      "loss": 0.4341,
+      "step": 6812
+    },
+    {
+      "epoch": 6.157181571815718,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.832589083045827e-05,
+      "loss": 0.4367,
+      "step": 6816
+    },
+    {
+      "epoch": 6.160794941282746,
+      "grad_norm": 0.373046875,
+      "learning_rate": 3.818246149423548e-05,
+      "loss": 0.4402,
+      "step": 6820
+    },
+    {
+      "epoch": 6.164408310749774,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.8039261893226106e-05,
+      "loss": 0.4786,
+      "step": 6824
+    },
+    {
+      "epoch": 6.168021680216802,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.789629232164157e-05,
+      "loss": 0.4196,
+      "step": 6828
+    },
+    {
+      "epoch": 6.171635049683831,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.775355307322063e-05,
+      "loss": 0.3895,
+      "step": 6832
+    },
+    {
+      "epoch": 6.175248419150858,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.76110444412288e-05,
+      "loss": 0.3994,
+      "step": 6836
+    },
+    {
+      "epoch": 6.178861788617886,
+      "grad_norm": 0.373046875,
+      "learning_rate": 3.7468766718457794e-05,
+      "loss": 0.4655,
+      "step": 6840
+    },
+    {
+      "epoch": 6.182475158084914,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.732672019722501e-05,
+      "loss": 0.4477,
+      "step": 6844
+    },
+    {
+      "epoch": 6.186088527551942,
+      "grad_norm": 0.3828125,
+      "learning_rate": 3.718490516937271e-05,
+      "loss": 0.4314,
+      "step": 6848
+    },
+    {
+      "epoch": 6.1897018970189706,
+      "grad_norm": 0.38671875,
+      "learning_rate": 3.7043321926267576e-05,
+      "loss": 0.4109,
+      "step": 6852
+    },
+    {
+      "epoch": 6.193315266485998,
+      "grad_norm": 0.375,
+      "learning_rate": 3.6901970758800064e-05,
+      "loss": 0.4238,
+      "step": 6856
+    },
+    {
+      "epoch": 6.196928635953026,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.676085195738384e-05,
+      "loss": 0.4011,
+      "step": 6860
+    },
+    {
+      "epoch": 6.200542005420054,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.661996581195518e-05,
+      "loss": 0.4163,
+      "step": 6864
+    },
+    {
+      "epoch": 6.204155374887082,
+      "grad_norm": 0.373046875,
+      "learning_rate": 3.647931261197233e-05,
+      "loss": 0.4219,
+      "step": 6868
+    },
+    {
+      "epoch": 6.2077687443541105,
+      "grad_norm": 0.396484375,
+      "learning_rate": 3.633889264641493e-05,
+      "loss": 0.4499,
+      "step": 6872
+    },
+    {
+      "epoch": 6.211382113821138,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.6198706203783414e-05,
+      "loss": 0.4499,
+      "step": 6876
+    },
+    {
+      "epoch": 6.214995483288166,
+      "grad_norm": 0.404296875,
+      "learning_rate": 3.6058753572098446e-05,
+      "loss": 0.4586,
+      "step": 6880
+    },
+    {
+      "epoch": 6.218608852755194,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.5919035038900376e-05,
+      "loss": 0.4048,
+      "step": 6884
+    },
+    {
+      "epoch": 6.222222222222222,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.577955089124848e-05,
+      "loss": 0.4284,
+      "step": 6888
+    },
+    {
+      "epoch": 6.2258355916892505,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.564030141572055e-05,
+      "loss": 0.414,
+      "step": 6892
+    },
+    {
+      "epoch": 6.229448961156278,
+      "grad_norm": 0.357421875,
+      "learning_rate": 3.5501286898412204e-05,
+      "loss": 0.4433,
+      "step": 6896
+    },
+    {
+      "epoch": 6.233062330623306,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.5362507624936265e-05,
+      "loss": 0.4141,
+      "step": 6900
+    },
+    {
+      "epoch": 6.236675700090334,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.522396388042238e-05,
+      "loss": 0.4694,
+      "step": 6904
+    },
+    {
+      "epoch": 6.240289069557362,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.50856559495162e-05,
+      "loss": 0.4377,
+      "step": 6908
+    },
+    {
+      "epoch": 6.2439024390243905,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.494758411637884e-05,
+      "loss": 0.4518,
+      "step": 6912
+    },
+    {
+      "epoch": 6.247515808491419,
+      "grad_norm": 0.3828125,
+      "learning_rate": 3.480974866468638e-05,
+      "loss": 0.4469,
+      "step": 6916
+    },
+    {
+      "epoch": 6.251129177958446,
+      "grad_norm": 0.373046875,
+      "learning_rate": 3.4672149877629346e-05,
+      "loss": 0.4501,
+      "step": 6920
+    },
+    {
+      "epoch": 6.254742547425474,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.453478803791184e-05,
+      "loss": 0.3961,
+      "step": 6924
+    },
+    {
+      "epoch": 6.258355916892502,
+      "grad_norm": 0.3828125,
+      "learning_rate": 3.439766342775127e-05,
+      "loss": 0.4447,
+      "step": 6928
+    },
+    {
+      "epoch": 6.26196928635953,
+      "grad_norm": 0.412109375,
+      "learning_rate": 3.42607763288776e-05,
+      "loss": 0.4497,
+      "step": 6932
+    },
+    {
+      "epoch": 6.265582655826559,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.412412702253277e-05,
+      "loss": 0.4258,
+      "step": 6936
+    },
+    {
+      "epoch": 6.269196025293586,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.39877157894703e-05,
+      "loss": 0.4462,
+      "step": 6940
+    },
+    {
+      "epoch": 6.272809394760614,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.385154290995443e-05,
+      "loss": 0.4711,
+      "step": 6944
+    },
+    {
+      "epoch": 6.276422764227642,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.371560866375976e-05,
+      "loss": 0.4123,
+      "step": 6948
+    },
+    {
+      "epoch": 6.28003613369467,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.3579913330170586e-05,
+      "loss": 0.4543,
+      "step": 6952
+    },
+    {
+      "epoch": 6.283649503161699,
+      "grad_norm": 0.37890625,
+      "learning_rate": 3.34444571879803e-05,
+      "loss": 0.4503,
+      "step": 6956
+    },
+    {
+      "epoch": 6.287262872628727,
+      "grad_norm": 0.36328125,
+      "learning_rate": 3.330924051549101e-05,
+      "loss": 0.4416,
+      "step": 6960
+    },
+    {
+      "epoch": 6.290876242095754,
+      "grad_norm": 0.373046875,
+      "learning_rate": 3.317426359051268e-05,
+      "loss": 0.4267,
+      "step": 6964
+    },
+    {
+      "epoch": 6.294489611562782,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.303952669036274e-05,
+      "loss": 0.4078,
+      "step": 6968
+    },
+    {
+      "epoch": 6.29810298102981,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.290503009186545e-05,
+      "loss": 0.4495,
+      "step": 6972
+    },
+    {
+      "epoch": 6.3017163504968385,
+      "grad_norm": 0.38671875,
+      "learning_rate": 3.27707740713514e-05,
+      "loss": 0.4247,
+      "step": 6976
+    },
+    {
+      "epoch": 6.305329719963867,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.263675890465692e-05,
+      "loss": 0.4255,
+      "step": 6980
+    },
+    {
+      "epoch": 6.308943089430894,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.250298486712345e-05,
+      "loss": 0.448,
+      "step": 6984
+    },
+    {
+      "epoch": 6.312556458897922,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.2369452233596994e-05,
+      "loss": 0.3959,
+      "step": 6988
+    },
+    {
+      "epoch": 6.31616982836495,
+      "grad_norm": 0.400390625,
+      "learning_rate": 3.22361612784276e-05,
+      "loss": 0.4513,
+      "step": 6992
+    },
+    {
+      "epoch": 6.3197831978319785,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.210311227546884e-05,
+      "loss": 0.4498,
+      "step": 6996
+    },
+    {
+      "epoch": 6.323396567299007,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.197030549807708e-05,
+      "loss": 0.491,
+      "step": 7000
+    },
+    {
+      "epoch": 6.327009936766034,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.183774121911109e-05,
+      "loss": 0.436,
+      "step": 7004
+    },
+    {
+      "epoch": 6.330623306233062,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.1705419710931377e-05,
+      "loss": 0.438,
+      "step": 7008
+    },
+    {
+      "epoch": 6.33423667570009,
+      "grad_norm": 0.3515625,
+      "learning_rate": 3.1573341245399645e-05,
+      "loss": 0.4001,
+      "step": 7012
+    },
+    {
+      "epoch": 6.3378500451671185,
+      "grad_norm": 0.38671875,
+      "learning_rate": 3.1441506093878346e-05,
+      "loss": 0.4078,
+      "step": 7016
+    },
+    {
+      "epoch": 6.341463414634147,
+      "grad_norm": 0.36328125,
+      "learning_rate": 3.1309914527229934e-05,
+      "loss": 0.4194,
+      "step": 7020
+    },
+    {
+      "epoch": 6.345076784101174,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.117856681581645e-05,
+      "loss": 0.4423,
+      "step": 7024
+    },
+    {
+      "epoch": 6.348690153568202,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.10474632294989e-05,
+      "loss": 0.4444,
+      "step": 7028
+    },
+    {
+      "epoch": 6.35230352303523,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.091660403763669e-05,
+      "loss": 0.4364,
+      "step": 7032
+    },
+    {
+      "epoch": 6.355916892502258,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.078598950908724e-05,
+      "loss": 0.4639,
+      "step": 7036
+    },
+    {
+      "epoch": 6.359530261969287,
+      "grad_norm": 0.3515625,
+      "learning_rate": 3.0655619912205136e-05,
+      "loss": 0.425,
+      "step": 7040
+    },
+    {
+      "epoch": 6.363143631436314,
+      "grad_norm": 0.39453125,
+      "learning_rate": 3.052549551484184e-05,
+      "loss": 0.4536,
+      "step": 7044
+    },
+    {
+      "epoch": 6.366757000903342,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.0395616584344962e-05,
+      "loss": 0.4423,
+      "step": 7048
+    },
+    {
+      "epoch": 6.37037037037037,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.026598338755783e-05,
+      "loss": 0.4083,
+      "step": 7052
+    },
+    {
+      "epoch": 6.373983739837398,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.013659619081893e-05,
+      "loss": 0.4333,
+      "step": 7056
+    },
+    {
+      "epoch": 6.377597109304427,
+      "grad_norm": 0.404296875,
+      "learning_rate": 3.00074552599613e-05,
+      "loss": 0.4283,
+      "step": 7060
+    },
+    {
+      "epoch": 6.381210478771455,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.9878560860311967e-05,
+      "loss": 0.4207,
+      "step": 7064
+    },
+    {
+      "epoch": 6.384823848238482,
+      "grad_norm": 0.392578125,
+      "learning_rate": 2.9749913256691448e-05,
+      "loss": 0.4735,
+      "step": 7068
+    },
+    {
+      "epoch": 6.38843721770551,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.9621512713413348e-05,
+      "loss": 0.4453,
+      "step": 7072
+    },
+    {
+      "epoch": 6.392050587172538,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.9493359494283507e-05,
+      "loss": 0.4303,
+      "step": 7076
+    },
+    {
+      "epoch": 6.3956639566395665,
+      "grad_norm": 0.384765625,
+      "learning_rate": 2.93654538625997e-05,
+      "loss": 0.4261,
+      "step": 7080
+    },
+    {
+      "epoch": 6.399277326106595,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.9237796081150973e-05,
+      "loss": 0.4499,
+      "step": 7084
+    },
+    {
+      "epoch": 6.402890695573622,
+      "grad_norm": 0.392578125,
+      "learning_rate": 2.9110386412217146e-05,
+      "loss": 0.4481,
+      "step": 7088
+    },
+    {
+      "epoch": 6.40650406504065,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.898322511756841e-05,
+      "loss": 0.4148,
+      "step": 7092
+    },
+    {
+      "epoch": 6.410117434507678,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.8856312458464524e-05,
+      "loss": 0.4011,
+      "step": 7096
+    },
+    {
+      "epoch": 6.4137308039747065,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.8729648695654415e-05,
+      "loss": 0.4418,
+      "step": 7100
+    },
+    {
+      "epoch": 6.417344173441735,
+      "grad_norm": 0.375,
+      "learning_rate": 2.86032340893757e-05,
+      "loss": 0.5009,
+      "step": 7104
+    },
+    {
+      "epoch": 6.420957542908763,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.847706889935401e-05,
+      "loss": 0.4655,
+      "step": 7108
+    },
+    {
+      "epoch": 6.42457091237579,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.835115338480269e-05,
+      "loss": 0.4362,
+      "step": 7112
+    },
+    {
+      "epoch": 6.428184281842818,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.8225487804421925e-05,
+      "loss": 0.4247,
+      "step": 7116
+    },
+    {
+      "epoch": 6.4317976513098465,
+      "grad_norm": 0.384765625,
+      "learning_rate": 2.810007241639853e-05,
+      "loss": 0.4774,
+      "step": 7120
+    },
+    {
+      "epoch": 6.435411020776875,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.797490747840518e-05,
+      "loss": 0.4086,
+      "step": 7124
+    },
+    {
+      "epoch": 6.439024390243903,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.784999324760012e-05,
+      "loss": 0.4817,
+      "step": 7128
+    },
+    {
+      "epoch": 6.44263775971093,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.7725329980626414e-05,
+      "loss": 0.3906,
+      "step": 7132
+    },
+    {
+      "epoch": 6.446251129177958,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.7600917933611493e-05,
+      "loss": 0.4507,
+      "step": 7136
+    },
+    {
+      "epoch": 6.4498644986449865,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.747675736216669e-05,
+      "loss": 0.4406,
+      "step": 7140
+    },
+    {
+      "epoch": 6.453477868112015,
+      "grad_norm": 0.408203125,
+      "learning_rate": 2.7352848521386616e-05,
+      "loss": 0.4542,
+      "step": 7144
+    },
+    {
+      "epoch": 6.457091237579043,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.7229191665848797e-05,
+      "loss": 0.4176,
+      "step": 7148
+    },
+    {
+      "epoch": 6.46070460704607,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.7105787049612927e-05,
+      "loss": 0.3961,
+      "step": 7152
+    },
+    {
+      "epoch": 6.464317976513098,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.6982634926220487e-05,
+      "loss": 0.4227,
+      "step": 7156
+    },
+    {
+      "epoch": 6.467931345980126,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.6859735548694234e-05,
+      "loss": 0.4482,
+      "step": 7160
+    },
+    {
+      "epoch": 6.471544715447155,
+      "grad_norm": 0.400390625,
+      "learning_rate": 2.6737089169537566e-05,
+      "loss": 0.4689,
+      "step": 7164
+    },
+    {
+      "epoch": 6.475158084914183,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.6614696040734225e-05,
+      "loss": 0.419,
+      "step": 7168
+    },
+    {
+      "epoch": 6.47877145438121,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.64925564137475e-05,
+      "loss": 0.4143,
+      "step": 7172
+    },
+    {
+      "epoch": 6.482384823848238,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.6370670539519915e-05,
+      "loss": 0.4473,
+      "step": 7176
+    },
+    {
+      "epoch": 6.485998193315266,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.624903866847262e-05,
+      "loss": 0.4058,
+      "step": 7180
+    },
+    {
+      "epoch": 6.489611562782295,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.6127661050504865e-05,
+      "loss": 0.4485,
+      "step": 7184
+    },
+    {
+      "epoch": 6.493224932249323,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.6006537934993636e-05,
+      "loss": 0.4406,
+      "step": 7188
+    },
+    {
+      "epoch": 6.496838301716351,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.588566957079294e-05,
+      "loss": 0.4612,
+      "step": 7192
+    },
+    {
+      "epoch": 6.500451671183378,
+      "grad_norm": 0.39453125,
+      "learning_rate": 2.576505620623339e-05,
+      "loss": 0.4491,
+      "step": 7196
+    },
+    {
+      "epoch": 6.504065040650406,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.564469808912169e-05,
+      "loss": 0.3919,
+      "step": 7200
+    },
+    {
+      "epoch": 6.5076784101174345,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.5524595466740178e-05,
+      "loss": 0.4153,
+      "step": 7204
+    },
+    {
+      "epoch": 6.511291779584463,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.5404748585846203e-05,
+      "loss": 0.4253,
+      "step": 7208
+    },
+    {
+      "epoch": 6.514905149051491,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.528515769267168e-05,
+      "loss": 0.4433,
+      "step": 7212
+    },
+    {
+      "epoch": 6.518518518518518,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.5165823032922638e-05,
+      "loss": 0.4563,
+      "step": 7216
+    },
+    {
+      "epoch": 6.522131887985546,
+      "grad_norm": 0.375,
+      "learning_rate": 2.504674485177855e-05,
+      "loss": 0.4117,
+      "step": 7220
+    },
+    {
+      "epoch": 6.5257452574525745,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.492792339389213e-05,
+      "loss": 0.4138,
+      "step": 7224
+    },
+    {
+      "epoch": 6.529358626919603,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.480935890338845e-05,
+      "loss": 0.4567,
+      "step": 7228
+    },
+    {
+      "epoch": 6.532971996386631,
+      "grad_norm": 0.375,
+      "learning_rate": 2.4691051623864717e-05,
+      "loss": 0.4305,
+      "step": 7232
+    },
+    {
+      "epoch": 6.536585365853659,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.4573001798389664e-05,
+      "loss": 0.469,
+      "step": 7236
+    },
+    {
+      "epoch": 6.540198735320686,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.4455209669503045e-05,
+      "loss": 0.4539,
+      "step": 7240
+    },
+    {
+      "epoch": 6.5438121047877145,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.433767547921529e-05,
+      "loss": 0.4723,
+      "step": 7244
+    },
+    {
+      "epoch": 6.547425474254743,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.4220399469006717e-05,
+      "loss": 0.4098,
+      "step": 7248
+    },
+    {
+      "epoch": 6.551038843721771,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.4103381879827273e-05,
+      "loss": 0.444,
+      "step": 7252
+    },
+    {
+      "epoch": 6.554652213188799,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.398662295209595e-05,
+      "loss": 0.4071,
+      "step": 7256
+    },
+    {
+      "epoch": 6.558265582655826,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.387012292570032e-05,
+      "loss": 0.421,
+      "step": 7260
+    },
+    {
+      "epoch": 6.561878952122854,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.3753882039996042e-05,
+      "loss": 0.4232,
+      "step": 7264
+    },
+    {
+      "epoch": 6.565492321589883,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.3637900533806333e-05,
+      "loss": 0.3635,
+      "step": 7268
+    },
+    {
+      "epoch": 6.569105691056911,
+      "grad_norm": 0.384765625,
+      "learning_rate": 2.352217864542149e-05,
+      "loss": 0.4244,
+      "step": 7272
+    },
+    {
+      "epoch": 6.572719060523939,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.3406716612598386e-05,
+      "loss": 0.4546,
+      "step": 7276
+    },
+    {
+      "epoch": 6.576332429990966,
+      "grad_norm": 0.392578125,
+      "learning_rate": 2.3291514672560142e-05,
+      "loss": 0.445,
+      "step": 7280
+    },
+    {
+      "epoch": 6.579945799457994,
+      "grad_norm": 0.400390625,
+      "learning_rate": 2.317657306199536e-05,
+      "loss": 0.4254,
+      "step": 7284
+    },
+    {
+      "epoch": 6.583559168925023,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.3061892017057815e-05,
+      "loss": 0.3915,
+      "step": 7288
+    },
+    {
+      "epoch": 6.587172538392051,
+      "grad_norm": 0.408203125,
+      "learning_rate": 2.2947471773365954e-05,
+      "loss": 0.4793,
+      "step": 7292
+    },
+    {
+      "epoch": 6.590785907859079,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.283331256600238e-05,
+      "loss": 0.4429,
+      "step": 7296
+    },
+    {
+      "epoch": 6.594399277326106,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.2719414629513443e-05,
+      "loss": 0.4268,
+      "step": 7300
+    },
+    {
+      "epoch": 6.598012646793134,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.2605778197908604e-05,
+      "loss": 0.423,
+      "step": 7304
+    },
+    {
+      "epoch": 6.6016260162601625,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.24924035046601e-05,
+      "loss": 0.4656,
+      "step": 7308
+    },
+    {
+      "epoch": 6.605239385727191,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.2379290782702425e-05,
+      "loss": 0.4556,
+      "step": 7312
+    },
+    {
+      "epoch": 6.608852755194219,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.2266440264431747e-05,
+      "loss": 0.4469,
+      "step": 7316
+    },
+    {
+      "epoch": 6.612466124661246,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.2153852181705677e-05,
+      "loss": 0.4537,
+      "step": 7320
+    },
+    {
+      "epoch": 6.616079494128274,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.2041526765842522e-05,
+      "loss": 0.4403,
+      "step": 7324
+    },
+    {
+      "epoch": 6.6196928635953025,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.1929464247620953e-05,
+      "loss": 0.4521,
+      "step": 7328
+    },
+    {
+      "epoch": 6.623306233062331,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.1817664857279506e-05,
+      "loss": 0.4094,
+      "step": 7332
+    },
+    {
+      "epoch": 6.626919602529359,
+      "grad_norm": 0.375,
+      "learning_rate": 2.1706128824516083e-05,
+      "loss": 0.4411,
+      "step": 7336
+    },
+    {
+      "epoch": 6.630532971996386,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.1594856378487584e-05,
+      "loss": 0.4586,
+      "step": 7340
+    },
+    {
+      "epoch": 6.634146341463414,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.1483847747809286e-05,
+      "loss": 0.4866,
+      "step": 7344
+    },
+    {
+      "epoch": 6.6377597109304425,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.1373103160554426e-05,
+      "loss": 0.4324,
+      "step": 7348
+    },
+    {
+      "epoch": 6.641373080397471,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.126262284425379e-05,
+      "loss": 0.4169,
+      "step": 7352
+    },
+    {
+      "epoch": 6.644986449864499,
+      "grad_norm": 0.35546875,
+      "learning_rate": 2.1152407025895217e-05,
+      "loss": 0.4381,
+      "step": 7356
+    },
+    {
+      "epoch": 6.648599819331527,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.104245593192309e-05,
+      "loss": 0.432,
+      "step": 7360
+    },
+    {
+      "epoch": 6.652213188798554,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.0932769788237906e-05,
+      "loss": 0.4295,
+      "step": 7364
+    },
+    {
+      "epoch": 6.6558265582655824,
+      "grad_norm": 0.39453125,
+      "learning_rate": 2.0823348820195836e-05,
+      "loss": 0.4128,
+      "step": 7368
+    },
+    {
+      "epoch": 6.659439927732611,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.0714193252608125e-05,
+      "loss": 0.4225,
+      "step": 7372
+    },
+    {
+      "epoch": 6.663053297199639,
+      "grad_norm": 0.35546875,
+      "learning_rate": 2.0605303309740934e-05,
+      "loss": 0.4306,
+      "step": 7376
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 0.38671875,
+      "learning_rate": 2.049667921531452e-05,
+      "loss": 0.4465,
+      "step": 7380
+    },
+    {
+      "epoch": 6.670280036133695,
+      "grad_norm": 0.408203125,
+      "learning_rate": 2.0388321192502982e-05,
+      "loss": 0.4366,
+      "step": 7384
+    },
+    {
+      "epoch": 6.673893405600722,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.02802294639338e-05,
+      "loss": 0.4409,
+      "step": 7388
+    },
+    {
+      "epoch": 6.677506775067751,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.0172404251687246e-05,
+      "loss": 0.4679,
+      "step": 7392
+    },
+    {
+      "epoch": 6.681120144534779,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.0064845777296163e-05,
+      "loss": 0.442,
+      "step": 7396
+    },
+    {
+      "epoch": 6.684733514001807,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.9957554261745247e-05,
+      "loss": 0.4684,
+      "step": 7400
+    },
+    {
+      "epoch": 6.688346883468835,
+      "grad_norm": 0.357421875,
+      "learning_rate": 1.9850529925470748e-05,
+      "loss": 0.4214,
+      "step": 7404
+    },
+    {
+      "epoch": 6.691960252935862,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.9743772988359997e-05,
+      "loss": 0.4291,
+      "step": 7408
+    },
+    {
+      "epoch": 6.695573622402891,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.9637283669750902e-05,
+      "loss": 0.4313,
+      "step": 7412
+    },
+    {
+      "epoch": 6.699186991869919,
+      "grad_norm": 0.404296875,
+      "learning_rate": 1.953106218843159e-05,
+      "loss": 0.4548,
+      "step": 7416
+    },
+    {
+      "epoch": 6.702800361336947,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.942510876263987e-05,
+      "loss": 0.3669,
+      "step": 7420
+    },
+    {
+      "epoch": 6.706413730803975,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.931942361006279e-05,
+      "loss": 0.426,
+      "step": 7424
+    },
+    {
+      "epoch": 6.710027100271002,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.921400694783623e-05,
+      "loss": 0.451,
+      "step": 7428
+    },
+    {
+      "epoch": 6.7136404697380305,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.9108858992544525e-05,
+      "loss": 0.4351,
+      "step": 7432
+    },
+    {
+      "epoch": 6.717253839205059,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.900397996021981e-05,
+      "loss": 0.4537,
+      "step": 7436
+    },
+    {
+      "epoch": 6.720867208672087,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.8899370066341794e-05,
+      "loss": 0.4404,
+      "step": 7440
+    },
+    {
+      "epoch": 6.724480578139115,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.879502952583715e-05,
+      "loss": 0.3894,
+      "step": 7444
+    },
+    {
+      "epoch": 6.728093947606142,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.86909585530792e-05,
+      "loss": 0.4797,
+      "step": 7448
+    },
+    {
+      "epoch": 6.7317073170731705,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.8587157361887467e-05,
+      "loss": 0.4323,
+      "step": 7452
+    },
+    {
+      "epoch": 6.735320686540199,
+      "grad_norm": 0.375,
+      "learning_rate": 1.848362616552711e-05,
+      "loss": 0.4711,
+      "step": 7456
+    },
+    {
+      "epoch": 6.738934056007227,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.8380365176708595e-05,
+      "loss": 0.4414,
+      "step": 7460
+    },
+    {
+      "epoch": 6.742547425474255,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.827737460758723e-05,
+      "loss": 0.4656,
+      "step": 7464
+    },
+    {
+      "epoch": 6.746160794941282,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.817465466976274e-05,
+      "loss": 0.3938,
+      "step": 7468
+    },
+    {
+      "epoch": 6.7497741644083105,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.8072205574278868e-05,
+      "loss": 0.4171,
+      "step": 7472
+    },
+    {
+      "epoch": 6.753387533875339,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.7970027531622817e-05,
+      "loss": 0.4572,
+      "step": 7476
+    },
+    {
+      "epoch": 6.757000903342367,
+      "grad_norm": 0.345703125,
+      "learning_rate": 1.786812075172495e-05,
+      "loss": 0.4105,
+      "step": 7480
+    },
+    {
+      "epoch": 6.760614272809395,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.776648544395826e-05,
+      "loss": 0.4421,
+      "step": 7484
+    },
+    {
+      "epoch": 6.764227642276423,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.7665121817137996e-05,
+      "loss": 0.4284,
+      "step": 7488
+    },
+    {
+      "epoch": 6.76784101174345,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.7564030079521312e-05,
+      "loss": 0.4323,
+      "step": 7492
+    },
+    {
+      "epoch": 6.771454381210479,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.746321043880664e-05,
+      "loss": 0.4626,
+      "step": 7496
+    },
+    {
+      "epoch": 6.775067750677507,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.7362663102133418e-05,
+      "loss": 0.4493,
+      "step": 7500
+    },
+    {
+      "epoch": 6.778681120144535,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.726238827608158e-05,
+      "loss": 0.4453,
+      "step": 7504
+    },
+    {
+      "epoch": 6.782294489611563,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.7162386166671278e-05,
+      "loss": 0.4169,
+      "step": 7508
+    },
+    {
+      "epoch": 6.78590785907859,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.7062656979362243e-05,
+      "loss": 0.413,
+      "step": 7512
+    },
+    {
+      "epoch": 6.789521228545619,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.6963200919053527e-05,
+      "loss": 0.413,
+      "step": 7516
+    },
+    {
+      "epoch": 6.793134598012647,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.6864018190083005e-05,
+      "loss": 0.4069,
+      "step": 7520
+    },
+    {
+      "epoch": 6.796747967479675,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.676510899622696e-05,
+      "loss": 0.4229,
+      "step": 7524
+    },
+    {
+      "epoch": 6.800361336946703,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.6666473540699772e-05,
+      "loss": 0.4883,
+      "step": 7528
+    },
+    {
+      "epoch": 6.803974706413731,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.6568112026153312e-05,
+      "loss": 0.4513,
+      "step": 7532
+    },
+    {
+      "epoch": 6.8075880758807585,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.6470024654676666e-05,
+      "loss": 0.44,
+      "step": 7536
+    },
+    {
+      "epoch": 6.811201445347787,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.637221162779568e-05,
+      "loss": 0.4285,
+      "step": 7540
+    },
+    {
+      "epoch": 6.814814814814815,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.627467314647248e-05,
+      "loss": 0.424,
+      "step": 7544
+    },
+    {
+      "epoch": 6.818428184281843,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.6177409411105262e-05,
+      "loss": 0.4215,
+      "step": 7548
+    },
+    {
+      "epoch": 6.822041553748871,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.6080420621527634e-05,
+      "loss": 0.3964,
+      "step": 7552
+    },
+    {
+      "epoch": 6.8256549232158985,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.5983706977008327e-05,
+      "loss": 0.435,
+      "step": 7556
+    },
+    {
+      "epoch": 6.829268292682927,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.5887268676250775e-05,
+      "loss": 0.4214,
+      "step": 7560
+    },
+    {
+      "epoch": 6.832881662149955,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.5791105917392704e-05,
+      "loss": 0.4442,
+      "step": 7564
+    },
+    {
+      "epoch": 6.836495031616983,
+      "grad_norm": 0.375,
+      "learning_rate": 1.5695218898005775e-05,
+      "loss": 0.42,
+      "step": 7568
+    },
+    {
+      "epoch": 6.840108401084011,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.5599607815095043e-05,
+      "loss": 0.4371,
+      "step": 7572
+    },
+    {
+      "epoch": 6.8437217705510385,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.5504272865098717e-05,
+      "loss": 0.4163,
+      "step": 7576
+    },
+    {
+      "epoch": 6.847335140018067,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.5409214243887558e-05,
+      "loss": 0.4216,
+      "step": 7580
+    },
+    {
+      "epoch": 6.850948509485095,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.531443214676476e-05,
+      "loss": 0.4511,
+      "step": 7584
+    },
+    {
+      "epoch": 6.854561878952123,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.5219926768465269e-05,
+      "loss": 0.4452,
+      "step": 7588
+    },
+    {
+      "epoch": 6.858175248419151,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.5125698303155514e-05,
+      "loss": 0.4575,
+      "step": 7592
+    },
+    {
+      "epoch": 6.861788617886178,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.5031746944433005e-05,
+      "loss": 0.4174,
+      "step": 7596
+    },
+    {
+      "epoch": 6.865401987353207,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.493807288532588e-05,
+      "loss": 0.4515,
+      "step": 7600
+    },
+    {
+      "epoch": 6.869015356820235,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.4844676318292659e-05,
+      "loss": 0.4567,
+      "step": 7604
+    },
+    {
+      "epoch": 6.872628726287263,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.4751557435221629e-05,
+      "loss": 0.4487,
+      "step": 7608
+    },
+    {
+      "epoch": 6.876242095754291,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.4658716427430584e-05,
+      "loss": 0.4181,
+      "step": 7612
+    },
+    {
+      "epoch": 6.879855465221318,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.456615348566646e-05,
+      "loss": 0.447,
+      "step": 7616
+    },
+    {
+      "epoch": 6.883468834688347,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.4473868800104799e-05,
+      "loss": 0.4173,
+      "step": 7620
+    },
+    {
+      "epoch": 6.887082204155375,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.4381862560349582e-05,
+      "loss": 0.4065,
+      "step": 7624
+    },
+    {
+      "epoch": 6.890695573622403,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.4290134955432587e-05,
+      "loss": 0.4809,
+      "step": 7628
+    },
+    {
+      "epoch": 6.894308943089431,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.4198686173813157e-05,
+      "loss": 0.448,
+      "step": 7632
+    },
+    {
+      "epoch": 6.897922312556459,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.4107516403377817e-05,
+      "loss": 0.4446,
+      "step": 7636
+    },
+    {
+      "epoch": 6.9015356820234866,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.4016625831439764e-05,
+      "loss": 0.4578,
+      "step": 7640
+    },
+    {
+      "epoch": 6.905149051490515,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.3926014644738693e-05,
+      "loss": 0.4206,
+      "step": 7644
+    },
+    {
+      "epoch": 6.908762420957543,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.383568302944018e-05,
+      "loss": 0.4116,
+      "step": 7648
+    },
+    {
+      "epoch": 6.912375790424571,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.374563117113544e-05,
+      "loss": 0.4477,
+      "step": 7652
+    },
+    {
+      "epoch": 6.915989159891599,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.365585925484089e-05,
+      "loss": 0.4452,
+      "step": 7656
+    },
+    {
+      "epoch": 6.919602529358627,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.3566367464997856e-05,
+      "loss": 0.4501,
+      "step": 7660
+    },
+    {
+      "epoch": 6.923215898825655,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.3477155985472065e-05,
+      "loss": 0.4455,
+      "step": 7664
+    },
+    {
+      "epoch": 6.926829268292683,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.3388224999553365e-05,
+      "loss": 0.4419,
+      "step": 7668
+    },
+    {
+      "epoch": 6.930442637759711,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.32995746899553e-05,
+      "loss": 0.4685,
+      "step": 7672
+    },
+    {
+      "epoch": 6.934056007226739,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.3211205238814709e-05,
+      "loss": 0.4227,
+      "step": 7676
+    },
+    {
+      "epoch": 6.937669376693767,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.312311682769151e-05,
+      "loss": 0.4738,
+      "step": 7680
+    },
+    {
+      "epoch": 6.941282746160795,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.3035309637568081e-05,
+      "loss": 0.4349,
+      "step": 7684
+    },
+    {
+      "epoch": 6.944896115627823,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.2947783848849098e-05,
+      "loss": 0.4168,
+      "step": 7688
+    },
+    {
+      "epoch": 6.948509485094851,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.2860539641361028e-05,
+      "loss": 0.4453,
+      "step": 7692
+    },
+    {
+      "epoch": 6.952122854561879,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.2773577194351824e-05,
+      "loss": 0.4207,
+      "step": 7696
+    },
+    {
+      "epoch": 6.955736224028907,
+      "grad_norm": 0.375,
+      "learning_rate": 1.2686896686490633e-05,
+      "loss": 0.3943,
+      "step": 7700
+    },
+    {
+      "epoch": 6.959349593495935,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.2600498295867211e-05,
+      "loss": 0.4065,
+      "step": 7704
+    },
+    {
+      "epoch": 6.962962962962963,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.2514382199991767e-05,
+      "loss": 0.4228,
+      "step": 7708
+    },
+    {
+      "epoch": 6.966576332429991,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.2428548575794506e-05,
+      "loss": 0.4366,
+      "step": 7712
+    },
+    {
+      "epoch": 6.970189701897019,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.2342997599625259e-05,
+      "loss": 0.4551,
+      "step": 7716
+    },
+    {
+      "epoch": 6.973803071364047,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.2257729447253178e-05,
+      "loss": 0.4836,
+      "step": 7720
+    },
+    {
+      "epoch": 6.977416440831075,
+      "grad_norm": 0.375,
+      "learning_rate": 1.2172744293866328e-05,
+      "loss": 0.4464,
+      "step": 7724
+    },
+    {
+      "epoch": 6.981029810298103,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.2088042314071316e-05,
+      "loss": 0.4342,
+      "step": 7728
+    },
+    {
+      "epoch": 6.984643179765131,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.2003623681892926e-05,
+      "loss": 0.4355,
+      "step": 7732
+    },
+    {
+      "epoch": 6.988256549232159,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.1919488570773899e-05,
+      "loss": 0.4295,
+      "step": 7736
+    },
+    {
+      "epoch": 6.991869918699187,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.1835637153574378e-05,
+      "loss": 0.457,
+      "step": 7740
+    },
+    {
+      "epoch": 6.995483288166215,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.1752069602571623e-05,
+      "loss": 0.4613,
+      "step": 7744
+    },
+    {
+      "epoch": 6.999096657633243,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.1668786089459741e-05,
+      "loss": 0.4617,
+      "step": 7748
+    },
+    {
+      "epoch": 7.002710027100271,
+      "grad_norm": 0.375,
+      "learning_rate": 1.158578678534921e-05,
+      "loss": 0.4494,
+      "step": 7752
+    },
+    {
+      "epoch": 7.006323396567299,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.1503071860766655e-05,
+      "loss": 0.4519,
+      "step": 7756
+    },
+    {
+      "epoch": 7.009936766034327,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.142064148565437e-05,
+      "loss": 0.4355,
+      "step": 7760
+    },
+    {
+      "epoch": 7.013550135501355,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.1338495829370031e-05,
+      "loss": 0.3998,
+      "step": 7764
+    },
+    {
+      "epoch": 7.017163504968383,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.1256635060686369e-05,
+      "loss": 0.4348,
+      "step": 7768
+    },
+    {
+      "epoch": 7.020776874435411,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.1175059347790766e-05,
+      "loss": 0.4116,
+      "step": 7772
+    },
+    {
+      "epoch": 7.024390243902439,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.1093768858284985e-05,
+      "loss": 0.399,
+      "step": 7776
+    },
+    {
+      "epoch": 7.028003613369467,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.1012763759184761e-05,
+      "loss": 0.4358,
+      "step": 7780
+    },
+    {
+      "epoch": 7.031616982836495,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.093204421691945e-05,
+      "loss": 0.4335,
+      "step": 7784
+    },
+    {
+      "epoch": 7.035230352303523,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.0851610397331756e-05,
+      "loss": 0.4048,
+      "step": 7788
+    },
+    {
+      "epoch": 7.038843721770551,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.0771462465677306e-05,
+      "loss": 0.4309,
+      "step": 7792
+    },
+    {
+      "epoch": 7.042457091237579,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.0691600586624427e-05,
+      "loss": 0.4309,
+      "step": 7796
+    },
+    {
+      "epoch": 7.046070460704607,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.0612024924253687e-05,
+      "loss": 0.4151,
+      "step": 7800
+    },
+    {
+      "epoch": 7.049683830171635,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.0532735642057588e-05,
+      "loss": 0.4171,
+      "step": 7804
+    },
+    {
+      "epoch": 7.053297199638663,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.045373290294027e-05,
+      "loss": 0.4623,
+      "step": 7808
+    },
+    {
+      "epoch": 7.056910569105691,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.0375016869217196e-05,
+      "loss": 0.4726,
+      "step": 7812
+    },
+    {
+      "epoch": 7.060523938572719,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.0296587702614695e-05,
+      "loss": 0.4241,
+      "step": 7816
+    },
+    {
+      "epoch": 7.064137308039747,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.021844556426979e-05,
+      "loss": 0.4337,
+      "step": 7820
+    },
+    {
+      "epoch": 7.067750677506775,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.0140590614729721e-05,
+      "loss": 0.4324,
+      "step": 7824
+    },
+    {
+      "epoch": 7.0713640469738035,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.0063023013951705e-05,
+      "loss": 0.4542,
+      "step": 7828
+    },
+    {
+      "epoch": 7.074977416440831,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.985742921302658e-06,
+      "loss": 0.4335,
+      "step": 7832
+    },
+    {
+      "epoch": 7.078590785907859,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.908750495558693e-06,
+      "loss": 0.4224,
+      "step": 7836
+    },
+    {
+      "epoch": 7.082204155374887,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.832045894904922e-06,
+      "loss": 0.4485,
+      "step": 7840
+    },
+    {
+      "epoch": 7.085817524841915,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.75562927693515e-06,
+      "loss": 0.4516,
+      "step": 7844
+    },
+    {
+      "epoch": 7.0894308943089435,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.679500798651424e-06,
+      "loss": 0.4164,
+      "step": 7848
+    },
+    {
+      "epoch": 7.093044263775971,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.60366061646391e-06,
+      "loss": 0.4383,
+      "step": 7852
+    },
+    {
+      "epoch": 7.096657633242999,
+      "grad_norm": 0.376953125,
+      "learning_rate": 9.52810888619035e-06,
+      "loss": 0.4478,
+      "step": 7856
+    },
+    {
+      "epoch": 7.100271002710027,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.452845763055894e-06,
+      "loss": 0.3991,
+      "step": 7860
+    },
+    {
+      "epoch": 7.103884372177055,
+      "grad_norm": 0.365234375,
+      "learning_rate": 9.3778714016927e-06,
+      "loss": 0.4196,
+      "step": 7864
+    },
+    {
+      "epoch": 7.107497741644083,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.303185956139664e-06,
+      "loss": 0.409,
+      "step": 7868
+    },
+    {
+      "epoch": 7.111111111111111,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.228789579842127e-06,
+      "loss": 0.4081,
+      "step": 7872
+    },
+    {
+      "epoch": 7.114724480578139,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.154682425651488e-06,
+      "loss": 0.4448,
+      "step": 7876
+    },
+    {
+      "epoch": 7.118337850045167,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.08086464582492e-06,
+      "loss": 0.4208,
+      "step": 7880
+    },
+    {
+      "epoch": 7.121951219512195,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.007336392025038e-06,
+      "loss": 0.4266,
+      "step": 7884
+    },
+    {
+      "epoch": 7.125564588979223,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.93409781531967e-06,
+      "loss": 0.4273,
+      "step": 7888
+    },
+    {
+      "epoch": 7.129177958446251,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.861149066181445e-06,
+      "loss": 0.4212,
+      "step": 7892
+    },
+    {
+      "epoch": 7.132791327913279,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.788490294487543e-06,
+      "loss": 0.3971,
+      "step": 7896
+    },
+    {
+      "epoch": 7.136404697380307,
+      "grad_norm": 0.37890625,
+      "learning_rate": 8.716121649519354e-06,
+      "loss": 0.4149,
+      "step": 7900
+    },
+    {
+      "epoch": 7.140018066847335,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.64404327996217e-06,
+      "loss": 0.4063,
+      "step": 7904
+    },
+    {
+      "epoch": 7.143631436314363,
+      "grad_norm": 0.373046875,
+      "learning_rate": 8.57225533390497e-06,
+      "loss": 0.4607,
+      "step": 7908
+    },
+    {
+      "epoch": 7.1472448057813915,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.500757958839954e-06,
+      "loss": 0.4243,
+      "step": 7912
+    },
+    {
+      "epoch": 7.150858175248419,
+      "grad_norm": 0.373046875,
+      "learning_rate": 8.429551301662386e-06,
+      "loss": 0.4444,
+      "step": 7916
+    },
+    {
+      "epoch": 7.154471544715447,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.35863550867022e-06,
+      "loss": 0.4668,
+      "step": 7920
+    },
+    {
+      "epoch": 7.158084914182475,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.288010725563759e-06,
+      "loss": 0.4559,
+      "step": 7924
+    },
+    {
+      "epoch": 7.161698283649503,
+      "grad_norm": 0.365234375,
+      "learning_rate": 8.217677097445535e-06,
+      "loss": 0.426,
+      "step": 7928
+    },
+    {
+      "epoch": 7.1653116531165315,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.147634768819788e-06,
+      "loss": 0.449,
+      "step": 7932
+    },
+    {
+      "epoch": 7.168925022583559,
+      "grad_norm": 0.37890625,
+      "learning_rate": 8.077883883592256e-06,
+      "loss": 0.4111,
+      "step": 7936
+    },
+    {
+      "epoch": 7.172538392050587,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.008424585069972e-06,
+      "loss": 0.4177,
+      "step": 7940
+    },
+    {
+      "epoch": 7.176151761517615,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.939257015960782e-06,
+      "loss": 0.4485,
+      "step": 7944
+    },
+    {
+      "epoch": 7.179765130984643,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.870381318373293e-06,
+      "loss": 0.4079,
+      "step": 7948
+    },
+    {
+      "epoch": 7.1833785004516715,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.801797633816325e-06,
+      "loss": 0.423,
+      "step": 7952
+    },
+    {
+      "epoch": 7.186991869918699,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.733506103198784e-06,
+      "loss": 0.4961,
+      "step": 7956
+    },
+    {
+      "epoch": 7.190605239385727,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.665506866829347e-06,
+      "loss": 0.4292,
+      "step": 7960
+    },
+    {
+      "epoch": 7.194218608852755,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.597800064416143e-06,
+      "loss": 0.4689,
+      "step": 7964
+    },
+    {
+      "epoch": 7.197831978319783,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.530385835066499e-06,
+      "loss": 0.4048,
+      "step": 7968
+    },
+    {
+      "epoch": 7.2014453477868114,
+      "grad_norm": 0.388671875,
+      "learning_rate": 7.463264317286571e-06,
+      "loss": 0.4369,
+      "step": 7972
+    },
+    {
+      "epoch": 7.20505871725384,
+      "grad_norm": 0.3828125,
+      "learning_rate": 7.396435648981214e-06,
+      "loss": 0.4248,
+      "step": 7976
+    },
+    {
+      "epoch": 7.208672086720867,
+      "grad_norm": 0.38671875,
+      "learning_rate": 7.329899967453495e-06,
+      "loss": 0.4861,
+      "step": 7980
+    },
+    {
+      "epoch": 7.212285456187895,
+      "grad_norm": 0.3828125,
+      "learning_rate": 7.263657409404666e-06,
+      "loss": 0.4476,
+      "step": 7984
+    },
+    {
+      "epoch": 7.215898825654923,
+      "grad_norm": 0.353515625,
+      "learning_rate": 7.197708110933642e-06,
+      "loss": 0.4006,
+      "step": 7988
+    },
+    {
+      "epoch": 7.219512195121951,
+      "grad_norm": 0.375,
+      "learning_rate": 7.1320522075368205e-06,
+      "loss": 0.4444,
+      "step": 7992
+    },
+    {
+      "epoch": 7.22312556458898,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.0666898341078305e-06,
+      "loss": 0.4436,
+      "step": 7996
+    },
+    {
+      "epoch": 7.226738934056007,
+      "grad_norm": 0.357421875,
+      "learning_rate": 7.001621124937218e-06,
+      "loss": 0.4231,
+      "step": 8000
+    },
+    {
+      "epoch": 7.230352303523035,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.936846213712211e-06,
+      "loss": 0.437,
+      "step": 8004
+    },
+    {
+      "epoch": 7.233965672990063,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.872365233516369e-06,
+      "loss": 0.4139,
+      "step": 8008
+    },
+    {
+      "epoch": 7.237579042457091,
+      "grad_norm": 0.380859375,
+      "learning_rate": 6.8081783168293876e-06,
+      "loss": 0.4395,
+      "step": 8012
+    },
+    {
+      "epoch": 7.2411924119241196,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.744285595526761e-06,
+      "loss": 0.4561,
+      "step": 8016
+    },
+    {
+      "epoch": 7.244805781391147,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.680687200879531e-06,
+      "loss": 0.4554,
+      "step": 8020
+    },
+    {
+      "epoch": 7.248419150858175,
+      "grad_norm": 0.3515625,
+      "learning_rate": 6.617383263554127e-06,
+      "loss": 0.4205,
+      "step": 8024
+    },
+    {
+      "epoch": 7.252032520325203,
+      "grad_norm": 0.42578125,
+      "learning_rate": 6.554373913611893e-06,
+      "loss": 0.4264,
+      "step": 8028
+    },
+    {
+      "epoch": 7.255645889792231,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.49165928050896e-06,
+      "loss": 0.429,
+      "step": 8032
+    },
+    {
+      "epoch": 7.2592592592592595,
+      "grad_norm": 0.384765625,
+      "learning_rate": 6.429239493095939e-06,
+      "loss": 0.4016,
+      "step": 8036
+    },
+    {
+      "epoch": 7.262872628726287,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.367114679617713e-06,
+      "loss": 0.4297,
+      "step": 8040
+    },
+    {
+      "epoch": 7.266485998193315,
+      "grad_norm": 0.38671875,
+      "learning_rate": 6.305284967713081e-06,
+      "loss": 0.4781,
+      "step": 8044
+    },
+    {
+      "epoch": 7.270099367660343,
+      "grad_norm": 0.396484375,
+      "learning_rate": 6.243750484414528e-06,
+      "loss": 0.4424,
+      "step": 8048
+    },
+    {
+      "epoch": 7.273712737127371,
+      "grad_norm": 0.369140625,
+      "learning_rate": 6.182511356148023e-06,
+      "loss": 0.4723,
+      "step": 8052
+    },
+    {
+      "epoch": 7.2773261065943995,
+      "grad_norm": 0.37109375,
+      "learning_rate": 6.121567708732639e-06,
+      "loss": 0.4389,
+      "step": 8056
+    },
+    {
+      "epoch": 7.280939476061428,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.0609196673804814e-06,
+      "loss": 0.4257,
+      "step": 8060
+    },
+    {
+      "epoch": 7.284552845528455,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.0005673566961784e-06,
+      "loss": 0.4441,
+      "step": 8064
+    },
+    {
+      "epoch": 7.288166214995483,
+      "grad_norm": 0.375,
+      "learning_rate": 5.940510900676926e-06,
+      "loss": 0.4201,
+      "step": 8068
+    },
+    {
+      "epoch": 7.291779584462511,
+      "grad_norm": 0.38671875,
+      "learning_rate": 5.8807504227118884e-06,
+      "loss": 0.4438,
+      "step": 8072
+    },
+    {
+      "epoch": 7.2953929539295395,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.821286045582235e-06,
+      "loss": 0.4535,
+      "step": 8076
+    },
+    {
+      "epoch": 7.299006323396568,
+      "grad_norm": 0.3828125,
+      "learning_rate": 5.762117891460788e-06,
+      "loss": 0.4425,
+      "step": 8080
+    },
+    {
+      "epoch": 7.302619692863595,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.703246081911739e-06,
+      "loss": 0.4208,
+      "step": 8084
+    },
+    {
+      "epoch": 7.306233062330623,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.644670737890417e-06,
+      "loss": 0.4197,
+      "step": 8088
+    },
+    {
+      "epoch": 7.309846431797651,
+      "grad_norm": 0.384765625,
+      "learning_rate": 5.586391979743038e-06,
+      "loss": 0.4257,
+      "step": 8092
+    },
+    {
+      "epoch": 7.313459801264679,
+      "grad_norm": 0.365234375,
+      "learning_rate": 5.528409927206473e-06,
+      "loss": 0.4205,
+      "step": 8096
+    },
+    {
+      "epoch": 7.317073170731708,
+      "grad_norm": 0.365234375,
+      "learning_rate": 5.47072469940803e-06,
+      "loss": 0.4079,
+      "step": 8100
+    },
+    {
+      "epoch": 7.320686540198736,
+      "grad_norm": 0.3828125,
+      "learning_rate": 5.413336414865138e-06,
+      "loss": 0.4147,
+      "step": 8104
+    },
+    {
+      "epoch": 7.324299909665763,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.356245191485148e-06,
+      "loss": 0.4364,
+      "step": 8108
+    },
+    {
+      "epoch": 7.327913279132791,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.299451146565048e-06,
+      "loss": 0.3964,
+      "step": 8112
+    },
+    {
+      "epoch": 7.331526648599819,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.242954396791349e-06,
+      "loss": 0.4233,
+      "step": 8116
+    },
+    {
+      "epoch": 7.335140018066848,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.1867550582396666e-06,
+      "loss": 0.4263,
+      "step": 8120
+    },
+    {
+      "epoch": 7.338753387533876,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.130853246374589e-06,
+      "loss": 0.456,
+      "step": 8124
+    },
+    {
+      "epoch": 7.342366757000903,
+      "grad_norm": 0.376953125,
+      "learning_rate": 5.075249076049459e-06,
+      "loss": 0.4169,
+      "step": 8128
+    },
+    {
+      "epoch": 7.345980126467931,
+      "grad_norm": 0.3828125,
+      "learning_rate": 5.0199426615060264e-06,
+      "loss": 0.4212,
+      "step": 8132
+    },
+    {
+      "epoch": 7.349593495934959,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.964934116374364e-06,
+      "loss": 0.4339,
+      "step": 8136
+    },
+    {
+      "epoch": 7.3532068654019875,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.910223553672482e-06,
+      "loss": 0.4498,
+      "step": 8140
+    },
+    {
+      "epoch": 7.356820234869016,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.855811085806266e-06,
+      "loss": 0.4583,
+      "step": 8144
+    },
+    {
+      "epoch": 7.360433604336043,
+      "grad_norm": 0.3984375,
+      "learning_rate": 4.80169682456904e-06,
+      "loss": 0.4343,
+      "step": 8148
+    },
+    {
+      "epoch": 7.364046973803071,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.747880881141502e-06,
+      "loss": 0.4173,
+      "step": 8152
+    },
+    {
+      "epoch": 7.367660343270099,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.694363366091441e-06,
+      "loss": 0.4739,
+      "step": 8156
+    },
+    {
+      "epoch": 7.3712737127371275,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.641144389373519e-06,
+      "loss": 0.4291,
+      "step": 8160
+    },
+    {
+      "epoch": 7.374887082204156,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.5882240603290045e-06,
+      "loss": 0.4318,
+      "step": 8164
+    },
+    {
+      "epoch": 7.378500451671183,
+      "grad_norm": 0.3984375,
+      "learning_rate": 4.535602487685558e-06,
+      "loss": 0.4472,
+      "step": 8168
+    },
+    {
+      "epoch": 7.382113821138211,
+      "grad_norm": 0.400390625,
+      "learning_rate": 4.483279779557114e-06,
+      "loss": 0.4248,
+      "step": 8172
+    },
+    {
+      "epoch": 7.385727190605239,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.431256043443515e-06,
+      "loss": 0.4148,
+      "step": 8176
+    },
+    {
+      "epoch": 7.3893405600722675,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.379531386230328e-06,
+      "loss": 0.3603,
+      "step": 8180
+    },
+    {
+      "epoch": 7.392953929539296,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.328105914188695e-06,
+      "loss": 0.4299,
+      "step": 8184
+    },
+    {
+      "epoch": 7.396567299006323,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.276979732975017e-06,
+      "loss": 0.4313,
+      "step": 8188
+    },
+    {
+      "epoch": 7.400180668473351,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.2261529476308345e-06,
+      "loss": 0.417,
+      "step": 8192
+    },
+    {
+      "epoch": 7.403794037940379,
+      "grad_norm": 0.39453125,
+      "learning_rate": 4.175625662582566e-06,
+      "loss": 0.4687,
+      "step": 8196
+    },
+    {
+      "epoch": 7.407407407407407,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.1253979816412206e-06,
+      "loss": 0.4499,
+      "step": 8200
+    },
+    {
+      "epoch": 7.411020776874436,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.075470008002335e-06,
+      "loss": 0.4295,
+      "step": 8204
+    },
+    {
+      "epoch": 7.414634146341464,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.0258418442456195e-06,
+      "loss": 0.4032,
+      "step": 8208
+    },
+    {
+      "epoch": 7.418247515808491,
+      "grad_norm": 0.375,
+      "learning_rate": 3.976513592334863e-06,
+      "loss": 0.3908,
+      "step": 8212
+    },
+    {
+      "epoch": 7.421860885275519,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.9274853536175935e-06,
+      "loss": 0.4193,
+      "step": 8216
+    },
+    {
+      "epoch": 7.425474254742547,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.878757228825086e-06,
+      "loss": 0.4653,
+      "step": 8220
+    },
+    {
+      "epoch": 7.429087624209576,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.830329318071823e-06,
+      "loss": 0.416,
+      "step": 8224
+    },
+    {
+      "epoch": 7.432700993676604,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.7822017208556154e-06,
+      "loss": 0.4302,
+      "step": 8228
+    },
+    {
+      "epoch": 7.436314363143631,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.73437453605725e-06,
+      "loss": 0.4451,
+      "step": 8232
+    },
+    {
+      "epoch": 7.439927732610659,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.6868478619402564e-06,
+      "loss": 0.4333,
+      "step": 8236
+    },
+    {
+      "epoch": 7.443541102077687,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.6396217961507936e-06,
+      "loss": 0.4261,
+      "step": 8240
+    },
+    {
+      "epoch": 7.4471544715447155,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.592696435717346e-06,
+      "loss": 0.4147,
+      "step": 8244
+    },
+    {
+      "epoch": 7.450767841011744,
+      "grad_norm": 0.3828125,
+      "learning_rate": 3.5460718770506434e-06,
+      "loss": 0.4278,
+      "step": 8248
+    },
+    {
+      "epoch": 7.454381210478772,
+      "grad_norm": 0.41796875,
+      "learning_rate": 3.4997482159433764e-06,
+      "loss": 0.4272,
+      "step": 8252
+    },
+    {
+      "epoch": 7.457994579945799,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.45372554757003e-06,
+      "loss": 0.4055,
+      "step": 8256
+    },
+    {
+      "epoch": 7.461607949412827,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.4080039664866497e-06,
+      "loss": 0.4364,
+      "step": 8260
+    },
+    {
+      "epoch": 7.4652213188798555,
+      "grad_norm": 0.404296875,
+      "learning_rate": 3.3625835666307445e-06,
+      "loss": 0.4313,
+      "step": 8264
+    },
+    {
+      "epoch": 7.468834688346884,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.317464441320983e-06,
+      "loss": 0.4039,
+      "step": 8268
+    },
+    {
+      "epoch": 7.472448057813912,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.2726466832570464e-06,
+      "loss": 0.4671,
+      "step": 8272
+    },
+    {
+      "epoch": 7.476061427280939,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.228130384519495e-06,
+      "loss": 0.4394,
+      "step": 8276
+    },
+    {
+      "epoch": 7.479674796747967,
+      "grad_norm": 0.404296875,
+      "learning_rate": 3.1839156365694506e-06,
+      "loss": 0.4598,
+      "step": 8280
+    },
+    {
+      "epoch": 7.4832881662149955,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.1400025302485145e-06,
+      "loss": 0.4358,
+      "step": 8284
+    },
+    {
+      "epoch": 7.486901535682024,
+      "grad_norm": 0.34375,
+      "learning_rate": 3.0963911557785824e-06,
+      "loss": 0.4136,
+      "step": 8288
+    },
+    {
+      "epoch": 7.490514905149052,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.053081602761548e-06,
+      "loss": 0.3874,
+      "step": 8292
+    },
+    {
+      "epoch": 7.494128274616079,
+      "grad_norm": 0.357421875,
+      "learning_rate": 3.0100739601793154e-06,
+      "loss": 0.4364,
+      "step": 8296
+    },
+    {
+      "epoch": 7.497741644083107,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.96736831639337e-06,
+      "loss": 0.4315,
+      "step": 8300
+    },
+    {
+      "epoch": 7.5013550135501355,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.92496475914476e-06,
+      "loss": 0.4645,
+      "step": 8304
+    },
+    {
+      "epoch": 7.504968383017164,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.8828633755539623e-06,
+      "loss": 0.4215,
+      "step": 8308
+    },
+    {
+      "epoch": 7.508581752484192,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.8410642521205007e-06,
+      "loss": 0.4524,
+      "step": 8312
+    },
+    {
+      "epoch": 7.512195121951219,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.7995674747230136e-06,
+      "loss": 0.4665,
+      "step": 8316
+    },
+    {
+      "epoch": 7.515808491418247,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.758373128618818e-06,
+      "loss": 0.4556,
+      "step": 8320
+    },
+    {
+      "epoch": 7.519421860885275,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.717481298443963e-06,
+      "loss": 0.4349,
+      "step": 8324
+    },
+    {
+      "epoch": 7.523035230352304,
+      "grad_norm": 0.375,
+      "learning_rate": 2.6768920682129604e-06,
+      "loss": 0.4189,
+      "step": 8328
+    },
+    {
+      "epoch": 7.526648599819332,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.636605521318569e-06,
+      "loss": 0.4243,
+      "step": 8332
+    },
+    {
+      "epoch": 7.530261969286359,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.5966217405316956e-06,
+      "loss": 0.4516,
+      "step": 8336
+    },
+    {
+      "epoch": 7.533875338753387,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.5569408080011624e-06,
+      "loss": 0.4545,
+      "step": 8340
+    },
+    {
+      "epoch": 7.537488708220415,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.5175628052536532e-06,
+      "loss": 0.4271,
+      "step": 8344
+    },
+    {
+      "epoch": 7.541102077687444,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.478487813193386e-06,
+      "loss": 0.4578,
+      "step": 8348
+    },
+    {
+      "epoch": 7.544715447154472,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.439715912102075e-06,
+      "loss": 0.4533,
+      "step": 8352
+    },
+    {
+      "epoch": 7.5483288166215,
+      "grad_norm": 0.375,
+      "learning_rate": 2.4012471816386827e-06,
+      "loss": 0.4591,
+      "step": 8356
+    },
+    {
+      "epoch": 7.551942186088527,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.3630817008393033e-06,
+      "loss": 0.445,
+      "step": 8360
+    },
+    {
+      "epoch": 7.555555555555555,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.3252195481170133e-06,
+      "loss": 0.4664,
+      "step": 8364
+    },
+    {
+      "epoch": 7.5591689250225835,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.2876608012616537e-06,
+      "loss": 0.393,
+      "step": 8368
+    },
+    {
+      "epoch": 7.562782294489612,
+      "grad_norm": 0.404296875,
+      "learning_rate": 2.2504055374397144e-06,
+      "loss": 0.4321,
+      "step": 8372
+    },
+    {
+      "epoch": 7.56639566395664,
+      "grad_norm": 0.35546875,
+      "learning_rate": 2.2134538331941677e-06,
+      "loss": 0.4305,
+      "step": 8376
+    },
+    {
+      "epoch": 7.570009033423668,
+      "grad_norm": 0.375,
+      "learning_rate": 2.1768057644442673e-06,
+      "loss": 0.4251,
+      "step": 8380
+    },
+    {
+      "epoch": 7.573622402890695,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.1404614064855e-06,
+      "loss": 0.3862,
+      "step": 8384
+    },
+    {
+      "epoch": 7.5772357723577235,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.104420833989301e-06,
+      "loss": 0.4189,
+      "step": 8388
+    },
+    {
+      "epoch": 7.580849141824752,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.0686841210030223e-06,
+      "loss": 0.4428,
+      "step": 8392
+    },
+    {
+      "epoch": 7.58446251129178,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.033251340949632e-06,
+      "loss": 0.3944,
+      "step": 8396
+    },
+    {
+      "epoch": 7.588075880758808,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.9981225666277466e-06,
+      "loss": 0.4303,
+      "step": 8400
+    },
+    {
+      "epoch": 7.591689250225835,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.9632978702113335e-06,
+      "loss": 0.4411,
+      "step": 8404
+    },
+    {
+      "epoch": 7.5953026196928635,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.9287773232496084e-06,
+      "loss": 0.4177,
+      "step": 8408
+    },
+    {
+      "epoch": 7.598915989159892,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.8945609966669383e-06,
+      "loss": 0.4171,
+      "step": 8412
+    },
+    {
+      "epoch": 7.60252935862692,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.8606489607625897e-06,
+      "loss": 0.4203,
+      "step": 8416
+    },
+    {
+      "epoch": 7.606142728093948,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.827041285210712e-06,
+      "loss": 0.4233,
+      "step": 8420
+    },
+    {
+      "epoch": 7.609756097560975,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.7937380390601054e-06,
+      "loss": 0.4779,
+      "step": 8424
+    },
+    {
+      "epoch": 7.613369467028003,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.7607392907340868e-06,
+      "loss": 0.4273,
+      "step": 8428
+    },
+    {
+      "epoch": 7.616982836495032,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.7280451080303902e-06,
+      "loss": 0.4323,
+      "step": 8432
+    },
+    {
+      "epoch": 7.62059620596206,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.6956555581209674e-06,
+      "loss": 0.4783,
+      "step": 8436
+    },
+    {
+      "epoch": 7.624209575429088,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.6635707075519533e-06,
+      "loss": 0.4164,
+      "step": 8440
+    },
+    {
+      "epoch": 7.627822944896115,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.6317906222433675e-06,
+      "loss": 0.4485,
+      "step": 8444
+    },
+    {
+      "epoch": 7.631436314363143,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.60031536748918e-06,
+      "loss": 0.4338,
+      "step": 8448
+    },
+    {
+      "epoch": 7.635049683830172,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.5691450079569622e-06,
+      "loss": 0.4312,
+      "step": 8452
+    },
+    {
+      "epoch": 7.6386630532972,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.5382796076879201e-06,
+      "loss": 0.4701,
+      "step": 8456
+    },
+    {
+      "epoch": 7.642276422764228,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.5077192300967268e-06,
+      "loss": 0.4259,
+      "step": 8460
+    },
+    {
+      "epoch": 7.645889792231255,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.477463937971307e-06,
+      "loss": 0.4558,
+      "step": 8464
+    },
+    {
+      "epoch": 7.649503161698283,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.4475137934728543e-06,
+      "loss": 0.3784,
+      "step": 8468
+    },
+    {
+      "epoch": 7.6531165311653115,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.4178688581354957e-06,
+      "loss": 0.4166,
+      "step": 8472
+    },
+    {
+      "epoch": 7.65672990063234,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.3885291928664278e-06,
+      "loss": 0.4005,
+      "step": 8476
+    },
+    {
+      "epoch": 7.660343270099368,
+      "grad_norm": 0.34765625,
+      "learning_rate": 1.359494857945548e-06,
+      "loss": 0.4405,
+      "step": 8480
+    },
+    {
+      "epoch": 7.663956639566395,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.3307659130255056e-06,
+      "loss": 0.4088,
+      "step": 8484
+    },
+    {
+      "epoch": 7.667570009033423,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.3023424171314523e-06,
+      "loss": 0.45,
+      "step": 8488
+    },
+    {
+      "epoch": 7.6711833785004515,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.2742244286610081e-06,
+      "loss": 0.4228,
+      "step": 8492
+    },
+    {
+      "epoch": 7.67479674796748,
+      "grad_norm": 0.349609375,
+      "learning_rate": 1.2464120053841454e-06,
+      "loss": 0.4276,
+      "step": 8496
+    },
+    {
+      "epoch": 7.678410117434508,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.2189052044429547e-06,
+      "loss": 0.3931,
+      "step": 8500
+    }
+  ],
+  "logging_steps": 4,
+  "max_steps": 8856,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 8,
+  "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": 5.560613094763266e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}