servo-groot-yam-v3 / trainer_state.json
npow's picture
Publish Servo finetuned checkpoint weights
020976d verified
Raw
History Blame Contribute Delete
421 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 30000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"grad_norm": 0.4512493908405304,
"learning_rate": 6.000000000000001e-07,
"loss": 1.2821,
"step": 10
},
{
"grad_norm": 0.5115783214569092,
"learning_rate": 1.2666666666666667e-06,
"loss": 1.2975,
"step": 20
},
{
"grad_norm": 0.3366401791572571,
"learning_rate": 1.9333333333333336e-06,
"loss": 1.2857,
"step": 30
},
{
"grad_norm": 0.5007226467132568,
"learning_rate": 2.6e-06,
"loss": 1.2773,
"step": 40
},
{
"grad_norm": 0.8038211464881897,
"learning_rate": 3.2666666666666666e-06,
"loss": 1.272,
"step": 50
},
{
"grad_norm": 0.821021556854248,
"learning_rate": 3.9333333333333335e-06,
"loss": 1.2149,
"step": 60
},
{
"grad_norm": 0.6811080574989319,
"learning_rate": 4.6e-06,
"loss": 1.1639,
"step": 70
},
{
"grad_norm": 0.9182567000389099,
"learning_rate": 5.266666666666667e-06,
"loss": 1.1259,
"step": 80
},
{
"grad_norm": 0.8028368353843689,
"learning_rate": 5.933333333333334e-06,
"loss": 1.1003,
"step": 90
},
{
"grad_norm": 0.7042647004127502,
"learning_rate": 6.6e-06,
"loss": 1.0847,
"step": 100
},
{
"grad_norm": 0.8734721541404724,
"learning_rate": 7.266666666666668e-06,
"loss": 1.0713,
"step": 110
},
{
"grad_norm": 1.1694245338439941,
"learning_rate": 7.933333333333334e-06,
"loss": 1.0582,
"step": 120
},
{
"grad_norm": 0.9218332767486572,
"learning_rate": 8.599999999999999e-06,
"loss": 1.0578,
"step": 130
},
{
"grad_norm": 0.8858274221420288,
"learning_rate": 9.266666666666667e-06,
"loss": 1.063,
"step": 140
},
{
"grad_norm": 0.8232358694076538,
"learning_rate": 9.933333333333334e-06,
"loss": 1.0582,
"step": 150
},
{
"grad_norm": 0.914615273475647,
"learning_rate": 1.06e-05,
"loss": 1.0543,
"step": 160
},
{
"grad_norm": 0.9622145891189575,
"learning_rate": 1.1266666666666667e-05,
"loss": 1.056,
"step": 170
},
{
"grad_norm": 1.021627426147461,
"learning_rate": 1.1933333333333333e-05,
"loss": 1.0338,
"step": 180
},
{
"grad_norm": 1.2972939014434814,
"learning_rate": 1.2600000000000001e-05,
"loss": 1.0088,
"step": 190
},
{
"grad_norm": 1.4146981239318848,
"learning_rate": 1.3266666666666666e-05,
"loss": 0.9705,
"step": 200
},
{
"grad_norm": 1.5866602659225464,
"learning_rate": 1.3933333333333334e-05,
"loss": 0.9017,
"step": 210
},
{
"grad_norm": 1.7686281204223633,
"learning_rate": 1.4599999999999999e-05,
"loss": 0.8424,
"step": 220
},
{
"grad_norm": 1.9282951354980469,
"learning_rate": 1.5266666666666667e-05,
"loss": 0.7887,
"step": 230
},
{
"grad_norm": 1.9342823028564453,
"learning_rate": 1.5933333333333332e-05,
"loss": 0.7247,
"step": 240
},
{
"grad_norm": 2.591855049133301,
"learning_rate": 1.66e-05,
"loss": 0.6411,
"step": 250
},
{
"grad_norm": 1.9724550247192383,
"learning_rate": 1.726666666666667e-05,
"loss": 0.5867,
"step": 260
},
{
"grad_norm": 2.4636716842651367,
"learning_rate": 1.7933333333333337e-05,
"loss": 0.5366,
"step": 270
},
{
"grad_norm": 2.4244909286499023,
"learning_rate": 1.86e-05,
"loss": 0.4723,
"step": 280
},
{
"grad_norm": 2.0608270168304443,
"learning_rate": 1.926666666666667e-05,
"loss": 0.4499,
"step": 290
},
{
"grad_norm": 2.2768301963806152,
"learning_rate": 1.9933333333333334e-05,
"loss": 0.3946,
"step": 300
},
{
"grad_norm": 1.8742576837539673,
"learning_rate": 2.06e-05,
"loss": 0.3877,
"step": 310
},
{
"grad_norm": 2.06176495552063,
"learning_rate": 2.1266666666666667e-05,
"loss": 0.3564,
"step": 320
},
{
"grad_norm": 2.683349132537842,
"learning_rate": 2.1933333333333332e-05,
"loss": 0.3257,
"step": 330
},
{
"grad_norm": 2.7738142013549805,
"learning_rate": 2.26e-05,
"loss": 0.2898,
"step": 340
},
{
"grad_norm": 2.178647041320801,
"learning_rate": 2.326666666666667e-05,
"loss": 0.2693,
"step": 350
},
{
"grad_norm": 1.875063419342041,
"learning_rate": 2.3933333333333337e-05,
"loss": 0.2636,
"step": 360
},
{
"grad_norm": 1.8726601600646973,
"learning_rate": 2.46e-05,
"loss": 0.2559,
"step": 370
},
{
"grad_norm": 2.1812222003936768,
"learning_rate": 2.5266666666666666e-05,
"loss": 0.244,
"step": 380
},
{
"grad_norm": 1.806279182434082,
"learning_rate": 2.5933333333333338e-05,
"loss": 0.2344,
"step": 390
},
{
"grad_norm": 2.1745898723602295,
"learning_rate": 2.6600000000000003e-05,
"loss": 0.2004,
"step": 400
},
{
"grad_norm": 1.4280179738998413,
"learning_rate": 2.7266666666666668e-05,
"loss": 0.1931,
"step": 410
},
{
"grad_norm": 1.8820170164108276,
"learning_rate": 2.7933333333333332e-05,
"loss": 0.179,
"step": 420
},
{
"grad_norm": 1.6634247303009033,
"learning_rate": 2.86e-05,
"loss": 0.1823,
"step": 430
},
{
"grad_norm": 2.1898722648620605,
"learning_rate": 2.926666666666667e-05,
"loss": 0.1811,
"step": 440
},
{
"grad_norm": 1.4655977487564087,
"learning_rate": 2.9933333333333337e-05,
"loss": 0.1707,
"step": 450
},
{
"grad_norm": 1.6052281856536865,
"learning_rate": 3.06e-05,
"loss": 0.1666,
"step": 460
},
{
"grad_norm": 1.804250717163086,
"learning_rate": 3.126666666666666e-05,
"loss": 0.1758,
"step": 470
},
{
"grad_norm": 1.7997639179229736,
"learning_rate": 3.1933333333333335e-05,
"loss": 0.155,
"step": 480
},
{
"grad_norm": 2.390489339828491,
"learning_rate": 3.26e-05,
"loss": 0.1566,
"step": 490
},
{
"grad_norm": 1.5680009126663208,
"learning_rate": 3.326666666666667e-05,
"loss": 0.1382,
"step": 500
},
{
"grad_norm": 1.4433048963546753,
"learning_rate": 3.3933333333333336e-05,
"loss": 0.1533,
"step": 510
},
{
"grad_norm": 1.4073946475982666,
"learning_rate": 3.46e-05,
"loss": 0.1413,
"step": 520
},
{
"grad_norm": 1.7121227979660034,
"learning_rate": 3.526666666666667e-05,
"loss": 0.1499,
"step": 530
},
{
"grad_norm": 1.8003591299057007,
"learning_rate": 3.593333333333334e-05,
"loss": 0.1408,
"step": 540
},
{
"grad_norm": 1.7915265560150146,
"learning_rate": 3.66e-05,
"loss": 0.1401,
"step": 550
},
{
"grad_norm": 1.6718388795852661,
"learning_rate": 3.726666666666667e-05,
"loss": 0.1439,
"step": 560
},
{
"grad_norm": 1.81702721118927,
"learning_rate": 3.793333333333334e-05,
"loss": 0.1396,
"step": 570
},
{
"grad_norm": 1.4860053062438965,
"learning_rate": 3.86e-05,
"loss": 0.1434,
"step": 580
},
{
"grad_norm": 1.5475099086761475,
"learning_rate": 3.926666666666667e-05,
"loss": 0.1421,
"step": 590
},
{
"grad_norm": 1.3822091817855835,
"learning_rate": 3.993333333333333e-05,
"loss": 0.1425,
"step": 600
},
{
"grad_norm": 1.3974543809890747,
"learning_rate": 4.0600000000000004e-05,
"loss": 0.1477,
"step": 610
},
{
"grad_norm": 1.8676766157150269,
"learning_rate": 4.126666666666667e-05,
"loss": 0.1297,
"step": 620
},
{
"grad_norm": 1.474252462387085,
"learning_rate": 4.1933333333333334e-05,
"loss": 0.1226,
"step": 630
},
{
"grad_norm": 1.0427896976470947,
"learning_rate": 4.26e-05,
"loss": 0.1229,
"step": 640
},
{
"grad_norm": 1.2088690996170044,
"learning_rate": 4.3266666666666664e-05,
"loss": 0.1271,
"step": 650
},
{
"grad_norm": 1.3452211618423462,
"learning_rate": 4.3933333333333335e-05,
"loss": 0.1296,
"step": 660
},
{
"grad_norm": 1.5475252866744995,
"learning_rate": 4.46e-05,
"loss": 0.1641,
"step": 670
},
{
"grad_norm": 1.4876209497451782,
"learning_rate": 4.526666666666667e-05,
"loss": 0.1429,
"step": 680
},
{
"grad_norm": 1.2453781366348267,
"learning_rate": 4.5933333333333336e-05,
"loss": 0.1307,
"step": 690
},
{
"grad_norm": 1.0960428714752197,
"learning_rate": 4.660000000000001e-05,
"loss": 0.1359,
"step": 700
},
{
"grad_norm": 1.6948680877685547,
"learning_rate": 4.726666666666667e-05,
"loss": 0.1648,
"step": 710
},
{
"grad_norm": 1.1291084289550781,
"learning_rate": 4.793333333333334e-05,
"loss": 0.14,
"step": 720
},
{
"grad_norm": 1.2624635696411133,
"learning_rate": 4.86e-05,
"loss": 0.1436,
"step": 730
},
{
"grad_norm": 1.4393306970596313,
"learning_rate": 4.926666666666667e-05,
"loss": 0.1263,
"step": 740
},
{
"grad_norm": 1.1790522336959839,
"learning_rate": 4.993333333333334e-05,
"loss": 0.1474,
"step": 750
},
{
"grad_norm": 1.2757195234298706,
"learning_rate": 5.0600000000000003e-05,
"loss": 0.1282,
"step": 760
},
{
"grad_norm": 0.8269880414009094,
"learning_rate": 5.1266666666666675e-05,
"loss": 0.1338,
"step": 770
},
{
"grad_norm": 1.3963230848312378,
"learning_rate": 5.193333333333333e-05,
"loss": 0.13,
"step": 780
},
{
"grad_norm": 1.0048385858535767,
"learning_rate": 5.2600000000000005e-05,
"loss": 0.1253,
"step": 790
},
{
"grad_norm": 1.1798450946807861,
"learning_rate": 5.326666666666666e-05,
"loss": 0.1274,
"step": 800
},
{
"grad_norm": 1.077629566192627,
"learning_rate": 5.3933333333333334e-05,
"loss": 0.1442,
"step": 810
},
{
"grad_norm": 1.1865150928497314,
"learning_rate": 5.4600000000000006e-05,
"loss": 0.1316,
"step": 820
},
{
"grad_norm": 1.089298129081726,
"learning_rate": 5.5266666666666664e-05,
"loss": 0.1377,
"step": 830
},
{
"grad_norm": 1.1403353214263916,
"learning_rate": 5.5933333333333335e-05,
"loss": 0.1209,
"step": 840
},
{
"grad_norm": 1.0784872770309448,
"learning_rate": 5.66e-05,
"loss": 0.136,
"step": 850
},
{
"grad_norm": 1.1640064716339111,
"learning_rate": 5.726666666666667e-05,
"loss": 0.1218,
"step": 860
},
{
"grad_norm": 1.3921116590499878,
"learning_rate": 5.7933333333333337e-05,
"loss": 0.1364,
"step": 870
},
{
"grad_norm": 1.3123397827148438,
"learning_rate": 5.86e-05,
"loss": 0.1259,
"step": 880
},
{
"grad_norm": 0.9924466609954834,
"learning_rate": 5.926666666666667e-05,
"loss": 0.1139,
"step": 890
},
{
"grad_norm": 1.2900114059448242,
"learning_rate": 5.9933333333333345e-05,
"loss": 0.1294,
"step": 900
},
{
"grad_norm": 0.8750925064086914,
"learning_rate": 6.06e-05,
"loss": 0.1083,
"step": 910
},
{
"grad_norm": 0.9233385324478149,
"learning_rate": 6.126666666666667e-05,
"loss": 0.1237,
"step": 920
},
{
"grad_norm": 0.8136280179023743,
"learning_rate": 6.193333333333333e-05,
"loss": 0.1227,
"step": 930
},
{
"grad_norm": 0.954278826713562,
"learning_rate": 6.26e-05,
"loss": 0.1104,
"step": 940
},
{
"grad_norm": 0.9921590685844421,
"learning_rate": 6.326666666666667e-05,
"loss": 0.1279,
"step": 950
},
{
"grad_norm": 0.9617112278938293,
"learning_rate": 6.393333333333333e-05,
"loss": 0.1097,
"step": 960
},
{
"grad_norm": 0.9765651226043701,
"learning_rate": 6.460000000000001e-05,
"loss": 0.1146,
"step": 970
},
{
"grad_norm": 0.9349573850631714,
"learning_rate": 6.526666666666666e-05,
"loss": 0.1169,
"step": 980
},
{
"grad_norm": 0.7948217391967773,
"learning_rate": 6.593333333333334e-05,
"loss": 0.1158,
"step": 990
},
{
"grad_norm": 0.9776331782341003,
"learning_rate": 6.66e-05,
"loss": 0.1071,
"step": 1000
},
{
"grad_norm": 0.7821135520935059,
"learning_rate": 6.726666666666667e-05,
"loss": 0.1079,
"step": 1010
},
{
"grad_norm": 1.0361899137496948,
"learning_rate": 6.793333333333334e-05,
"loss": 0.1111,
"step": 1020
},
{
"grad_norm": 0.9012498259544373,
"learning_rate": 6.860000000000001e-05,
"loss": 0.1116,
"step": 1030
},
{
"grad_norm": 0.9130027890205383,
"learning_rate": 6.926666666666667e-05,
"loss": 0.1104,
"step": 1040
},
{
"grad_norm": 0.661876380443573,
"learning_rate": 6.993333333333334e-05,
"loss": 0.1016,
"step": 1050
},
{
"grad_norm": 0.893703818321228,
"learning_rate": 7.06e-05,
"loss": 0.1139,
"step": 1060
},
{
"grad_norm": 0.9637819528579712,
"learning_rate": 7.126666666666667e-05,
"loss": 0.1118,
"step": 1070
},
{
"grad_norm": 1.069870948791504,
"learning_rate": 7.193333333333334e-05,
"loss": 0.1169,
"step": 1080
},
{
"grad_norm": 0.9310933947563171,
"learning_rate": 7.26e-05,
"loss": 0.1046,
"step": 1090
},
{
"grad_norm": 0.9750380516052246,
"learning_rate": 7.326666666666667e-05,
"loss": 0.1129,
"step": 1100
},
{
"grad_norm": 0.7587231397628784,
"learning_rate": 7.393333333333333e-05,
"loss": 0.1204,
"step": 1110
},
{
"grad_norm": 0.8045339584350586,
"learning_rate": 7.46e-05,
"loss": 0.1135,
"step": 1120
},
{
"grad_norm": 1.0460385084152222,
"learning_rate": 7.526666666666668e-05,
"loss": 0.1208,
"step": 1130
},
{
"grad_norm": 0.8942551612854004,
"learning_rate": 7.593333333333334e-05,
"loss": 0.1117,
"step": 1140
},
{
"grad_norm": 0.9932526350021362,
"learning_rate": 7.66e-05,
"loss": 0.1223,
"step": 1150
},
{
"grad_norm": 0.8706763982772827,
"learning_rate": 7.726666666666667e-05,
"loss": 0.1228,
"step": 1160
},
{
"grad_norm": 0.6985638737678528,
"learning_rate": 7.793333333333333e-05,
"loss": 0.1114,
"step": 1170
},
{
"grad_norm": 0.9824487566947937,
"learning_rate": 7.860000000000001e-05,
"loss": 0.113,
"step": 1180
},
{
"grad_norm": 0.7363618612289429,
"learning_rate": 7.926666666666666e-05,
"loss": 0.1147,
"step": 1190
},
{
"grad_norm": 0.7442812919616699,
"learning_rate": 7.993333333333334e-05,
"loss": 0.1238,
"step": 1200
},
{
"grad_norm": 0.6828699111938477,
"learning_rate": 8.060000000000001e-05,
"loss": 0.1162,
"step": 1210
},
{
"grad_norm": 0.7698734998703003,
"learning_rate": 8.126666666666667e-05,
"loss": 0.1175,
"step": 1220
},
{
"grad_norm": 0.8658196330070496,
"learning_rate": 8.193333333333334e-05,
"loss": 0.122,
"step": 1230
},
{
"grad_norm": 0.7327065467834473,
"learning_rate": 8.26e-05,
"loss": 0.1085,
"step": 1240
},
{
"grad_norm": 0.7281940579414368,
"learning_rate": 8.326666666666667e-05,
"loss": 0.1133,
"step": 1250
},
{
"grad_norm": 0.6572296023368835,
"learning_rate": 8.393333333333335e-05,
"loss": 0.1083,
"step": 1260
},
{
"grad_norm": 0.8149025440216064,
"learning_rate": 8.46e-05,
"loss": 0.1226,
"step": 1270
},
{
"grad_norm": 0.6859647631645203,
"learning_rate": 8.526666666666667e-05,
"loss": 0.1159,
"step": 1280
},
{
"grad_norm": 0.7365497350692749,
"learning_rate": 8.593333333333333e-05,
"loss": 0.1143,
"step": 1290
},
{
"grad_norm": 0.8521119356155396,
"learning_rate": 8.66e-05,
"loss": 0.1108,
"step": 1300
},
{
"grad_norm": 0.7932054996490479,
"learning_rate": 8.726666666666667e-05,
"loss": 0.0996,
"step": 1310
},
{
"grad_norm": 0.9650532007217407,
"learning_rate": 8.793333333333333e-05,
"loss": 0.1173,
"step": 1320
},
{
"grad_norm": 0.7747259736061096,
"learning_rate": 8.86e-05,
"loss": 0.1078,
"step": 1330
},
{
"grad_norm": 0.7914727926254272,
"learning_rate": 8.926666666666668e-05,
"loss": 0.1121,
"step": 1340
},
{
"grad_norm": 0.8803926706314087,
"learning_rate": 8.993333333333334e-05,
"loss": 0.0989,
"step": 1350
},
{
"grad_norm": 0.8079709410667419,
"learning_rate": 9.06e-05,
"loss": 0.1095,
"step": 1360
},
{
"grad_norm": 0.9866183996200562,
"learning_rate": 9.126666666666667e-05,
"loss": 0.0985,
"step": 1370
},
{
"grad_norm": 0.7971082329750061,
"learning_rate": 9.193333333333334e-05,
"loss": 0.1033,
"step": 1380
},
{
"grad_norm": 0.7286013960838318,
"learning_rate": 9.260000000000001e-05,
"loss": 0.1041,
"step": 1390
},
{
"grad_norm": 0.8173230886459351,
"learning_rate": 9.326666666666667e-05,
"loss": 0.1069,
"step": 1400
},
{
"grad_norm": 0.6183474659919739,
"learning_rate": 9.393333333333334e-05,
"loss": 0.0953,
"step": 1410
},
{
"grad_norm": 0.7262031435966492,
"learning_rate": 9.46e-05,
"loss": 0.1143,
"step": 1420
},
{
"grad_norm": 0.5853284001350403,
"learning_rate": 9.526666666666667e-05,
"loss": 0.1062,
"step": 1430
},
{
"grad_norm": 0.7501056790351868,
"learning_rate": 9.593333333333334e-05,
"loss": 0.1054,
"step": 1440
},
{
"grad_norm": 0.7188894152641296,
"learning_rate": 9.66e-05,
"loss": 0.1083,
"step": 1450
},
{
"grad_norm": 0.7067384123802185,
"learning_rate": 9.726666666666667e-05,
"loss": 0.1025,
"step": 1460
},
{
"grad_norm": 0.778515636920929,
"learning_rate": 9.793333333333333e-05,
"loss": 0.106,
"step": 1470
},
{
"grad_norm": 0.6316561698913574,
"learning_rate": 9.86e-05,
"loss": 0.1184,
"step": 1480
},
{
"grad_norm": 0.6245169639587402,
"learning_rate": 9.926666666666668e-05,
"loss": 0.1001,
"step": 1490
},
{
"grad_norm": 0.6359462738037109,
"learning_rate": 9.993333333333334e-05,
"loss": 0.1231,
"step": 1500
},
{
"grad_norm": 0.6508445739746094,
"learning_rate": 9.999997539434007e-05,
"loss": 0.1101,
"step": 1510
},
{
"grad_norm": 0.5554887652397156,
"learning_rate": 9.999989033776898e-05,
"loss": 0.108,
"step": 1520
},
{
"grad_norm": 0.49556964635849,
"learning_rate": 9.999974452661641e-05,
"loss": 0.1116,
"step": 1530
},
{
"grad_norm": 0.7269011735916138,
"learning_rate": 9.999953796105959e-05,
"loss": 0.1207,
"step": 1540
},
{
"grad_norm": 0.7191609740257263,
"learning_rate": 9.999927064134949e-05,
"loss": 0.1143,
"step": 1550
},
{
"grad_norm": 0.5663521885871887,
"learning_rate": 9.999894256781095e-05,
"loss": 0.1046,
"step": 1560
},
{
"grad_norm": 0.7881866097450256,
"learning_rate": 9.99985537408426e-05,
"loss": 0.1022,
"step": 1570
},
{
"grad_norm": 0.654952347278595,
"learning_rate": 9.999810416091688e-05,
"loss": 0.1027,
"step": 1580
},
{
"grad_norm": 0.634846031665802,
"learning_rate": 9.99975938285801e-05,
"loss": 0.1056,
"step": 1590
},
{
"grad_norm": 0.743867039680481,
"learning_rate": 9.999702274445236e-05,
"loss": 0.1009,
"step": 1600
},
{
"grad_norm": 0.5587864518165588,
"learning_rate": 9.999639090922756e-05,
"loss": 0.0997,
"step": 1610
},
{
"grad_norm": 0.5834605097770691,
"learning_rate": 9.999569832367346e-05,
"loss": 0.1008,
"step": 1620
},
{
"grad_norm": 0.5587565302848816,
"learning_rate": 9.999494498863162e-05,
"loss": 0.0987,
"step": 1630
},
{
"grad_norm": 0.7109138369560242,
"learning_rate": 9.99941309050174e-05,
"loss": 0.1056,
"step": 1640
},
{
"grad_norm": 0.6822052597999573,
"learning_rate": 9.999325607381999e-05,
"loss": 0.1079,
"step": 1650
},
{
"grad_norm": 0.522294282913208,
"learning_rate": 9.999232049610238e-05,
"loss": 0.0899,
"step": 1660
},
{
"grad_norm": 0.5994908213615417,
"learning_rate": 9.999132417300142e-05,
"loss": 0.0967,
"step": 1670
},
{
"grad_norm": 0.5230586528778076,
"learning_rate": 9.99902671057277e-05,
"loss": 0.1027,
"step": 1680
},
{
"grad_norm": 0.6761882901191711,
"learning_rate": 9.998914929556569e-05,
"loss": 0.1023,
"step": 1690
},
{
"grad_norm": 0.5694276094436646,
"learning_rate": 9.998797074387361e-05,
"loss": 0.0934,
"step": 1700
},
{
"grad_norm": 0.5761187672615051,
"learning_rate": 9.99867314520835e-05,
"loss": 0.0864,
"step": 1710
},
{
"grad_norm": 0.6426172256469727,
"learning_rate": 9.998543142170126e-05,
"loss": 0.1012,
"step": 1720
},
{
"grad_norm": 0.6870645880699158,
"learning_rate": 9.99840706543065e-05,
"loss": 0.095,
"step": 1730
},
{
"grad_norm": 0.649592399597168,
"learning_rate": 9.998264915155274e-05,
"loss": 0.0917,
"step": 1740
},
{
"grad_norm": 0.7904589176177979,
"learning_rate": 9.998116691516718e-05,
"loss": 0.1081,
"step": 1750
},
{
"grad_norm": 0.6588676571846008,
"learning_rate": 9.997962394695091e-05,
"loss": 0.1177,
"step": 1760
},
{
"grad_norm": 0.8387064337730408,
"learning_rate": 9.997802024877875e-05,
"loss": 0.1136,
"step": 1770
},
{
"grad_norm": 0.6109852194786072,
"learning_rate": 9.99763558225994e-05,
"loss": 0.1092,
"step": 1780
},
{
"grad_norm": 0.7083513736724854,
"learning_rate": 9.997463067043526e-05,
"loss": 0.1241,
"step": 1790
},
{
"grad_norm": 0.7610124349594116,
"learning_rate": 9.997284479438253e-05,
"loss": 0.1091,
"step": 1800
},
{
"grad_norm": 0.5530665516853333,
"learning_rate": 9.997099819661127e-05,
"loss": 0.1158,
"step": 1810
},
{
"grad_norm": 0.5753591060638428,
"learning_rate": 9.996909087936524e-05,
"loss": 0.1033,
"step": 1820
},
{
"grad_norm": 0.4656466245651245,
"learning_rate": 9.996712284496201e-05,
"loss": 0.0934,
"step": 1830
},
{
"grad_norm": 0.5710314512252808,
"learning_rate": 9.996509409579293e-05,
"loss": 0.094,
"step": 1840
},
{
"grad_norm": 0.4732897877693176,
"learning_rate": 9.996300463432312e-05,
"loss": 0.0982,
"step": 1850
},
{
"grad_norm": 0.5390840172767639,
"learning_rate": 9.996085446309148e-05,
"loss": 0.1018,
"step": 1860
},
{
"grad_norm": 0.5768323540687561,
"learning_rate": 9.995864358471066e-05,
"loss": 0.1021,
"step": 1870
},
{
"grad_norm": 0.49832701683044434,
"learning_rate": 9.99563720018671e-05,
"loss": 0.0904,
"step": 1880
},
{
"grad_norm": 0.5812219977378845,
"learning_rate": 9.995403971732098e-05,
"loss": 0.0947,
"step": 1890
},
{
"grad_norm": 0.48234522342681885,
"learning_rate": 9.995164673390625e-05,
"loss": 0.0988,
"step": 1900
},
{
"grad_norm": 0.6657654643058777,
"learning_rate": 9.994919305453059e-05,
"loss": 0.0973,
"step": 1910
},
{
"grad_norm": 0.49641579389572144,
"learning_rate": 9.994667868217548e-05,
"loss": 0.0938,
"step": 1920
},
{
"grad_norm": 0.6323332786560059,
"learning_rate": 9.99441036198961e-05,
"loss": 0.1005,
"step": 1930
},
{
"grad_norm": 0.654400646686554,
"learning_rate": 9.99414678708214e-05,
"loss": 0.0909,
"step": 1940
},
{
"grad_norm": 0.6005545258522034,
"learning_rate": 9.993877143815407e-05,
"loss": 0.0976,
"step": 1950
},
{
"grad_norm": 0.5324005484580994,
"learning_rate": 9.993601432517053e-05,
"loss": 0.0831,
"step": 1960
},
{
"grad_norm": 0.4393467605113983,
"learning_rate": 9.993319653522091e-05,
"loss": 0.098,
"step": 1970
},
{
"grad_norm": 0.6097012162208557,
"learning_rate": 9.993031807172911e-05,
"loss": 0.1022,
"step": 1980
},
{
"grad_norm": 0.601715087890625,
"learning_rate": 9.992737893819273e-05,
"loss": 0.0862,
"step": 1990
},
{
"grad_norm": 0.4945420026779175,
"learning_rate": 9.992437913818312e-05,
"loss": 0.0826,
"step": 2000
},
{
"grad_norm": 0.4312385618686676,
"learning_rate": 9.992131867534526e-05,
"loss": 0.0835,
"step": 2010
},
{
"grad_norm": 0.591633677482605,
"learning_rate": 9.991819755339796e-05,
"loss": 0.0933,
"step": 2020
},
{
"grad_norm": 0.5529596209526062,
"learning_rate": 9.991501577613365e-05,
"loss": 0.0932,
"step": 2030
},
{
"grad_norm": 0.5599483251571655,
"learning_rate": 9.99117733474185e-05,
"loss": 0.0952,
"step": 2040
},
{
"grad_norm": 0.4217069745063782,
"learning_rate": 9.990847027119234e-05,
"loss": 0.0842,
"step": 2050
},
{
"grad_norm": 0.5155817270278931,
"learning_rate": 9.990510655146877e-05,
"loss": 0.0831,
"step": 2060
},
{
"grad_norm": 0.4740280210971832,
"learning_rate": 9.990168219233496e-05,
"loss": 0.0803,
"step": 2070
},
{
"grad_norm": 0.563309371471405,
"learning_rate": 9.989819719795188e-05,
"loss": 0.0779,
"step": 2080
},
{
"grad_norm": 0.44790732860565186,
"learning_rate": 9.989465157255412e-05,
"loss": 0.0743,
"step": 2090
},
{
"grad_norm": 0.5462765693664551,
"learning_rate": 9.989104532044994e-05,
"loss": 0.0683,
"step": 2100
},
{
"grad_norm": 0.5009019374847412,
"learning_rate": 9.988737844602128e-05,
"loss": 0.0777,
"step": 2110
},
{
"grad_norm": 0.4951901137828827,
"learning_rate": 9.988365095372372e-05,
"loss": 0.0855,
"step": 2120
},
{
"grad_norm": 0.4020739197731018,
"learning_rate": 9.987986284808654e-05,
"loss": 0.0865,
"step": 2130
},
{
"grad_norm": 0.6208875775337219,
"learning_rate": 9.987601413371264e-05,
"loss": 0.0901,
"step": 2140
},
{
"grad_norm": 0.4745827615261078,
"learning_rate": 9.987210481527855e-05,
"loss": 0.0911,
"step": 2150
},
{
"grad_norm": 0.5301135778427124,
"learning_rate": 9.98681348975345e-05,
"loss": 0.0842,
"step": 2160
},
{
"grad_norm": 0.7118707895278931,
"learning_rate": 9.986410438530427e-05,
"loss": 0.0922,
"step": 2170
},
{
"grad_norm": 0.6230258941650391,
"learning_rate": 9.986001328348534e-05,
"loss": 0.0973,
"step": 2180
},
{
"grad_norm": 0.4599594175815582,
"learning_rate": 9.985586159704878e-05,
"loss": 0.0916,
"step": 2190
},
{
"grad_norm": 0.5305163860321045,
"learning_rate": 9.985164933103929e-05,
"loss": 0.0851,
"step": 2200
},
{
"grad_norm": 0.5047589540481567,
"learning_rate": 9.984737649057513e-05,
"loss": 0.0881,
"step": 2210
},
{
"grad_norm": 0.5656068921089172,
"learning_rate": 9.984304308084827e-05,
"loss": 0.0794,
"step": 2220
},
{
"grad_norm": 0.7826402187347412,
"learning_rate": 9.983864910712416e-05,
"loss": 0.0874,
"step": 2230
},
{
"grad_norm": 0.72354656457901,
"learning_rate": 9.98341945747419e-05,
"loss": 0.0858,
"step": 2240
},
{
"grad_norm": 0.46766701340675354,
"learning_rate": 9.98296794891142e-05,
"loss": 0.0923,
"step": 2250
},
{
"grad_norm": 0.5706295371055603,
"learning_rate": 9.982510385572725e-05,
"loss": 0.0895,
"step": 2260
},
{
"grad_norm": 0.6010457873344421,
"learning_rate": 9.982046768014094e-05,
"loss": 0.0866,
"step": 2270
},
{
"grad_norm": 0.6102373600006104,
"learning_rate": 9.981577096798863e-05,
"loss": 0.099,
"step": 2280
},
{
"grad_norm": 0.43630605936050415,
"learning_rate": 9.981101372497727e-05,
"loss": 0.094,
"step": 2290
},
{
"grad_norm": 0.5948781371116638,
"learning_rate": 9.980619595688737e-05,
"loss": 0.0895,
"step": 2300
},
{
"grad_norm": 0.5268418192863464,
"learning_rate": 9.980131766957295e-05,
"loss": 0.0944,
"step": 2310
},
{
"grad_norm": 0.5282972455024719,
"learning_rate": 9.979637886896163e-05,
"loss": 0.0916,
"step": 2320
},
{
"grad_norm": 1.2789876461029053,
"learning_rate": 9.979137956105447e-05,
"loss": 0.1069,
"step": 2330
},
{
"grad_norm": 0.5114489793777466,
"learning_rate": 9.978631975192613e-05,
"loss": 0.0896,
"step": 2340
},
{
"grad_norm": 0.6944462656974792,
"learning_rate": 9.978119944772475e-05,
"loss": 0.108,
"step": 2350
},
{
"grad_norm": 0.4593367278575897,
"learning_rate": 9.977601865467197e-05,
"loss": 0.0931,
"step": 2360
},
{
"grad_norm": 0.5884009003639221,
"learning_rate": 9.977077737906297e-05,
"loss": 0.0962,
"step": 2370
},
{
"grad_norm": 0.6174680590629578,
"learning_rate": 9.976547562726636e-05,
"loss": 0.0867,
"step": 2380
},
{
"grad_norm": 0.48397818207740784,
"learning_rate": 9.976011340572429e-05,
"loss": 0.1015,
"step": 2390
},
{
"grad_norm": 0.5150822401046753,
"learning_rate": 9.975469072095237e-05,
"loss": 0.0903,
"step": 2400
},
{
"grad_norm": 0.5585728883743286,
"learning_rate": 9.974920757953965e-05,
"loss": 0.088,
"step": 2410
},
{
"grad_norm": 0.49235379695892334,
"learning_rate": 9.97436639881487e-05,
"loss": 0.086,
"step": 2420
},
{
"grad_norm": 0.4590860903263092,
"learning_rate": 9.973805995351545e-05,
"loss": 0.0823,
"step": 2430
},
{
"grad_norm": 0.5391248464584351,
"learning_rate": 9.973239548244939e-05,
"loss": 0.0867,
"step": 2440
},
{
"grad_norm": 0.4855298101902008,
"learning_rate": 9.972667058183333e-05,
"loss": 0.0853,
"step": 2450
},
{
"grad_norm": 0.5823668241500854,
"learning_rate": 9.972088525862362e-05,
"loss": 0.0872,
"step": 2460
},
{
"grad_norm": 0.44840821623802185,
"learning_rate": 9.971503951984995e-05,
"loss": 0.0826,
"step": 2470
},
{
"grad_norm": 0.4621039927005768,
"learning_rate": 9.970913337261543e-05,
"loss": 0.0828,
"step": 2480
},
{
"grad_norm": 0.4865065813064575,
"learning_rate": 9.97031668240966e-05,
"loss": 0.0703,
"step": 2490
},
{
"grad_norm": 0.5840463638305664,
"learning_rate": 9.969713988154339e-05,
"loss": 0.0882,
"step": 2500
},
{
"grad_norm": 0.44769713282585144,
"learning_rate": 9.969105255227906e-05,
"loss": 0.0878,
"step": 2510
},
{
"grad_norm": 0.47158923745155334,
"learning_rate": 9.968490484370035e-05,
"loss": 0.0784,
"step": 2520
},
{
"grad_norm": 0.5088538527488708,
"learning_rate": 9.967869676327726e-05,
"loss": 0.0775,
"step": 2530
},
{
"grad_norm": 0.574734628200531,
"learning_rate": 9.96724283185532e-05,
"loss": 0.0802,
"step": 2540
},
{
"grad_norm": 0.47324708104133606,
"learning_rate": 9.966609951714494e-05,
"loss": 0.0913,
"step": 2550
},
{
"grad_norm": 0.4296296536922455,
"learning_rate": 9.965971036674255e-05,
"loss": 0.0819,
"step": 2560
},
{
"grad_norm": 0.4795040488243103,
"learning_rate": 9.965326087510947e-05,
"loss": 0.0842,
"step": 2570
},
{
"grad_norm": 0.3793949484825134,
"learning_rate": 9.964675105008244e-05,
"loss": 0.0874,
"step": 2580
},
{
"grad_norm": 0.5109375715255737,
"learning_rate": 9.964018089957147e-05,
"loss": 0.0813,
"step": 2590
},
{
"grad_norm": 0.4429580271244049,
"learning_rate": 9.963355043155996e-05,
"loss": 0.076,
"step": 2600
},
{
"grad_norm": 0.42398586869239807,
"learning_rate": 9.962685965410455e-05,
"loss": 0.0772,
"step": 2610
},
{
"grad_norm": 0.538895845413208,
"learning_rate": 9.962010857533514e-05,
"loss": 0.0779,
"step": 2620
},
{
"grad_norm": 0.536656379699707,
"learning_rate": 9.961329720345493e-05,
"loss": 0.0829,
"step": 2630
},
{
"grad_norm": 0.5466274619102478,
"learning_rate": 9.96064255467404e-05,
"loss": 0.0741,
"step": 2640
},
{
"grad_norm": 0.526410698890686,
"learning_rate": 9.959949361354126e-05,
"loss": 0.0865,
"step": 2650
},
{
"grad_norm": 0.39437776803970337,
"learning_rate": 9.959250141228045e-05,
"loss": 0.0942,
"step": 2660
},
{
"grad_norm": 0.3603488802909851,
"learning_rate": 9.958544895145414e-05,
"loss": 0.0818,
"step": 2670
},
{
"grad_norm": 0.43313562870025635,
"learning_rate": 9.957833623963177e-05,
"loss": 0.0822,
"step": 2680
},
{
"grad_norm": 0.48792722821235657,
"learning_rate": 9.957116328545593e-05,
"loss": 0.0807,
"step": 2690
},
{
"grad_norm": 0.5208479762077332,
"learning_rate": 9.956393009764244e-05,
"loss": 0.0852,
"step": 2700
},
{
"grad_norm": 0.49033123254776,
"learning_rate": 9.955663668498032e-05,
"loss": 0.0819,
"step": 2710
},
{
"grad_norm": 0.45797768235206604,
"learning_rate": 9.954928305633173e-05,
"loss": 0.0784,
"step": 2720
},
{
"grad_norm": 0.5387983322143555,
"learning_rate": 9.954186922063204e-05,
"loss": 0.0862,
"step": 2730
},
{
"grad_norm": 0.4782763421535492,
"learning_rate": 9.953439518688974e-05,
"loss": 0.0838,
"step": 2740
},
{
"grad_norm": 0.4697781503200531,
"learning_rate": 9.952686096418652e-05,
"loss": 0.0797,
"step": 2750
},
{
"grad_norm": 0.5175778865814209,
"learning_rate": 9.951926656167714e-05,
"loss": 0.0767,
"step": 2760
},
{
"grad_norm": 0.4419208765029907,
"learning_rate": 9.951161198858953e-05,
"loss": 0.0805,
"step": 2770
},
{
"grad_norm": 0.4175063371658325,
"learning_rate": 9.950389725422471e-05,
"loss": 0.0799,
"step": 2780
},
{
"grad_norm": 0.5536662340164185,
"learning_rate": 9.949612236795682e-05,
"loss": 0.0729,
"step": 2790
},
{
"grad_norm": 0.3647864758968353,
"learning_rate": 9.948828733923305e-05,
"loss": 0.0803,
"step": 2800
},
{
"grad_norm": 0.4188004434108734,
"learning_rate": 9.948039217757374e-05,
"loss": 0.0832,
"step": 2810
},
{
"grad_norm": 0.5175153613090515,
"learning_rate": 9.947243689257225e-05,
"loss": 0.077,
"step": 2820
},
{
"grad_norm": 0.38628271222114563,
"learning_rate": 9.946442149389497e-05,
"loss": 0.0871,
"step": 2830
},
{
"grad_norm": 0.42856937646865845,
"learning_rate": 9.945634599128139e-05,
"loss": 0.0737,
"step": 2840
},
{
"grad_norm": 0.51496821641922,
"learning_rate": 9.944821039454402e-05,
"loss": 0.0826,
"step": 2850
},
{
"grad_norm": 0.44280192255973816,
"learning_rate": 9.944001471356835e-05,
"loss": 0.0798,
"step": 2860
},
{
"grad_norm": 0.43873319029808044,
"learning_rate": 9.94317589583129e-05,
"loss": 0.0773,
"step": 2870
},
{
"grad_norm": 0.48948147892951965,
"learning_rate": 9.942344313880922e-05,
"loss": 0.08,
"step": 2880
},
{
"grad_norm": 0.4801962971687317,
"learning_rate": 9.941506726516179e-05,
"loss": 0.0805,
"step": 2890
},
{
"grad_norm": 0.4290546178817749,
"learning_rate": 9.94066313475481e-05,
"loss": 0.0802,
"step": 2900
},
{
"grad_norm": 0.48247382044792175,
"learning_rate": 9.939813539621857e-05,
"loss": 0.0885,
"step": 2910
},
{
"grad_norm": 0.584211528301239,
"learning_rate": 9.93895794214966e-05,
"loss": 0.0771,
"step": 2920
},
{
"grad_norm": 0.5174719095230103,
"learning_rate": 9.938096343377852e-05,
"loss": 0.0781,
"step": 2930
},
{
"grad_norm": 0.49908751249313354,
"learning_rate": 9.937228744353353e-05,
"loss": 0.073,
"step": 2940
},
{
"grad_norm": 0.37439218163490295,
"learning_rate": 9.936355146130379e-05,
"loss": 0.0743,
"step": 2950
},
{
"grad_norm": 0.5374181270599365,
"learning_rate": 9.935475549770435e-05,
"loss": 0.0816,
"step": 2960
},
{
"grad_norm": 0.44372794032096863,
"learning_rate": 9.934589956342315e-05,
"loss": 0.0787,
"step": 2970
},
{
"grad_norm": 0.5551127195358276,
"learning_rate": 9.933698366922093e-05,
"loss": 0.0762,
"step": 2980
},
{
"grad_norm": 0.4737628698348999,
"learning_rate": 9.93280078259314e-05,
"loss": 0.0774,
"step": 2990
},
{
"grad_norm": 0.516853392124176,
"learning_rate": 9.931897204446104e-05,
"loss": 0.0837,
"step": 3000
},
{
"grad_norm": 0.4667161703109741,
"learning_rate": 9.930987633578915e-05,
"loss": 0.0728,
"step": 3010
},
{
"grad_norm": 0.612838625907898,
"learning_rate": 9.93007207109679e-05,
"loss": 0.0858,
"step": 3020
},
{
"grad_norm": 0.4260261356830597,
"learning_rate": 9.929150518112224e-05,
"loss": 0.081,
"step": 3030
},
{
"grad_norm": 0.5185019969940186,
"learning_rate": 9.928222975744991e-05,
"loss": 0.07,
"step": 3040
},
{
"grad_norm": 0.4650256037712097,
"learning_rate": 9.92728944512214e-05,
"loss": 0.0842,
"step": 3050
},
{
"grad_norm": 0.5540300607681274,
"learning_rate": 9.926349927378001e-05,
"loss": 0.0841,
"step": 3060
},
{
"grad_norm": 0.39831820130348206,
"learning_rate": 9.925404423654174e-05,
"loss": 0.0889,
"step": 3070
},
{
"grad_norm": 0.45987674593925476,
"learning_rate": 9.924452935099537e-05,
"loss": 0.0793,
"step": 3080
},
{
"grad_norm": 0.43055635690689087,
"learning_rate": 9.92349546287024e-05,
"loss": 0.0809,
"step": 3090
},
{
"grad_norm": 0.39429405331611633,
"learning_rate": 9.9225320081297e-05,
"loss": 0.0829,
"step": 3100
},
{
"grad_norm": 0.5204262733459473,
"learning_rate": 9.921562572048606e-05,
"loss": 0.0924,
"step": 3110
},
{
"grad_norm": 0.4526405334472656,
"learning_rate": 9.920587155804913e-05,
"loss": 0.0804,
"step": 3120
},
{
"grad_norm": 0.42921537160873413,
"learning_rate": 9.919605760583845e-05,
"loss": 0.0828,
"step": 3130
},
{
"grad_norm": 0.3426622450351715,
"learning_rate": 9.91861838757789e-05,
"loss": 0.0713,
"step": 3140
},
{
"grad_norm": 0.3761134743690491,
"learning_rate": 9.917625037986798e-05,
"loss": 0.0685,
"step": 3150
},
{
"grad_norm": 0.43040111660957336,
"learning_rate": 9.916625713017583e-05,
"loss": 0.0718,
"step": 3160
},
{
"grad_norm": 0.4879830777645111,
"learning_rate": 9.915620413884519e-05,
"loss": 0.072,
"step": 3170
},
{
"grad_norm": 0.38001909852027893,
"learning_rate": 9.914609141809139e-05,
"loss": 0.0746,
"step": 3180
},
{
"grad_norm": 0.4578431248664856,
"learning_rate": 9.913591898020235e-05,
"loss": 0.0756,
"step": 3190
},
{
"grad_norm": 0.5365106463432312,
"learning_rate": 9.912568683753853e-05,
"loss": 0.0812,
"step": 3200
},
{
"grad_norm": 0.4154616594314575,
"learning_rate": 9.911539500253295e-05,
"loss": 0.0774,
"step": 3210
},
{
"grad_norm": 0.42652618885040283,
"learning_rate": 9.910504348769118e-05,
"loss": 0.072,
"step": 3220
},
{
"grad_norm": 0.5053730010986328,
"learning_rate": 9.909463230559127e-05,
"loss": 0.074,
"step": 3230
},
{
"grad_norm": 0.42999640107154846,
"learning_rate": 9.908416146888376e-05,
"loss": 0.0726,
"step": 3240
},
{
"grad_norm": 0.3914642333984375,
"learning_rate": 9.907363099029175e-05,
"loss": 0.0719,
"step": 3250
},
{
"grad_norm": 0.4248952865600586,
"learning_rate": 9.906304088261073e-05,
"loss": 0.0873,
"step": 3260
},
{
"grad_norm": 0.4257753789424896,
"learning_rate": 9.905239115870872e-05,
"loss": 0.0805,
"step": 3270
},
{
"grad_norm": 0.3652375638484955,
"learning_rate": 9.90416818315261e-05,
"loss": 0.0792,
"step": 3280
},
{
"grad_norm": 0.5478135347366333,
"learning_rate": 9.903091291407573e-05,
"loss": 0.0793,
"step": 3290
},
{
"grad_norm": 0.4161733090877533,
"learning_rate": 9.902008441944286e-05,
"loss": 0.0698,
"step": 3300
},
{
"grad_norm": 0.39713406562805176,
"learning_rate": 9.900919636078512e-05,
"loss": 0.0746,
"step": 3310
},
{
"grad_norm": 0.4765084683895111,
"learning_rate": 9.899824875133255e-05,
"loss": 0.0699,
"step": 3320
},
{
"grad_norm": 0.3772863447666168,
"learning_rate": 9.898724160438749e-05,
"loss": 0.0774,
"step": 3330
},
{
"grad_norm": 0.4268409013748169,
"learning_rate": 9.89761749333247e-05,
"loss": 0.0675,
"step": 3340
},
{
"grad_norm": 0.4206852614879608,
"learning_rate": 9.896504875159122e-05,
"loss": 0.0782,
"step": 3350
},
{
"grad_norm": 0.3789820373058319,
"learning_rate": 9.89538630727064e-05,
"loss": 0.0682,
"step": 3360
},
{
"grad_norm": 0.5044154524803162,
"learning_rate": 9.894261791026189e-05,
"loss": 0.0808,
"step": 3370
},
{
"grad_norm": 0.5173647403717041,
"learning_rate": 9.893131327792165e-05,
"loss": 0.0826,
"step": 3380
},
{
"grad_norm": 0.4938333034515381,
"learning_rate": 9.891994918942182e-05,
"loss": 0.0837,
"step": 3390
},
{
"grad_norm": 0.4987143874168396,
"learning_rate": 9.890852565857092e-05,
"loss": 0.0902,
"step": 3400
},
{
"grad_norm": 0.483456015586853,
"learning_rate": 9.889704269924954e-05,
"loss": 0.078,
"step": 3410
},
{
"grad_norm": 0.5119736194610596,
"learning_rate": 9.888550032541059e-05,
"loss": 0.0814,
"step": 3420
},
{
"grad_norm": 0.3884252607822418,
"learning_rate": 9.887389855107916e-05,
"loss": 0.0765,
"step": 3430
},
{
"grad_norm": 0.5105825662612915,
"learning_rate": 9.886223739035248e-05,
"loss": 0.0778,
"step": 3440
},
{
"grad_norm": 0.4057926535606384,
"learning_rate": 9.885051685739997e-05,
"loss": 0.0693,
"step": 3450
},
{
"grad_norm": 0.426470011472702,
"learning_rate": 9.883873696646316e-05,
"loss": 0.0767,
"step": 3460
},
{
"grad_norm": 0.3430192768573761,
"learning_rate": 9.882689773185575e-05,
"loss": 0.0702,
"step": 3470
},
{
"grad_norm": 0.479816198348999,
"learning_rate": 9.881499916796353e-05,
"loss": 0.0724,
"step": 3480
},
{
"grad_norm": 0.4488897919654846,
"learning_rate": 9.880304128924434e-05,
"loss": 0.0716,
"step": 3490
},
{
"grad_norm": 0.45463094115257263,
"learning_rate": 9.879102411022817e-05,
"loss": 0.0842,
"step": 3500
},
{
"grad_norm": 0.37502074241638184,
"learning_rate": 9.877894764551703e-05,
"loss": 0.0747,
"step": 3510
},
{
"grad_norm": 0.5187013745307922,
"learning_rate": 9.876681190978494e-05,
"loss": 0.0765,
"step": 3520
},
{
"grad_norm": 0.4604604244232178,
"learning_rate": 9.875461691777797e-05,
"loss": 0.0766,
"step": 3530
},
{
"grad_norm": 0.3916610777378082,
"learning_rate": 9.874236268431417e-05,
"loss": 0.0836,
"step": 3540
},
{
"grad_norm": 0.32828935980796814,
"learning_rate": 9.873004922428361e-05,
"loss": 0.0804,
"step": 3550
},
{
"grad_norm": 0.46902748942375183,
"learning_rate": 9.871767655264829e-05,
"loss": 0.0768,
"step": 3560
},
{
"grad_norm": 0.42918306589126587,
"learning_rate": 9.87052446844422e-05,
"loss": 0.0833,
"step": 3570
},
{
"grad_norm": 0.42306581139564514,
"learning_rate": 9.869275363477122e-05,
"loss": 0.0749,
"step": 3580
},
{
"grad_norm": 0.38532283902168274,
"learning_rate": 9.868020341881312e-05,
"loss": 0.0734,
"step": 3590
},
{
"grad_norm": 0.31471458077430725,
"learning_rate": 9.866759405181765e-05,
"loss": 0.0752,
"step": 3600
},
{
"grad_norm": 0.41112542152404785,
"learning_rate": 9.865492554910633e-05,
"loss": 0.0719,
"step": 3610
},
{
"grad_norm": 0.46945077180862427,
"learning_rate": 9.864219792607262e-05,
"loss": 0.0764,
"step": 3620
},
{
"grad_norm": 0.33219555020332336,
"learning_rate": 9.862941119818177e-05,
"loss": 0.0744,
"step": 3630
},
{
"grad_norm": 0.45927107334136963,
"learning_rate": 9.861656538097086e-05,
"loss": 0.0792,
"step": 3640
},
{
"grad_norm": 0.37848377227783203,
"learning_rate": 9.860366049004877e-05,
"loss": 0.0806,
"step": 3650
},
{
"grad_norm": 0.4966573119163513,
"learning_rate": 9.859069654109615e-05,
"loss": 0.0668,
"step": 3660
},
{
"grad_norm": 0.47644102573394775,
"learning_rate": 9.857767354986545e-05,
"loss": 0.0791,
"step": 3670
},
{
"grad_norm": 0.3492504954338074,
"learning_rate": 9.856459153218078e-05,
"loss": 0.076,
"step": 3680
},
{
"grad_norm": 0.42598026990890503,
"learning_rate": 9.855145050393808e-05,
"loss": 0.0782,
"step": 3690
},
{
"grad_norm": 0.3982851505279541,
"learning_rate": 9.85382504811049e-05,
"loss": 0.0728,
"step": 3700
},
{
"grad_norm": 0.36638298630714417,
"learning_rate": 9.852499147972054e-05,
"loss": 0.0722,
"step": 3710
},
{
"grad_norm": 0.4078195095062256,
"learning_rate": 9.851167351589592e-05,
"loss": 0.0773,
"step": 3720
},
{
"grad_norm": 0.4698351323604584,
"learning_rate": 9.849829660581363e-05,
"loss": 0.0709,
"step": 3730
},
{
"grad_norm": 0.3940131366252899,
"learning_rate": 9.848486076572787e-05,
"loss": 0.0713,
"step": 3740
},
{
"grad_norm": 0.5406202077865601,
"learning_rate": 9.847136601196446e-05,
"loss": 0.0784,
"step": 3750
},
{
"grad_norm": 0.4469393193721771,
"learning_rate": 9.845781236092078e-05,
"loss": 0.0747,
"step": 3760
},
{
"grad_norm": 0.5607311129570007,
"learning_rate": 9.844419982906583e-05,
"loss": 0.0797,
"step": 3770
},
{
"grad_norm": 0.4935024082660675,
"learning_rate": 9.843052843294008e-05,
"loss": 0.0793,
"step": 3780
},
{
"grad_norm": 0.40715348720550537,
"learning_rate": 9.841679818915559e-05,
"loss": 0.0698,
"step": 3790
},
{
"grad_norm": 0.39498183131217957,
"learning_rate": 9.840300911439591e-05,
"loss": 0.0799,
"step": 3800
},
{
"grad_norm": 0.350668728351593,
"learning_rate": 9.838916122541603e-05,
"loss": 0.0759,
"step": 3810
},
{
"grad_norm": 0.41015708446502686,
"learning_rate": 9.837525453904246e-05,
"loss": 0.0786,
"step": 3820
},
{
"grad_norm": 0.2985064685344696,
"learning_rate": 9.836128907217314e-05,
"loss": 0.0752,
"step": 3830
},
{
"grad_norm": 0.458590567111969,
"learning_rate": 9.834726484177743e-05,
"loss": 0.0724,
"step": 3840
},
{
"grad_norm": 0.540894091129303,
"learning_rate": 9.833318186489609e-05,
"loss": 0.0709,
"step": 3850
},
{
"grad_norm": 0.4432045519351959,
"learning_rate": 9.831904015864126e-05,
"loss": 0.086,
"step": 3860
},
{
"grad_norm": 0.43406978249549866,
"learning_rate": 9.830483974019645e-05,
"loss": 0.0776,
"step": 3870
},
{
"grad_norm": 0.43287715315818787,
"learning_rate": 9.82905806268165e-05,
"loss": 0.0774,
"step": 3880
},
{
"grad_norm": 0.4447200894355774,
"learning_rate": 9.82762628358276e-05,
"loss": 0.0699,
"step": 3890
},
{
"grad_norm": 0.4370230734348297,
"learning_rate": 9.826188638462718e-05,
"loss": 0.0736,
"step": 3900
},
{
"grad_norm": 0.4798665940761566,
"learning_rate": 9.824745129068402e-05,
"loss": 0.0732,
"step": 3910
},
{
"grad_norm": 0.428438276052475,
"learning_rate": 9.82329575715381e-05,
"loss": 0.072,
"step": 3920
},
{
"grad_norm": 0.3836239278316498,
"learning_rate": 9.821840524480066e-05,
"loss": 0.0696,
"step": 3930
},
{
"grad_norm": 0.4632021188735962,
"learning_rate": 9.820379432815414e-05,
"loss": 0.068,
"step": 3940
},
{
"grad_norm": 0.4033900499343872,
"learning_rate": 9.81891248393522e-05,
"loss": 0.0703,
"step": 3950
},
{
"grad_norm": 0.43214908242225647,
"learning_rate": 9.817439679621963e-05,
"loss": 0.0652,
"step": 3960
},
{
"grad_norm": 0.4888061583042145,
"learning_rate": 9.815961021665243e-05,
"loss": 0.0712,
"step": 3970
},
{
"grad_norm": 0.4061157703399658,
"learning_rate": 9.814476511861763e-05,
"loss": 0.066,
"step": 3980
},
{
"grad_norm": 0.43131521344184875,
"learning_rate": 9.812986152015348e-05,
"loss": 0.0749,
"step": 3990
},
{
"grad_norm": 0.3810648024082184,
"learning_rate": 9.811489943936922e-05,
"loss": 0.0765,
"step": 4000
},
{
"grad_norm": 0.4778054654598236,
"learning_rate": 9.809987889444522e-05,
"loss": 0.0776,
"step": 4010
},
{
"grad_norm": 0.38132381439208984,
"learning_rate": 9.808479990363282e-05,
"loss": 0.0731,
"step": 4020
},
{
"grad_norm": 0.42138880491256714,
"learning_rate": 9.806966248525445e-05,
"loss": 0.0827,
"step": 4030
},
{
"grad_norm": 0.346861869096756,
"learning_rate": 9.805446665770348e-05,
"loss": 0.0735,
"step": 4040
},
{
"grad_norm": 0.4531242847442627,
"learning_rate": 9.803921243944429e-05,
"loss": 0.0739,
"step": 4050
},
{
"grad_norm": 0.4656151831150055,
"learning_rate": 9.802389984901218e-05,
"loss": 0.0817,
"step": 4060
},
{
"grad_norm": 0.3624925911426544,
"learning_rate": 9.80085289050134e-05,
"loss": 0.0739,
"step": 4070
},
{
"grad_norm": 0.4334985017776489,
"learning_rate": 9.799309962612508e-05,
"loss": 0.068,
"step": 4080
},
{
"grad_norm": 0.3184892237186432,
"learning_rate": 9.797761203109527e-05,
"loss": 0.0733,
"step": 4090
},
{
"grad_norm": 0.33551061153411865,
"learning_rate": 9.796206613874283e-05,
"loss": 0.0736,
"step": 4100
},
{
"grad_norm": 0.3884878158569336,
"learning_rate": 9.794646196795754e-05,
"loss": 0.0724,
"step": 4110
},
{
"grad_norm": 0.4005712866783142,
"learning_rate": 9.793079953769987e-05,
"loss": 0.0703,
"step": 4120
},
{
"grad_norm": 0.4162188768386841,
"learning_rate": 9.79150788670012e-05,
"loss": 0.073,
"step": 4130
},
{
"grad_norm": 0.41469624638557434,
"learning_rate": 9.78992999749636e-05,
"loss": 0.0724,
"step": 4140
},
{
"grad_norm": 0.38397979736328125,
"learning_rate": 9.788346288075994e-05,
"loss": 0.0771,
"step": 4150
},
{
"grad_norm": 0.44882339239120483,
"learning_rate": 9.786756760363373e-05,
"loss": 0.0716,
"step": 4160
},
{
"grad_norm": 0.45670127868652344,
"learning_rate": 9.78516141628993e-05,
"loss": 0.0753,
"step": 4170
},
{
"grad_norm": 0.3398466110229492,
"learning_rate": 9.783560257794154e-05,
"loss": 0.0767,
"step": 4180
},
{
"grad_norm": 0.3378835618495941,
"learning_rate": 9.781953286821603e-05,
"loss": 0.072,
"step": 4190
},
{
"grad_norm": 0.3285226821899414,
"learning_rate": 9.780340505324901e-05,
"loss": 0.0699,
"step": 4200
},
{
"grad_norm": 0.4315303862094879,
"learning_rate": 9.778721915263727e-05,
"loss": 0.07,
"step": 4210
},
{
"grad_norm": 0.35673606395721436,
"learning_rate": 9.777097518604824e-05,
"loss": 0.0678,
"step": 4220
},
{
"grad_norm": 0.38063254952430725,
"learning_rate": 9.775467317321984e-05,
"loss": 0.0724,
"step": 4230
},
{
"grad_norm": 0.4448537230491638,
"learning_rate": 9.773831313396055e-05,
"loss": 0.0791,
"step": 4240
},
{
"grad_norm": 0.40598979592323303,
"learning_rate": 9.77218950881494e-05,
"loss": 0.0793,
"step": 4250
},
{
"grad_norm": 0.4415881037712097,
"learning_rate": 9.770541905573583e-05,
"loss": 0.0815,
"step": 4260
},
{
"grad_norm": 0.4739032983779907,
"learning_rate": 9.768888505673976e-05,
"loss": 0.0763,
"step": 4270
},
{
"grad_norm": 0.41163259744644165,
"learning_rate": 9.767229311125162e-05,
"loss": 0.0717,
"step": 4280
},
{
"grad_norm": 0.42694783210754395,
"learning_rate": 9.765564323943211e-05,
"loss": 0.0703,
"step": 4290
},
{
"grad_norm": 0.36140456795692444,
"learning_rate": 9.763893546151244e-05,
"loss": 0.0731,
"step": 4300
},
{
"grad_norm": 0.343685120344162,
"learning_rate": 9.762216979779412e-05,
"loss": 0.0629,
"step": 4310
},
{
"grad_norm": 0.3439638912677765,
"learning_rate": 9.760534626864902e-05,
"loss": 0.0762,
"step": 4320
},
{
"grad_norm": 0.3824269473552704,
"learning_rate": 9.758846489451931e-05,
"loss": 0.078,
"step": 4330
},
{
"grad_norm": 0.37309131026268005,
"learning_rate": 9.757152569591748e-05,
"loss": 0.0742,
"step": 4340
},
{
"grad_norm": 0.4569232165813446,
"learning_rate": 9.75545286934262e-05,
"loss": 0.0717,
"step": 4350
},
{
"grad_norm": 0.34614187479019165,
"learning_rate": 9.753747390769847e-05,
"loss": 0.068,
"step": 4360
},
{
"grad_norm": 0.4518280327320099,
"learning_rate": 9.752036135945744e-05,
"loss": 0.0792,
"step": 4370
},
{
"grad_norm": 0.4086059331893921,
"learning_rate": 9.750319106949649e-05,
"loss": 0.072,
"step": 4380
},
{
"grad_norm": 0.41269636154174805,
"learning_rate": 9.748596305867913e-05,
"loss": 0.0825,
"step": 4390
},
{
"grad_norm": 0.4099424183368683,
"learning_rate": 9.746867734793903e-05,
"loss": 0.0838,
"step": 4400
},
{
"grad_norm": 0.43975910544395447,
"learning_rate": 9.745133395827993e-05,
"loss": 0.0794,
"step": 4410
},
{
"grad_norm": 0.4421364665031433,
"learning_rate": 9.743393291077572e-05,
"loss": 0.0777,
"step": 4420
},
{
"grad_norm": 0.3365815281867981,
"learning_rate": 9.741647422657028e-05,
"loss": 0.0762,
"step": 4430
},
{
"grad_norm": 0.4021996855735779,
"learning_rate": 9.739895792687758e-05,
"loss": 0.0783,
"step": 4440
},
{
"grad_norm": 0.3530466854572296,
"learning_rate": 9.738138403298157e-05,
"loss": 0.0743,
"step": 4450
},
{
"grad_norm": 0.3247053623199463,
"learning_rate": 9.736375256623619e-05,
"loss": 0.0753,
"step": 4460
},
{
"grad_norm": 0.4123826324939728,
"learning_rate": 9.734606354806533e-05,
"loss": 0.0698,
"step": 4470
},
{
"grad_norm": 0.3126823902130127,
"learning_rate": 9.73283169999628e-05,
"loss": 0.0704,
"step": 4480
},
{
"grad_norm": 0.3824714720249176,
"learning_rate": 9.731051294349238e-05,
"loss": 0.0745,
"step": 4490
},
{
"grad_norm": 0.4448513984680176,
"learning_rate": 9.729265140028762e-05,
"loss": 0.0734,
"step": 4500
},
{
"grad_norm": 0.3782031834125519,
"learning_rate": 9.727473239205201e-05,
"loss": 0.079,
"step": 4510
},
{
"grad_norm": 0.42318156361579895,
"learning_rate": 9.725675594055883e-05,
"loss": 0.0735,
"step": 4520
},
{
"grad_norm": 0.3395531177520752,
"learning_rate": 9.723872206765116e-05,
"loss": 0.0711,
"step": 4530
},
{
"grad_norm": 0.3714655935764313,
"learning_rate": 9.722063079524185e-05,
"loss": 0.0683,
"step": 4540
},
{
"grad_norm": 0.37211620807647705,
"learning_rate": 9.720248214531351e-05,
"loss": 0.0648,
"step": 4550
},
{
"grad_norm": 0.2871655821800232,
"learning_rate": 9.718427613991848e-05,
"loss": 0.0641,
"step": 4560
},
{
"grad_norm": 0.3858413100242615,
"learning_rate": 9.716601280117873e-05,
"loss": 0.0715,
"step": 4570
},
{
"grad_norm": 0.4011073410511017,
"learning_rate": 9.714769215128596e-05,
"loss": 0.0657,
"step": 4580
},
{
"grad_norm": 0.40221312642097473,
"learning_rate": 9.712931421250152e-05,
"loss": 0.0681,
"step": 4590
},
{
"grad_norm": 0.47440239787101746,
"learning_rate": 9.711087900715627e-05,
"loss": 0.0689,
"step": 4600
},
{
"grad_norm": 0.4292180836200714,
"learning_rate": 9.709238655765078e-05,
"loss": 0.0678,
"step": 4610
},
{
"grad_norm": 0.3444955050945282,
"learning_rate": 9.707383688645511e-05,
"loss": 0.0646,
"step": 4620
},
{
"grad_norm": 0.45068204402923584,
"learning_rate": 9.705523001610883e-05,
"loss": 0.0609,
"step": 4630
},
{
"grad_norm": 0.4344935119152069,
"learning_rate": 9.703656596922107e-05,
"loss": 0.0647,
"step": 4640
},
{
"grad_norm": 0.4532392919063568,
"learning_rate": 9.70178447684704e-05,
"loss": 0.0718,
"step": 4650
},
{
"grad_norm": 0.5870494842529297,
"learning_rate": 9.699906643660483e-05,
"loss": 0.0728,
"step": 4660
},
{
"grad_norm": 0.42196017503738403,
"learning_rate": 9.698023099644185e-05,
"loss": 0.0662,
"step": 4670
},
{
"grad_norm": 0.35073211789131165,
"learning_rate": 9.696133847086823e-05,
"loss": 0.0574,
"step": 4680
},
{
"grad_norm": 0.38236382603645325,
"learning_rate": 9.694238888284022e-05,
"loss": 0.0655,
"step": 4690
},
{
"grad_norm": 0.4481329619884491,
"learning_rate": 9.692338225538333e-05,
"loss": 0.075,
"step": 4700
},
{
"grad_norm": 0.48714524507522583,
"learning_rate": 9.690431861159241e-05,
"loss": 0.0693,
"step": 4710
},
{
"grad_norm": 0.478758841753006,
"learning_rate": 9.688519797463161e-05,
"loss": 0.0794,
"step": 4720
},
{
"grad_norm": 0.3997093439102173,
"learning_rate": 9.686602036773426e-05,
"loss": 0.0761,
"step": 4730
},
{
"grad_norm": 0.48385292291641235,
"learning_rate": 9.684678581420302e-05,
"loss": 0.074,
"step": 4740
},
{
"grad_norm": 0.4378373324871063,
"learning_rate": 9.682749433740962e-05,
"loss": 0.0764,
"step": 4750
},
{
"grad_norm": 0.33325397968292236,
"learning_rate": 9.680814596079507e-05,
"loss": 0.0734,
"step": 4760
},
{
"grad_norm": 0.4322454333305359,
"learning_rate": 9.678874070786945e-05,
"loss": 0.0712,
"step": 4770
},
{
"grad_norm": 0.3594813942909241,
"learning_rate": 9.676927860221199e-05,
"loss": 0.0696,
"step": 4780
},
{
"grad_norm": 0.3405745327472687,
"learning_rate": 9.674975966747097e-05,
"loss": 0.0693,
"step": 4790
},
{
"grad_norm": 0.3336330056190491,
"learning_rate": 9.673018392736374e-05,
"loss": 0.0694,
"step": 4800
},
{
"grad_norm": 0.3852177858352661,
"learning_rate": 9.671055140567667e-05,
"loss": 0.0651,
"step": 4810
},
{
"grad_norm": 0.45991280674934387,
"learning_rate": 9.669086212626511e-05,
"loss": 0.0701,
"step": 4820
},
{
"grad_norm": 0.5722672939300537,
"learning_rate": 9.667111611305341e-05,
"loss": 0.0744,
"step": 4830
},
{
"grad_norm": 0.2540041506290436,
"learning_rate": 9.665131339003486e-05,
"loss": 0.0804,
"step": 4840
},
{
"grad_norm": 0.30723729729652405,
"learning_rate": 9.663145398127158e-05,
"loss": 0.067,
"step": 4850
},
{
"grad_norm": 0.4877570867538452,
"learning_rate": 9.661153791089467e-05,
"loss": 0.0669,
"step": 4860
},
{
"grad_norm": 0.32467806339263916,
"learning_rate": 9.659156520310402e-05,
"loss": 0.0658,
"step": 4870
},
{
"grad_norm": 0.40045493841171265,
"learning_rate": 9.657153588216835e-05,
"loss": 0.0731,
"step": 4880
},
{
"grad_norm": 0.34688499569892883,
"learning_rate": 9.655144997242516e-05,
"loss": 0.0723,
"step": 4890
},
{
"grad_norm": 0.43813446164131165,
"learning_rate": 9.653130749828075e-05,
"loss": 0.075,
"step": 4900
},
{
"grad_norm": 0.3956025540828705,
"learning_rate": 9.65111084842101e-05,
"loss": 0.0705,
"step": 4910
},
{
"grad_norm": 0.4464576840400696,
"learning_rate": 9.649085295475695e-05,
"loss": 0.0711,
"step": 4920
},
{
"grad_norm": 0.42245548963546753,
"learning_rate": 9.647054093453365e-05,
"loss": 0.0796,
"step": 4930
},
{
"grad_norm": 0.40897077322006226,
"learning_rate": 9.645017244822123e-05,
"loss": 0.0739,
"step": 4940
},
{
"grad_norm": 0.4206048548221588,
"learning_rate": 9.642974752056931e-05,
"loss": 0.0736,
"step": 4950
},
{
"grad_norm": 0.4242159426212311,
"learning_rate": 9.640926617639613e-05,
"loss": 0.0713,
"step": 4960
},
{
"grad_norm": 0.35714834928512573,
"learning_rate": 9.638872844058843e-05,
"loss": 0.0817,
"step": 4970
},
{
"grad_norm": 0.44501522183418274,
"learning_rate": 9.63681343381015e-05,
"loss": 0.0817,
"step": 4980
},
{
"grad_norm": 0.42664507031440735,
"learning_rate": 9.634748389395914e-05,
"loss": 0.0798,
"step": 4990
},
{
"grad_norm": 0.42096149921417236,
"learning_rate": 9.632677713325353e-05,
"loss": 0.0741,
"step": 5000
},
{
"grad_norm": 0.29973748326301575,
"learning_rate": 9.63060140811454e-05,
"loss": 0.0752,
"step": 5010
},
{
"grad_norm": 0.3298620283603668,
"learning_rate": 9.628519476286379e-05,
"loss": 0.074,
"step": 5020
},
{
"grad_norm": 0.38388893008232117,
"learning_rate": 9.626431920370612e-05,
"loss": 0.0698,
"step": 5030
},
{
"grad_norm": 0.33063802123069763,
"learning_rate": 9.624338742903819e-05,
"loss": 0.068,
"step": 5040
},
{
"grad_norm": 0.4108198285102844,
"learning_rate": 9.622239946429406e-05,
"loss": 0.069,
"step": 5050
},
{
"grad_norm": 0.3624817728996277,
"learning_rate": 9.620135533497609e-05,
"loss": 0.0673,
"step": 5060
},
{
"grad_norm": 0.367994487285614,
"learning_rate": 9.61802550666549e-05,
"loss": 0.0565,
"step": 5070
},
{
"grad_norm": 0.41292113065719604,
"learning_rate": 9.615909868496928e-05,
"loss": 0.0684,
"step": 5080
},
{
"grad_norm": 0.4170876741409302,
"learning_rate": 9.613788621562622e-05,
"loss": 0.0667,
"step": 5090
},
{
"grad_norm": 0.2896060049533844,
"learning_rate": 9.611661768440093e-05,
"loss": 0.0675,
"step": 5100
},
{
"grad_norm": 0.43625274300575256,
"learning_rate": 9.609529311713661e-05,
"loss": 0.0681,
"step": 5110
},
{
"grad_norm": 0.3629418909549713,
"learning_rate": 9.607391253974466e-05,
"loss": 0.0726,
"step": 5120
},
{
"grad_norm": 0.3434334695339203,
"learning_rate": 9.605247597820448e-05,
"loss": 0.0584,
"step": 5130
},
{
"grad_norm": 0.29164981842041016,
"learning_rate": 9.603098345856354e-05,
"loss": 0.064,
"step": 5140
},
{
"grad_norm": 0.4003913998603821,
"learning_rate": 9.600943500693725e-05,
"loss": 0.0633,
"step": 5150
},
{
"grad_norm": 0.34255415201187134,
"learning_rate": 9.598783064950902e-05,
"loss": 0.0657,
"step": 5160
},
{
"grad_norm": 0.4544960558414459,
"learning_rate": 9.596617041253018e-05,
"loss": 0.0567,
"step": 5170
},
{
"grad_norm": 0.43727031350135803,
"learning_rate": 9.594445432231996e-05,
"loss": 0.0605,
"step": 5180
},
{
"grad_norm": 0.3787287175655365,
"learning_rate": 9.592268240526547e-05,
"loss": 0.0717,
"step": 5190
},
{
"grad_norm": 0.3631057143211365,
"learning_rate": 9.590085468782162e-05,
"loss": 0.0739,
"step": 5200
},
{
"grad_norm": 0.4738662540912628,
"learning_rate": 9.587897119651116e-05,
"loss": 0.0712,
"step": 5210
},
{
"grad_norm": 0.4778849482536316,
"learning_rate": 9.585703195792459e-05,
"loss": 0.0693,
"step": 5220
},
{
"grad_norm": 0.40974536538124084,
"learning_rate": 9.583503699872016e-05,
"loss": 0.0684,
"step": 5230
},
{
"grad_norm": 0.35605180263519287,
"learning_rate": 9.581298634562381e-05,
"loss": 0.0664,
"step": 5240
},
{
"grad_norm": 0.3348657786846161,
"learning_rate": 9.579088002542917e-05,
"loss": 0.0655,
"step": 5250
},
{
"grad_norm": 0.32003048062324524,
"learning_rate": 9.57687180649975e-05,
"loss": 0.0697,
"step": 5260
},
{
"grad_norm": 0.3447810411453247,
"learning_rate": 9.574650049125768e-05,
"loss": 0.071,
"step": 5270
},
{
"grad_norm": 0.31894874572753906,
"learning_rate": 9.572422733120614e-05,
"loss": 0.0676,
"step": 5280
},
{
"grad_norm": 0.3575884997844696,
"learning_rate": 9.570189861190689e-05,
"loss": 0.071,
"step": 5290
},
{
"grad_norm": 0.3461254835128784,
"learning_rate": 9.56795143604914e-05,
"loss": 0.07,
"step": 5300
},
{
"grad_norm": 0.3771500885486603,
"learning_rate": 9.565707460415869e-05,
"loss": 0.0681,
"step": 5310
},
{
"grad_norm": 0.3387458324432373,
"learning_rate": 9.563457937017515e-05,
"loss": 0.0711,
"step": 5320
},
{
"grad_norm": 0.38215750455856323,
"learning_rate": 9.56120286858746e-05,
"loss": 0.0599,
"step": 5330
},
{
"grad_norm": 0.3428252637386322,
"learning_rate": 9.558942257865829e-05,
"loss": 0.0667,
"step": 5340
},
{
"grad_norm": 0.3720010817050934,
"learning_rate": 9.556676107599472e-05,
"loss": 0.0655,
"step": 5350
},
{
"grad_norm": 0.39170342683792114,
"learning_rate": 9.554404420541978e-05,
"loss": 0.0724,
"step": 5360
},
{
"grad_norm": 0.48484688997268677,
"learning_rate": 9.55212719945366e-05,
"loss": 0.0696,
"step": 5370
},
{
"grad_norm": 0.4146574139595032,
"learning_rate": 9.549844447101559e-05,
"loss": 0.066,
"step": 5380
},
{
"grad_norm": 0.4974808394908905,
"learning_rate": 9.547556166259433e-05,
"loss": 0.0622,
"step": 5390
},
{
"grad_norm": 0.5124579668045044,
"learning_rate": 9.545262359707756e-05,
"loss": 0.068,
"step": 5400
},
{
"grad_norm": 0.4220755696296692,
"learning_rate": 9.542963030233724e-05,
"loss": 0.0667,
"step": 5410
},
{
"grad_norm": 0.5311270356178284,
"learning_rate": 9.540658180631237e-05,
"loss": 0.0627,
"step": 5420
},
{
"grad_norm": 0.571665346622467,
"learning_rate": 9.538347813700904e-05,
"loss": 0.0655,
"step": 5430
},
{
"grad_norm": 0.36707204580307007,
"learning_rate": 9.536031932250036e-05,
"loss": 0.0652,
"step": 5440
},
{
"grad_norm": 0.5425820350646973,
"learning_rate": 9.533710539092653e-05,
"loss": 0.0708,
"step": 5450
},
{
"grad_norm": 0.41234076023101807,
"learning_rate": 9.531383637049464e-05,
"loss": 0.0626,
"step": 5460
},
{
"grad_norm": 0.3595837950706482,
"learning_rate": 9.529051228947875e-05,
"loss": 0.0631,
"step": 5470
},
{
"grad_norm": 0.33364495635032654,
"learning_rate": 9.52671331762198e-05,
"loss": 0.0598,
"step": 5480
},
{
"grad_norm": 0.3618021607398987,
"learning_rate": 9.524369905912565e-05,
"loss": 0.065,
"step": 5490
},
{
"grad_norm": 0.4057350158691406,
"learning_rate": 9.522020996667092e-05,
"loss": 0.0656,
"step": 5500
},
{
"grad_norm": 0.45496687293052673,
"learning_rate": 9.519666592739709e-05,
"loss": 0.073,
"step": 5510
},
{
"grad_norm": 0.32082778215408325,
"learning_rate": 9.517306696991241e-05,
"loss": 0.0635,
"step": 5520
},
{
"grad_norm": 0.3926721215248108,
"learning_rate": 9.51494131228918e-05,
"loss": 0.0757,
"step": 5530
},
{
"grad_norm": 0.4289970397949219,
"learning_rate": 9.512570441507695e-05,
"loss": 0.0685,
"step": 5540
},
{
"grad_norm": 0.3503778874874115,
"learning_rate": 9.510194087527615e-05,
"loss": 0.0675,
"step": 5550
},
{
"grad_norm": 0.4926631450653076,
"learning_rate": 9.507812253236435e-05,
"loss": 0.0711,
"step": 5560
},
{
"grad_norm": 0.394659161567688,
"learning_rate": 9.505424941528309e-05,
"loss": 0.0651,
"step": 5570
},
{
"grad_norm": 0.3866186738014221,
"learning_rate": 9.503032155304046e-05,
"loss": 0.0686,
"step": 5580
},
{
"grad_norm": 0.4077188968658447,
"learning_rate": 9.500633897471106e-05,
"loss": 0.0652,
"step": 5590
},
{
"grad_norm": 0.38112786412239075,
"learning_rate": 9.498230170943596e-05,
"loss": 0.0667,
"step": 5600
},
{
"grad_norm": 0.4371899962425232,
"learning_rate": 9.495820978642275e-05,
"loss": 0.0673,
"step": 5610
},
{
"grad_norm": 0.3777690529823303,
"learning_rate": 9.493406323494535e-05,
"loss": 0.0641,
"step": 5620
},
{
"grad_norm": 0.41798272728919983,
"learning_rate": 9.490986208434413e-05,
"loss": 0.0603,
"step": 5630
},
{
"grad_norm": 0.4356377422809601,
"learning_rate": 9.488560636402577e-05,
"loss": 0.0755,
"step": 5640
},
{
"grad_norm": 0.30686861276626587,
"learning_rate": 9.486129610346321e-05,
"loss": 0.0648,
"step": 5650
},
{
"grad_norm": 0.34002140164375305,
"learning_rate": 9.483693133219576e-05,
"loss": 0.0724,
"step": 5660
},
{
"grad_norm": 0.4165436029434204,
"learning_rate": 9.481251207982888e-05,
"loss": 0.0673,
"step": 5670
},
{
"grad_norm": 0.347219854593277,
"learning_rate": 9.47880383760343e-05,
"loss": 0.0691,
"step": 5680
},
{
"grad_norm": 0.3736044466495514,
"learning_rate": 9.476351025054983e-05,
"loss": 0.0761,
"step": 5690
},
{
"grad_norm": 0.4664802551269531,
"learning_rate": 9.473892773317952e-05,
"loss": 0.0679,
"step": 5700
},
{
"grad_norm": 0.36449456214904785,
"learning_rate": 9.471429085379338e-05,
"loss": 0.0729,
"step": 5710
},
{
"grad_norm": 0.46617311239242554,
"learning_rate": 9.468959964232757e-05,
"loss": 0.0775,
"step": 5720
},
{
"grad_norm": 0.32528772950172424,
"learning_rate": 9.466485412878425e-05,
"loss": 0.0733,
"step": 5730
},
{
"grad_norm": 0.3369789123535156,
"learning_rate": 9.464005434323154e-05,
"loss": 0.0645,
"step": 5740
},
{
"grad_norm": 0.3734840154647827,
"learning_rate": 9.461520031580352e-05,
"loss": 0.0667,
"step": 5750
},
{
"grad_norm": 0.33726829290390015,
"learning_rate": 9.459029207670019e-05,
"loss": 0.0671,
"step": 5760
},
{
"grad_norm": 0.3473425507545471,
"learning_rate": 9.456532965618737e-05,
"loss": 0.0653,
"step": 5770
},
{
"grad_norm": 0.32953932881355286,
"learning_rate": 9.454031308459681e-05,
"loss": 0.0692,
"step": 5780
},
{
"grad_norm": 0.28829506039619446,
"learning_rate": 9.451524239232595e-05,
"loss": 0.0641,
"step": 5790
},
{
"grad_norm": 0.3854352533817291,
"learning_rate": 9.449011760983809e-05,
"loss": 0.0641,
"step": 5800
},
{
"grad_norm": 0.4310356676578522,
"learning_rate": 9.446493876766218e-05,
"loss": 0.063,
"step": 5810
},
{
"grad_norm": 0.30458804965019226,
"learning_rate": 9.44397058963929e-05,
"loss": 0.059,
"step": 5820
},
{
"grad_norm": 0.4184858798980713,
"learning_rate": 9.441441902669056e-05,
"loss": 0.0676,
"step": 5830
},
{
"grad_norm": 0.3030279278755188,
"learning_rate": 9.43890781892811e-05,
"loss": 0.0716,
"step": 5840
},
{
"grad_norm": 0.28139039874076843,
"learning_rate": 9.436368341495603e-05,
"loss": 0.0706,
"step": 5850
},
{
"grad_norm": 0.33913949131965637,
"learning_rate": 9.43382347345724e-05,
"loss": 0.071,
"step": 5860
},
{
"grad_norm": 0.3131044805049896,
"learning_rate": 9.431273217905273e-05,
"loss": 0.0713,
"step": 5870
},
{
"grad_norm": 0.4301745593547821,
"learning_rate": 9.428717577938504e-05,
"loss": 0.0788,
"step": 5880
},
{
"grad_norm": 0.41206052899360657,
"learning_rate": 9.426156556662276e-05,
"loss": 0.0684,
"step": 5890
},
{
"grad_norm": 0.4445531666278839,
"learning_rate": 9.423590157188474e-05,
"loss": 0.0772,
"step": 5900
},
{
"grad_norm": 0.33571889996528625,
"learning_rate": 9.421018382635513e-05,
"loss": 0.0774,
"step": 5910
},
{
"grad_norm": 0.3417772054672241,
"learning_rate": 9.418441236128343e-05,
"loss": 0.0725,
"step": 5920
},
{
"grad_norm": 0.2538796365261078,
"learning_rate": 9.41585872079844e-05,
"loss": 0.0685,
"step": 5930
},
{
"grad_norm": 0.3100067675113678,
"learning_rate": 9.413270839783802e-05,
"loss": 0.0717,
"step": 5940
},
{
"grad_norm": 0.2729007303714752,
"learning_rate": 9.41067759622895e-05,
"loss": 0.0753,
"step": 5950
},
{
"grad_norm": 0.37981194257736206,
"learning_rate": 9.408078993284917e-05,
"loss": 0.0741,
"step": 5960
},
{
"grad_norm": 0.34475141763687134,
"learning_rate": 9.405475034109254e-05,
"loss": 0.0725,
"step": 5970
},
{
"grad_norm": 0.42219117283821106,
"learning_rate": 9.402865721866015e-05,
"loss": 0.071,
"step": 5980
},
{
"grad_norm": 0.3952416777610779,
"learning_rate": 9.400251059725762e-05,
"loss": 0.0711,
"step": 5990
},
{
"grad_norm": 0.37309595942497253,
"learning_rate": 9.397631050865554e-05,
"loss": 0.0687,
"step": 6000
},
{
"grad_norm": 0.31540486216545105,
"learning_rate": 9.395005698468949e-05,
"loss": 0.0647,
"step": 6010
},
{
"grad_norm": 0.43652501702308655,
"learning_rate": 9.392375005726e-05,
"loss": 0.0645,
"step": 6020
},
{
"grad_norm": 0.3055417835712433,
"learning_rate": 9.389738975833242e-05,
"loss": 0.0634,
"step": 6030
},
{
"grad_norm": 0.32249680161476135,
"learning_rate": 9.387097611993707e-05,
"loss": 0.0647,
"step": 6040
},
{
"grad_norm": 0.3303893804550171,
"learning_rate": 9.384450917416894e-05,
"loss": 0.0582,
"step": 6050
},
{
"grad_norm": 0.3033483922481537,
"learning_rate": 9.381798895318792e-05,
"loss": 0.0587,
"step": 6060
},
{
"grad_norm": 0.40831151604652405,
"learning_rate": 9.379141548921853e-05,
"loss": 0.0555,
"step": 6070
},
{
"grad_norm": 0.3445760905742645,
"learning_rate": 9.376478881455009e-05,
"loss": 0.0641,
"step": 6080
},
{
"grad_norm": 0.3816492259502411,
"learning_rate": 9.373810896153647e-05,
"loss": 0.0642,
"step": 6090
},
{
"grad_norm": 0.3756420612335205,
"learning_rate": 9.371137596259623e-05,
"loss": 0.0561,
"step": 6100
},
{
"grad_norm": 0.3469025492668152,
"learning_rate": 9.368458985021248e-05,
"loss": 0.0608,
"step": 6110
},
{
"grad_norm": 0.2778720259666443,
"learning_rate": 9.365775065693287e-05,
"loss": 0.0645,
"step": 6120
},
{
"grad_norm": 0.3746732473373413,
"learning_rate": 9.363085841536957e-05,
"loss": 0.0666,
"step": 6130
},
{
"grad_norm": 0.32475602626800537,
"learning_rate": 9.360391315819917e-05,
"loss": 0.059,
"step": 6140
},
{
"grad_norm": 0.34035053849220276,
"learning_rate": 9.357691491816269e-05,
"loss": 0.0545,
"step": 6150
},
{
"grad_norm": 0.3032926917076111,
"learning_rate": 9.354986372806557e-05,
"loss": 0.0647,
"step": 6160
},
{
"grad_norm": 0.32066160440444946,
"learning_rate": 9.352275962077752e-05,
"loss": 0.0599,
"step": 6170
},
{
"grad_norm": 0.4139830470085144,
"learning_rate": 9.349560262923262e-05,
"loss": 0.0754,
"step": 6180
},
{
"grad_norm": 0.3689512312412262,
"learning_rate": 9.346839278642913e-05,
"loss": 0.0739,
"step": 6190
},
{
"grad_norm": 0.35650965571403503,
"learning_rate": 9.344113012542963e-05,
"loss": 0.0638,
"step": 6200
},
{
"grad_norm": 0.45089250802993774,
"learning_rate": 9.341381467936079e-05,
"loss": 0.0674,
"step": 6210
},
{
"grad_norm": 0.4204792380332947,
"learning_rate": 9.338644648141346e-05,
"loss": 0.0632,
"step": 6220
},
{
"grad_norm": 0.5111079216003418,
"learning_rate": 9.335902556484257e-05,
"loss": 0.0618,
"step": 6230
},
{
"grad_norm": 0.3371949791908264,
"learning_rate": 9.333155196296712e-05,
"loss": 0.0588,
"step": 6240
},
{
"grad_norm": 0.3017164170742035,
"learning_rate": 9.330402570917016e-05,
"loss": 0.0584,
"step": 6250
},
{
"grad_norm": 0.3659664988517761,
"learning_rate": 9.327644683689865e-05,
"loss": 0.0611,
"step": 6260
},
{
"grad_norm": 0.355772465467453,
"learning_rate": 9.324881537966354e-05,
"loss": 0.0532,
"step": 6270
},
{
"grad_norm": 0.2984813451766968,
"learning_rate": 9.322113137103964e-05,
"loss": 0.0583,
"step": 6280
},
{
"grad_norm": 0.3575005829334259,
"learning_rate": 9.319339484466564e-05,
"loss": 0.0614,
"step": 6290
},
{
"grad_norm": 0.40738996863365173,
"learning_rate": 9.316560583424403e-05,
"loss": 0.0632,
"step": 6300
},
{
"grad_norm": 0.3470759391784668,
"learning_rate": 9.313776437354108e-05,
"loss": 0.0562,
"step": 6310
},
{
"grad_norm": 0.2865963280200958,
"learning_rate": 9.31098704963868e-05,
"loss": 0.0612,
"step": 6320
},
{
"grad_norm": 0.2873695194721222,
"learning_rate": 9.308192423667485e-05,
"loss": 0.0604,
"step": 6330
},
{
"grad_norm": 0.36493730545043945,
"learning_rate": 9.305392562836262e-05,
"loss": 0.0601,
"step": 6340
},
{
"grad_norm": 0.3458423912525177,
"learning_rate": 9.3025874705471e-05,
"loss": 0.0546,
"step": 6350
},
{
"grad_norm": 0.3104035556316376,
"learning_rate": 9.299777150208457e-05,
"loss": 0.0604,
"step": 6360
},
{
"grad_norm": 0.28149229288101196,
"learning_rate": 9.296961605235133e-05,
"loss": 0.059,
"step": 6370
},
{
"grad_norm": 0.34341859817504883,
"learning_rate": 9.29414083904828e-05,
"loss": 0.0568,
"step": 6380
},
{
"grad_norm": 0.2640079855918884,
"learning_rate": 9.291314855075397e-05,
"loss": 0.0589,
"step": 6390
},
{
"grad_norm": 0.28648123145103455,
"learning_rate": 9.288483656750322e-05,
"loss": 0.0584,
"step": 6400
},
{
"grad_norm": 0.3198994994163513,
"learning_rate": 9.285647247513225e-05,
"loss": 0.0585,
"step": 6410
},
{
"grad_norm": 0.5129627585411072,
"learning_rate": 9.282805630810614e-05,
"loss": 0.0599,
"step": 6420
},
{
"grad_norm": 0.27086517214775085,
"learning_rate": 9.279958810095317e-05,
"loss": 0.0585,
"step": 6430
},
{
"grad_norm": 0.32081013917922974,
"learning_rate": 9.277106788826494e-05,
"loss": 0.06,
"step": 6440
},
{
"grad_norm": 0.36721181869506836,
"learning_rate": 9.274249570469618e-05,
"loss": 0.0533,
"step": 6450
},
{
"grad_norm": 0.43546050786972046,
"learning_rate": 9.271387158496476e-05,
"loss": 0.0654,
"step": 6460
},
{
"grad_norm": 0.3249818682670593,
"learning_rate": 9.268519556385173e-05,
"loss": 0.0699,
"step": 6470
},
{
"grad_norm": 0.34717804193496704,
"learning_rate": 9.265646767620113e-05,
"loss": 0.0599,
"step": 6480
},
{
"grad_norm": 0.33604496717453003,
"learning_rate": 9.262768795692006e-05,
"loss": 0.0621,
"step": 6490
},
{
"grad_norm": 0.4002475440502167,
"learning_rate": 9.259885644097861e-05,
"loss": 0.061,
"step": 6500
},
{
"grad_norm": 0.329664021730423,
"learning_rate": 9.256997316340976e-05,
"loss": 0.0651,
"step": 6510
},
{
"grad_norm": 0.3436298072338104,
"learning_rate": 9.254103815930943e-05,
"loss": 0.0666,
"step": 6520
},
{
"grad_norm": 0.3798339068889618,
"learning_rate": 9.251205146383636e-05,
"loss": 0.0673,
"step": 6530
},
{
"grad_norm": 0.2892748713493347,
"learning_rate": 9.248301311221216e-05,
"loss": 0.0608,
"step": 6540
},
{
"grad_norm": 0.30561935901641846,
"learning_rate": 9.245392313972115e-05,
"loss": 0.0668,
"step": 6550
},
{
"grad_norm": 0.35236406326293945,
"learning_rate": 9.242478158171037e-05,
"loss": 0.0617,
"step": 6560
},
{
"grad_norm": 0.3107627034187317,
"learning_rate": 9.239558847358957e-05,
"loss": 0.0674,
"step": 6570
},
{
"grad_norm": 0.35042592883110046,
"learning_rate": 9.236634385083114e-05,
"loss": 0.0677,
"step": 6580
},
{
"grad_norm": 0.2890722155570984,
"learning_rate": 9.233704774897005e-05,
"loss": 0.0614,
"step": 6590
},
{
"grad_norm": 0.26071658730506897,
"learning_rate": 9.230770020360382e-05,
"loss": 0.0628,
"step": 6600
},
{
"grad_norm": 0.30993548035621643,
"learning_rate": 9.227830125039247e-05,
"loss": 0.0603,
"step": 6610
},
{
"grad_norm": 0.3092069923877716,
"learning_rate": 9.224885092505853e-05,
"loss": 0.0617,
"step": 6620
},
{
"grad_norm": 0.2374105006456375,
"learning_rate": 9.221934926338691e-05,
"loss": 0.0648,
"step": 6630
},
{
"grad_norm": 0.29087743163108826,
"learning_rate": 9.21897963012249e-05,
"loss": 0.0596,
"step": 6640
},
{
"grad_norm": 0.2983260452747345,
"learning_rate": 9.216019207448217e-05,
"loss": 0.0616,
"step": 6650
},
{
"grad_norm": 0.2889538109302521,
"learning_rate": 9.21305366191306e-05,
"loss": 0.0597,
"step": 6660
},
{
"grad_norm": 0.3775623142719269,
"learning_rate": 9.21008299712044e-05,
"loss": 0.0723,
"step": 6670
},
{
"grad_norm": 0.32277101278305054,
"learning_rate": 9.207107216679993e-05,
"loss": 0.0773,
"step": 6680
},
{
"grad_norm": 0.35921409726142883,
"learning_rate": 9.204126324207574e-05,
"loss": 0.0649,
"step": 6690
},
{
"grad_norm": 0.37834393978118896,
"learning_rate": 9.201140323325248e-05,
"loss": 0.0731,
"step": 6700
},
{
"grad_norm": 0.35319387912750244,
"learning_rate": 9.198149217661287e-05,
"loss": 0.0677,
"step": 6710
},
{
"grad_norm": 0.4476366937160492,
"learning_rate": 9.195153010850167e-05,
"loss": 0.0652,
"step": 6720
},
{
"grad_norm": 0.3607640862464905,
"learning_rate": 9.192151706532562e-05,
"loss": 0.0664,
"step": 6730
},
{
"grad_norm": 0.3416297137737274,
"learning_rate": 9.189145308355337e-05,
"loss": 0.0727,
"step": 6740
},
{
"grad_norm": 0.3925589621067047,
"learning_rate": 9.186133819971555e-05,
"loss": 0.0673,
"step": 6750
},
{
"grad_norm": 0.33648642897605896,
"learning_rate": 9.183117245040455e-05,
"loss": 0.0697,
"step": 6760
},
{
"grad_norm": 0.2551674544811249,
"learning_rate": 9.180095587227458e-05,
"loss": 0.0654,
"step": 6770
},
{
"grad_norm": 0.36581864953041077,
"learning_rate": 9.177068850204167e-05,
"loss": 0.0653,
"step": 6780
},
{
"grad_norm": 0.3501262366771698,
"learning_rate": 9.174037037648351e-05,
"loss": 0.0654,
"step": 6790
},
{
"grad_norm": 0.3031077980995178,
"learning_rate": 9.171000153243947e-05,
"loss": 0.0716,
"step": 6800
},
{
"grad_norm": 0.30600833892822266,
"learning_rate": 9.167958200681057e-05,
"loss": 0.0592,
"step": 6810
},
{
"grad_norm": 0.36297503113746643,
"learning_rate": 9.164911183655943e-05,
"loss": 0.0695,
"step": 6820
},
{
"grad_norm": 0.27861258387565613,
"learning_rate": 9.161859105871013e-05,
"loss": 0.0668,
"step": 6830
},
{
"grad_norm": 0.3474787175655365,
"learning_rate": 9.158801971034832e-05,
"loss": 0.0666,
"step": 6840
},
{
"grad_norm": 0.2959362268447876,
"learning_rate": 9.155739782862107e-05,
"loss": 0.062,
"step": 6850
},
{
"grad_norm": 0.3768981993198395,
"learning_rate": 9.152672545073687e-05,
"loss": 0.0686,
"step": 6860
},
{
"grad_norm": 0.26318952441215515,
"learning_rate": 9.149600261396551e-05,
"loss": 0.0673,
"step": 6870
},
{
"grad_norm": 0.4874945282936096,
"learning_rate": 9.146522935563817e-05,
"loss": 0.068,
"step": 6880
},
{
"grad_norm": 0.372440904378891,
"learning_rate": 9.143440571314723e-05,
"loss": 0.0674,
"step": 6890
},
{
"grad_norm": 0.30360472202301025,
"learning_rate": 9.140353172394636e-05,
"loss": 0.0664,
"step": 6900
},
{
"grad_norm": 0.293652206659317,
"learning_rate": 9.137260742555032e-05,
"loss": 0.0655,
"step": 6910
},
{
"grad_norm": 0.3741999566555023,
"learning_rate": 9.13416328555351e-05,
"loss": 0.0699,
"step": 6920
},
{
"grad_norm": 0.2632998526096344,
"learning_rate": 9.131060805153769e-05,
"loss": 0.0578,
"step": 6930
},
{
"grad_norm": 0.3971377909183502,
"learning_rate": 9.127953305125618e-05,
"loss": 0.0613,
"step": 6940
},
{
"grad_norm": 0.30452463030815125,
"learning_rate": 9.124840789244958e-05,
"loss": 0.07,
"step": 6950
},
{
"grad_norm": 0.3832702040672302,
"learning_rate": 9.121723261293794e-05,
"loss": 0.0675,
"step": 6960
},
{
"grad_norm": 0.44045209884643555,
"learning_rate": 9.118600725060214e-05,
"loss": 0.0644,
"step": 6970
},
{
"grad_norm": 0.3051941394805908,
"learning_rate": 9.115473184338392e-05,
"loss": 0.0628,
"step": 6980
},
{
"grad_norm": 0.39422881603240967,
"learning_rate": 9.11234064292859e-05,
"loss": 0.0638,
"step": 6990
},
{
"grad_norm": 0.36388933658599854,
"learning_rate": 9.109203104637137e-05,
"loss": 0.0607,
"step": 7000
},
{
"grad_norm": 0.3603878915309906,
"learning_rate": 9.10606057327644e-05,
"loss": 0.0628,
"step": 7010
},
{
"grad_norm": 0.32500094175338745,
"learning_rate": 9.10291305266497e-05,
"loss": 0.0579,
"step": 7020
},
{
"grad_norm": 0.3371680676937103,
"learning_rate": 9.099760546627261e-05,
"loss": 0.0609,
"step": 7030
},
{
"grad_norm": 0.4003782570362091,
"learning_rate": 9.096603058993907e-05,
"loss": 0.0645,
"step": 7040
},
{
"grad_norm": 0.39239737391471863,
"learning_rate": 9.093440593601552e-05,
"loss": 0.0746,
"step": 7050
},
{
"grad_norm": 0.3393862843513489,
"learning_rate": 9.090273154292889e-05,
"loss": 0.0674,
"step": 7060
},
{
"grad_norm": 0.28776049613952637,
"learning_rate": 9.087100744916656e-05,
"loss": 0.0663,
"step": 7070
},
{
"grad_norm": 0.32731860876083374,
"learning_rate": 9.08392336932763e-05,
"loss": 0.0598,
"step": 7080
},
{
"grad_norm": 0.35297343134880066,
"learning_rate": 9.080741031386619e-05,
"loss": 0.066,
"step": 7090
},
{
"grad_norm": 0.31639593839645386,
"learning_rate": 9.077553734960468e-05,
"loss": 0.0648,
"step": 7100
},
{
"grad_norm": 0.35920313000679016,
"learning_rate": 9.074361483922041e-05,
"loss": 0.0669,
"step": 7110
},
{
"grad_norm": 0.3878105580806732,
"learning_rate": 9.071164282150224e-05,
"loss": 0.0654,
"step": 7120
},
{
"grad_norm": 0.40422526001930237,
"learning_rate": 9.067962133529919e-05,
"loss": 0.0642,
"step": 7130
},
{
"grad_norm": 0.32659009099006653,
"learning_rate": 9.064755041952036e-05,
"loss": 0.0642,
"step": 7140
},
{
"grad_norm": 0.32331034541130066,
"learning_rate": 9.061543011313497e-05,
"loss": 0.0665,
"step": 7150
},
{
"grad_norm": 0.3575258255004883,
"learning_rate": 9.05832604551722e-05,
"loss": 0.0683,
"step": 7160
},
{
"grad_norm": 0.3281266689300537,
"learning_rate": 9.055104148472124e-05,
"loss": 0.0633,
"step": 7170
},
{
"grad_norm": 0.33890095353126526,
"learning_rate": 9.051877324093114e-05,
"loss": 0.0623,
"step": 7180
},
{
"grad_norm": 0.3232220709323883,
"learning_rate": 9.04864557630109e-05,
"loss": 0.0664,
"step": 7190
},
{
"grad_norm": 0.37060031294822693,
"learning_rate": 9.045408909022928e-05,
"loss": 0.0582,
"step": 7200
},
{
"grad_norm": 0.2787524461746216,
"learning_rate": 9.042167326191484e-05,
"loss": 0.055,
"step": 7210
},
{
"grad_norm": 0.3439229726791382,
"learning_rate": 9.038920831745587e-05,
"loss": 0.0637,
"step": 7220
},
{
"grad_norm": 0.30662214756011963,
"learning_rate": 9.035669429630036e-05,
"loss": 0.0544,
"step": 7230
},
{
"grad_norm": 0.3291890025138855,
"learning_rate": 9.032413123795588e-05,
"loss": 0.0571,
"step": 7240
},
{
"grad_norm": 0.34164825081825256,
"learning_rate": 9.029151918198962e-05,
"loss": 0.0526,
"step": 7250
},
{
"grad_norm": 0.2659841477870941,
"learning_rate": 9.025885816802833e-05,
"loss": 0.0565,
"step": 7260
},
{
"grad_norm": 0.27595120668411255,
"learning_rate": 9.02261482357582e-05,
"loss": 0.0539,
"step": 7270
},
{
"grad_norm": 0.3696479797363281,
"learning_rate": 9.019338942492485e-05,
"loss": 0.05,
"step": 7280
},
{
"grad_norm": 0.32906869053840637,
"learning_rate": 9.01605817753334e-05,
"loss": 0.0539,
"step": 7290
},
{
"grad_norm": 0.6042653322219849,
"learning_rate": 9.012772532684818e-05,
"loss": 0.0616,
"step": 7300
},
{
"grad_norm": 0.3706338405609131,
"learning_rate": 9.00948201193929e-05,
"loss": 0.0594,
"step": 7310
},
{
"grad_norm": 0.4325748383998871,
"learning_rate": 9.006186619295047e-05,
"loss": 0.0531,
"step": 7320
},
{
"grad_norm": 0.37009015679359436,
"learning_rate": 9.002886358756305e-05,
"loss": 0.0568,
"step": 7330
},
{
"grad_norm": 0.4368363916873932,
"learning_rate": 8.999581234333188e-05,
"loss": 0.0487,
"step": 7340
},
{
"grad_norm": 0.289333701133728,
"learning_rate": 8.996271250041734e-05,
"loss": 0.0507,
"step": 7350
},
{
"grad_norm": 0.40612152218818665,
"learning_rate": 8.99295640990389e-05,
"loss": 0.0494,
"step": 7360
},
{
"grad_norm": 0.3836275339126587,
"learning_rate": 8.989636717947494e-05,
"loss": 0.05,
"step": 7370
},
{
"grad_norm": 0.45196664333343506,
"learning_rate": 8.986312178206291e-05,
"loss": 0.0523,
"step": 7380
},
{
"grad_norm": 0.27916866540908813,
"learning_rate": 8.982982794719904e-05,
"loss": 0.0593,
"step": 7390
},
{
"grad_norm": 0.33737024664878845,
"learning_rate": 8.979648571533852e-05,
"loss": 0.0505,
"step": 7400
},
{
"grad_norm": 0.3450508713722229,
"learning_rate": 8.976309512699528e-05,
"loss": 0.0454,
"step": 7410
},
{
"grad_norm": 0.2986001670360565,
"learning_rate": 8.972965622274205e-05,
"loss": 0.0663,
"step": 7420
},
{
"grad_norm": 0.9776252508163452,
"learning_rate": 8.969616904321026e-05,
"loss": 0.0744,
"step": 7430
},
{
"grad_norm": 0.3753082752227783,
"learning_rate": 8.966263362908998e-05,
"loss": 0.0719,
"step": 7440
},
{
"grad_norm": 0.3799479603767395,
"learning_rate": 8.962905002112989e-05,
"loss": 0.0659,
"step": 7450
},
{
"grad_norm": 0.3383823037147522,
"learning_rate": 8.959541826013724e-05,
"loss": 0.067,
"step": 7460
},
{
"grad_norm": 0.3402482569217682,
"learning_rate": 8.956173838697781e-05,
"loss": 0.0673,
"step": 7470
},
{
"grad_norm": 0.38396984338760376,
"learning_rate": 8.952801044257581e-05,
"loss": 0.071,
"step": 7480
},
{
"grad_norm": 0.2706429958343506,
"learning_rate": 8.949423446791387e-05,
"loss": 0.0696,
"step": 7490
},
{
"grad_norm": 0.3861299753189087,
"learning_rate": 8.9460410504033e-05,
"loss": 0.0713,
"step": 7500
},
{
"grad_norm": 0.3712531626224518,
"learning_rate": 8.942653859203248e-05,
"loss": 0.0568,
"step": 7510
},
{
"grad_norm": 0.32684022188186646,
"learning_rate": 8.93926187730699e-05,
"loss": 0.0637,
"step": 7520
},
{
"grad_norm": 0.39768844842910767,
"learning_rate": 8.935865108836102e-05,
"loss": 0.0675,
"step": 7530
},
{
"grad_norm": 0.3049660325050354,
"learning_rate": 8.932463557917981e-05,
"loss": 0.056,
"step": 7540
},
{
"grad_norm": 0.26355111598968506,
"learning_rate": 8.929057228685829e-05,
"loss": 0.0644,
"step": 7550
},
{
"grad_norm": 0.2769785225391388,
"learning_rate": 8.925646125278658e-05,
"loss": 0.0716,
"step": 7560
},
{
"grad_norm": 0.3128792345523834,
"learning_rate": 8.92223025184128e-05,
"loss": 0.0719,
"step": 7570
},
{
"grad_norm": 0.5121272206306458,
"learning_rate": 8.918809612524303e-05,
"loss": 0.0768,
"step": 7580
},
{
"grad_norm": 0.34691956639289856,
"learning_rate": 8.91538421148413e-05,
"loss": 0.072,
"step": 7590
},
{
"grad_norm": 0.36319229006767273,
"learning_rate": 8.91195405288294e-05,
"loss": 0.0765,
"step": 7600
},
{
"grad_norm": 0.3029498755931854,
"learning_rate": 8.908519140888705e-05,
"loss": 0.0693,
"step": 7610
},
{
"grad_norm": 0.4379647672176361,
"learning_rate": 8.905079479675164e-05,
"loss": 0.0649,
"step": 7620
},
{
"grad_norm": 0.338838666677475,
"learning_rate": 8.901635073421831e-05,
"loss": 0.0642,
"step": 7630
},
{
"grad_norm": 0.41508030891418457,
"learning_rate": 8.898185926313983e-05,
"loss": 0.0754,
"step": 7640
},
{
"grad_norm": 0.2787676453590393,
"learning_rate": 8.894732042542659e-05,
"loss": 0.0719,
"step": 7650
},
{
"grad_norm": 0.39652198553085327,
"learning_rate": 8.891273426304654e-05,
"loss": 0.0666,
"step": 7660
},
{
"grad_norm": 0.2888587713241577,
"learning_rate": 8.887810081802515e-05,
"loss": 0.0685,
"step": 7670
},
{
"grad_norm": 0.33755621314048767,
"learning_rate": 8.884342013244529e-05,
"loss": 0.0646,
"step": 7680
},
{
"grad_norm": 0.3627069592475891,
"learning_rate": 8.880869224844725e-05,
"loss": 0.0642,
"step": 7690
},
{
"grad_norm": 0.3541169762611389,
"learning_rate": 8.877391720822875e-05,
"loss": 0.0593,
"step": 7700
},
{
"grad_norm": 0.34673547744750977,
"learning_rate": 8.873909505404468e-05,
"loss": 0.0638,
"step": 7710
},
{
"grad_norm": 0.28731051087379456,
"learning_rate": 8.870422582820726e-05,
"loss": 0.0574,
"step": 7720
},
{
"grad_norm": 0.36393386125564575,
"learning_rate": 8.866930957308588e-05,
"loss": 0.0687,
"step": 7730
},
{
"grad_norm": 0.35560142993927,
"learning_rate": 8.86343463311071e-05,
"loss": 0.0555,
"step": 7740
},
{
"grad_norm": 0.29869282245635986,
"learning_rate": 8.859933614475452e-05,
"loss": 0.0586,
"step": 7750
},
{
"grad_norm": 0.34030094742774963,
"learning_rate": 8.85642790565689e-05,
"loss": 0.0573,
"step": 7760
},
{
"grad_norm": 0.3658068776130676,
"learning_rate": 8.852917510914782e-05,
"loss": 0.0564,
"step": 7770
},
{
"grad_norm": 0.4289948046207428,
"learning_rate": 8.84940243451459e-05,
"loss": 0.0645,
"step": 7780
},
{
"grad_norm": 0.29169508814811707,
"learning_rate": 8.845882680727469e-05,
"loss": 0.0571,
"step": 7790
},
{
"grad_norm": 0.2572537064552307,
"learning_rate": 8.842358253830246e-05,
"loss": 0.0587,
"step": 7800
},
{
"grad_norm": 0.3173958361148834,
"learning_rate": 8.838829158105434e-05,
"loss": 0.0593,
"step": 7810
},
{
"grad_norm": 0.35048696398735046,
"learning_rate": 8.835295397841217e-05,
"loss": 0.0599,
"step": 7820
},
{
"grad_norm": 0.27995210886001587,
"learning_rate": 8.831756977331447e-05,
"loss": 0.0681,
"step": 7830
},
{
"grad_norm": 0.28225505352020264,
"learning_rate": 8.828213900875638e-05,
"loss": 0.0593,
"step": 7840
},
{
"grad_norm": 0.36020296812057495,
"learning_rate": 8.824666172778964e-05,
"loss": 0.0545,
"step": 7850
},
{
"grad_norm": 0.26851361989974976,
"learning_rate": 8.821113797352245e-05,
"loss": 0.0506,
"step": 7860
},
{
"grad_norm": 0.3186512291431427,
"learning_rate": 8.817556778911956e-05,
"loss": 0.0478,
"step": 7870
},
{
"grad_norm": 0.3279949426651001,
"learning_rate": 8.81399512178021e-05,
"loss": 0.0572,
"step": 7880
},
{
"grad_norm": 0.3330066502094269,
"learning_rate": 8.810428830284752e-05,
"loss": 0.059,
"step": 7890
},
{
"grad_norm": 0.3072631359100342,
"learning_rate": 8.806857908758967e-05,
"loss": 0.0539,
"step": 7900
},
{
"grad_norm": 0.24302351474761963,
"learning_rate": 8.80328236154186e-05,
"loss": 0.0542,
"step": 7910
},
{
"grad_norm": 0.41512331366539,
"learning_rate": 8.799702192978056e-05,
"loss": 0.0544,
"step": 7920
},
{
"grad_norm": 0.3629709780216217,
"learning_rate": 8.7961174074178e-05,
"loss": 0.0564,
"step": 7930
},
{
"grad_norm": 0.395085871219635,
"learning_rate": 8.792528009216942e-05,
"loss": 0.0557,
"step": 7940
},
{
"grad_norm": 0.3090662658214569,
"learning_rate": 8.788934002736943e-05,
"loss": 0.0614,
"step": 7950
},
{
"grad_norm": 0.5340372920036316,
"learning_rate": 8.785335392344857e-05,
"loss": 0.0823,
"step": 7960
},
{
"grad_norm": 0.36461347341537476,
"learning_rate": 8.781732182413336e-05,
"loss": 0.0706,
"step": 7970
},
{
"grad_norm": 0.4251161515712738,
"learning_rate": 8.778124377320618e-05,
"loss": 0.0732,
"step": 7980
},
{
"grad_norm": 0.3521801829338074,
"learning_rate": 8.774511981450529e-05,
"loss": 0.0619,
"step": 7990
},
{
"grad_norm": 0.2793223559856415,
"learning_rate": 8.770894999192467e-05,
"loss": 0.0567,
"step": 8000
},
{
"grad_norm": 0.265826016664505,
"learning_rate": 8.767273434941412e-05,
"loss": 0.0655,
"step": 8010
},
{
"grad_norm": 0.4201524555683136,
"learning_rate": 8.7636472930979e-05,
"loss": 0.0632,
"step": 8020
},
{
"grad_norm": 0.347598671913147,
"learning_rate": 8.76001657806804e-05,
"loss": 0.0719,
"step": 8030
},
{
"grad_norm": 0.30251580476760864,
"learning_rate": 8.756381294263491e-05,
"loss": 0.0621,
"step": 8040
},
{
"grad_norm": 0.3600787818431854,
"learning_rate": 8.752741446101464e-05,
"loss": 0.0619,
"step": 8050
},
{
"grad_norm": 0.3036717176437378,
"learning_rate": 8.749097038004722e-05,
"loss": 0.0622,
"step": 8060
},
{
"grad_norm": 0.25727060437202454,
"learning_rate": 8.745448074401561e-05,
"loss": 0.0567,
"step": 8070
},
{
"grad_norm": 0.2790777385234833,
"learning_rate": 8.741794559725818e-05,
"loss": 0.0555,
"step": 8080
},
{
"grad_norm": 0.321492075920105,
"learning_rate": 8.738136498416856e-05,
"loss": 0.0522,
"step": 8090
},
{
"grad_norm": 0.3006642460823059,
"learning_rate": 8.734473894919564e-05,
"loss": 0.0539,
"step": 8100
},
{
"grad_norm": 0.27951598167419434,
"learning_rate": 8.730806753684353e-05,
"loss": 0.0565,
"step": 8110
},
{
"grad_norm": 0.3235524892807007,
"learning_rate": 8.727135079167144e-05,
"loss": 0.0532,
"step": 8120
},
{
"grad_norm": 0.35913369059562683,
"learning_rate": 8.723458875829368e-05,
"loss": 0.0547,
"step": 8130
},
{
"grad_norm": 0.2691338062286377,
"learning_rate": 8.719778148137958e-05,
"loss": 0.0565,
"step": 8140
},
{
"grad_norm": 0.31909283995628357,
"learning_rate": 8.716092900565346e-05,
"loss": 0.0554,
"step": 8150
},
{
"grad_norm": 0.29553696513175964,
"learning_rate": 8.712403137589456e-05,
"loss": 0.0528,
"step": 8160
},
{
"grad_norm": 0.34499192237854004,
"learning_rate": 8.708708863693697e-05,
"loss": 0.058,
"step": 8170
},
{
"grad_norm": 0.383587121963501,
"learning_rate": 8.70501008336696e-05,
"loss": 0.0562,
"step": 8180
},
{
"grad_norm": 0.2699386477470398,
"learning_rate": 8.701306801103611e-05,
"loss": 0.0536,
"step": 8190
},
{
"grad_norm": 0.3846524953842163,
"learning_rate": 8.697599021403489e-05,
"loss": 0.0575,
"step": 8200
},
{
"grad_norm": 0.27478575706481934,
"learning_rate": 8.693886748771896e-05,
"loss": 0.0513,
"step": 8210
},
{
"grad_norm": 0.297850638628006,
"learning_rate": 8.690169987719593e-05,
"loss": 0.0576,
"step": 8220
},
{
"grad_norm": 0.2798923850059509,
"learning_rate": 8.686448742762791e-05,
"loss": 0.0594,
"step": 8230
},
{
"grad_norm": 0.31219011545181274,
"learning_rate": 8.682723018423161e-05,
"loss": 0.0656,
"step": 8240
},
{
"grad_norm": 0.379766583442688,
"learning_rate": 8.678992819227804e-05,
"loss": 0.0617,
"step": 8250
},
{
"grad_norm": 0.40430745482444763,
"learning_rate": 8.675258149709265e-05,
"loss": 0.0611,
"step": 8260
},
{
"grad_norm": 0.3074914216995239,
"learning_rate": 8.67151901440552e-05,
"loss": 0.075,
"step": 8270
},
{
"grad_norm": 0.42521771788597107,
"learning_rate": 8.667775417859971e-05,
"loss": 0.065,
"step": 8280
},
{
"grad_norm": 0.28691840171813965,
"learning_rate": 8.664027364621441e-05,
"loss": 0.0672,
"step": 8290
},
{
"grad_norm": 0.22753511369228363,
"learning_rate": 8.660274859244168e-05,
"loss": 0.064,
"step": 8300
},
{
"grad_norm": 0.3184296190738678,
"learning_rate": 8.656517906287799e-05,
"loss": 0.0694,
"step": 8310
},
{
"grad_norm": 0.3687664866447449,
"learning_rate": 8.652756510317387e-05,
"loss": 0.0641,
"step": 8320
},
{
"grad_norm": 0.3146474063396454,
"learning_rate": 8.648990675903382e-05,
"loss": 0.0621,
"step": 8330
},
{
"grad_norm": 0.30919939279556274,
"learning_rate": 8.64522040762163e-05,
"loss": 0.0611,
"step": 8340
},
{
"grad_norm": 0.30580487847328186,
"learning_rate": 8.64144571005336e-05,
"loss": 0.0631,
"step": 8350
},
{
"grad_norm": 0.28198421001434326,
"learning_rate": 8.637666587785184e-05,
"loss": 0.0561,
"step": 8360
},
{
"grad_norm": 0.3170509934425354,
"learning_rate": 8.633883045409096e-05,
"loss": 0.0571,
"step": 8370
},
{
"grad_norm": 0.2576914131641388,
"learning_rate": 8.630095087522457e-05,
"loss": 0.0565,
"step": 8380
},
{
"grad_norm": 0.3463984727859497,
"learning_rate": 8.62630271872799e-05,
"loss": 0.0558,
"step": 8390
},
{
"grad_norm": 0.2870137393474579,
"learning_rate": 8.62250594363378e-05,
"loss": 0.0519,
"step": 8400
},
{
"grad_norm": 0.26863113045692444,
"learning_rate": 8.61870476685327e-05,
"loss": 0.0612,
"step": 8410
},
{
"grad_norm": 0.2688433527946472,
"learning_rate": 8.614899193005248e-05,
"loss": 0.0533,
"step": 8420
},
{
"grad_norm": 0.24542585015296936,
"learning_rate": 8.611089226713842e-05,
"loss": 0.0552,
"step": 8430
},
{
"grad_norm": 0.2753819227218628,
"learning_rate": 8.607274872608521e-05,
"loss": 0.0517,
"step": 8440
},
{
"grad_norm": 0.38737908005714417,
"learning_rate": 8.603456135324088e-05,
"loss": 0.0501,
"step": 8450
},
{
"grad_norm": 0.299326628446579,
"learning_rate": 8.599633019500665e-05,
"loss": 0.0482,
"step": 8460
},
{
"grad_norm": 0.37694159150123596,
"learning_rate": 8.595805529783702e-05,
"loss": 0.056,
"step": 8470
},
{
"grad_norm": 0.35342279076576233,
"learning_rate": 8.591973670823958e-05,
"loss": 0.0592,
"step": 8480
},
{
"grad_norm": 0.30702999234199524,
"learning_rate": 8.588137447277503e-05,
"loss": 0.0609,
"step": 8490
},
{
"grad_norm": 0.39492323994636536,
"learning_rate": 8.584296863805709e-05,
"loss": 0.0522,
"step": 8500
},
{
"grad_norm": 0.307187020778656,
"learning_rate": 8.580451925075248e-05,
"loss": 0.0608,
"step": 8510
},
{
"grad_norm": 0.3392636179924011,
"learning_rate": 8.576602635758085e-05,
"loss": 0.0574,
"step": 8520
},
{
"grad_norm": 0.29386869072914124,
"learning_rate": 8.572749000531469e-05,
"loss": 0.059,
"step": 8530
},
{
"grad_norm": 0.2934737205505371,
"learning_rate": 8.568891024077926e-05,
"loss": 0.0567,
"step": 8540
},
{
"grad_norm": 0.2253584861755371,
"learning_rate": 8.565028711085265e-05,
"loss": 0.0586,
"step": 8550
},
{
"grad_norm": 0.26662448048591614,
"learning_rate": 8.561162066246561e-05,
"loss": 0.0527,
"step": 8560
},
{
"grad_norm": 0.3468407988548279,
"learning_rate": 8.557291094260152e-05,
"loss": 0.059,
"step": 8570
},
{
"grad_norm": 0.3342597782611847,
"learning_rate": 8.553415799829629e-05,
"loss": 0.0583,
"step": 8580
},
{
"grad_norm": 0.38965821266174316,
"learning_rate": 8.549536187663847e-05,
"loss": 0.0548,
"step": 8590
},
{
"grad_norm": 0.31282347440719604,
"learning_rate": 8.545652262476897e-05,
"loss": 0.0646,
"step": 8600
},
{
"grad_norm": 0.30918484926223755,
"learning_rate": 8.541764028988115e-05,
"loss": 0.055,
"step": 8610
},
{
"grad_norm": 0.3459490239620209,
"learning_rate": 8.537871491922071e-05,
"loss": 0.0621,
"step": 8620
},
{
"grad_norm": 0.29841867089271545,
"learning_rate": 8.533974656008566e-05,
"loss": 0.0582,
"step": 8630
},
{
"grad_norm": 0.30868518352508545,
"learning_rate": 8.530073525982621e-05,
"loss": 0.0573,
"step": 8640
},
{
"grad_norm": 0.34344035387039185,
"learning_rate": 8.526168106584476e-05,
"loss": 0.0568,
"step": 8650
},
{
"grad_norm": 0.30905619263648987,
"learning_rate": 8.522258402559586e-05,
"loss": 0.0533,
"step": 8660
},
{
"grad_norm": 0.2781258523464203,
"learning_rate": 8.518344418658613e-05,
"loss": 0.0562,
"step": 8670
},
{
"grad_norm": 0.30840253829956055,
"learning_rate": 8.51442615963741e-05,
"loss": 0.0595,
"step": 8680
},
{
"grad_norm": 0.24609558284282684,
"learning_rate": 8.510503630257034e-05,
"loss": 0.0584,
"step": 8690
},
{
"grad_norm": 0.31283825635910034,
"learning_rate": 8.50657683528373e-05,
"loss": 0.0573,
"step": 8700
},
{
"grad_norm": 0.29947957396507263,
"learning_rate": 8.502645779488922e-05,
"loss": 0.0528,
"step": 8710
},
{
"grad_norm": 0.2313683182001114,
"learning_rate": 8.498710467649213e-05,
"loss": 0.062,
"step": 8720
},
{
"grad_norm": 0.26798510551452637,
"learning_rate": 8.494770904546381e-05,
"loss": 0.0604,
"step": 8730
},
{
"grad_norm": 0.3512875735759735,
"learning_rate": 8.490827094967363e-05,
"loss": 0.0595,
"step": 8740
},
{
"grad_norm": 0.24629729986190796,
"learning_rate": 8.486879043704263e-05,
"loss": 0.0585,
"step": 8750
},
{
"grad_norm": 0.25759562849998474,
"learning_rate": 8.482926755554332e-05,
"loss": 0.0628,
"step": 8760
},
{
"grad_norm": 0.3520289957523346,
"learning_rate": 8.478970235319975e-05,
"loss": 0.0587,
"step": 8770
},
{
"grad_norm": 0.31175631284713745,
"learning_rate": 8.475009487808737e-05,
"loss": 0.0595,
"step": 8780
},
{
"grad_norm": 0.2952142357826233,
"learning_rate": 8.471044517833298e-05,
"loss": 0.0661,
"step": 8790
},
{
"grad_norm": 0.3652871251106262,
"learning_rate": 8.467075330211474e-05,
"loss": 0.0581,
"step": 8800
},
{
"grad_norm": 0.32049500942230225,
"learning_rate": 8.463101929766198e-05,
"loss": 0.0613,
"step": 8810
},
{
"grad_norm": 0.3296147882938385,
"learning_rate": 8.459124321325528e-05,
"loss": 0.0587,
"step": 8820
},
{
"grad_norm": 0.2919204831123352,
"learning_rate": 8.455142509722634e-05,
"loss": 0.0607,
"step": 8830
},
{
"grad_norm": 0.2918052077293396,
"learning_rate": 8.451156499795792e-05,
"loss": 0.0499,
"step": 8840
},
{
"grad_norm": 0.37306520342826843,
"learning_rate": 8.447166296388381e-05,
"loss": 0.0605,
"step": 8850
},
{
"grad_norm": 0.27683860063552856,
"learning_rate": 8.443171904348872e-05,
"loss": 0.0585,
"step": 8860
},
{
"grad_norm": 0.2575588822364807,
"learning_rate": 8.43917332853083e-05,
"loss": 0.051,
"step": 8870
},
{
"grad_norm": 0.35811111330986023,
"learning_rate": 8.435170573792901e-05,
"loss": 0.0505,
"step": 8880
},
{
"grad_norm": 0.41832399368286133,
"learning_rate": 8.431163644998808e-05,
"loss": 0.0496,
"step": 8890
},
{
"grad_norm": 0.3170270621776581,
"learning_rate": 8.42715254701735e-05,
"loss": 0.0557,
"step": 8900
},
{
"grad_norm": 0.33597618341445923,
"learning_rate": 8.42313728472239e-05,
"loss": 0.0541,
"step": 8910
},
{
"grad_norm": 0.23453223705291748,
"learning_rate": 8.419117862992844e-05,
"loss": 0.0524,
"step": 8920
},
{
"grad_norm": 0.2725987732410431,
"learning_rate": 8.415094286712694e-05,
"loss": 0.0529,
"step": 8930
},
{
"grad_norm": 0.31563571095466614,
"learning_rate": 8.411066560770965e-05,
"loss": 0.0567,
"step": 8940
},
{
"grad_norm": 0.23979544639587402,
"learning_rate": 8.407034690061721e-05,
"loss": 0.0578,
"step": 8950
},
{
"grad_norm": 0.3043804168701172,
"learning_rate": 8.402998679484067e-05,
"loss": 0.0672,
"step": 8960
},
{
"grad_norm": 0.27323442697525024,
"learning_rate": 8.398958533942135e-05,
"loss": 0.0503,
"step": 8970
},
{
"grad_norm": 0.2699633538722992,
"learning_rate": 8.394914258345083e-05,
"loss": 0.0565,
"step": 8980
},
{
"grad_norm": 0.3449034094810486,
"learning_rate": 8.390865857607089e-05,
"loss": 0.0528,
"step": 8990
},
{
"grad_norm": 0.3420887887477875,
"learning_rate": 8.386813336647339e-05,
"loss": 0.0597,
"step": 9000
},
{
"grad_norm": 0.4537886679172516,
"learning_rate": 8.38275670039003e-05,
"loss": 0.0548,
"step": 9010
},
{
"grad_norm": 0.3082842230796814,
"learning_rate": 8.378695953764356e-05,
"loss": 0.0615,
"step": 9020
},
{
"grad_norm": 0.2827955484390259,
"learning_rate": 8.374631101704509e-05,
"loss": 0.0569,
"step": 9030
},
{
"grad_norm": 0.27589523792266846,
"learning_rate": 8.370562149149665e-05,
"loss": 0.0605,
"step": 9040
},
{
"grad_norm": 0.3450038731098175,
"learning_rate": 8.366489101043989e-05,
"loss": 0.0628,
"step": 9050
},
{
"grad_norm": 0.24143576622009277,
"learning_rate": 8.362411962336614e-05,
"loss": 0.0558,
"step": 9060
},
{
"grad_norm": 0.36401116847991943,
"learning_rate": 8.358330737981651e-05,
"loss": 0.0606,
"step": 9070
},
{
"grad_norm": 0.3450113534927368,
"learning_rate": 8.35424543293817e-05,
"loss": 0.0633,
"step": 9080
},
{
"grad_norm": 0.37859871983528137,
"learning_rate": 8.350156052170206e-05,
"loss": 0.0562,
"step": 9090
},
{
"grad_norm": 0.30864763259887695,
"learning_rate": 8.346062600646738e-05,
"loss": 0.0617,
"step": 9100
},
{
"grad_norm": 0.30922940373420715,
"learning_rate": 8.341965083341697e-05,
"loss": 0.0605,
"step": 9110
},
{
"grad_norm": 0.2579199969768524,
"learning_rate": 8.337863505233953e-05,
"loss": 0.0622,
"step": 9120
},
{
"grad_norm": 0.2974727153778076,
"learning_rate": 8.33375787130731e-05,
"loss": 0.0482,
"step": 9130
},
{
"grad_norm": 0.3074272871017456,
"learning_rate": 8.3296481865505e-05,
"loss": 0.0508,
"step": 9140
},
{
"grad_norm": 0.36350566148757935,
"learning_rate": 8.325534455957181e-05,
"loss": 0.0557,
"step": 9150
},
{
"grad_norm": 0.3459492325782776,
"learning_rate": 8.321416684525917e-05,
"loss": 0.0517,
"step": 9160
},
{
"grad_norm": 0.2856054902076721,
"learning_rate": 8.317294877260193e-05,
"loss": 0.0515,
"step": 9170
},
{
"grad_norm": 0.3308728039264679,
"learning_rate": 8.313169039168394e-05,
"loss": 0.0519,
"step": 9180
},
{
"grad_norm": 0.34493395686149597,
"learning_rate": 8.3090391752638e-05,
"loss": 0.0508,
"step": 9190
},
{
"grad_norm": 0.30624109506607056,
"learning_rate": 8.304905290564586e-05,
"loss": 0.0592,
"step": 9200
},
{
"grad_norm": 0.3027808666229248,
"learning_rate": 8.300767390093813e-05,
"loss": 0.0526,
"step": 9210
},
{
"grad_norm": 0.33821627497673035,
"learning_rate": 8.296625478879417e-05,
"loss": 0.0606,
"step": 9220
},
{
"grad_norm": 0.31960609555244446,
"learning_rate": 8.292479561954214e-05,
"loss": 0.0521,
"step": 9230
},
{
"grad_norm": 0.2194732427597046,
"learning_rate": 8.288329644355883e-05,
"loss": 0.0517,
"step": 9240
},
{
"grad_norm": 0.2659290134906769,
"learning_rate": 8.284175731126965e-05,
"loss": 0.0548,
"step": 9250
},
{
"grad_norm": 0.269540935754776,
"learning_rate": 8.280017827314854e-05,
"loss": 0.0551,
"step": 9260
},
{
"grad_norm": 0.29007139801979065,
"learning_rate": 8.2758559379718e-05,
"loss": 0.0526,
"step": 9270
},
{
"grad_norm": 0.2816387712955475,
"learning_rate": 8.271690068154887e-05,
"loss": 0.051,
"step": 9280
},
{
"grad_norm": 0.3523404598236084,
"learning_rate": 8.26752022292604e-05,
"loss": 0.0515,
"step": 9290
},
{
"grad_norm": 0.24975639581680298,
"learning_rate": 8.263346407352018e-05,
"loss": 0.0525,
"step": 9300
},
{
"grad_norm": 0.32613706588745117,
"learning_rate": 8.259168626504395e-05,
"loss": 0.0563,
"step": 9310
},
{
"grad_norm": 0.261671245098114,
"learning_rate": 8.254986885459572e-05,
"loss": 0.0527,
"step": 9320
},
{
"grad_norm": 0.3823186159133911,
"learning_rate": 8.250801189298757e-05,
"loss": 0.0603,
"step": 9330
},
{
"grad_norm": 0.21251723170280457,
"learning_rate": 8.246611543107967e-05,
"loss": 0.0531,
"step": 9340
},
{
"grad_norm": 0.3855661153793335,
"learning_rate": 8.242417951978018e-05,
"loss": 0.057,
"step": 9350
},
{
"grad_norm": 0.338233083486557,
"learning_rate": 8.238220421004517e-05,
"loss": 0.0652,
"step": 9360
},
{
"grad_norm": 0.46015167236328125,
"learning_rate": 8.234018955287859e-05,
"loss": 0.066,
"step": 9370
},
{
"grad_norm": 0.34608346223831177,
"learning_rate": 8.229813559933224e-05,
"loss": 0.0634,
"step": 9380
},
{
"grad_norm": 0.3083241581916809,
"learning_rate": 8.22560424005056e-05,
"loss": 0.0605,
"step": 9390
},
{
"grad_norm": 0.29086363315582275,
"learning_rate": 8.22139100075459e-05,
"loss": 0.0526,
"step": 9400
},
{
"grad_norm": 0.3538160026073456,
"learning_rate": 8.217173847164799e-05,
"loss": 0.0535,
"step": 9410
},
{
"grad_norm": 0.2763722836971283,
"learning_rate": 8.212952784405422e-05,
"loss": 0.0574,
"step": 9420
},
{
"grad_norm": 0.3092942535877228,
"learning_rate": 8.208727817605451e-05,
"loss": 0.0531,
"step": 9430
},
{
"grad_norm": 0.34929075837135315,
"learning_rate": 8.204498951898618e-05,
"loss": 0.0607,
"step": 9440
},
{
"grad_norm": 0.3502836227416992,
"learning_rate": 8.200266192423395e-05,
"loss": 0.0549,
"step": 9450
},
{
"grad_norm": 0.3439253568649292,
"learning_rate": 8.196029544322981e-05,
"loss": 0.0603,
"step": 9460
},
{
"grad_norm": 0.2901812791824341,
"learning_rate": 8.191789012745305e-05,
"loss": 0.0591,
"step": 9470
},
{
"grad_norm": 0.37335893511772156,
"learning_rate": 8.187544602843014e-05,
"loss": 0.0562,
"step": 9480
},
{
"grad_norm": 0.31019797921180725,
"learning_rate": 8.183296319773466e-05,
"loss": 0.0572,
"step": 9490
},
{
"grad_norm": 0.3309723138809204,
"learning_rate": 8.179044168698721e-05,
"loss": 0.0574,
"step": 9500
},
{
"grad_norm": 0.26634976267814636,
"learning_rate": 8.174788154785548e-05,
"loss": 0.0549,
"step": 9510
},
{
"grad_norm": 0.3002128303050995,
"learning_rate": 8.170528283205403e-05,
"loss": 0.0615,
"step": 9520
},
{
"grad_norm": 0.41828247904777527,
"learning_rate": 8.166264559134435e-05,
"loss": 0.0609,
"step": 9530
},
{
"grad_norm": 0.3085687458515167,
"learning_rate": 8.161996987753467e-05,
"loss": 0.0625,
"step": 9540
},
{
"grad_norm": 0.31141144037246704,
"learning_rate": 8.157725574248e-05,
"loss": 0.0648,
"step": 9550
},
{
"grad_norm": 0.3100118637084961,
"learning_rate": 8.153450323808209e-05,
"loss": 0.0571,
"step": 9560
},
{
"grad_norm": 0.37268757820129395,
"learning_rate": 8.149171241628923e-05,
"loss": 0.0625,
"step": 9570
},
{
"grad_norm": 0.3809780776500702,
"learning_rate": 8.144888332909631e-05,
"loss": 0.0621,
"step": 9580
},
{
"grad_norm": 0.33155569434165955,
"learning_rate": 8.14060160285447e-05,
"loss": 0.0553,
"step": 9590
},
{
"grad_norm": 0.2909473180770874,
"learning_rate": 8.136311056672223e-05,
"loss": 0.0544,
"step": 9600
},
{
"grad_norm": 0.3208588659763336,
"learning_rate": 8.132016699576308e-05,
"loss": 0.0572,
"step": 9610
},
{
"grad_norm": 0.2708950340747833,
"learning_rate": 8.127718536784771e-05,
"loss": 0.0495,
"step": 9620
},
{
"grad_norm": 0.31380999088287354,
"learning_rate": 8.12341657352029e-05,
"loss": 0.0532,
"step": 9630
},
{
"grad_norm": 0.25318339467048645,
"learning_rate": 8.119110815010151e-05,
"loss": 0.0446,
"step": 9640
},
{
"grad_norm": 0.36635565757751465,
"learning_rate": 8.114801266486261e-05,
"loss": 0.0494,
"step": 9650
},
{
"grad_norm": 0.34168368577957153,
"learning_rate": 8.110487933185122e-05,
"loss": 0.0516,
"step": 9660
},
{
"grad_norm": 0.2991236448287964,
"learning_rate": 8.106170820347847e-05,
"loss": 0.0516,
"step": 9670
},
{
"grad_norm": 0.33494940400123596,
"learning_rate": 8.101849933220132e-05,
"loss": 0.0532,
"step": 9680
},
{
"grad_norm": 0.3717610836029053,
"learning_rate": 8.097525277052264e-05,
"loss": 0.0574,
"step": 9690
},
{
"grad_norm": 0.2837487459182739,
"learning_rate": 8.093196857099106e-05,
"loss": 0.0489,
"step": 9700
},
{
"grad_norm": 0.23697471618652344,
"learning_rate": 8.088864678620096e-05,
"loss": 0.0549,
"step": 9710
},
{
"grad_norm": 0.26986512541770935,
"learning_rate": 8.084528746879243e-05,
"loss": 0.0571,
"step": 9720
},
{
"grad_norm": 0.3337372839450836,
"learning_rate": 8.080189067145108e-05,
"loss": 0.0589,
"step": 9730
},
{
"grad_norm": 0.30603867769241333,
"learning_rate": 8.075845644690813e-05,
"loss": 0.0543,
"step": 9740
},
{
"grad_norm": 0.3038812577724457,
"learning_rate": 8.07149848479403e-05,
"loss": 0.059,
"step": 9750
},
{
"grad_norm": 0.28275421261787415,
"learning_rate": 8.067147592736962e-05,
"loss": 0.0526,
"step": 9760
},
{
"grad_norm": 0.297232061624527,
"learning_rate": 8.062792973806357e-05,
"loss": 0.0522,
"step": 9770
},
{
"grad_norm": 0.2925558090209961,
"learning_rate": 8.058434633293485e-05,
"loss": 0.049,
"step": 9780
},
{
"grad_norm": 0.27258849143981934,
"learning_rate": 8.054072576494143e-05,
"loss": 0.0549,
"step": 9790
},
{
"grad_norm": 0.2997378706932068,
"learning_rate": 8.04970680870864e-05,
"loss": 0.0502,
"step": 9800
},
{
"grad_norm": 0.3086252212524414,
"learning_rate": 8.045337335241793e-05,
"loss": 0.0574,
"step": 9810
},
{
"grad_norm": 0.32624274492263794,
"learning_rate": 8.040964161402932e-05,
"loss": 0.0566,
"step": 9820
},
{
"grad_norm": 0.32253265380859375,
"learning_rate": 8.036587292505869e-05,
"loss": 0.0588,
"step": 9830
},
{
"grad_norm": 0.2653786540031433,
"learning_rate": 8.032206733868912e-05,
"loss": 0.0602,
"step": 9840
},
{
"grad_norm": 0.33071067929267883,
"learning_rate": 8.027822490814858e-05,
"loss": 0.0603,
"step": 9850
},
{
"grad_norm": 0.29406294226646423,
"learning_rate": 8.023434568670972e-05,
"loss": 0.0592,
"step": 9860
},
{
"grad_norm": 0.3208906054496765,
"learning_rate": 8.019042972768992e-05,
"loss": 0.0557,
"step": 9870
},
{
"grad_norm": 0.2623123526573181,
"learning_rate": 8.014647708445124e-05,
"loss": 0.0566,
"step": 9880
},
{
"grad_norm": 0.34910595417022705,
"learning_rate": 8.010248781040027e-05,
"loss": 0.0629,
"step": 9890
},
{
"grad_norm": 0.33865371346473694,
"learning_rate": 8.005846195898814e-05,
"loss": 0.0563,
"step": 9900
},
{
"grad_norm": 0.33840838074684143,
"learning_rate": 8.00143995837104e-05,
"loss": 0.053,
"step": 9910
},
{
"grad_norm": 0.4690181612968445,
"learning_rate": 7.997030073810698e-05,
"loss": 0.0555,
"step": 9920
},
{
"grad_norm": 0.3174230754375458,
"learning_rate": 7.992616547576217e-05,
"loss": 0.0534,
"step": 9930
},
{
"grad_norm": 0.2898480296134949,
"learning_rate": 7.988199385030445e-05,
"loss": 0.0606,
"step": 9940
},
{
"grad_norm": 0.2929697036743164,
"learning_rate": 7.98377859154065e-05,
"loss": 0.0599,
"step": 9950
},
{
"grad_norm": 0.38996651768684387,
"learning_rate": 7.979354172478516e-05,
"loss": 0.0554,
"step": 9960
},
{
"grad_norm": 0.3246240019798279,
"learning_rate": 7.974926133220127e-05,
"loss": 0.0547,
"step": 9970
},
{
"grad_norm": 0.31366217136383057,
"learning_rate": 7.970494479145967e-05,
"loss": 0.0544,
"step": 9980
},
{
"grad_norm": 0.2537480592727661,
"learning_rate": 7.966059215640919e-05,
"loss": 0.059,
"step": 9990
},
{
"grad_norm": 0.2626407742500305,
"learning_rate": 7.961620348094241e-05,
"loss": 0.0554,
"step": 10000
},
{
"grad_norm": 0.2957606613636017,
"learning_rate": 7.957177881899577e-05,
"loss": 0.0577,
"step": 10010
},
{
"grad_norm": 0.30218595266342163,
"learning_rate": 7.952731822454944e-05,
"loss": 0.0554,
"step": 10020
},
{
"grad_norm": 0.4015539884567261,
"learning_rate": 7.948282175162722e-05,
"loss": 0.0545,
"step": 10030
},
{
"grad_norm": 0.3823576867580414,
"learning_rate": 7.943828945429652e-05,
"loss": 0.0557,
"step": 10040
},
{
"grad_norm": 0.32471245527267456,
"learning_rate": 7.939372138666827e-05,
"loss": 0.0543,
"step": 10050
},
{
"grad_norm": 0.23842814564704895,
"learning_rate": 7.934911760289692e-05,
"loss": 0.0551,
"step": 10060
},
{
"grad_norm": 0.2753767669200897,
"learning_rate": 7.930447815718022e-05,
"loss": 0.058,
"step": 10070
},
{
"grad_norm": 0.2852107286453247,
"learning_rate": 7.925980310375934e-05,
"loss": 0.0536,
"step": 10080
},
{
"grad_norm": 0.2800537943840027,
"learning_rate": 7.921509249691865e-05,
"loss": 0.0527,
"step": 10090
},
{
"grad_norm": 0.3370341956615448,
"learning_rate": 7.917034639098579e-05,
"loss": 0.0567,
"step": 10100
},
{
"grad_norm": 0.34083855152130127,
"learning_rate": 7.912556484033146e-05,
"loss": 0.0539,
"step": 10110
},
{
"grad_norm": 0.23762837052345276,
"learning_rate": 7.908074789936952e-05,
"loss": 0.051,
"step": 10120
},
{
"grad_norm": 0.278666615486145,
"learning_rate": 7.903589562255673e-05,
"loss": 0.0534,
"step": 10130
},
{
"grad_norm": 0.3003731667995453,
"learning_rate": 7.899100806439285e-05,
"loss": 0.0551,
"step": 10140
},
{
"grad_norm": 0.25763794779777527,
"learning_rate": 7.894608527942049e-05,
"loss": 0.0547,
"step": 10150
},
{
"grad_norm": 0.2735033333301544,
"learning_rate": 7.89011273222251e-05,
"loss": 0.055,
"step": 10160
},
{
"grad_norm": 0.2910924553871155,
"learning_rate": 7.885613424743482e-05,
"loss": 0.0588,
"step": 10170
},
{
"grad_norm": 0.23881179094314575,
"learning_rate": 7.881110610972044e-05,
"loss": 0.0551,
"step": 10180
},
{
"grad_norm": 0.20886707305908203,
"learning_rate": 7.876604296379544e-05,
"loss": 0.0533,
"step": 10190
},
{
"grad_norm": 0.3306140899658203,
"learning_rate": 7.872094486441579e-05,
"loss": 0.0579,
"step": 10200
},
{
"grad_norm": 0.23585818707942963,
"learning_rate": 7.867581186637991e-05,
"loss": 0.0621,
"step": 10210
},
{
"grad_norm": 0.32251736521720886,
"learning_rate": 7.863064402452866e-05,
"loss": 0.0546,
"step": 10220
},
{
"grad_norm": 0.29432475566864014,
"learning_rate": 7.858544139374524e-05,
"loss": 0.0589,
"step": 10230
},
{
"grad_norm": 0.2866153419017792,
"learning_rate": 7.854020402895508e-05,
"loss": 0.0552,
"step": 10240
},
{
"grad_norm": 0.3613327443599701,
"learning_rate": 7.849493198512587e-05,
"loss": 0.0536,
"step": 10250
},
{
"grad_norm": 0.3706571161746979,
"learning_rate": 7.84496253172674e-05,
"loss": 0.0578,
"step": 10260
},
{
"grad_norm": 0.26008591055870056,
"learning_rate": 7.840428408043155e-05,
"loss": 0.0543,
"step": 10270
},
{
"grad_norm": 0.296407550573349,
"learning_rate": 7.835890832971218e-05,
"loss": 0.0622,
"step": 10280
},
{
"grad_norm": 0.2628116309642792,
"learning_rate": 7.831349812024512e-05,
"loss": 0.058,
"step": 10290
},
{
"grad_norm": 0.35624197125434875,
"learning_rate": 7.826805350720807e-05,
"loss": 0.0506,
"step": 10300
},
{
"grad_norm": 0.344029039144516,
"learning_rate": 7.822257454582049e-05,
"loss": 0.0589,
"step": 10310
},
{
"grad_norm": 0.3383975625038147,
"learning_rate": 7.817706129134363e-05,
"loss": 0.0653,
"step": 10320
},
{
"grad_norm": 0.37289443612098694,
"learning_rate": 7.813151379908036e-05,
"loss": 0.0524,
"step": 10330
},
{
"grad_norm": 0.32714107632637024,
"learning_rate": 7.808593212437523e-05,
"loss": 0.0539,
"step": 10340
},
{
"grad_norm": 0.29376092553138733,
"learning_rate": 7.804031632261421e-05,
"loss": 0.056,
"step": 10350
},
{
"grad_norm": 0.27209872007369995,
"learning_rate": 7.799466644922484e-05,
"loss": 0.0477,
"step": 10360
},
{
"grad_norm": 0.3670084476470947,
"learning_rate": 7.794898255967602e-05,
"loss": 0.0572,
"step": 10370
},
{
"grad_norm": 0.3195774555206299,
"learning_rate": 7.790326470947795e-05,
"loss": 0.0583,
"step": 10380
},
{
"grad_norm": 0.345571905374527,
"learning_rate": 7.785751295418217e-05,
"loss": 0.0551,
"step": 10390
},
{
"grad_norm": 0.29521915316581726,
"learning_rate": 7.781172734938136e-05,
"loss": 0.0581,
"step": 10400
},
{
"grad_norm": 0.2645949125289917,
"learning_rate": 7.776590795070933e-05,
"loss": 0.051,
"step": 10410
},
{
"grad_norm": 0.26568496227264404,
"learning_rate": 7.772005481384099e-05,
"loss": 0.0551,
"step": 10420
},
{
"grad_norm": 0.2961045503616333,
"learning_rate": 7.767416799449223e-05,
"loss": 0.0519,
"step": 10430
},
{
"grad_norm": 0.27615880966186523,
"learning_rate": 7.762824754841984e-05,
"loss": 0.0545,
"step": 10440
},
{
"grad_norm": 0.28159958124160767,
"learning_rate": 7.758229353142152e-05,
"loss": 0.0529,
"step": 10450
},
{
"grad_norm": 0.24424229562282562,
"learning_rate": 7.753630599933572e-05,
"loss": 0.0476,
"step": 10460
},
{
"grad_norm": 0.28863927721977234,
"learning_rate": 7.749028500804161e-05,
"loss": 0.0487,
"step": 10470
},
{
"grad_norm": 0.3881639540195465,
"learning_rate": 7.744423061345906e-05,
"loss": 0.0492,
"step": 10480
},
{
"grad_norm": 0.29462048411369324,
"learning_rate": 7.73981428715485e-05,
"loss": 0.0502,
"step": 10490
},
{
"grad_norm": 0.3249775767326355,
"learning_rate": 7.735202183831085e-05,
"loss": 0.0529,
"step": 10500
},
{
"grad_norm": 0.26566651463508606,
"learning_rate": 7.730586756978757e-05,
"loss": 0.0547,
"step": 10510
},
{
"grad_norm": 0.3077145218849182,
"learning_rate": 7.72596801220604e-05,
"loss": 0.0561,
"step": 10520
},
{
"grad_norm": 0.272254079580307,
"learning_rate": 7.721345955125147e-05,
"loss": 0.0541,
"step": 10530
},
{
"grad_norm": 0.24374987185001373,
"learning_rate": 7.71672059135231e-05,
"loss": 0.0499,
"step": 10540
},
{
"grad_norm": 0.2858620285987854,
"learning_rate": 7.712091926507787e-05,
"loss": 0.046,
"step": 10550
},
{
"grad_norm": 0.2695977985858917,
"learning_rate": 7.70745996621584e-05,
"loss": 0.048,
"step": 10560
},
{
"grad_norm": 0.2367054671049118,
"learning_rate": 7.702824716104735e-05,
"loss": 0.0482,
"step": 10570
},
{
"grad_norm": 0.28835031390190125,
"learning_rate": 7.698186181806743e-05,
"loss": 0.0481,
"step": 10580
},
{
"grad_norm": 0.3011409044265747,
"learning_rate": 7.693544368958116e-05,
"loss": 0.0571,
"step": 10590
},
{
"grad_norm": 0.22110246121883392,
"learning_rate": 7.688899283199096e-05,
"loss": 0.0494,
"step": 10600
},
{
"grad_norm": 0.2895663380622864,
"learning_rate": 7.684250930173901e-05,
"loss": 0.0537,
"step": 10610
},
{
"grad_norm": 0.26278984546661377,
"learning_rate": 7.679599315530718e-05,
"loss": 0.046,
"step": 10620
},
{
"grad_norm": 0.2954424321651459,
"learning_rate": 7.674944444921695e-05,
"loss": 0.0568,
"step": 10630
},
{
"grad_norm": 0.2598683536052704,
"learning_rate": 7.670286324002944e-05,
"loss": 0.0505,
"step": 10640
},
{
"grad_norm": 0.27540767192840576,
"learning_rate": 7.665624958434514e-05,
"loss": 0.0493,
"step": 10650
},
{
"grad_norm": 0.27130767703056335,
"learning_rate": 7.66096035388041e-05,
"loss": 0.0541,
"step": 10660
},
{
"grad_norm": 0.3153267502784729,
"learning_rate": 7.656292516008563e-05,
"loss": 0.0523,
"step": 10670
},
{
"grad_norm": 0.30076658725738525,
"learning_rate": 7.651621450490837e-05,
"loss": 0.0517,
"step": 10680
},
{
"grad_norm": 0.2649375796318054,
"learning_rate": 7.646947163003017e-05,
"loss": 0.045,
"step": 10690
},
{
"grad_norm": 0.23141999542713165,
"learning_rate": 7.642269659224803e-05,
"loss": 0.0513,
"step": 10700
},
{
"grad_norm": 0.4023504853248596,
"learning_rate": 7.637588944839803e-05,
"loss": 0.0563,
"step": 10710
},
{
"grad_norm": 0.29016342759132385,
"learning_rate": 7.632905025535528e-05,
"loss": 0.0508,
"step": 10720
},
{
"grad_norm": 0.2949693202972412,
"learning_rate": 7.628217907003378e-05,
"loss": 0.0504,
"step": 10730
},
{
"grad_norm": 0.3682021200656891,
"learning_rate": 7.623527594938649e-05,
"loss": 0.0583,
"step": 10740
},
{
"grad_norm": 0.33373114466667175,
"learning_rate": 7.618834095040509e-05,
"loss": 0.0558,
"step": 10750
},
{
"grad_norm": 0.30852210521698,
"learning_rate": 7.614137413012e-05,
"loss": 0.0539,
"step": 10760
},
{
"grad_norm": 0.3553842604160309,
"learning_rate": 7.609437554560042e-05,
"loss": 0.0533,
"step": 10770
},
{
"grad_norm": 0.28333380818367004,
"learning_rate": 7.604734525395398e-05,
"loss": 0.0485,
"step": 10780
},
{
"grad_norm": 0.30952954292297363,
"learning_rate": 7.600028331232699e-05,
"loss": 0.0546,
"step": 10790
},
{
"grad_norm": 0.36419758200645447,
"learning_rate": 7.595318977790408e-05,
"loss": 0.0486,
"step": 10800
},
{
"grad_norm": 0.4144981801509857,
"learning_rate": 7.590606470790836e-05,
"loss": 0.0561,
"step": 10810
},
{
"grad_norm": 0.38663849234580994,
"learning_rate": 7.585890815960125e-05,
"loss": 0.058,
"step": 10820
},
{
"grad_norm": 0.33041051030158997,
"learning_rate": 7.581172019028238e-05,
"loss": 0.0558,
"step": 10830
},
{
"grad_norm": 0.28932985663414,
"learning_rate": 7.576450085728959e-05,
"loss": 0.0522,
"step": 10840
},
{
"grad_norm": 0.26190751791000366,
"learning_rate": 7.571725021799885e-05,
"loss": 0.0504,
"step": 10850
},
{
"grad_norm": 0.2685620188713074,
"learning_rate": 7.566996832982409e-05,
"loss": 0.0471,
"step": 10860
},
{
"grad_norm": 0.2793464660644531,
"learning_rate": 7.56226552502173e-05,
"loss": 0.0544,
"step": 10870
},
{
"grad_norm": 0.3538641631603241,
"learning_rate": 7.557531103666833e-05,
"loss": 0.0539,
"step": 10880
},
{
"grad_norm": 0.33281540870666504,
"learning_rate": 7.552793574670485e-05,
"loss": 0.0534,
"step": 10890
},
{
"grad_norm": 0.32615673542022705,
"learning_rate": 7.54805294378923e-05,
"loss": 0.0551,
"step": 10900
},
{
"grad_norm": 0.2785199284553528,
"learning_rate": 7.543309216783384e-05,
"loss": 0.0513,
"step": 10910
},
{
"grad_norm": 0.3233277201652527,
"learning_rate": 7.53856239941702e-05,
"loss": 0.0483,
"step": 10920
},
{
"grad_norm": 0.2624678313732147,
"learning_rate": 7.533812497457972e-05,
"loss": 0.0538,
"step": 10930
},
{
"grad_norm": 0.30161261558532715,
"learning_rate": 7.529059516677814e-05,
"loss": 0.0549,
"step": 10940
},
{
"grad_norm": 0.3129892945289612,
"learning_rate": 7.524303462851872e-05,
"loss": 0.0512,
"step": 10950
},
{
"grad_norm": 0.2751290202140808,
"learning_rate": 7.519544341759192e-05,
"loss": 0.0516,
"step": 10960
},
{
"grad_norm": 0.3377801775932312,
"learning_rate": 7.514782159182562e-05,
"loss": 0.0512,
"step": 10970
},
{
"grad_norm": 0.2826242446899414,
"learning_rate": 7.510016920908481e-05,
"loss": 0.0489,
"step": 10980
},
{
"grad_norm": 0.391279935836792,
"learning_rate": 7.505248632727158e-05,
"loss": 0.0581,
"step": 10990
},
{
"grad_norm": 0.3448256850242615,
"learning_rate": 7.50047730043252e-05,
"loss": 0.0561,
"step": 11000
},
{
"grad_norm": 0.2837821841239929,
"learning_rate": 7.495702929822183e-05,
"loss": 0.051,
"step": 11010
},
{
"grad_norm": 0.25449007749557495,
"learning_rate": 7.490925526697455e-05,
"loss": 0.0537,
"step": 11020
},
{
"grad_norm": 0.29622283577919006,
"learning_rate": 7.486145096863334e-05,
"loss": 0.0544,
"step": 11030
},
{
"grad_norm": 0.28924697637557983,
"learning_rate": 7.481361646128491e-05,
"loss": 0.046,
"step": 11040
},
{
"grad_norm": 0.23967409133911133,
"learning_rate": 7.476575180305271e-05,
"loss": 0.0513,
"step": 11050
},
{
"grad_norm": 0.24662648141384125,
"learning_rate": 7.471785705209682e-05,
"loss": 0.0455,
"step": 11060
},
{
"grad_norm": 0.238308385014534,
"learning_rate": 7.466993226661387e-05,
"loss": 0.0498,
"step": 11070
},
{
"grad_norm": 0.3129105269908905,
"learning_rate": 7.462197750483699e-05,
"loss": 0.0557,
"step": 11080
},
{
"grad_norm": 0.3350023627281189,
"learning_rate": 7.457399282503574e-05,
"loss": 0.0506,
"step": 11090
},
{
"grad_norm": 0.2854120433330536,
"learning_rate": 7.452597828551604e-05,
"loss": 0.0536,
"step": 11100
},
{
"grad_norm": 0.3024255931377411,
"learning_rate": 7.447793394462006e-05,
"loss": 0.054,
"step": 11110
},
{
"grad_norm": 0.2673250138759613,
"learning_rate": 7.442985986072624e-05,
"loss": 0.0527,
"step": 11120
},
{
"grad_norm": 0.2633294463157654,
"learning_rate": 7.438175609224908e-05,
"loss": 0.0506,
"step": 11130
},
{
"grad_norm": 0.2514217495918274,
"learning_rate": 7.433362269763924e-05,
"loss": 0.06,
"step": 11140
},
{
"grad_norm": 0.26395174860954285,
"learning_rate": 7.428545973538329e-05,
"loss": 0.0568,
"step": 11150
},
{
"grad_norm": 0.26760855317115784,
"learning_rate": 7.42372672640038e-05,
"loss": 0.0559,
"step": 11160
},
{
"grad_norm": 0.24121910333633423,
"learning_rate": 7.418904534205917e-05,
"loss": 0.0524,
"step": 11170
},
{
"grad_norm": 0.2505238950252533,
"learning_rate": 7.414079402814356e-05,
"loss": 0.0553,
"step": 11180
},
{
"grad_norm": 0.23185905814170837,
"learning_rate": 7.409251338088687e-05,
"loss": 0.0541,
"step": 11190
},
{
"grad_norm": 0.32485470175743103,
"learning_rate": 7.404420345895467e-05,
"loss": 0.061,
"step": 11200
},
{
"grad_norm": 0.29086700081825256,
"learning_rate": 7.399586432104804e-05,
"loss": 0.0614,
"step": 11210
},
{
"grad_norm": 0.23832561075687408,
"learning_rate": 7.394749602590358e-05,
"loss": 0.058,
"step": 11220
},
{
"grad_norm": 0.2901882827281952,
"learning_rate": 7.389909863229336e-05,
"loss": 0.0642,
"step": 11230
},
{
"grad_norm": 0.2207191437482834,
"learning_rate": 7.385067219902477e-05,
"loss": 0.0525,
"step": 11240
},
{
"grad_norm": 0.29125672578811646,
"learning_rate": 7.380221678494049e-05,
"loss": 0.0608,
"step": 11250
},
{
"grad_norm": 0.2869950532913208,
"learning_rate": 7.37537324489184e-05,
"loss": 0.0509,
"step": 11260
},
{
"grad_norm": 0.2855561375617981,
"learning_rate": 7.370521924987155e-05,
"loss": 0.0549,
"step": 11270
},
{
"grad_norm": 0.253984272480011,
"learning_rate": 7.365667724674806e-05,
"loss": 0.0547,
"step": 11280
},
{
"grad_norm": 0.3805805444717407,
"learning_rate": 7.360810649853105e-05,
"loss": 0.0529,
"step": 11290
},
{
"grad_norm": 0.298690527677536,
"learning_rate": 7.355950706423853e-05,
"loss": 0.0518,
"step": 11300
},
{
"grad_norm": 0.23842504620552063,
"learning_rate": 7.351087900292342e-05,
"loss": 0.0495,
"step": 11310
},
{
"grad_norm": 0.2545222043991089,
"learning_rate": 7.346222237367339e-05,
"loss": 0.0451,
"step": 11320
},
{
"grad_norm": 0.276802122592926,
"learning_rate": 7.341353723561081e-05,
"loss": 0.0512,
"step": 11330
},
{
"grad_norm": 0.28438711166381836,
"learning_rate": 7.336482364789277e-05,
"loss": 0.046,
"step": 11340
},
{
"grad_norm": 0.22939875721931458,
"learning_rate": 7.331608166971082e-05,
"loss": 0.0433,
"step": 11350
},
{
"grad_norm": 0.21980120241641998,
"learning_rate": 7.326731136029108e-05,
"loss": 0.0393,
"step": 11360
},
{
"grad_norm": 0.23749177157878876,
"learning_rate": 7.321851277889408e-05,
"loss": 0.0485,
"step": 11370
},
{
"grad_norm": 0.2049066126346588,
"learning_rate": 7.316968598481469e-05,
"loss": 0.0427,
"step": 11380
},
{
"grad_norm": 0.26963067054748535,
"learning_rate": 7.312083103738206e-05,
"loss": 0.0427,
"step": 11390
},
{
"grad_norm": 0.2776131331920624,
"learning_rate": 7.307194799595958e-05,
"loss": 0.0427,
"step": 11400
},
{
"grad_norm": 0.23154111206531525,
"learning_rate": 7.302303691994473e-05,
"loss": 0.0481,
"step": 11410
},
{
"grad_norm": 0.2440502941608429,
"learning_rate": 7.29740978687691e-05,
"loss": 0.0507,
"step": 11420
},
{
"grad_norm": 0.26325440406799316,
"learning_rate": 7.292513090189825e-05,
"loss": 0.0454,
"step": 11430
},
{
"grad_norm": 0.25071603059768677,
"learning_rate": 7.287613607883163e-05,
"loss": 0.0492,
"step": 11440
},
{
"grad_norm": 0.20199261605739594,
"learning_rate": 7.282711345910263e-05,
"loss": 0.0448,
"step": 11450
},
{
"grad_norm": 0.3465411365032196,
"learning_rate": 7.27780631022783e-05,
"loss": 0.054,
"step": 11460
},
{
"grad_norm": 0.24728655815124512,
"learning_rate": 7.272898506795948e-05,
"loss": 0.0514,
"step": 11470
},
{
"grad_norm": 0.16234205663204193,
"learning_rate": 7.267987941578058e-05,
"loss": 0.048,
"step": 11480
},
{
"grad_norm": 0.24996469914913177,
"learning_rate": 7.263074620540962e-05,
"loss": 0.0496,
"step": 11490
},
{
"grad_norm": 0.3259834945201874,
"learning_rate": 7.25815854965481e-05,
"loss": 0.0523,
"step": 11500
},
{
"grad_norm": 0.33556458353996277,
"learning_rate": 7.253239734893089e-05,
"loss": 0.0463,
"step": 11510
},
{
"grad_norm": 0.28204345703125,
"learning_rate": 7.248318182232623e-05,
"loss": 0.0517,
"step": 11520
},
{
"grad_norm": 0.30604133009910583,
"learning_rate": 7.243393897653565e-05,
"loss": 0.0484,
"step": 11530
},
{
"grad_norm": 0.23240718245506287,
"learning_rate": 7.23846688713938e-05,
"loss": 0.0455,
"step": 11540
},
{
"grad_norm": 0.26025390625,
"learning_rate": 7.233537156676854e-05,
"loss": 0.0505,
"step": 11550
},
{
"grad_norm": 0.2843422591686249,
"learning_rate": 7.228604712256076e-05,
"loss": 0.0534,
"step": 11560
},
{
"grad_norm": 0.2540539801120758,
"learning_rate": 7.223669559870426e-05,
"loss": 0.0488,
"step": 11570
},
{
"grad_norm": 0.3193459212779999,
"learning_rate": 7.218731705516585e-05,
"loss": 0.0555,
"step": 11580
},
{
"grad_norm": 0.29101163148880005,
"learning_rate": 7.21379115519451e-05,
"loss": 0.0588,
"step": 11590
},
{
"grad_norm": 0.3162247836589813,
"learning_rate": 7.20884791490743e-05,
"loss": 0.0583,
"step": 11600
},
{
"grad_norm": 0.24247194826602936,
"learning_rate": 7.203901990661857e-05,
"loss": 0.0541,
"step": 11610
},
{
"grad_norm": 0.25254520773887634,
"learning_rate": 7.198953388467549e-05,
"loss": 0.0532,
"step": 11620
},
{
"grad_norm": 0.26075753569602966,
"learning_rate": 7.194002114337526e-05,
"loss": 0.0517,
"step": 11630
},
{
"grad_norm": 0.28647321462631226,
"learning_rate": 7.189048174288054e-05,
"loss": 0.0536,
"step": 11640
},
{
"grad_norm": 0.2290695309638977,
"learning_rate": 7.184091574338636e-05,
"loss": 0.0505,
"step": 11650
},
{
"grad_norm": 0.27686601877212524,
"learning_rate": 7.179132320512009e-05,
"loss": 0.0512,
"step": 11660
},
{
"grad_norm": 0.2697356343269348,
"learning_rate": 7.174170418834134e-05,
"loss": 0.049,
"step": 11670
},
{
"grad_norm": 0.2826821506023407,
"learning_rate": 7.169205875334189e-05,
"loss": 0.0479,
"step": 11680
},
{
"grad_norm": 0.25433579087257385,
"learning_rate": 7.164238696044562e-05,
"loss": 0.0498,
"step": 11690
},
{
"grad_norm": 0.25723183155059814,
"learning_rate": 7.159268887000846e-05,
"loss": 0.0532,
"step": 11700
},
{
"grad_norm": 0.3720298707485199,
"learning_rate": 7.154296454241827e-05,
"loss": 0.0548,
"step": 11710
},
{
"grad_norm": 0.2573554813861847,
"learning_rate": 7.149321403809479e-05,
"loss": 0.0473,
"step": 11720
},
{
"grad_norm": 0.3220192790031433,
"learning_rate": 7.144343741748956e-05,
"loss": 0.0461,
"step": 11730
},
{
"grad_norm": 0.21101906895637512,
"learning_rate": 7.139363474108589e-05,
"loss": 0.049,
"step": 11740
},
{
"grad_norm": 0.21838796138763428,
"learning_rate": 7.134380606939874e-05,
"loss": 0.0474,
"step": 11750
},
{
"grad_norm": 0.2753686010837555,
"learning_rate": 7.12939514629746e-05,
"loss": 0.0529,
"step": 11760
},
{
"grad_norm": 0.2979985475540161,
"learning_rate": 7.124407098239155e-05,
"loss": 0.044,
"step": 11770
},
{
"grad_norm": 0.2618331015110016,
"learning_rate": 7.119416468825908e-05,
"loss": 0.0477,
"step": 11780
},
{
"grad_norm": 0.2861156165599823,
"learning_rate": 7.114423264121804e-05,
"loss": 0.051,
"step": 11790
},
{
"grad_norm": 0.23823212087154388,
"learning_rate": 7.109427490194056e-05,
"loss": 0.0474,
"step": 11800
},
{
"grad_norm": 0.30353397130966187,
"learning_rate": 7.104429153113001e-05,
"loss": 0.0426,
"step": 11810
},
{
"grad_norm": 0.2628457546234131,
"learning_rate": 7.099428258952092e-05,
"loss": 0.0492,
"step": 11820
},
{
"grad_norm": 0.29817619919776917,
"learning_rate": 7.094424813787883e-05,
"loss": 0.0491,
"step": 11830
},
{
"grad_norm": 0.2458418756723404,
"learning_rate": 7.089418823700034e-05,
"loss": 0.0445,
"step": 11840
},
{
"grad_norm": 0.24208581447601318,
"learning_rate": 7.084410294771298e-05,
"loss": 0.0508,
"step": 11850
},
{
"grad_norm": 0.3362455666065216,
"learning_rate": 7.079399233087504e-05,
"loss": 0.0498,
"step": 11860
},
{
"grad_norm": 0.29550471901893616,
"learning_rate": 7.074385644737568e-05,
"loss": 0.0564,
"step": 11870
},
{
"grad_norm": 0.3065795600414276,
"learning_rate": 7.069369535813473e-05,
"loss": 0.0499,
"step": 11880
},
{
"grad_norm": 0.2721593379974365,
"learning_rate": 7.06435091241026e-05,
"loss": 0.0553,
"step": 11890
},
{
"grad_norm": 0.2991701066493988,
"learning_rate": 7.059329780626034e-05,
"loss": 0.0567,
"step": 11900
},
{
"grad_norm": 0.3451690375804901,
"learning_rate": 7.054306146561944e-05,
"loss": 0.057,
"step": 11910
},
{
"grad_norm": 0.27689701318740845,
"learning_rate": 7.049280016322178e-05,
"loss": 0.0527,
"step": 11920
},
{
"grad_norm": 0.23240388929843903,
"learning_rate": 7.044251396013957e-05,
"loss": 0.0533,
"step": 11930
},
{
"grad_norm": 0.34827688336372375,
"learning_rate": 7.039220291747528e-05,
"loss": 0.0595,
"step": 11940
},
{
"grad_norm": 0.337442547082901,
"learning_rate": 7.034186709636159e-05,
"loss": 0.0559,
"step": 11950
},
{
"grad_norm": 0.2699776291847229,
"learning_rate": 7.029150655796129e-05,
"loss": 0.0524,
"step": 11960
},
{
"grad_norm": 0.23389935493469238,
"learning_rate": 7.024112136346712e-05,
"loss": 0.0614,
"step": 11970
},
{
"grad_norm": 0.3306303918361664,
"learning_rate": 7.01907115741019e-05,
"loss": 0.0524,
"step": 11980
},
{
"grad_norm": 0.3217543065547943,
"learning_rate": 7.014027725111825e-05,
"loss": 0.0508,
"step": 11990
},
{
"grad_norm": 0.27304449677467346,
"learning_rate": 7.008981845579864e-05,
"loss": 0.0492,
"step": 12000
},
{
"grad_norm": 0.2922233045101166,
"learning_rate": 7.003933524945528e-05,
"loss": 0.0501,
"step": 12010
},
{
"grad_norm": 0.2665089964866638,
"learning_rate": 6.998882769342998e-05,
"loss": 0.0551,
"step": 12020
},
{
"grad_norm": 0.3386456370353699,
"learning_rate": 6.993829584909423e-05,
"loss": 0.0551,
"step": 12030
},
{
"grad_norm": 0.3269071877002716,
"learning_rate": 6.988773977784895e-05,
"loss": 0.0528,
"step": 12040
},
{
"grad_norm": 0.30471569299697876,
"learning_rate": 6.983715954112454e-05,
"loss": 0.0591,
"step": 12050
},
{
"grad_norm": 0.2722010016441345,
"learning_rate": 6.978655520038079e-05,
"loss": 0.059,
"step": 12060
},
{
"grad_norm": 0.24576807022094727,
"learning_rate": 6.97359268171067e-05,
"loss": 0.0498,
"step": 12070
},
{
"grad_norm": 0.3330242931842804,
"learning_rate": 6.968527445282056e-05,
"loss": 0.05,
"step": 12080
},
{
"grad_norm": 0.36369574069976807,
"learning_rate": 6.963459816906977e-05,
"loss": 0.0471,
"step": 12090
},
{
"grad_norm": 0.281669020652771,
"learning_rate": 6.958389802743077e-05,
"loss": 0.0505,
"step": 12100
},
{
"grad_norm": 0.24405209720134735,
"learning_rate": 6.953317408950903e-05,
"loss": 0.0475,
"step": 12110
},
{
"grad_norm": 0.2823590636253357,
"learning_rate": 6.94824264169389e-05,
"loss": 0.0519,
"step": 12120
},
{
"grad_norm": 0.25073617696762085,
"learning_rate": 6.94316550713836e-05,
"loss": 0.0468,
"step": 12130
},
{
"grad_norm": 0.18889234960079193,
"learning_rate": 6.938086011453513e-05,
"loss": 0.0486,
"step": 12140
},
{
"grad_norm": 0.21192562580108643,
"learning_rate": 6.93300416081141e-05,
"loss": 0.0464,
"step": 12150
},
{
"grad_norm": 0.3048607409000397,
"learning_rate": 6.927919961386984e-05,
"loss": 0.0576,
"step": 12160
},
{
"grad_norm": 0.33697861433029175,
"learning_rate": 6.922833419358013e-05,
"loss": 0.0525,
"step": 12170
},
{
"grad_norm": 0.29246458411216736,
"learning_rate": 6.917744540905124e-05,
"loss": 0.0539,
"step": 12180
},
{
"grad_norm": 0.30944833159446716,
"learning_rate": 6.912653332211787e-05,
"loss": 0.0577,
"step": 12190
},
{
"grad_norm": 0.2835339903831482,
"learning_rate": 6.9075597994643e-05,
"loss": 0.0497,
"step": 12200
},
{
"grad_norm": 0.29097044467926025,
"learning_rate": 6.902463948851787e-05,
"loss": 0.06,
"step": 12210
},
{
"grad_norm": 0.3251795470714569,
"learning_rate": 6.897365786566184e-05,
"loss": 0.0516,
"step": 12220
},
{
"grad_norm": 0.3100990355014801,
"learning_rate": 6.892265318802242e-05,
"loss": 0.0613,
"step": 12230
},
{
"grad_norm": 0.2956431806087494,
"learning_rate": 6.887162551757507e-05,
"loss": 0.0506,
"step": 12240
},
{
"grad_norm": 0.2606987953186035,
"learning_rate": 6.882057491632326e-05,
"loss": 0.0553,
"step": 12250
},
{
"grad_norm": 0.2829456329345703,
"learning_rate": 6.876950144629824e-05,
"loss": 0.052,
"step": 12260
},
{
"grad_norm": 0.3198920488357544,
"learning_rate": 6.87184051695591e-05,
"loss": 0.0557,
"step": 12270
},
{
"grad_norm": 0.3255424499511719,
"learning_rate": 6.866728614819268e-05,
"loss": 0.0555,
"step": 12280
},
{
"grad_norm": 0.324792742729187,
"learning_rate": 6.861614444431337e-05,
"loss": 0.0551,
"step": 12290
},
{
"grad_norm": 0.2300267219543457,
"learning_rate": 6.856498012006318e-05,
"loss": 0.0518,
"step": 12300
},
{
"grad_norm": 0.3206157982349396,
"learning_rate": 6.851379323761157e-05,
"loss": 0.0541,
"step": 12310
},
{
"grad_norm": 0.23384563624858856,
"learning_rate": 6.846258385915545e-05,
"loss": 0.0515,
"step": 12320
},
{
"grad_norm": 0.27740728855133057,
"learning_rate": 6.841135204691902e-05,
"loss": 0.0513,
"step": 12330
},
{
"grad_norm": 0.29553917050361633,
"learning_rate": 6.836009786315377e-05,
"loss": 0.0526,
"step": 12340
},
{
"grad_norm": 0.263065367937088,
"learning_rate": 6.830882137013839e-05,
"loss": 0.0525,
"step": 12350
},
{
"grad_norm": 0.3303307890892029,
"learning_rate": 6.825752263017863e-05,
"loss": 0.0491,
"step": 12360
},
{
"grad_norm": 0.2839193344116211,
"learning_rate": 6.820620170560731e-05,
"loss": 0.049,
"step": 12370
},
{
"grad_norm": 0.23940905928611755,
"learning_rate": 6.815485865878418e-05,
"loss": 0.0561,
"step": 12380
},
{
"grad_norm": 0.28745949268341064,
"learning_rate": 6.810349355209588e-05,
"loss": 0.048,
"step": 12390
},
{
"grad_norm": 0.2896861135959625,
"learning_rate": 6.805210644795588e-05,
"loss": 0.0574,
"step": 12400
},
{
"grad_norm": 0.30991312861442566,
"learning_rate": 6.800069740880433e-05,
"loss": 0.0543,
"step": 12410
},
{
"grad_norm": 0.23973147571086884,
"learning_rate": 6.794926649710807e-05,
"loss": 0.0514,
"step": 12420
},
{
"grad_norm": 0.2702583968639374,
"learning_rate": 6.789781377536052e-05,
"loss": 0.052,
"step": 12430
},
{
"grad_norm": 0.29322734475135803,
"learning_rate": 6.784633930608158e-05,
"loss": 0.0482,
"step": 12440
},
{
"grad_norm": 0.2758643329143524,
"learning_rate": 6.779484315181759e-05,
"loss": 0.0474,
"step": 12450
},
{
"grad_norm": 0.3207719624042511,
"learning_rate": 6.774332537514122e-05,
"loss": 0.0523,
"step": 12460
},
{
"grad_norm": 0.19675292074680328,
"learning_rate": 6.769178603865143e-05,
"loss": 0.0492,
"step": 12470
},
{
"grad_norm": 0.297958642244339,
"learning_rate": 6.764022520497337e-05,
"loss": 0.0528,
"step": 12480
},
{
"grad_norm": 0.23101019859313965,
"learning_rate": 6.758864293675833e-05,
"loss": 0.0476,
"step": 12490
},
{
"grad_norm": 0.3055208921432495,
"learning_rate": 6.753703929668362e-05,
"loss": 0.0484,
"step": 12500
},
{
"grad_norm": 0.2861189842224121,
"learning_rate": 6.748541434745253e-05,
"loss": 0.0448,
"step": 12510
},
{
"grad_norm": 0.2901809811592102,
"learning_rate": 6.743376815179424e-05,
"loss": 0.047,
"step": 12520
},
{
"grad_norm": 0.25986236333847046,
"learning_rate": 6.738210077246376e-05,
"loss": 0.0473,
"step": 12530
},
{
"grad_norm": 0.37869927287101746,
"learning_rate": 6.733041227224181e-05,
"loss": 0.0516,
"step": 12540
},
{
"grad_norm": 0.2770848572254181,
"learning_rate": 6.72787027139348e-05,
"loss": 0.0532,
"step": 12550
},
{
"grad_norm": 0.2677651643753052,
"learning_rate": 6.72269721603747e-05,
"loss": 0.0501,
"step": 12560
},
{
"grad_norm": 0.3006638288497925,
"learning_rate": 6.717522067441904e-05,
"loss": 0.0575,
"step": 12570
},
{
"grad_norm": 0.24701638519763947,
"learning_rate": 6.712344831895074e-05,
"loss": 0.0521,
"step": 12580
},
{
"grad_norm": 0.27238336205482483,
"learning_rate": 6.707165515687811e-05,
"loss": 0.0494,
"step": 12590
},
{
"grad_norm": 0.2701241075992584,
"learning_rate": 6.70198412511347e-05,
"loss": 0.0561,
"step": 12600
},
{
"grad_norm": 0.39735496044158936,
"learning_rate": 6.69680066646793e-05,
"loss": 0.0499,
"step": 12610
},
{
"grad_norm": 0.3241919279098511,
"learning_rate": 6.691615146049583e-05,
"loss": 0.0537,
"step": 12620
},
{
"grad_norm": 0.27666524052619934,
"learning_rate": 6.686427570159324e-05,
"loss": 0.0509,
"step": 12630
},
{
"grad_norm": 0.2497592270374298,
"learning_rate": 6.681237945100548e-05,
"loss": 0.0504,
"step": 12640
},
{
"grad_norm": 0.2674764096736908,
"learning_rate": 6.676046277179139e-05,
"loss": 0.0486,
"step": 12650
},
{
"grad_norm": 0.2923271358013153,
"learning_rate": 6.670852572703462e-05,
"loss": 0.0485,
"step": 12660
},
{
"grad_norm": 0.26970750093460083,
"learning_rate": 6.665656837984359e-05,
"loss": 0.0506,
"step": 12670
},
{
"grad_norm": 0.3077962100505829,
"learning_rate": 6.660459079335135e-05,
"loss": 0.046,
"step": 12680
},
{
"grad_norm": 0.22250962257385254,
"learning_rate": 6.655259303071558e-05,
"loss": 0.0526,
"step": 12690
},
{
"grad_norm": 0.268349826335907,
"learning_rate": 6.650057515511848e-05,
"loss": 0.0471,
"step": 12700
},
{
"grad_norm": 0.2600545585155487,
"learning_rate": 6.644853722976667e-05,
"loss": 0.045,
"step": 12710
},
{
"grad_norm": 0.23383289575576782,
"learning_rate": 6.639647931789113e-05,
"loss": 0.0477,
"step": 12720
},
{
"grad_norm": 0.3337390720844269,
"learning_rate": 6.634440148274713e-05,
"loss": 0.0461,
"step": 12730
},
{
"grad_norm": 0.3223837614059448,
"learning_rate": 6.629230378761415e-05,
"loss": 0.0463,
"step": 12740
},
{
"grad_norm": 0.24198704957962036,
"learning_rate": 6.624018629579581e-05,
"loss": 0.0459,
"step": 12750
},
{
"grad_norm": 0.24410076439380646,
"learning_rate": 6.618804907061976e-05,
"loss": 0.046,
"step": 12760
},
{
"grad_norm": 0.47915250062942505,
"learning_rate": 6.613589217543766e-05,
"loss": 0.044,
"step": 12770
},
{
"grad_norm": 0.2951572835445404,
"learning_rate": 6.608371567362504e-05,
"loss": 0.0529,
"step": 12780
},
{
"grad_norm": 0.30561885237693787,
"learning_rate": 6.60315196285813e-05,
"loss": 0.055,
"step": 12790
},
{
"grad_norm": 0.2287045270204544,
"learning_rate": 6.597930410372954e-05,
"loss": 0.0566,
"step": 12800
},
{
"grad_norm": 0.25188148021698,
"learning_rate": 6.592706916251653e-05,
"loss": 0.0542,
"step": 12810
},
{
"grad_norm": 0.2927015721797943,
"learning_rate": 6.587481486841267e-05,
"loss": 0.0539,
"step": 12820
},
{
"grad_norm": 0.2053963840007782,
"learning_rate": 6.582254128491184e-05,
"loss": 0.048,
"step": 12830
},
{
"grad_norm": 0.2623174786567688,
"learning_rate": 6.577024847553139e-05,
"loss": 0.0525,
"step": 12840
},
{
"grad_norm": 0.285585880279541,
"learning_rate": 6.571793650381201e-05,
"loss": 0.043,
"step": 12850
},
{
"grad_norm": 0.314848929643631,
"learning_rate": 6.566560543331766e-05,
"loss": 0.0534,
"step": 12860
},
{
"grad_norm": 0.26681679487228394,
"learning_rate": 6.561325532763554e-05,
"loss": 0.051,
"step": 12870
},
{
"grad_norm": 0.3126400113105774,
"learning_rate": 6.556088625037597e-05,
"loss": 0.0504,
"step": 12880
},
{
"grad_norm": 0.26152631640434265,
"learning_rate": 6.55084982651723e-05,
"loss": 0.0467,
"step": 12890
},
{
"grad_norm": 0.23513339459896088,
"learning_rate": 6.54560914356809e-05,
"loss": 0.0453,
"step": 12900
},
{
"grad_norm": 0.24291107058525085,
"learning_rate": 6.540366582558097e-05,
"loss": 0.0463,
"step": 12910
},
{
"grad_norm": 0.22658753395080566,
"learning_rate": 6.53512214985746e-05,
"loss": 0.0515,
"step": 12920
},
{
"grad_norm": 0.2757647931575775,
"learning_rate": 6.529875851838658e-05,
"loss": 0.0434,
"step": 12930
},
{
"grad_norm": 0.26386114954948425,
"learning_rate": 6.52462769487644e-05,
"loss": 0.0424,
"step": 12940
},
{
"grad_norm": 0.24191245436668396,
"learning_rate": 6.519377685347807e-05,
"loss": 0.0459,
"step": 12950
},
{
"grad_norm": 0.23351170122623444,
"learning_rate": 6.514125829632021e-05,
"loss": 0.0454,
"step": 12960
},
{
"grad_norm": 0.2373272180557251,
"learning_rate": 6.508872134110577e-05,
"loss": 0.0439,
"step": 12970
},
{
"grad_norm": 0.31740468740463257,
"learning_rate": 6.503616605167213e-05,
"loss": 0.0434,
"step": 12980
},
{
"grad_norm": 0.3169281780719757,
"learning_rate": 6.498359249187893e-05,
"loss": 0.0447,
"step": 12990
},
{
"grad_norm": 0.2216937243938446,
"learning_rate": 6.493100072560799e-05,
"loss": 0.0438,
"step": 13000
},
{
"grad_norm": 0.2945482134819031,
"learning_rate": 6.487839081676327e-05,
"loss": 0.0465,
"step": 13010
},
{
"grad_norm": 0.2810530364513397,
"learning_rate": 6.482576282927076e-05,
"loss": 0.0445,
"step": 13020
},
{
"grad_norm": 0.21927085518836975,
"learning_rate": 6.477311682707845e-05,
"loss": 0.0436,
"step": 13030
},
{
"grad_norm": 0.24121566116809845,
"learning_rate": 6.472045287415616e-05,
"loss": 0.0443,
"step": 13040
},
{
"grad_norm": 0.2199757993221283,
"learning_rate": 6.466777103449558e-05,
"loss": 0.0394,
"step": 13050
},
{
"grad_norm": 0.19569247961044312,
"learning_rate": 6.461507137211012e-05,
"loss": 0.0442,
"step": 13060
},
{
"grad_norm": 0.24014554917812347,
"learning_rate": 6.456235395103483e-05,
"loss": 0.0364,
"step": 13070
},
{
"grad_norm": 0.3624272644519806,
"learning_rate": 6.450961883532634e-05,
"loss": 0.0402,
"step": 13080
},
{
"grad_norm": 0.23237080872058868,
"learning_rate": 6.445686608906283e-05,
"loss": 0.0433,
"step": 13090
},
{
"grad_norm": 0.24174578487873077,
"learning_rate": 6.44040957763438e-05,
"loss": 0.0425,
"step": 13100
},
{
"grad_norm": 0.26301848888397217,
"learning_rate": 6.435130796129018e-05,
"loss": 0.0436,
"step": 13110
},
{
"grad_norm": 0.2683694660663605,
"learning_rate": 6.429850270804416e-05,
"loss": 0.0501,
"step": 13120
},
{
"grad_norm": 0.2611674964427948,
"learning_rate": 6.424568008076909e-05,
"loss": 0.0479,
"step": 13130
},
{
"grad_norm": 0.3324321508407593,
"learning_rate": 6.419284014364943e-05,
"loss": 0.0462,
"step": 13140
},
{
"grad_norm": 0.31166189908981323,
"learning_rate": 6.41399829608907e-05,
"loss": 0.0444,
"step": 13150
},
{
"grad_norm": 0.27646604180336,
"learning_rate": 6.408710859671938e-05,
"loss": 0.0485,
"step": 13160
},
{
"grad_norm": 0.2932283282279968,
"learning_rate": 6.403421711538278e-05,
"loss": 0.0497,
"step": 13170
},
{
"grad_norm": 0.242189422249794,
"learning_rate": 6.398130858114903e-05,
"loss": 0.0503,
"step": 13180
},
{
"grad_norm": 0.26972296833992004,
"learning_rate": 6.392838305830701e-05,
"loss": 0.0441,
"step": 13190
},
{
"grad_norm": 0.2916775345802307,
"learning_rate": 6.387544061116621e-05,
"loss": 0.0554,
"step": 13200
},
{
"grad_norm": 0.25495779514312744,
"learning_rate": 6.382248130405671e-05,
"loss": 0.0488,
"step": 13210
},
{
"grad_norm": 0.24461768567562103,
"learning_rate": 6.376950520132905e-05,
"loss": 0.0481,
"step": 13220
},
{
"grad_norm": 0.31973975896835327,
"learning_rate": 6.371651236735418e-05,
"loss": 0.049,
"step": 13230
},
{
"grad_norm": 0.24903327226638794,
"learning_rate": 6.366350286652341e-05,
"loss": 0.0517,
"step": 13240
},
{
"grad_norm": 0.2802121341228485,
"learning_rate": 6.361047676324827e-05,
"loss": 0.0563,
"step": 13250
},
{
"grad_norm": 0.3128843307495117,
"learning_rate": 6.355743412196047e-05,
"loss": 0.053,
"step": 13260
},
{
"grad_norm": 0.2942280173301697,
"learning_rate": 6.350437500711184e-05,
"loss": 0.0532,
"step": 13270
},
{
"grad_norm": 0.2966551184654236,
"learning_rate": 6.345129948317418e-05,
"loss": 0.0556,
"step": 13280
},
{
"grad_norm": 0.2549077570438385,
"learning_rate": 6.33982076146393e-05,
"loss": 0.0559,
"step": 13290
},
{
"grad_norm": 0.3053489327430725,
"learning_rate": 6.334509946601879e-05,
"loss": 0.0556,
"step": 13300
},
{
"grad_norm": 0.24328504502773285,
"learning_rate": 6.329197510184406e-05,
"loss": 0.0461,
"step": 13310
},
{
"grad_norm": 0.29686078429222107,
"learning_rate": 6.323883458666624e-05,
"loss": 0.0519,
"step": 13320
},
{
"grad_norm": 0.35195133090019226,
"learning_rate": 6.318567798505605e-05,
"loss": 0.0509,
"step": 13330
},
{
"grad_norm": 0.3170219659805298,
"learning_rate": 6.313250536160378e-05,
"loss": 0.0449,
"step": 13340
},
{
"grad_norm": 0.2436753511428833,
"learning_rate": 6.307931678091917e-05,
"loss": 0.044,
"step": 13350
},
{
"grad_norm": 0.23700007796287537,
"learning_rate": 6.302611230763138e-05,
"loss": 0.0499,
"step": 13360
},
{
"grad_norm": 0.4010815918445587,
"learning_rate": 6.297289200638887e-05,
"loss": 0.0446,
"step": 13370
},
{
"grad_norm": 0.1880156397819519,
"learning_rate": 6.29196559418593e-05,
"loss": 0.0477,
"step": 13380
},
{
"grad_norm": 0.2685961425304413,
"learning_rate": 6.286640417872951e-05,
"loss": 0.0473,
"step": 13390
},
{
"grad_norm": 0.2100432962179184,
"learning_rate": 6.281313678170542e-05,
"loss": 0.0413,
"step": 13400
},
{
"grad_norm": 0.2987616956233978,
"learning_rate": 6.275985381551194e-05,
"loss": 0.0441,
"step": 13410
},
{
"grad_norm": 0.2511388063430786,
"learning_rate": 6.270655534489292e-05,
"loss": 0.0443,
"step": 13420
},
{
"grad_norm": 0.258575439453125,
"learning_rate": 6.265324143461098e-05,
"loss": 0.042,
"step": 13430
},
{
"grad_norm": 0.26680201292037964,
"learning_rate": 6.259991214944758e-05,
"loss": 0.0487,
"step": 13440
},
{
"grad_norm": 0.36215871572494507,
"learning_rate": 6.254656755420283e-05,
"loss": 0.0484,
"step": 13450
},
{
"grad_norm": 0.2670304477214813,
"learning_rate": 6.249320771369545e-05,
"loss": 0.047,
"step": 13460
},
{
"grad_norm": 0.21186518669128418,
"learning_rate": 6.243983269276263e-05,
"loss": 0.0439,
"step": 13470
},
{
"grad_norm": 0.23457102477550507,
"learning_rate": 6.238644255626012e-05,
"loss": 0.0412,
"step": 13480
},
{
"grad_norm": 0.2891707122325897,
"learning_rate": 6.233303736906193e-05,
"loss": 0.0545,
"step": 13490
},
{
"grad_norm": 0.30372804403305054,
"learning_rate": 6.227961719606039e-05,
"loss": 0.048,
"step": 13500
},
{
"grad_norm": 0.2036457359790802,
"learning_rate": 6.222618210216608e-05,
"loss": 0.0504,
"step": 13510
},
{
"grad_norm": 0.2237347960472107,
"learning_rate": 6.217273215230767e-05,
"loss": 0.0477,
"step": 13520
},
{
"grad_norm": 0.2197960615158081,
"learning_rate": 6.211926741143188e-05,
"loss": 0.0477,
"step": 13530
},
{
"grad_norm": 0.22308778762817383,
"learning_rate": 6.20657879445034e-05,
"loss": 0.0518,
"step": 13540
},
{
"grad_norm": 0.25409528613090515,
"learning_rate": 6.201229381650485e-05,
"loss": 0.0477,
"step": 13550
},
{
"grad_norm": 0.4137152433395386,
"learning_rate": 6.195878509243661e-05,
"loss": 0.059,
"step": 13560
},
{
"grad_norm": 0.3241106867790222,
"learning_rate": 6.190526183731685e-05,
"loss": 0.052,
"step": 13570
},
{
"grad_norm": 0.3466551601886749,
"learning_rate": 6.185172411618138e-05,
"loss": 0.0558,
"step": 13580
},
{
"grad_norm": 0.2961535155773163,
"learning_rate": 6.179817199408355e-05,
"loss": 0.051,
"step": 13590
},
{
"grad_norm": 0.2600025534629822,
"learning_rate": 6.174460553609426e-05,
"loss": 0.0526,
"step": 13600
},
{
"grad_norm": 0.3617904782295227,
"learning_rate": 6.16910248073018e-05,
"loss": 0.0491,
"step": 13610
},
{
"grad_norm": 0.3671341836452484,
"learning_rate": 6.16374298728118e-05,
"loss": 0.0547,
"step": 13620
},
{
"grad_norm": 0.3040948510169983,
"learning_rate": 6.158382079774716e-05,
"loss": 0.0496,
"step": 13630
},
{
"grad_norm": 0.23346646130084991,
"learning_rate": 6.153019764724799e-05,
"loss": 0.0516,
"step": 13640
},
{
"grad_norm": 0.2637227773666382,
"learning_rate": 6.147656048647144e-05,
"loss": 0.0499,
"step": 13650
},
{
"grad_norm": 0.287783682346344,
"learning_rate": 6.142290938059173e-05,
"loss": 0.0463,
"step": 13660
},
{
"grad_norm": 0.24072909355163574,
"learning_rate": 6.136924439480001e-05,
"loss": 0.0452,
"step": 13670
},
{
"grad_norm": 0.29946663975715637,
"learning_rate": 6.13155655943043e-05,
"loss": 0.0443,
"step": 13680
},
{
"grad_norm": 0.2755713164806366,
"learning_rate": 6.126187304432941e-05,
"loss": 0.0437,
"step": 13690
},
{
"grad_norm": 0.324932724237442,
"learning_rate": 6.120816681011684e-05,
"loss": 0.0436,
"step": 13700
},
{
"grad_norm": 0.28711050748825073,
"learning_rate": 6.115444695692474e-05,
"loss": 0.044,
"step": 13710
},
{
"grad_norm": 0.20507220923900604,
"learning_rate": 6.110071355002779e-05,
"loss": 0.0375,
"step": 13720
},
{
"grad_norm": 0.3771602213382721,
"learning_rate": 6.104696665471714e-05,
"loss": 0.0469,
"step": 13730
},
{
"grad_norm": 0.20338517427444458,
"learning_rate": 6.099320633630036e-05,
"loss": 0.046,
"step": 13740
},
{
"grad_norm": 0.25987479090690613,
"learning_rate": 6.093943266010128e-05,
"loss": 0.0509,
"step": 13750
},
{
"grad_norm": 0.2442474365234375,
"learning_rate": 6.088564569146e-05,
"loss": 0.0452,
"step": 13760
},
{
"grad_norm": 0.2892291843891144,
"learning_rate": 6.083184549573274e-05,
"loss": 0.0483,
"step": 13770
},
{
"grad_norm": 0.22794269025325775,
"learning_rate": 6.077803213829184e-05,
"loss": 0.0464,
"step": 13780
},
{
"grad_norm": 0.3164249360561371,
"learning_rate": 6.0724205684525594e-05,
"loss": 0.0463,
"step": 13790
},
{
"grad_norm": 0.2796657085418701,
"learning_rate": 6.0670366199838215e-05,
"loss": 0.0471,
"step": 13800
},
{
"grad_norm": 0.30009639263153076,
"learning_rate": 6.061651374964974e-05,
"loss": 0.0525,
"step": 13810
},
{
"grad_norm": 0.34995946288108826,
"learning_rate": 6.0562648399396004e-05,
"loss": 0.0451,
"step": 13820
},
{
"grad_norm": 0.3073829710483551,
"learning_rate": 6.050877021452845e-05,
"loss": 0.0503,
"step": 13830
},
{
"grad_norm": 0.253593772649765,
"learning_rate": 6.04548792605142e-05,
"loss": 0.0472,
"step": 13840
},
{
"grad_norm": 0.25738525390625,
"learning_rate": 6.04009756028358e-05,
"loss": 0.0472,
"step": 13850
},
{
"grad_norm": 0.2966947555541992,
"learning_rate": 6.0347059306991294e-05,
"loss": 0.0484,
"step": 13860
},
{
"grad_norm": 0.22222433984279633,
"learning_rate": 6.029313043849407e-05,
"loss": 0.0427,
"step": 13870
},
{
"grad_norm": 0.3048725426197052,
"learning_rate": 6.023918906287279e-05,
"loss": 0.0427,
"step": 13880
},
{
"grad_norm": 0.18616923689842224,
"learning_rate": 6.01852352456713e-05,
"loss": 0.0442,
"step": 13890
},
{
"grad_norm": 0.2609843313694,
"learning_rate": 6.013126905244858e-05,
"loss": 0.0447,
"step": 13900
},
{
"grad_norm": 0.34710702300071716,
"learning_rate": 6.007729054877864e-05,
"loss": 0.0396,
"step": 13910
},
{
"grad_norm": 0.20360657572746277,
"learning_rate": 6.0023299800250466e-05,
"loss": 0.0508,
"step": 13920
},
{
"grad_norm": 0.23097991943359375,
"learning_rate": 5.9969296872467894e-05,
"loss": 0.0421,
"step": 13930
},
{
"grad_norm": 0.20531237125396729,
"learning_rate": 5.9915281831049585e-05,
"loss": 0.041,
"step": 13940
},
{
"grad_norm": 0.3211476802825928,
"learning_rate": 5.98612547416289e-05,
"loss": 0.043,
"step": 13950
},
{
"grad_norm": 0.23077180981636047,
"learning_rate": 5.9807215669853855e-05,
"loss": 0.0443,
"step": 13960
},
{
"grad_norm": 0.25315532088279724,
"learning_rate": 5.9753164681387e-05,
"loss": 0.043,
"step": 13970
},
{
"grad_norm": 0.22051264345645905,
"learning_rate": 5.969910184190539e-05,
"loss": 0.0403,
"step": 13980
},
{
"grad_norm": 0.3214477598667145,
"learning_rate": 5.9645027217100477e-05,
"loss": 0.0376,
"step": 13990
},
{
"grad_norm": 0.2828483581542969,
"learning_rate": 5.959094087267804e-05,
"loss": 0.0396,
"step": 14000
},
{
"grad_norm": 0.310552179813385,
"learning_rate": 5.9536842874358065e-05,
"loss": 0.0459,
"step": 14010
},
{
"grad_norm": 0.3362694978713989,
"learning_rate": 5.9482733287874734e-05,
"loss": 0.0434,
"step": 14020
},
{
"grad_norm": 0.2320353388786316,
"learning_rate": 5.942861217897631e-05,
"loss": 0.0399,
"step": 14030
},
{
"grad_norm": 0.20843276381492615,
"learning_rate": 5.9374479613425e-05,
"loss": 0.0405,
"step": 14040
},
{
"grad_norm": 0.21404658257961273,
"learning_rate": 5.932033565699704e-05,
"loss": 0.0414,
"step": 14050
},
{
"grad_norm": 0.263860821723938,
"learning_rate": 5.926618037548237e-05,
"loss": 0.0459,
"step": 14060
},
{
"grad_norm": 0.34168803691864014,
"learning_rate": 5.9212013834684824e-05,
"loss": 0.0487,
"step": 14070
},
{
"grad_norm": 0.23258757591247559,
"learning_rate": 5.9157836100421824e-05,
"loss": 0.0421,
"step": 14080
},
{
"grad_norm": 0.2945728302001953,
"learning_rate": 5.910364723852444e-05,
"loss": 0.0438,
"step": 14090
},
{
"grad_norm": 0.27724385261535645,
"learning_rate": 5.904944731483723e-05,
"loss": 0.0454,
"step": 14100
},
{
"grad_norm": 0.24319922924041748,
"learning_rate": 5.899523639521825e-05,
"loss": 0.0439,
"step": 14110
},
{
"grad_norm": 0.2116938829421997,
"learning_rate": 5.894101454553883e-05,
"loss": 0.0444,
"step": 14120
},
{
"grad_norm": 0.25595414638519287,
"learning_rate": 5.888678183168368e-05,
"loss": 0.0506,
"step": 14130
},
{
"grad_norm": 0.2684904932975769,
"learning_rate": 5.883253831955061e-05,
"loss": 0.0516,
"step": 14140
},
{
"grad_norm": 0.3018367886543274,
"learning_rate": 5.8778284075050625e-05,
"loss": 0.0501,
"step": 14150
},
{
"grad_norm": 0.2546306252479553,
"learning_rate": 5.872401916410777e-05,
"loss": 0.044,
"step": 14160
},
{
"grad_norm": 0.31291893124580383,
"learning_rate": 5.8669743652659014e-05,
"loss": 0.0522,
"step": 14170
},
{
"grad_norm": 0.20264975726604462,
"learning_rate": 5.86154576066542e-05,
"loss": 0.0453,
"step": 14180
},
{
"grad_norm": 0.32889530062675476,
"learning_rate": 5.856116109205602e-05,
"loss": 0.0453,
"step": 14190
},
{
"grad_norm": 0.2896175682544708,
"learning_rate": 5.850685417483983e-05,
"loss": 0.048,
"step": 14200
},
{
"grad_norm": 0.29567596316337585,
"learning_rate": 5.8452536920993684e-05,
"loss": 0.0466,
"step": 14210
},
{
"grad_norm": 0.245882049202919,
"learning_rate": 5.839820939651817e-05,
"loss": 0.046,
"step": 14220
},
{
"grad_norm": 0.328325480222702,
"learning_rate": 5.834387166742632e-05,
"loss": 0.046,
"step": 14230
},
{
"grad_norm": 0.27349409461021423,
"learning_rate": 5.828952379974364e-05,
"loss": 0.0404,
"step": 14240
},
{
"grad_norm": 0.2633655071258545,
"learning_rate": 5.8235165859507864e-05,
"loss": 0.0439,
"step": 14250
},
{
"grad_norm": 0.2326846420764923,
"learning_rate": 5.818079791276907e-05,
"loss": 0.0449,
"step": 14260
},
{
"grad_norm": 0.2080424427986145,
"learning_rate": 5.8126420025589415e-05,
"loss": 0.0421,
"step": 14270
},
{
"grad_norm": 0.28082093596458435,
"learning_rate": 5.807203226404313e-05,
"loss": 0.0461,
"step": 14280
},
{
"grad_norm": 0.2617836892604828,
"learning_rate": 5.801763469421652e-05,
"loss": 0.0408,
"step": 14290
},
{
"grad_norm": 0.22883723676204681,
"learning_rate": 5.796322738220774e-05,
"loss": 0.0431,
"step": 14300
},
{
"grad_norm": 0.2687304615974426,
"learning_rate": 5.79088103941268e-05,
"loss": 0.0421,
"step": 14310
},
{
"grad_norm": 0.3234752416610718,
"learning_rate": 5.785438379609549e-05,
"loss": 0.0462,
"step": 14320
},
{
"grad_norm": 0.2790320813655853,
"learning_rate": 5.779994765424724e-05,
"loss": 0.0435,
"step": 14330
},
{
"grad_norm": 0.29228901863098145,
"learning_rate": 5.7745502034727114e-05,
"loss": 0.0515,
"step": 14340
},
{
"grad_norm": 0.3126240670681,
"learning_rate": 5.769104700369164e-05,
"loss": 0.0398,
"step": 14350
},
{
"grad_norm": 0.2869804799556732,
"learning_rate": 5.7636582627308854e-05,
"loss": 0.0498,
"step": 14360
},
{
"grad_norm": 0.2043546587228775,
"learning_rate": 5.7582108971758095e-05,
"loss": 0.0429,
"step": 14370
},
{
"grad_norm": 0.24439789354801178,
"learning_rate": 5.752762610322999e-05,
"loss": 0.0443,
"step": 14380
},
{
"grad_norm": 0.25771400332450867,
"learning_rate": 5.747313408792636e-05,
"loss": 0.0458,
"step": 14390
},
{
"grad_norm": 0.2352185994386673,
"learning_rate": 5.741863299206014e-05,
"loss": 0.0429,
"step": 14400
},
{
"grad_norm": 0.22542239725589752,
"learning_rate": 5.73641228818553e-05,
"loss": 0.0449,
"step": 14410
},
{
"grad_norm": 0.2165168821811676,
"learning_rate": 5.730960382354676e-05,
"loss": 0.045,
"step": 14420
},
{
"grad_norm": 0.22130000591278076,
"learning_rate": 5.725507588338035e-05,
"loss": 0.0384,
"step": 14430
},
{
"grad_norm": 0.25066035985946655,
"learning_rate": 5.7200539127612604e-05,
"loss": 0.042,
"step": 14440
},
{
"grad_norm": 0.3225209414958954,
"learning_rate": 5.714599362251087e-05,
"loss": 0.052,
"step": 14450
},
{
"grad_norm": 0.28190383315086365,
"learning_rate": 5.709143943435307e-05,
"loss": 0.0475,
"step": 14460
},
{
"grad_norm": 0.32284918427467346,
"learning_rate": 5.7036876629427646e-05,
"loss": 0.0527,
"step": 14470
},
{
"grad_norm": 0.29126057028770447,
"learning_rate": 5.698230527403361e-05,
"loss": 0.0534,
"step": 14480
},
{
"grad_norm": 0.3034134805202484,
"learning_rate": 5.6927725434480274e-05,
"loss": 0.0496,
"step": 14490
},
{
"grad_norm": 0.26183196902275085,
"learning_rate": 5.687313717708728e-05,
"loss": 0.0495,
"step": 14500
},
{
"grad_norm": 0.25203219056129456,
"learning_rate": 5.681854056818453e-05,
"loss": 0.0522,
"step": 14510
},
{
"grad_norm": 0.23825468122959137,
"learning_rate": 5.676393567411205e-05,
"loss": 0.0476,
"step": 14520
},
{
"grad_norm": 0.27571746706962585,
"learning_rate": 5.670932256121991e-05,
"loss": 0.0492,
"step": 14530
},
{
"grad_norm": 0.26615071296691895,
"learning_rate": 5.6654701295868216e-05,
"loss": 0.0539,
"step": 14540
},
{
"grad_norm": 0.25037598609924316,
"learning_rate": 5.660007194442697e-05,
"loss": 0.0521,
"step": 14550
},
{
"grad_norm": 0.2499276101589203,
"learning_rate": 5.6545434573275946e-05,
"loss": 0.0435,
"step": 14560
},
{
"grad_norm": 0.28505125641822815,
"learning_rate": 5.649078924880472e-05,
"loss": 0.0552,
"step": 14570
},
{
"grad_norm": 0.2607171833515167,
"learning_rate": 5.643613603741252e-05,
"loss": 0.0491,
"step": 14580
},
{
"grad_norm": 0.24682322144508362,
"learning_rate": 5.6381475005508154e-05,
"loss": 0.0504,
"step": 14590
},
{
"grad_norm": 0.19607535004615784,
"learning_rate": 5.632680621950992e-05,
"loss": 0.0462,
"step": 14600
},
{
"grad_norm": 0.21294380724430084,
"learning_rate": 5.627212974584555e-05,
"loss": 0.0492,
"step": 14610
},
{
"grad_norm": 0.2033899873495102,
"learning_rate": 5.62174456509521e-05,
"loss": 0.0446,
"step": 14620
},
{
"grad_norm": 0.30011358857154846,
"learning_rate": 5.616275400127594e-05,
"loss": 0.0568,
"step": 14630
},
{
"grad_norm": 0.265842467546463,
"learning_rate": 5.610805486327254e-05,
"loss": 0.0511,
"step": 14640
},
{
"grad_norm": 0.32071059942245483,
"learning_rate": 5.6053348303406536e-05,
"loss": 0.0451,
"step": 14650
},
{
"grad_norm": 0.2244894653558731,
"learning_rate": 5.599863438815156e-05,
"loss": 0.0421,
"step": 14660
},
{
"grad_norm": 0.25378966331481934,
"learning_rate": 5.594391318399017e-05,
"loss": 0.0418,
"step": 14670
},
{
"grad_norm": 0.2519189417362213,
"learning_rate": 5.588918475741377e-05,
"loss": 0.0435,
"step": 14680
},
{
"grad_norm": 0.320303738117218,
"learning_rate": 5.5834449174922586e-05,
"loss": 0.0503,
"step": 14690
},
{
"grad_norm": 0.3209528923034668,
"learning_rate": 5.57797065030255e-05,
"loss": 0.0561,
"step": 14700
},
{
"grad_norm": 0.3086570203304291,
"learning_rate": 5.5724956808240016e-05,
"loss": 0.0612,
"step": 14710
},
{
"grad_norm": 0.2290535569190979,
"learning_rate": 5.567020015709219e-05,
"loss": 0.0551,
"step": 14720
},
{
"grad_norm": 0.2647378146648407,
"learning_rate": 5.561543661611649e-05,
"loss": 0.0536,
"step": 14730
},
{
"grad_norm": 0.2938559055328369,
"learning_rate": 5.556066625185583e-05,
"loss": 0.0523,
"step": 14740
},
{
"grad_norm": 0.2859813868999481,
"learning_rate": 5.550588913086131e-05,
"loss": 0.0521,
"step": 14750
},
{
"grad_norm": 0.24138957262039185,
"learning_rate": 5.545110531969234e-05,
"loss": 0.0499,
"step": 14760
},
{
"grad_norm": 0.2977292537689209,
"learning_rate": 5.539631488491641e-05,
"loss": 0.055,
"step": 14770
},
{
"grad_norm": 0.25103700160980225,
"learning_rate": 5.534151789310904e-05,
"loss": 0.05,
"step": 14780
},
{
"grad_norm": 0.26145505905151367,
"learning_rate": 5.528671441085376e-05,
"loss": 0.0509,
"step": 14790
},
{
"grad_norm": 0.26307588815689087,
"learning_rate": 5.523190450474197e-05,
"loss": 0.0427,
"step": 14800
},
{
"grad_norm": 0.2919415831565857,
"learning_rate": 5.5177088241372865e-05,
"loss": 0.0569,
"step": 14810
},
{
"grad_norm": 0.28256168961524963,
"learning_rate": 5.512226568735338e-05,
"loss": 0.05,
"step": 14820
},
{
"grad_norm": 0.244222030043602,
"learning_rate": 5.506743690929809e-05,
"loss": 0.0508,
"step": 14830
},
{
"grad_norm": 0.2430955022573471,
"learning_rate": 5.501260197382912e-05,
"loss": 0.0473,
"step": 14840
},
{
"grad_norm": 0.21149927377700806,
"learning_rate": 5.49577609475761e-05,
"loss": 0.0509,
"step": 14850
},
{
"grad_norm": 0.2540222704410553,
"learning_rate": 5.490291389717603e-05,
"loss": 0.0515,
"step": 14860
},
{
"grad_norm": 0.28227245807647705,
"learning_rate": 5.484806088927329e-05,
"loss": 0.0495,
"step": 14870
},
{
"grad_norm": 0.28444305062294006,
"learning_rate": 5.479320199051942e-05,
"loss": 0.0508,
"step": 14880
},
{
"grad_norm": 0.2179258167743683,
"learning_rate": 5.473833726757314e-05,
"loss": 0.0449,
"step": 14890
},
{
"grad_norm": 0.2866404354572296,
"learning_rate": 5.4683466787100304e-05,
"loss": 0.0456,
"step": 14900
},
{
"grad_norm": 0.2782478928565979,
"learning_rate": 5.462859061577369e-05,
"loss": 0.0462,
"step": 14910
},
{
"grad_norm": 0.26032379269599915,
"learning_rate": 5.4573708820273026e-05,
"loss": 0.0528,
"step": 14920
},
{
"grad_norm": 0.2662164270877838,
"learning_rate": 5.451882146728488e-05,
"loss": 0.0459,
"step": 14930
},
{
"grad_norm": 0.3212851285934448,
"learning_rate": 5.446392862350255e-05,
"loss": 0.0462,
"step": 14940
},
{
"grad_norm": 0.2576570510864258,
"learning_rate": 5.440903035562603e-05,
"loss": 0.0523,
"step": 14950
},
{
"grad_norm": 0.2582419514656067,
"learning_rate": 5.435412673036188e-05,
"loss": 0.0417,
"step": 14960
},
{
"grad_norm": 0.26304998993873596,
"learning_rate": 5.429921781442318e-05,
"loss": 0.0442,
"step": 14970
},
{
"grad_norm": 0.26007649302482605,
"learning_rate": 5.424430367452945e-05,
"loss": 0.0454,
"step": 14980
},
{
"grad_norm": 0.2555650472640991,
"learning_rate": 5.418938437740655e-05,
"loss": 0.0503,
"step": 14990
},
{
"grad_norm": 0.267459899187088,
"learning_rate": 5.413445998978658e-05,
"loss": 0.0482,
"step": 15000
},
{
"grad_norm": 0.23930881917476654,
"learning_rate": 5.407953057840789e-05,
"loss": 0.0515,
"step": 15010
},
{
"grad_norm": 0.30673158168792725,
"learning_rate": 5.4024596210014853e-05,
"loss": 0.042,
"step": 15020
},
{
"grad_norm": 0.25517454743385315,
"learning_rate": 5.3969656951357935e-05,
"loss": 0.0509,
"step": 15030
},
{
"grad_norm": 0.2126220017671585,
"learning_rate": 5.3914712869193504e-05,
"loss": 0.0447,
"step": 15040
},
{
"grad_norm": 0.2095569521188736,
"learning_rate": 5.385976403028381e-05,
"loss": 0.0435,
"step": 15050
},
{
"grad_norm": 0.24438205361366272,
"learning_rate": 5.3804810501396875e-05,
"loss": 0.0415,
"step": 15060
},
{
"grad_norm": 0.23538653552532196,
"learning_rate": 5.374985234930639e-05,
"loss": 0.0434,
"step": 15070
},
{
"grad_norm": 0.302910715341568,
"learning_rate": 5.3694889640791724e-05,
"loss": 0.04,
"step": 15080
},
{
"grad_norm": 0.20624373853206635,
"learning_rate": 5.3639922442637736e-05,
"loss": 0.0391,
"step": 15090
},
{
"grad_norm": 0.22013568878173828,
"learning_rate": 5.358495082163476e-05,
"loss": 0.0408,
"step": 15100
},
{
"grad_norm": 0.22129015624523163,
"learning_rate": 5.35299748445785e-05,
"loss": 0.0501,
"step": 15110
},
{
"grad_norm": 0.21036967635154724,
"learning_rate": 5.3474994578269944e-05,
"loss": 0.0496,
"step": 15120
},
{
"grad_norm": 0.2544102966785431,
"learning_rate": 5.34200100895153e-05,
"loss": 0.0434,
"step": 15130
},
{
"grad_norm": 0.21975572407245636,
"learning_rate": 5.3365021445125916e-05,
"loss": 0.0435,
"step": 15140
},
{
"grad_norm": 0.21073217689990997,
"learning_rate": 5.331002871191817e-05,
"loss": 0.0417,
"step": 15150
},
{
"grad_norm": 0.30627772212028503,
"learning_rate": 5.3255031956713443e-05,
"loss": 0.0397,
"step": 15160
},
{
"grad_norm": 0.24781733751296997,
"learning_rate": 5.320003124633795e-05,
"loss": 0.0387,
"step": 15170
},
{
"grad_norm": 0.2073890119791031,
"learning_rate": 5.314502664762274e-05,
"loss": 0.0396,
"step": 15180
},
{
"grad_norm": 0.202210932970047,
"learning_rate": 5.3090018227403604e-05,
"loss": 0.0446,
"step": 15190
},
{
"grad_norm": 0.2902548909187317,
"learning_rate": 5.3035006052520955e-05,
"loss": 0.0475,
"step": 15200
},
{
"grad_norm": 0.21991859376430511,
"learning_rate": 5.297999018981976e-05,
"loss": 0.0446,
"step": 15210
},
{
"grad_norm": 0.26398995518684387,
"learning_rate": 5.2924970706149505e-05,
"loss": 0.0446,
"step": 15220
},
{
"grad_norm": 0.22249983251094818,
"learning_rate": 5.2869947668364014e-05,
"loss": 0.0446,
"step": 15230
},
{
"grad_norm": 0.28279614448547363,
"learning_rate": 5.2814921143321504e-05,
"loss": 0.0494,
"step": 15240
},
{
"grad_norm": 0.26672565937042236,
"learning_rate": 5.275989119788436e-05,
"loss": 0.0491,
"step": 15250
},
{
"grad_norm": 0.17791648209095,
"learning_rate": 5.270485789891919e-05,
"loss": 0.0478,
"step": 15260
},
{
"grad_norm": 0.21051760017871857,
"learning_rate": 5.264982131329661e-05,
"loss": 0.0473,
"step": 15270
},
{
"grad_norm": 0.27387237548828125,
"learning_rate": 5.259478150789128e-05,
"loss": 0.0481,
"step": 15280
},
{
"grad_norm": 0.26533058285713196,
"learning_rate": 5.253973854958173e-05,
"loss": 0.0507,
"step": 15290
},
{
"grad_norm": 0.2458713799715042,
"learning_rate": 5.2484692505250376e-05,
"loss": 0.045,
"step": 15300
},
{
"grad_norm": 0.23784229159355164,
"learning_rate": 5.242964344178333e-05,
"loss": 0.0542,
"step": 15310
},
{
"grad_norm": 0.22673584520816803,
"learning_rate": 5.237459142607041e-05,
"loss": 0.0547,
"step": 15320
},
{
"grad_norm": 0.29306861758232117,
"learning_rate": 5.231953652500498e-05,
"loss": 0.0505,
"step": 15330
},
{
"grad_norm": 0.2488688826560974,
"learning_rate": 5.2264478805483976e-05,
"loss": 0.0491,
"step": 15340
},
{
"grad_norm": 0.20069773495197296,
"learning_rate": 5.220941833440768e-05,
"loss": 0.0449,
"step": 15350
},
{
"grad_norm": 0.2538835406303406,
"learning_rate": 5.215435517867977e-05,
"loss": 0.0487,
"step": 15360
},
{
"grad_norm": 0.2568768262863159,
"learning_rate": 5.209928940520719e-05,
"loss": 0.044,
"step": 15370
},
{
"grad_norm": 0.18513739109039307,
"learning_rate": 5.204422108090004e-05,
"loss": 0.0418,
"step": 15380
},
{
"grad_norm": 0.20990408957004547,
"learning_rate": 5.19891502726715e-05,
"loss": 0.0435,
"step": 15390
},
{
"grad_norm": 0.2529078722000122,
"learning_rate": 5.193407704743782e-05,
"loss": 0.0485,
"step": 15400
},
{
"grad_norm": 0.23657166957855225,
"learning_rate": 5.187900147211815e-05,
"loss": 0.0415,
"step": 15410
},
{
"grad_norm": 0.27921801805496216,
"learning_rate": 5.182392361363453e-05,
"loss": 0.0494,
"step": 15420
},
{
"grad_norm": 0.20698505640029907,
"learning_rate": 5.176884353891172e-05,
"loss": 0.0409,
"step": 15430
},
{
"grad_norm": 0.21160903573036194,
"learning_rate": 5.171376131487722e-05,
"loss": 0.0444,
"step": 15440
},
{
"grad_norm": 0.22327765822410583,
"learning_rate": 5.165867700846113e-05,
"loss": 0.045,
"step": 15450
},
{
"grad_norm": 0.2398146688938141,
"learning_rate": 5.160359068659606e-05,
"loss": 0.0442,
"step": 15460
},
{
"grad_norm": 0.2598781883716583,
"learning_rate": 5.154850241621712e-05,
"loss": 0.0496,
"step": 15470
},
{
"grad_norm": 0.2717367708683014,
"learning_rate": 5.149341226426172e-05,
"loss": 0.0444,
"step": 15480
},
{
"grad_norm": 0.3123072683811188,
"learning_rate": 5.14383202976696e-05,
"loss": 0.0503,
"step": 15490
},
{
"grad_norm": 0.28409379720687866,
"learning_rate": 5.1383226583382684e-05,
"loss": 0.0469,
"step": 15500
},
{
"grad_norm": 0.23946945369243622,
"learning_rate": 5.132813118834504e-05,
"loss": 0.0458,
"step": 15510
},
{
"grad_norm": 0.2553609311580658,
"learning_rate": 5.1273034179502777e-05,
"loss": 0.0453,
"step": 15520
},
{
"grad_norm": 0.31236058473587036,
"learning_rate": 5.1217935623803945e-05,
"loss": 0.0494,
"step": 15530
},
{
"grad_norm": 0.2789974510669708,
"learning_rate": 5.1162835588198476e-05,
"loss": 0.0445,
"step": 15540
},
{
"grad_norm": 0.26257187128067017,
"learning_rate": 5.110773413963813e-05,
"loss": 0.0467,
"step": 15550
},
{
"grad_norm": 0.2072436660528183,
"learning_rate": 5.105263134507636e-05,
"loss": 0.0394,
"step": 15560
},
{
"grad_norm": 0.22758962213993073,
"learning_rate": 5.099752727146824e-05,
"loss": 0.0409,
"step": 15570
},
{
"grad_norm": 0.2439352124929428,
"learning_rate": 5.094242198577042e-05,
"loss": 0.0429,
"step": 15580
},
{
"grad_norm": 0.2788987159729004,
"learning_rate": 5.088731555494102e-05,
"loss": 0.0444,
"step": 15590
},
{
"grad_norm": 0.22255167365074158,
"learning_rate": 5.0832208045939556e-05,
"loss": 0.0395,
"step": 15600
},
{
"grad_norm": 0.23059864342212677,
"learning_rate": 5.0777099525726844e-05,
"loss": 0.0404,
"step": 15610
},
{
"grad_norm": 0.36181753873825073,
"learning_rate": 5.072199006126494e-05,
"loss": 0.045,
"step": 15620
},
{
"grad_norm": 0.240495502948761,
"learning_rate": 5.0666879719517026e-05,
"loss": 0.0404,
"step": 15630
},
{
"grad_norm": 0.2812608480453491,
"learning_rate": 5.061176856744737e-05,
"loss": 0.0442,
"step": 15640
},
{
"grad_norm": 0.22425560653209686,
"learning_rate": 5.0556656672021205e-05,
"loss": 0.0503,
"step": 15650
},
{
"grad_norm": 0.19794771075248718,
"learning_rate": 5.050154410020472e-05,
"loss": 0.0483,
"step": 15660
},
{
"grad_norm": 0.2336088865995407,
"learning_rate": 5.044643091896485e-05,
"loss": 0.0476,
"step": 15670
},
{
"grad_norm": 0.2400697022676468,
"learning_rate": 5.039131719526932e-05,
"loss": 0.0466,
"step": 15680
},
{
"grad_norm": 0.252143532037735,
"learning_rate": 5.03362029960865e-05,
"loss": 0.0458,
"step": 15690
},
{
"grad_norm": 0.252275675535202,
"learning_rate": 5.028108838838533e-05,
"loss": 0.0391,
"step": 15700
},
{
"grad_norm": 0.24177294969558716,
"learning_rate": 5.0225973439135276e-05,
"loss": 0.0432,
"step": 15710
},
{
"grad_norm": 0.2244001030921936,
"learning_rate": 5.017085821530617e-05,
"loss": 0.0464,
"step": 15720
},
{
"grad_norm": 0.3090827763080597,
"learning_rate": 5.011574278386822e-05,
"loss": 0.0464,
"step": 15730
},
{
"grad_norm": 0.22903846204280853,
"learning_rate": 5.006062721179189e-05,
"loss": 0.0507,
"step": 15740
},
{
"grad_norm": 0.30591559410095215,
"learning_rate": 5.000551156604777e-05,
"loss": 0.0513,
"step": 15750
},
{
"grad_norm": 0.28165003657341003,
"learning_rate": 4.9950395913606594e-05,
"loss": 0.0476,
"step": 15760
},
{
"grad_norm": 0.3013670742511749,
"learning_rate": 4.989528032143903e-05,
"loss": 0.0551,
"step": 15770
},
{
"grad_norm": 0.23625996708869934,
"learning_rate": 4.984016485651578e-05,
"loss": 0.0494,
"step": 15780
},
{
"grad_norm": 0.24937473237514496,
"learning_rate": 4.9785049585807274e-05,
"loss": 0.0454,
"step": 15790
},
{
"grad_norm": 0.21752090752124786,
"learning_rate": 4.972993457628381e-05,
"loss": 0.051,
"step": 15800
},
{
"grad_norm": 0.2985188364982605,
"learning_rate": 4.9674819894915306e-05,
"loss": 0.05,
"step": 15810
},
{
"grad_norm": 0.23297394812107086,
"learning_rate": 4.9619705608671265e-05,
"loss": 0.0502,
"step": 15820
},
{
"grad_norm": 0.29793471097946167,
"learning_rate": 4.956459178452079e-05,
"loss": 0.0478,
"step": 15830
},
{
"grad_norm": 0.2011656016111374,
"learning_rate": 4.950947848943235e-05,
"loss": 0.0451,
"step": 15840
},
{
"grad_norm": 0.2071039378643036,
"learning_rate": 4.9454365790373805e-05,
"loss": 0.0472,
"step": 15850
},
{
"grad_norm": 0.2766331434249878,
"learning_rate": 4.939925375431226e-05,
"loss": 0.0442,
"step": 15860
},
{
"grad_norm": 0.22968849539756775,
"learning_rate": 4.9344142448214046e-05,
"loss": 0.0426,
"step": 15870
},
{
"grad_norm": 0.30736660957336426,
"learning_rate": 4.928903193904461e-05,
"loss": 0.0485,
"step": 15880
},
{
"grad_norm": 0.26015669107437134,
"learning_rate": 4.923392229376842e-05,
"loss": 0.0406,
"step": 15890
},
{
"grad_norm": 0.28995513916015625,
"learning_rate": 4.9178813579348865e-05,
"loss": 0.0489,
"step": 15900
},
{
"grad_norm": 0.2618240714073181,
"learning_rate": 4.9123705862748253e-05,
"loss": 0.05,
"step": 15910
},
{
"grad_norm": 0.29167214035987854,
"learning_rate": 4.9068599210927627e-05,
"loss": 0.0484,
"step": 15920
},
{
"grad_norm": 0.24498440325260162,
"learning_rate": 4.901349369084681e-05,
"loss": 0.0473,
"step": 15930
},
{
"grad_norm": 0.20655056834220886,
"learning_rate": 4.8958389369464156e-05,
"loss": 0.041,
"step": 15940
},
{
"grad_norm": 0.2142612189054489,
"learning_rate": 4.890328631373666e-05,
"loss": 0.0409,
"step": 15950
},
{
"grad_norm": 0.2597513496875763,
"learning_rate": 4.88481845906197e-05,
"loss": 0.0446,
"step": 15960
},
{
"grad_norm": 0.2989039719104767,
"learning_rate": 4.879308426706707e-05,
"loss": 0.0418,
"step": 15970
},
{
"grad_norm": 0.22752147912979126,
"learning_rate": 4.873798541003084e-05,
"loss": 0.0395,
"step": 15980
},
{
"grad_norm": 0.29173240065574646,
"learning_rate": 4.868288808646136e-05,
"loss": 0.0359,
"step": 15990
},
{
"grad_norm": 0.29574379324913025,
"learning_rate": 4.862779236330705e-05,
"loss": 0.0401,
"step": 16000
},
{
"grad_norm": 0.2588016390800476,
"learning_rate": 4.8572698307514395e-05,
"loss": 0.0363,
"step": 16010
},
{
"grad_norm": 0.2854311466217041,
"learning_rate": 4.85176059860279e-05,
"loss": 0.0365,
"step": 16020
},
{
"grad_norm": 0.23890909552574158,
"learning_rate": 4.846251546578989e-05,
"loss": 0.041,
"step": 16030
},
{
"grad_norm": 0.2235451340675354,
"learning_rate": 4.840742681374058e-05,
"loss": 0.0412,
"step": 16040
},
{
"grad_norm": 0.2570871114730835,
"learning_rate": 4.835234009681787e-05,
"loss": 0.0422,
"step": 16050
},
{
"grad_norm": 0.21320363879203796,
"learning_rate": 4.8297255381957296e-05,
"loss": 0.0393,
"step": 16060
},
{
"grad_norm": 0.27716270089149475,
"learning_rate": 4.824217273609198e-05,
"loss": 0.0423,
"step": 16070
},
{
"grad_norm": 0.238595649600029,
"learning_rate": 4.8187092226152555e-05,
"loss": 0.0405,
"step": 16080
},
{
"grad_norm": 0.2758939266204834,
"learning_rate": 4.8132013919067016e-05,
"loss": 0.0391,
"step": 16090
},
{
"grad_norm": 0.20341502130031586,
"learning_rate": 4.807693788176071e-05,
"loss": 0.0414,
"step": 16100
},
{
"grad_norm": 0.2437739074230194,
"learning_rate": 4.8021864181156214e-05,
"loss": 0.0386,
"step": 16110
},
{
"grad_norm": 0.23328454792499542,
"learning_rate": 4.796679288417326e-05,
"loss": 0.0388,
"step": 16120
},
{
"grad_norm": 0.24193072319030762,
"learning_rate": 4.791172405772866e-05,
"loss": 0.0386,
"step": 16130
},
{
"grad_norm": 0.22620069980621338,
"learning_rate": 4.785665776873625e-05,
"loss": 0.0424,
"step": 16140
},
{
"grad_norm": 0.2430485486984253,
"learning_rate": 4.7801594084106763e-05,
"loss": 0.0412,
"step": 16150
},
{
"grad_norm": 0.23368537425994873,
"learning_rate": 4.774653307074775e-05,
"loss": 0.0466,
"step": 16160
},
{
"grad_norm": 0.2368520200252533,
"learning_rate": 4.769147479556355e-05,
"loss": 0.0436,
"step": 16170
},
{
"grad_norm": 0.22956649959087372,
"learning_rate": 4.763641932545515e-05,
"loss": 0.0488,
"step": 16180
},
{
"grad_norm": 0.23883633315563202,
"learning_rate": 4.758136672732013e-05,
"loss": 0.0463,
"step": 16190
},
{
"grad_norm": 0.2455056756734848,
"learning_rate": 4.752631706805261e-05,
"loss": 0.0468,
"step": 16200
},
{
"grad_norm": 0.2396029829978943,
"learning_rate": 4.7471270414543114e-05,
"loss": 0.0428,
"step": 16210
},
{
"grad_norm": 0.1973467320203781,
"learning_rate": 4.741622683367849e-05,
"loss": 0.0443,
"step": 16220
},
{
"grad_norm": 0.2058270424604416,
"learning_rate": 4.736118639234191e-05,
"loss": 0.0429,
"step": 16230
},
{
"grad_norm": 0.2541133463382721,
"learning_rate": 4.730614915741267e-05,
"loss": 0.0433,
"step": 16240
},
{
"grad_norm": 0.2561813294887543,
"learning_rate": 4.7251115195766234e-05,
"loss": 0.0412,
"step": 16250
},
{
"grad_norm": 0.21074876189231873,
"learning_rate": 4.719608457427404e-05,
"loss": 0.0482,
"step": 16260
},
{
"grad_norm": 0.2540050446987152,
"learning_rate": 4.714105735980347e-05,
"loss": 0.0457,
"step": 16270
},
{
"grad_norm": 0.26263827085494995,
"learning_rate": 4.70860336192178e-05,
"loss": 0.045,
"step": 16280
},
{
"grad_norm": 0.20934392511844635,
"learning_rate": 4.7031013419376034e-05,
"loss": 0.0484,
"step": 16290
},
{
"grad_norm": 0.19207070767879486,
"learning_rate": 4.6975996827132926e-05,
"loss": 0.0438,
"step": 16300
},
{
"grad_norm": 0.2288825660943985,
"learning_rate": 4.692098390933883e-05,
"loss": 0.0448,
"step": 16310
},
{
"grad_norm": 0.25006332993507385,
"learning_rate": 4.6865974732839615e-05,
"loss": 0.0452,
"step": 16320
},
{
"grad_norm": 0.20641279220581055,
"learning_rate": 4.6810969364476615e-05,
"loss": 0.0448,
"step": 16330
},
{
"grad_norm": 0.23286370933055878,
"learning_rate": 4.675596787108653e-05,
"loss": 0.0417,
"step": 16340
},
{
"grad_norm": 0.1981956511735916,
"learning_rate": 4.670097031950139e-05,
"loss": 0.043,
"step": 16350
},
{
"grad_norm": 0.23176626861095428,
"learning_rate": 4.664597677654838e-05,
"loss": 0.0456,
"step": 16360
},
{
"grad_norm": 0.26896607875823975,
"learning_rate": 4.659098730904986e-05,
"loss": 0.0431,
"step": 16370
},
{
"grad_norm": 0.2587151825428009,
"learning_rate": 4.6536001983823204e-05,
"loss": 0.0444,
"step": 16380
},
{
"grad_norm": 0.22970648109912872,
"learning_rate": 4.648102086768078e-05,
"loss": 0.0416,
"step": 16390
},
{
"grad_norm": 0.2982827126979828,
"learning_rate": 4.6426044027429786e-05,
"loss": 0.0429,
"step": 16400
},
{
"grad_norm": 0.2466629594564438,
"learning_rate": 4.637107152987234e-05,
"loss": 0.0434,
"step": 16410
},
{
"grad_norm": 0.32496169209480286,
"learning_rate": 4.631610344180515e-05,
"loss": 0.0469,
"step": 16420
},
{
"grad_norm": 0.2558490037918091,
"learning_rate": 4.626113983001965e-05,
"loss": 0.0384,
"step": 16430
},
{
"grad_norm": 0.27086713910102844,
"learning_rate": 4.620618076130182e-05,
"loss": 0.0469,
"step": 16440
},
{
"grad_norm": 0.2811828851699829,
"learning_rate": 4.6151226302432074e-05,
"loss": 0.0381,
"step": 16450
},
{
"grad_norm": 0.3162977397441864,
"learning_rate": 4.6096276520185304e-05,
"loss": 0.042,
"step": 16460
},
{
"grad_norm": 0.21985571086406708,
"learning_rate": 4.604133148133066e-05,
"loss": 0.0408,
"step": 16470
},
{
"grad_norm": 0.22086675465106964,
"learning_rate": 4.598639125263155e-05,
"loss": 0.0464,
"step": 16480
},
{
"grad_norm": 0.30571097135543823,
"learning_rate": 4.593145590084553e-05,
"loss": 0.0455,
"step": 16490
},
{
"grad_norm": 0.2800919711589813,
"learning_rate": 4.58765254927242e-05,
"loss": 0.0409,
"step": 16500
},
{
"grad_norm": 0.2297942340373993,
"learning_rate": 4.582160009501323e-05,
"loss": 0.0414,
"step": 16510
},
{
"grad_norm": 0.25374218821525574,
"learning_rate": 4.5766679774452143e-05,
"loss": 0.0421,
"step": 16520
},
{
"grad_norm": 0.18726161122322083,
"learning_rate": 4.571176459777431e-05,
"loss": 0.0464,
"step": 16530
},
{
"grad_norm": 0.2580954134464264,
"learning_rate": 4.5656854631706846e-05,
"loss": 0.0454,
"step": 16540
},
{
"grad_norm": 0.3155600130558014,
"learning_rate": 4.560194994297054e-05,
"loss": 0.0548,
"step": 16550
},
{
"grad_norm": 0.2772836983203888,
"learning_rate": 4.554705059827974e-05,
"loss": 0.0505,
"step": 16560
},
{
"grad_norm": 0.25732606649398804,
"learning_rate": 4.5492156664342365e-05,
"loss": 0.0493,
"step": 16570
},
{
"grad_norm": 0.23419468104839325,
"learning_rate": 4.5437268207859695e-05,
"loss": 0.0498,
"step": 16580
},
{
"grad_norm": 0.26506146788597107,
"learning_rate": 4.538238529552641e-05,
"loss": 0.0499,
"step": 16590
},
{
"grad_norm": 0.23580984771251678,
"learning_rate": 4.5327507994030394e-05,
"loss": 0.049,
"step": 16600
},
{
"grad_norm": 0.2190466672182083,
"learning_rate": 4.527263637005273e-05,
"loss": 0.0489,
"step": 16610
},
{
"grad_norm": 0.28061193227767944,
"learning_rate": 4.521777049026766e-05,
"loss": 0.0522,
"step": 16620
},
{
"grad_norm": 0.2862352132797241,
"learning_rate": 4.5162910421342386e-05,
"loss": 0.0465,
"step": 16630
},
{
"grad_norm": 0.2629992961883545,
"learning_rate": 4.510805622993706e-05,
"loss": 0.0496,
"step": 16640
},
{
"grad_norm": 0.2484590858221054,
"learning_rate": 4.505320798270467e-05,
"loss": 0.0535,
"step": 16650
},
{
"grad_norm": 0.17633068561553955,
"learning_rate": 4.499836574629105e-05,
"loss": 0.0394,
"step": 16660
},
{
"grad_norm": 0.2302699238061905,
"learning_rate": 4.494352958733465e-05,
"loss": 0.0441,
"step": 16670
},
{
"grad_norm": 0.1859251856803894,
"learning_rate": 4.488869957246663e-05,
"loss": 0.0425,
"step": 16680
},
{
"grad_norm": 0.2368263453245163,
"learning_rate": 4.4833875768310575e-05,
"loss": 0.0455,
"step": 16690
},
{
"grad_norm": 0.28583237528800964,
"learning_rate": 4.47790582414826e-05,
"loss": 0.045,
"step": 16700
},
{
"grad_norm": 0.30050718784332275,
"learning_rate": 4.472424705859114e-05,
"loss": 0.0385,
"step": 16710
},
{
"grad_norm": 0.20519119501113892,
"learning_rate": 4.466944228623701e-05,
"loss": 0.0424,
"step": 16720
},
{
"grad_norm": 0.24367737770080566,
"learning_rate": 4.461464399101312e-05,
"loss": 0.0364,
"step": 16730
},
{
"grad_norm": 0.20735490322113037,
"learning_rate": 4.45598522395046e-05,
"loss": 0.0321,
"step": 16740
},
{
"grad_norm": 0.2423200160264969,
"learning_rate": 4.4505067098288577e-05,
"loss": 0.0333,
"step": 16750
},
{
"grad_norm": 0.2585209012031555,
"learning_rate": 4.445028863393417e-05,
"loss": 0.043,
"step": 16760
},
{
"grad_norm": 0.273395836353302,
"learning_rate": 4.439551691300236e-05,
"loss": 0.043,
"step": 16770
},
{
"grad_norm": 0.2151714712381363,
"learning_rate": 4.434075200204599e-05,
"loss": 0.0404,
"step": 16780
},
{
"grad_norm": 0.23354794085025787,
"learning_rate": 4.428599396760956e-05,
"loss": 0.0452,
"step": 16790
},
{
"grad_norm": 0.2630499005317688,
"learning_rate": 4.423124287622926e-05,
"loss": 0.0496,
"step": 16800
},
{
"grad_norm": 0.24308457970619202,
"learning_rate": 4.417649879443282e-05,
"loss": 0.0532,
"step": 16810
},
{
"grad_norm": 0.20168504118919373,
"learning_rate": 4.4121761788739444e-05,
"loss": 0.0487,
"step": 16820
},
{
"grad_norm": 0.3582064211368561,
"learning_rate": 4.4067031925659805e-05,
"loss": 0.0447,
"step": 16830
},
{
"grad_norm": 0.2573299705982208,
"learning_rate": 4.4012309271695816e-05,
"loss": 0.0449,
"step": 16840
},
{
"grad_norm": 0.24972954392433167,
"learning_rate": 4.395759389334067e-05,
"loss": 0.0456,
"step": 16850
},
{
"grad_norm": 0.25737905502319336,
"learning_rate": 4.3902885857078685e-05,
"loss": 0.0415,
"step": 16860
},
{
"grad_norm": 0.28830280900001526,
"learning_rate": 4.384818522938531e-05,
"loss": 0.045,
"step": 16870
},
{
"grad_norm": 0.21057310700416565,
"learning_rate": 4.379349207672695e-05,
"loss": 0.0453,
"step": 16880
},
{
"grad_norm": 0.27277544140815735,
"learning_rate": 4.373880646556098e-05,
"loss": 0.0456,
"step": 16890
},
{
"grad_norm": 0.22695840895175934,
"learning_rate": 4.3684128462335533e-05,
"loss": 0.05,
"step": 16900
},
{
"grad_norm": 0.29053574800491333,
"learning_rate": 4.362945813348955e-05,
"loss": 0.0408,
"step": 16910
},
{
"grad_norm": 0.25185078382492065,
"learning_rate": 4.357479554545262e-05,
"loss": 0.0408,
"step": 16920
},
{
"grad_norm": 0.28470367193222046,
"learning_rate": 4.352014076464499e-05,
"loss": 0.0434,
"step": 16930
},
{
"grad_norm": 0.2341253161430359,
"learning_rate": 4.346549385747734e-05,
"loss": 0.0431,
"step": 16940
},
{
"grad_norm": 0.34226688742637634,
"learning_rate": 4.34108548903508e-05,
"loss": 0.0392,
"step": 16950
},
{
"grad_norm": 0.26250115036964417,
"learning_rate": 4.335622392965689e-05,
"loss": 0.0391,
"step": 16960
},
{
"grad_norm": 0.21088621020317078,
"learning_rate": 4.330160104177738e-05,
"loss": 0.0439,
"step": 16970
},
{
"grad_norm": 0.19729141891002655,
"learning_rate": 4.32469862930842e-05,
"loss": 0.0372,
"step": 16980
},
{
"grad_norm": 0.24273568391799927,
"learning_rate": 4.319237974993946e-05,
"loss": 0.0415,
"step": 16990
},
{
"grad_norm": 0.22975994646549225,
"learning_rate": 4.3137781478695236e-05,
"loss": 0.0439,
"step": 17000
},
{
"grad_norm": 0.2597068250179291,
"learning_rate": 4.308319154569358e-05,
"loss": 0.0406,
"step": 17010
},
{
"grad_norm": 0.22805127501487732,
"learning_rate": 4.302861001726641e-05,
"loss": 0.039,
"step": 17020
},
{
"grad_norm": 0.2874810993671417,
"learning_rate": 4.2974036959735415e-05,
"loss": 0.0402,
"step": 17030
},
{
"grad_norm": 0.2207222431898117,
"learning_rate": 4.2919472439412034e-05,
"loss": 0.0389,
"step": 17040
},
{
"grad_norm": 0.23963770270347595,
"learning_rate": 4.286491652259729e-05,
"loss": 0.0401,
"step": 17050
},
{
"grad_norm": 0.22334584593772888,
"learning_rate": 4.281036927558178e-05,
"loss": 0.0421,
"step": 17060
},
{
"grad_norm": 0.2502647638320923,
"learning_rate": 4.275583076464552e-05,
"loss": 0.0433,
"step": 17070
},
{
"grad_norm": 0.19369353353977203,
"learning_rate": 4.270130105605794e-05,
"loss": 0.0405,
"step": 17080
},
{
"grad_norm": 0.22441303730010986,
"learning_rate": 4.264678021607782e-05,
"loss": 0.0371,
"step": 17090
},
{
"grad_norm": 0.2386510819196701,
"learning_rate": 4.25922683109531e-05,
"loss": 0.0419,
"step": 17100
},
{
"grad_norm": 0.31641948223114014,
"learning_rate": 4.25377654069209e-05,
"loss": 0.043,
"step": 17110
},
{
"grad_norm": 0.24089911580085754,
"learning_rate": 4.248327157020737e-05,
"loss": 0.0395,
"step": 17120
},
{
"grad_norm": 0.23753131926059723,
"learning_rate": 4.242878686702763e-05,
"loss": 0.0464,
"step": 17130
},
{
"grad_norm": 0.24898654222488403,
"learning_rate": 4.23743113635858e-05,
"loss": 0.0406,
"step": 17140
},
{
"grad_norm": 0.2762834131717682,
"learning_rate": 4.2319845126074705e-05,
"loss": 0.0433,
"step": 17150
},
{
"grad_norm": 0.2910861670970917,
"learning_rate": 4.226538822067598e-05,
"loss": 0.0426,
"step": 17160
},
{
"grad_norm": 0.2944703996181488,
"learning_rate": 4.2210940713559896e-05,
"loss": 0.0427,
"step": 17170
},
{
"grad_norm": 0.26751646399497986,
"learning_rate": 4.21565026708853e-05,
"loss": 0.0452,
"step": 17180
},
{
"grad_norm": 0.22268252074718475,
"learning_rate": 4.2102074158799535e-05,
"loss": 0.036,
"step": 17190
},
{
"grad_norm": 0.2522592544555664,
"learning_rate": 4.204765524343841e-05,
"loss": 0.0408,
"step": 17200
},
{
"grad_norm": 0.18269868195056915,
"learning_rate": 4.1993245990926036e-05,
"loss": 0.0397,
"step": 17210
},
{
"grad_norm": 0.20358693599700928,
"learning_rate": 4.193884646737474e-05,
"loss": 0.0398,
"step": 17220
},
{
"grad_norm": 0.23927372694015503,
"learning_rate": 4.1884456738885124e-05,
"loss": 0.0358,
"step": 17230
},
{
"grad_norm": 0.24257494509220123,
"learning_rate": 4.1830076871545795e-05,
"loss": 0.0357,
"step": 17240
},
{
"grad_norm": 0.20762422680854797,
"learning_rate": 4.177570693143346e-05,
"loss": 0.037,
"step": 17250
},
{
"grad_norm": 0.2919173240661621,
"learning_rate": 4.172134698461271e-05,
"loss": 0.0392,
"step": 17260
},
{
"grad_norm": 0.33089086413383484,
"learning_rate": 4.166699709713599e-05,
"loss": 0.0398,
"step": 17270
},
{
"grad_norm": 0.23980920016765594,
"learning_rate": 4.1612657335043555e-05,
"loss": 0.0391,
"step": 17280
},
{
"grad_norm": 0.22947347164154053,
"learning_rate": 4.155832776436331e-05,
"loss": 0.0347,
"step": 17290
},
{
"grad_norm": 0.30801156163215637,
"learning_rate": 4.1504008451110845e-05,
"loss": 0.041,
"step": 17300
},
{
"grad_norm": 0.2813476026058197,
"learning_rate": 4.1449699461289225e-05,
"loss": 0.0405,
"step": 17310
},
{
"grad_norm": 0.2448110282421112,
"learning_rate": 4.139540086088901e-05,
"loss": 0.0393,
"step": 17320
},
{
"grad_norm": 0.2404831498861313,
"learning_rate": 4.1341112715888106e-05,
"loss": 0.0442,
"step": 17330
},
{
"grad_norm": 0.277383953332901,
"learning_rate": 4.128683509225172e-05,
"loss": 0.0434,
"step": 17340
},
{
"grad_norm": 0.2359265834093094,
"learning_rate": 4.123256805593231e-05,
"loss": 0.0459,
"step": 17350
},
{
"grad_norm": 0.2240133434534073,
"learning_rate": 4.117831167286943e-05,
"loss": 0.0451,
"step": 17360
},
{
"grad_norm": 0.26490435004234314,
"learning_rate": 4.1124066008989685e-05,
"loss": 0.0388,
"step": 17370
},
{
"grad_norm": 0.24470119178295135,
"learning_rate": 4.106983113020669e-05,
"loss": 0.0425,
"step": 17380
},
{
"grad_norm": 0.2700499892234802,
"learning_rate": 4.101560710242094e-05,
"loss": 0.0401,
"step": 17390
},
{
"grad_norm": 0.3246651291847229,
"learning_rate": 4.096139399151971e-05,
"loss": 0.0445,
"step": 17400
},
{
"grad_norm": 0.29312121868133545,
"learning_rate": 4.090719186337709e-05,
"loss": 0.0415,
"step": 17410
},
{
"grad_norm": 0.29748082160949707,
"learning_rate": 4.0853000783853746e-05,
"loss": 0.0414,
"step": 17420
},
{
"grad_norm": 0.23067064583301544,
"learning_rate": 4.0798820818796956e-05,
"loss": 0.0468,
"step": 17430
},
{
"grad_norm": 0.26061466336250305,
"learning_rate": 4.074465203404048e-05,
"loss": 0.0372,
"step": 17440
},
{
"grad_norm": 0.20375820994377136,
"learning_rate": 4.0690494495404495e-05,
"loss": 0.0404,
"step": 17450
},
{
"grad_norm": 0.21558482944965363,
"learning_rate": 4.0636348268695535e-05,
"loss": 0.0422,
"step": 17460
},
{
"grad_norm": 0.2138373851776123,
"learning_rate": 4.058221341970637e-05,
"loss": 0.0391,
"step": 17470
},
{
"grad_norm": 0.2216326743364334,
"learning_rate": 4.0528090014215945e-05,
"loss": 0.0433,
"step": 17480
},
{
"grad_norm": 0.24290354549884796,
"learning_rate": 4.047397811798929e-05,
"loss": 0.0389,
"step": 17490
},
{
"grad_norm": 0.23431578278541565,
"learning_rate": 4.041987779677745e-05,
"loss": 0.038,
"step": 17500
},
{
"grad_norm": 0.20056647062301636,
"learning_rate": 4.036578911631745e-05,
"loss": 0.0351,
"step": 17510
},
{
"grad_norm": 0.24972932040691376,
"learning_rate": 4.031171214233211e-05,
"loss": 0.0422,
"step": 17520
},
{
"grad_norm": 0.20535226166248322,
"learning_rate": 4.025764694053008e-05,
"loss": 0.0374,
"step": 17530
},
{
"grad_norm": 0.21940284967422485,
"learning_rate": 4.020359357660566e-05,
"loss": 0.0445,
"step": 17540
},
{
"grad_norm": 0.2667447030544281,
"learning_rate": 4.014955211623875e-05,
"loss": 0.0451,
"step": 17550
},
{
"grad_norm": 0.23712141811847687,
"learning_rate": 4.0095522625094895e-05,
"loss": 0.0472,
"step": 17560
},
{
"grad_norm": 0.18918225169181824,
"learning_rate": 4.004150516882497e-05,
"loss": 0.0409,
"step": 17570
},
{
"grad_norm": 0.20510989427566528,
"learning_rate": 3.99874998130653e-05,
"loss": 0.0422,
"step": 17580
},
{
"grad_norm": 0.23385252058506012,
"learning_rate": 3.9933506623437455e-05,
"loss": 0.0357,
"step": 17590
},
{
"grad_norm": 0.23668572306632996,
"learning_rate": 3.9879525665548276e-05,
"loss": 0.0402,
"step": 17600
},
{
"grad_norm": 0.18782325088977814,
"learning_rate": 3.9825557004989704e-05,
"loss": 0.0371,
"step": 17610
},
{
"grad_norm": 0.2863144278526306,
"learning_rate": 3.977160070733877e-05,
"loss": 0.0427,
"step": 17620
},
{
"grad_norm": 0.19398735463619232,
"learning_rate": 3.971765683815746e-05,
"loss": 0.039,
"step": 17630
},
{
"grad_norm": 0.1992606371641159,
"learning_rate": 3.966372546299265e-05,
"loss": 0.0388,
"step": 17640
},
{
"grad_norm": 0.17186053097248077,
"learning_rate": 3.960980664737604e-05,
"loss": 0.0399,
"step": 17650
},
{
"grad_norm": 0.21715551614761353,
"learning_rate": 3.955590045682408e-05,
"loss": 0.0381,
"step": 17660
},
{
"grad_norm": 0.24425223469734192,
"learning_rate": 3.950200695683788e-05,
"loss": 0.0442,
"step": 17670
},
{
"grad_norm": 0.16548076272010803,
"learning_rate": 3.944812621290314e-05,
"loss": 0.0375,
"step": 17680
},
{
"grad_norm": 0.18859930336475372,
"learning_rate": 3.939425829049002e-05,
"loss": 0.0397,
"step": 17690
},
{
"grad_norm": 0.20394188165664673,
"learning_rate": 3.9340403255053135e-05,
"loss": 0.0487,
"step": 17700
},
{
"grad_norm": 0.2106298804283142,
"learning_rate": 3.9286561172031407e-05,
"loss": 0.0392,
"step": 17710
},
{
"grad_norm": 0.24647602438926697,
"learning_rate": 3.923273210684809e-05,
"loss": 0.0383,
"step": 17720
},
{
"grad_norm": 0.23291322588920593,
"learning_rate": 3.9178916124910555e-05,
"loss": 0.0396,
"step": 17730
},
{
"grad_norm": 0.24432086944580078,
"learning_rate": 3.9125113291610274e-05,
"loss": 0.038,
"step": 17740
},
{
"grad_norm": 0.2440585047006607,
"learning_rate": 3.907132367232278e-05,
"loss": 0.0361,
"step": 17750
},
{
"grad_norm": 0.2175803780555725,
"learning_rate": 3.901754733240752e-05,
"loss": 0.0415,
"step": 17760
},
{
"grad_norm": 0.27778545022010803,
"learning_rate": 3.896378433720785e-05,
"loss": 0.0395,
"step": 17770
},
{
"grad_norm": 0.20533399283885956,
"learning_rate": 3.891003475205086e-05,
"loss": 0.0371,
"step": 17780
},
{
"grad_norm": 0.33764806389808655,
"learning_rate": 3.885629864224736e-05,
"loss": 0.0464,
"step": 17790
},
{
"grad_norm": 0.28361326456069946,
"learning_rate": 3.880257607309178e-05,
"loss": 0.0472,
"step": 17800
},
{
"grad_norm": 0.24788931012153625,
"learning_rate": 3.874886710986213e-05,
"loss": 0.0479,
"step": 17810
},
{
"grad_norm": 0.25397729873657227,
"learning_rate": 3.8695171817819834e-05,
"loss": 0.0458,
"step": 17820
},
{
"grad_norm": 0.2608543336391449,
"learning_rate": 3.8641490262209774e-05,
"loss": 0.0436,
"step": 17830
},
{
"grad_norm": 0.23443326354026794,
"learning_rate": 3.8587822508260084e-05,
"loss": 0.0444,
"step": 17840
},
{
"grad_norm": 0.26841938495635986,
"learning_rate": 3.853416862118214e-05,
"loss": 0.0513,
"step": 17850
},
{
"grad_norm": 0.21288099884986877,
"learning_rate": 3.848052866617049e-05,
"loss": 0.0364,
"step": 17860
},
{
"grad_norm": 0.24344845116138458,
"learning_rate": 3.84269027084027e-05,
"loss": 0.0422,
"step": 17870
},
{
"grad_norm": 0.24393054842948914,
"learning_rate": 3.83732908130394e-05,
"loss": 0.0425,
"step": 17880
},
{
"grad_norm": 0.22546300292015076,
"learning_rate": 3.831969304522409e-05,
"loss": 0.0399,
"step": 17890
},
{
"grad_norm": 0.22337380051612854,
"learning_rate": 3.826610947008312e-05,
"loss": 0.0445,
"step": 17900
},
{
"grad_norm": 0.3501909077167511,
"learning_rate": 3.821254015272559e-05,
"loss": 0.0442,
"step": 17910
},
{
"grad_norm": 0.15416522324085236,
"learning_rate": 3.8158985158243214e-05,
"loss": 0.0471,
"step": 17920
},
{
"grad_norm": 0.2972775399684906,
"learning_rate": 3.810544455171044e-05,
"loss": 0.0474,
"step": 17930
},
{
"grad_norm": 0.2955247759819031,
"learning_rate": 3.8051918398184114e-05,
"loss": 0.0433,
"step": 17940
},
{
"grad_norm": 0.1636299192905426,
"learning_rate": 3.799840676270356e-05,
"loss": 0.0406,
"step": 17950
},
{
"grad_norm": 0.2458428591489792,
"learning_rate": 3.794490971029048e-05,
"loss": 0.0434,
"step": 17960
},
{
"grad_norm": 0.25921162962913513,
"learning_rate": 3.789142730594881e-05,
"loss": 0.0472,
"step": 17970
},
{
"grad_norm": 0.24500936269760132,
"learning_rate": 3.783795961466471e-05,
"loss": 0.0415,
"step": 17980
},
{
"grad_norm": 0.25606709718704224,
"learning_rate": 3.778450670140651e-05,
"loss": 0.046,
"step": 17990
},
{
"grad_norm": 0.2675757110118866,
"learning_rate": 3.773106863112451e-05,
"loss": 0.0456,
"step": 18000
},
{
"grad_norm": 0.2798529267311096,
"learning_rate": 3.7677645468751e-05,
"loss": 0.0398,
"step": 18010
},
{
"grad_norm": 0.2074119597673416,
"learning_rate": 3.762423727920018e-05,
"loss": 0.0448,
"step": 18020
},
{
"grad_norm": 0.3195675015449524,
"learning_rate": 3.7570844127367995e-05,
"loss": 0.0457,
"step": 18030
},
{
"grad_norm": 0.23299692571163177,
"learning_rate": 3.751746607813221e-05,
"loss": 0.0424,
"step": 18040
},
{
"grad_norm": 0.19835485517978668,
"learning_rate": 3.746410319635217e-05,
"loss": 0.0435,
"step": 18050
},
{
"grad_norm": 0.2061367779970169,
"learning_rate": 3.7410755546868806e-05,
"loss": 0.0442,
"step": 18060
},
{
"grad_norm": 0.27388718724250793,
"learning_rate": 3.735742319450454e-05,
"loss": 0.0485,
"step": 18070
},
{
"grad_norm": 0.22543561458587646,
"learning_rate": 3.7304106204063184e-05,
"loss": 0.0453,
"step": 18080
},
{
"grad_norm": 0.2597689628601074,
"learning_rate": 3.725080464032996e-05,
"loss": 0.0425,
"step": 18090
},
{
"grad_norm": 0.18491318821907043,
"learning_rate": 3.719751856807125e-05,
"loss": 0.0441,
"step": 18100
},
{
"grad_norm": 0.27627354860305786,
"learning_rate": 3.71442480520347e-05,
"loss": 0.0431,
"step": 18110
},
{
"grad_norm": 0.18951763212680817,
"learning_rate": 3.709099315694897e-05,
"loss": 0.0412,
"step": 18120
},
{
"grad_norm": 0.21118535101413727,
"learning_rate": 3.703775394752378e-05,
"loss": 0.0425,
"step": 18130
},
{
"grad_norm": 0.20011115074157715,
"learning_rate": 3.698453048844983e-05,
"loss": 0.0461,
"step": 18140
},
{
"grad_norm": 0.20581305027008057,
"learning_rate": 3.693132284439861e-05,
"loss": 0.0368,
"step": 18150
},
{
"grad_norm": 0.21786653995513916,
"learning_rate": 3.6878131080022414e-05,
"loss": 0.0464,
"step": 18160
},
{
"grad_norm": 0.22326785326004028,
"learning_rate": 3.682495525995429e-05,
"loss": 0.0402,
"step": 18170
},
{
"grad_norm": 0.21070930361747742,
"learning_rate": 3.6771795448807846e-05,
"loss": 0.0463,
"step": 18180
},
{
"grad_norm": 0.2747941017150879,
"learning_rate": 3.671865171117724e-05,
"loss": 0.0434,
"step": 18190
},
{
"grad_norm": 0.1904992163181305,
"learning_rate": 3.666552411163718e-05,
"loss": 0.0392,
"step": 18200
},
{
"grad_norm": 0.2862382233142853,
"learning_rate": 3.661241271474269e-05,
"loss": 0.0402,
"step": 18210
},
{
"grad_norm": 0.22728949785232544,
"learning_rate": 3.6559317585029116e-05,
"loss": 0.0439,
"step": 18220
},
{
"grad_norm": 0.2740098536014557,
"learning_rate": 3.650623878701204e-05,
"loss": 0.0456,
"step": 18230
},
{
"grad_norm": 0.24038535356521606,
"learning_rate": 3.645317638518721e-05,
"loss": 0.0399,
"step": 18240
},
{
"grad_norm": 0.22618719935417175,
"learning_rate": 3.640013044403046e-05,
"loss": 0.0444,
"step": 18250
},
{
"grad_norm": 0.18376512825489044,
"learning_rate": 3.634710102799761e-05,
"loss": 0.0377,
"step": 18260
},
{
"grad_norm": 0.2520391345024109,
"learning_rate": 3.6294088201524394e-05,
"loss": 0.0431,
"step": 18270
},
{
"grad_norm": 0.18286025524139404,
"learning_rate": 3.62410920290264e-05,
"loss": 0.0353,
"step": 18280
},
{
"grad_norm": 0.25871551036834717,
"learning_rate": 3.6188112574898954e-05,
"loss": 0.0435,
"step": 18290
},
{
"grad_norm": 0.26164939999580383,
"learning_rate": 3.6135149903517115e-05,
"loss": 0.0391,
"step": 18300
},
{
"grad_norm": 0.20961543917655945,
"learning_rate": 3.608220407923552e-05,
"loss": 0.0388,
"step": 18310
},
{
"grad_norm": 0.20862069725990295,
"learning_rate": 3.602927516638833e-05,
"loss": 0.0368,
"step": 18320
},
{
"grad_norm": 0.2546931803226471,
"learning_rate": 3.5976363229289165e-05,
"loss": 0.0458,
"step": 18330
},
{
"grad_norm": 0.25214657187461853,
"learning_rate": 3.5923468332231e-05,
"loss": 0.0434,
"step": 18340
},
{
"grad_norm": 0.18178719282150269,
"learning_rate": 3.587059053948616e-05,
"loss": 0.0361,
"step": 18350
},
{
"grad_norm": 0.2568759620189667,
"learning_rate": 3.5817729915306134e-05,
"loss": 0.0429,
"step": 18360
},
{
"grad_norm": 0.2525114119052887,
"learning_rate": 3.5764886523921566e-05,
"loss": 0.0452,
"step": 18370
},
{
"grad_norm": 0.3183315098285675,
"learning_rate": 3.571206042954214e-05,
"loss": 0.0406,
"step": 18380
},
{
"grad_norm": 0.2480093538761139,
"learning_rate": 3.5659251696356566e-05,
"loss": 0.0409,
"step": 18390
},
{
"grad_norm": 0.23349300026893616,
"learning_rate": 3.56064603885324e-05,
"loss": 0.0485,
"step": 18400
},
{
"grad_norm": 0.20299144089221954,
"learning_rate": 3.5553686570216115e-05,
"loss": 0.0438,
"step": 18410
},
{
"grad_norm": 0.20766325294971466,
"learning_rate": 3.550093030553284e-05,
"loss": 0.0399,
"step": 18420
},
{
"grad_norm": 0.2589561641216278,
"learning_rate": 3.544819165858642e-05,
"loss": 0.0402,
"step": 18430
},
{
"grad_norm": 0.26660415530204773,
"learning_rate": 3.539547069345926e-05,
"loss": 0.0426,
"step": 18440
},
{
"grad_norm": 0.24452301859855652,
"learning_rate": 3.5342767474212344e-05,
"loss": 0.0454,
"step": 18450
},
{
"grad_norm": 0.253874808549881,
"learning_rate": 3.529008206488502e-05,
"loss": 0.0446,
"step": 18460
},
{
"grad_norm": 0.267327219247818,
"learning_rate": 3.5237414529495055e-05,
"loss": 0.0389,
"step": 18470
},
{
"grad_norm": 0.2717471122741699,
"learning_rate": 3.5184764932038454e-05,
"loss": 0.0365,
"step": 18480
},
{
"grad_norm": 0.25722411274909973,
"learning_rate": 3.513213333648945e-05,
"loss": 0.0398,
"step": 18490
},
{
"grad_norm": 0.2525750994682312,
"learning_rate": 3.5079519806800374e-05,
"loss": 0.0398,
"step": 18500
},
{
"grad_norm": 0.282569020986557,
"learning_rate": 3.502692440690165e-05,
"loss": 0.042,
"step": 18510
},
{
"grad_norm": 0.22381988167762756,
"learning_rate": 3.497434720070165e-05,
"loss": 0.0391,
"step": 18520
},
{
"grad_norm": 0.2427671253681183,
"learning_rate": 3.4921788252086616e-05,
"loss": 0.0412,
"step": 18530
},
{
"grad_norm": 0.2386632263660431,
"learning_rate": 3.486924762492065e-05,
"loss": 0.0413,
"step": 18540
},
{
"grad_norm": 0.17656196653842926,
"learning_rate": 3.4816725383045534e-05,
"loss": 0.0351,
"step": 18550
},
{
"grad_norm": 0.22033797204494476,
"learning_rate": 3.476422159028079e-05,
"loss": 0.0429,
"step": 18560
},
{
"grad_norm": 0.20949293673038483,
"learning_rate": 3.471173631042345e-05,
"loss": 0.0344,
"step": 18570
},
{
"grad_norm": 0.19709567725658417,
"learning_rate": 3.465926960724808e-05,
"loss": 0.0346,
"step": 18580
},
{
"grad_norm": 0.2954472303390503,
"learning_rate": 3.460682154450666e-05,
"loss": 0.0379,
"step": 18590
},
{
"grad_norm": 0.2521780729293823,
"learning_rate": 3.4554392185928564e-05,
"loss": 0.0391,
"step": 18600
},
{
"grad_norm": 0.20922695100307465,
"learning_rate": 3.450198159522037e-05,
"loss": 0.0376,
"step": 18610
},
{
"grad_norm": 0.20044288039207458,
"learning_rate": 3.444958983606592e-05,
"loss": 0.0377,
"step": 18620
},
{
"grad_norm": 0.23954859375953674,
"learning_rate": 3.439721697212612e-05,
"loss": 0.0375,
"step": 18630
},
{
"grad_norm": 0.25387296080589294,
"learning_rate": 3.4344863067038954e-05,
"loss": 0.0387,
"step": 18640
},
{
"grad_norm": 0.22904056310653687,
"learning_rate": 3.429252818441935e-05,
"loss": 0.0373,
"step": 18650
},
{
"grad_norm": 0.21606239676475525,
"learning_rate": 3.4240212387859094e-05,
"loss": 0.0421,
"step": 18660
},
{
"grad_norm": 0.3004136383533478,
"learning_rate": 3.418791574092686e-05,
"loss": 0.0451,
"step": 18670
},
{
"grad_norm": 0.19411419332027435,
"learning_rate": 3.413563830716796e-05,
"loss": 0.0342,
"step": 18680
},
{
"grad_norm": 0.22306843101978302,
"learning_rate": 3.408338015010445e-05,
"loss": 0.0373,
"step": 18690
},
{
"grad_norm": 0.2377527803182602,
"learning_rate": 3.40311413332349e-05,
"loss": 0.0458,
"step": 18700
},
{
"grad_norm": 0.23494932055473328,
"learning_rate": 3.3978921920034365e-05,
"loss": 0.0477,
"step": 18710
},
{
"grad_norm": 0.19891507923603058,
"learning_rate": 3.392672197395441e-05,
"loss": 0.0425,
"step": 18720
},
{
"grad_norm": 0.2535465359687805,
"learning_rate": 3.387454155842287e-05,
"loss": 0.0416,
"step": 18730
},
{
"grad_norm": 0.224177286028862,
"learning_rate": 3.382238073684386e-05,
"loss": 0.0462,
"step": 18740
},
{
"grad_norm": 0.24074696004390717,
"learning_rate": 3.377023957259771e-05,
"loss": 0.0446,
"step": 18750
},
{
"grad_norm": 0.3233257234096527,
"learning_rate": 3.3718118129040835e-05,
"loss": 0.0409,
"step": 18760
},
{
"grad_norm": 0.3109950125217438,
"learning_rate": 3.3666016469505724e-05,
"loss": 0.0422,
"step": 18770
},
{
"grad_norm": 0.2582498788833618,
"learning_rate": 3.361393465730079e-05,
"loss": 0.0386,
"step": 18780
},
{
"grad_norm": 0.23778612911701202,
"learning_rate": 3.3561872755710366e-05,
"loss": 0.0419,
"step": 18790
},
{
"grad_norm": 0.21542304754257202,
"learning_rate": 3.350983082799456e-05,
"loss": 0.0382,
"step": 18800
},
{
"grad_norm": 0.24762573838233948,
"learning_rate": 3.34578089373892e-05,
"loss": 0.0416,
"step": 18810
},
{
"grad_norm": 0.23513144254684448,
"learning_rate": 3.340580714710581e-05,
"loss": 0.0384,
"step": 18820
},
{
"grad_norm": 0.23195451498031616,
"learning_rate": 3.3353825520331466e-05,
"loss": 0.0372,
"step": 18830
},
{
"grad_norm": 0.18893930315971375,
"learning_rate": 3.330186412022876e-05,
"loss": 0.0357,
"step": 18840
},
{
"grad_norm": 0.20610663294792175,
"learning_rate": 3.324992300993568e-05,
"loss": 0.0344,
"step": 18850
},
{
"grad_norm": 0.2942042350769043,
"learning_rate": 3.319800225256556e-05,
"loss": 0.0387,
"step": 18860
},
{
"grad_norm": 0.25522539019584656,
"learning_rate": 3.314610191120702e-05,
"loss": 0.0403,
"step": 18870
},
{
"grad_norm": 0.22649113833904266,
"learning_rate": 3.30942220489239e-05,
"loss": 0.0331,
"step": 18880
},
{
"grad_norm": 0.22661210596561432,
"learning_rate": 3.3042362728755086e-05,
"loss": 0.0359,
"step": 18890
},
{
"grad_norm": 0.20929859578609467,
"learning_rate": 3.299052401371456e-05,
"loss": 0.0387,
"step": 18900
},
{
"grad_norm": 0.2471480816602707,
"learning_rate": 3.293870596679125e-05,
"loss": 0.0336,
"step": 18910
},
{
"grad_norm": 0.20952628552913666,
"learning_rate": 3.288690865094895e-05,
"loss": 0.0357,
"step": 18920
},
{
"grad_norm": 0.27363121509552,
"learning_rate": 3.283513212912632e-05,
"loss": 0.0362,
"step": 18930
},
{
"grad_norm": 0.20753365755081177,
"learning_rate": 3.2783376464236684e-05,
"loss": 0.0379,
"step": 18940
},
{
"grad_norm": 0.26773202419281006,
"learning_rate": 3.273164171916806e-05,
"loss": 0.0415,
"step": 18950
},
{
"grad_norm": 0.24766762554645538,
"learning_rate": 3.267992795678306e-05,
"loss": 0.0396,
"step": 18960
},
{
"grad_norm": 0.194326713681221,
"learning_rate": 3.2628235239918744e-05,
"loss": 0.0344,
"step": 18970
},
{
"grad_norm": 0.19778311252593994,
"learning_rate": 3.2576563631386695e-05,
"loss": 0.0385,
"step": 18980
},
{
"grad_norm": 0.21886137127876282,
"learning_rate": 3.252491319397275e-05,
"loss": 0.0357,
"step": 18990
},
{
"grad_norm": 0.23862069845199585,
"learning_rate": 3.247328399043706e-05,
"loss": 0.0412,
"step": 19000
},
{
"grad_norm": 0.24115510284900665,
"learning_rate": 3.242167608351399e-05,
"loss": 0.0399,
"step": 19010
},
{
"grad_norm": 0.22835344076156616,
"learning_rate": 3.2370089535911986e-05,
"loss": 0.043,
"step": 19020
},
{
"grad_norm": 0.1849687397480011,
"learning_rate": 3.23185244103136e-05,
"loss": 0.042,
"step": 19030
},
{
"grad_norm": 0.19990837574005127,
"learning_rate": 3.22669807693753e-05,
"loss": 0.0351,
"step": 19040
},
{
"grad_norm": 0.20480957627296448,
"learning_rate": 3.2215458675727495e-05,
"loss": 0.0368,
"step": 19050
},
{
"grad_norm": 0.18700067698955536,
"learning_rate": 3.216395819197438e-05,
"loss": 0.0334,
"step": 19060
},
{
"grad_norm": 0.21672746539115906,
"learning_rate": 3.211247938069387e-05,
"loss": 0.0335,
"step": 19070
},
{
"grad_norm": 0.311467707157135,
"learning_rate": 3.206102230443759e-05,
"loss": 0.0368,
"step": 19080
},
{
"grad_norm": 0.27216923236846924,
"learning_rate": 3.2009587025730764e-05,
"loss": 0.0361,
"step": 19090
},
{
"grad_norm": 0.2709263265132904,
"learning_rate": 3.195817360707207e-05,
"loss": 0.039,
"step": 19100
},
{
"grad_norm": 0.16970647871494293,
"learning_rate": 3.19067821109337e-05,
"loss": 0.0331,
"step": 19110
},
{
"grad_norm": 0.21941938996315002,
"learning_rate": 3.185541259976114e-05,
"loss": 0.0365,
"step": 19120
},
{
"grad_norm": 0.2659476697444916,
"learning_rate": 3.180406513597316e-05,
"loss": 0.0356,
"step": 19130
},
{
"grad_norm": 0.223378524184227,
"learning_rate": 3.1752739781961835e-05,
"loss": 0.0365,
"step": 19140
},
{
"grad_norm": 0.1896408200263977,
"learning_rate": 3.170143660009228e-05,
"loss": 0.0376,
"step": 19150
},
{
"grad_norm": 0.22322721779346466,
"learning_rate": 3.16501556527027e-05,
"loss": 0.0368,
"step": 19160
},
{
"grad_norm": 0.21433135867118835,
"learning_rate": 3.1598897002104265e-05,
"loss": 0.0384,
"step": 19170
},
{
"grad_norm": 0.22972038388252258,
"learning_rate": 3.154766071058108e-05,
"loss": 0.0329,
"step": 19180
},
{
"grad_norm": 0.2167443037033081,
"learning_rate": 3.149644684039008e-05,
"loss": 0.0363,
"step": 19190
},
{
"grad_norm": 0.2225545048713684,
"learning_rate": 3.144525545376095e-05,
"loss": 0.0347,
"step": 19200
},
{
"grad_norm": 0.3011263310909271,
"learning_rate": 3.139408661289603e-05,
"loss": 0.0402,
"step": 19210
},
{
"grad_norm": 0.25113537907600403,
"learning_rate": 3.134294037997032e-05,
"loss": 0.0422,
"step": 19220
},
{
"grad_norm": 0.26568353176116943,
"learning_rate": 3.129181681713127e-05,
"loss": 0.0468,
"step": 19230
},
{
"grad_norm": 0.21064133942127228,
"learning_rate": 3.1240715986498855e-05,
"loss": 0.0446,
"step": 19240
},
{
"grad_norm": 0.18753895163536072,
"learning_rate": 3.1189637950165394e-05,
"loss": 0.0465,
"step": 19250
},
{
"grad_norm": 0.22880344092845917,
"learning_rate": 3.1138582770195544e-05,
"loss": 0.0466,
"step": 19260
},
{
"grad_norm": 0.20228902995586395,
"learning_rate": 3.108755050862615e-05,
"loss": 0.0409,
"step": 19270
},
{
"grad_norm": 0.19551384449005127,
"learning_rate": 3.1036541227466204e-05,
"loss": 0.0474,
"step": 19280
},
{
"grad_norm": 0.2145748734474182,
"learning_rate": 3.0985554988696786e-05,
"loss": 0.0439,
"step": 19290
},
{
"grad_norm": 0.1807737499475479,
"learning_rate": 3.093459185427102e-05,
"loss": 0.044,
"step": 19300
},
{
"grad_norm": 0.22352388501167297,
"learning_rate": 3.0883651886113905e-05,
"loss": 0.0438,
"step": 19310
},
{
"grad_norm": 0.29523608088493347,
"learning_rate": 3.0832735146122294e-05,
"loss": 0.0377,
"step": 19320
},
{
"grad_norm": 0.2290971577167511,
"learning_rate": 3.078184169616485e-05,
"loss": 0.0387,
"step": 19330
},
{
"grad_norm": 0.18039393424987793,
"learning_rate": 3.073097159808187e-05,
"loss": 0.0395,
"step": 19340
},
{
"grad_norm": 0.19013889133930206,
"learning_rate": 3.068012491368537e-05,
"loss": 0.0401,
"step": 19350
},
{
"grad_norm": 0.20428170263767242,
"learning_rate": 3.062930170475885e-05,
"loss": 0.038,
"step": 19360
},
{
"grad_norm": 0.2568492889404297,
"learning_rate": 3.057850203305729e-05,
"loss": 0.0379,
"step": 19370
},
{
"grad_norm": 0.23973898589611053,
"learning_rate": 3.052772596030708e-05,
"loss": 0.0445,
"step": 19380
},
{
"grad_norm": 0.21464787423610687,
"learning_rate": 3.0476973548205943e-05,
"loss": 0.0447,
"step": 19390
},
{
"grad_norm": 0.19437824189662933,
"learning_rate": 3.0426244858422847e-05,
"loss": 0.044,
"step": 19400
},
{
"grad_norm": 0.21956686675548553,
"learning_rate": 3.0375539952597943e-05,
"loss": 0.0409,
"step": 19410
},
{
"grad_norm": 0.18171855807304382,
"learning_rate": 3.0324858892342468e-05,
"loss": 0.0389,
"step": 19420
},
{
"grad_norm": 0.26618847250938416,
"learning_rate": 3.0274201739238672e-05,
"loss": 0.0421,
"step": 19430
},
{
"grad_norm": 0.22888627648353577,
"learning_rate": 3.0223568554839786e-05,
"loss": 0.044,
"step": 19440
},
{
"grad_norm": 0.20647931098937988,
"learning_rate": 3.0172959400669886e-05,
"loss": 0.0443,
"step": 19450
},
{
"grad_norm": 0.22473642230033875,
"learning_rate": 3.0122374338223902e-05,
"loss": 0.0403,
"step": 19460
},
{
"grad_norm": 0.16114316880702972,
"learning_rate": 3.0071813428967427e-05,
"loss": 0.0455,
"step": 19470
},
{
"grad_norm": 0.24044319987297058,
"learning_rate": 3.0021276734336746e-05,
"loss": 0.0452,
"step": 19480
},
{
"grad_norm": 0.22676408290863037,
"learning_rate": 2.997076431573871e-05,
"loss": 0.0475,
"step": 19490
},
{
"grad_norm": 0.22266586124897003,
"learning_rate": 2.9920276234550636e-05,
"loss": 0.0375,
"step": 19500
},
{
"grad_norm": 0.21760885417461395,
"learning_rate": 2.986981255212035e-05,
"loss": 0.0382,
"step": 19510
},
{
"grad_norm": 0.18137460947036743,
"learning_rate": 2.9819373329765977e-05,
"loss": 0.0341,
"step": 19520
},
{
"grad_norm": 0.2684221863746643,
"learning_rate": 2.9768958628775902e-05,
"loss": 0.0408,
"step": 19530
},
{
"grad_norm": 0.2557636499404907,
"learning_rate": 2.9718568510408762e-05,
"loss": 0.0363,
"step": 19540
},
{
"grad_norm": 0.2818508446216583,
"learning_rate": 2.9668203035893272e-05,
"loss": 0.0448,
"step": 19550
},
{
"grad_norm": 0.2379492223262787,
"learning_rate": 2.9617862266428287e-05,
"loss": 0.0485,
"step": 19560
},
{
"grad_norm": 0.27195000648498535,
"learning_rate": 2.9567546263182556e-05,
"loss": 0.0452,
"step": 19570
},
{
"grad_norm": 0.2500018775463104,
"learning_rate": 2.951725508729476e-05,
"loss": 0.0381,
"step": 19580
},
{
"grad_norm": 0.24323666095733643,
"learning_rate": 2.946698879987344e-05,
"loss": 0.0369,
"step": 19590
},
{
"grad_norm": 0.1845945566892624,
"learning_rate": 2.9416747461996853e-05,
"loss": 0.0328,
"step": 19600
},
{
"grad_norm": 0.22040122747421265,
"learning_rate": 2.9366531134712973e-05,
"loss": 0.0382,
"step": 19610
},
{
"grad_norm": 0.21812698245048523,
"learning_rate": 2.9316339879039367e-05,
"loss": 0.0468,
"step": 19620
},
{
"grad_norm": 0.2688360810279846,
"learning_rate": 2.9266173755963167e-05,
"loss": 0.0419,
"step": 19630
},
{
"grad_norm": 0.20977988839149475,
"learning_rate": 2.9216032826440926e-05,
"loss": 0.0414,
"step": 19640
},
{
"grad_norm": 0.2208116203546524,
"learning_rate": 2.9165917151398592e-05,
"loss": 0.0369,
"step": 19650
},
{
"grad_norm": 0.1991991102695465,
"learning_rate": 2.9115826791731428e-05,
"loss": 0.0435,
"step": 19660
},
{
"grad_norm": 0.24117377400398254,
"learning_rate": 2.906576180830398e-05,
"loss": 0.039,
"step": 19670
},
{
"grad_norm": 0.23553824424743652,
"learning_rate": 2.9015722261949917e-05,
"loss": 0.0367,
"step": 19680
},
{
"grad_norm": 0.24124255776405334,
"learning_rate": 2.8965708213471986e-05,
"loss": 0.0356,
"step": 19690
},
{
"grad_norm": 0.2099500298500061,
"learning_rate": 2.8915719723641975e-05,
"loss": 0.045,
"step": 19700
},
{
"grad_norm": 0.21426641941070557,
"learning_rate": 2.8865756853200604e-05,
"loss": 0.0367,
"step": 19710
},
{
"grad_norm": 0.17906296253204346,
"learning_rate": 2.8815819662857503e-05,
"loss": 0.0387,
"step": 19720
},
{
"grad_norm": 0.20531953871250153,
"learning_rate": 2.876590821329105e-05,
"loss": 0.0374,
"step": 19730
},
{
"grad_norm": 0.25135716795921326,
"learning_rate": 2.871602256514836e-05,
"loss": 0.0413,
"step": 19740
},
{
"grad_norm": 0.21213361620903015,
"learning_rate": 2.8666162779045203e-05,
"loss": 0.0363,
"step": 19750
},
{
"grad_norm": 0.18084824085235596,
"learning_rate": 2.8616328915565904e-05,
"loss": 0.0421,
"step": 19760
},
{
"grad_norm": 0.28554266691207886,
"learning_rate": 2.856652103526334e-05,
"loss": 0.0412,
"step": 19770
},
{
"grad_norm": 0.18620534241199493,
"learning_rate": 2.8516739198658755e-05,
"loss": 0.0403,
"step": 19780
},
{
"grad_norm": 0.2547351121902466,
"learning_rate": 2.846698346624177e-05,
"loss": 0.042,
"step": 19790
},
{
"grad_norm": 0.19010309875011444,
"learning_rate": 2.841725389847032e-05,
"loss": 0.0366,
"step": 19800
},
{
"grad_norm": 0.1901090294122696,
"learning_rate": 2.8367550555770506e-05,
"loss": 0.0375,
"step": 19810
},
{
"grad_norm": 0.23061218857765198,
"learning_rate": 2.831787349853655e-05,
"loss": 0.0426,
"step": 19820
},
{
"grad_norm": 0.20305535197257996,
"learning_rate": 2.8268222787130806e-05,
"loss": 0.0376,
"step": 19830
},
{
"grad_norm": 0.19640764594078064,
"learning_rate": 2.821859848188355e-05,
"loss": 0.0424,
"step": 19840
},
{
"grad_norm": 0.2568477690219879,
"learning_rate": 2.816900064309299e-05,
"loss": 0.0427,
"step": 19850
},
{
"grad_norm": 0.29017311334609985,
"learning_rate": 2.811942933102517e-05,
"loss": 0.0478,
"step": 19860
},
{
"grad_norm": 0.20586654543876648,
"learning_rate": 2.806988460591391e-05,
"loss": 0.0437,
"step": 19870
},
{
"grad_norm": 0.1892215609550476,
"learning_rate": 2.802036652796074e-05,
"loss": 0.0378,
"step": 19880
},
{
"grad_norm": 0.16674864292144775,
"learning_rate": 2.797087515733478e-05,
"loss": 0.0327,
"step": 19890
},
{
"grad_norm": 0.23387303948402405,
"learning_rate": 2.7921410554172723e-05,
"loss": 0.0393,
"step": 19900
},
{
"grad_norm": 0.2262447625398636,
"learning_rate": 2.787197277857871e-05,
"loss": 0.0316,
"step": 19910
},
{
"grad_norm": 0.2698863744735718,
"learning_rate": 2.782256189062429e-05,
"loss": 0.0353,
"step": 19920
},
{
"grad_norm": 0.15189649164676666,
"learning_rate": 2.777317795034839e-05,
"loss": 0.0312,
"step": 19930
},
{
"grad_norm": 0.19718492031097412,
"learning_rate": 2.7723821017757112e-05,
"loss": 0.033,
"step": 19940
},
{
"grad_norm": 0.24574217200279236,
"learning_rate": 2.7674491152823822e-05,
"loss": 0.0389,
"step": 19950
},
{
"grad_norm": 0.1919129490852356,
"learning_rate": 2.7625188415488944e-05,
"loss": 0.0349,
"step": 19960
},
{
"grad_norm": 0.23063267767429352,
"learning_rate": 2.7575912865659924e-05,
"loss": 0.0386,
"step": 19970
},
{
"grad_norm": 0.19078946113586426,
"learning_rate": 2.7526664563211245e-05,
"loss": 0.0311,
"step": 19980
},
{
"grad_norm": 0.24912944436073303,
"learning_rate": 2.7477443567984224e-05,
"loss": 0.0403,
"step": 19990
},
{
"grad_norm": 0.22693023085594177,
"learning_rate": 2.7428249939787e-05,
"loss": 0.0322,
"step": 20000
},
{
"grad_norm": 0.21841545403003693,
"learning_rate": 2.7379083738394484e-05,
"loss": 0.032,
"step": 20010
},
{
"grad_norm": 0.20512725412845612,
"learning_rate": 2.732994502354823e-05,
"loss": 0.0357,
"step": 20020
},
{
"grad_norm": 0.2208089977502823,
"learning_rate": 2.72808338549564e-05,
"loss": 0.0365,
"step": 20030
},
{
"grad_norm": 0.25611817836761475,
"learning_rate": 2.723175029229374e-05,
"loss": 0.037,
"step": 20040
},
{
"grad_norm": 0.20526891946792603,
"learning_rate": 2.718269439520138e-05,
"loss": 0.0344,
"step": 20050
},
{
"grad_norm": 0.23398402333259583,
"learning_rate": 2.7133666223286858e-05,
"loss": 0.0356,
"step": 20060
},
{
"grad_norm": 0.25886890292167664,
"learning_rate": 2.7084665836124006e-05,
"loss": 0.0363,
"step": 20070
},
{
"grad_norm": 0.185447558760643,
"learning_rate": 2.703569329325296e-05,
"loss": 0.0341,
"step": 20080
},
{
"grad_norm": 0.23907871544361115,
"learning_rate": 2.698674865417994e-05,
"loss": 0.0374,
"step": 20090
},
{
"grad_norm": 0.1773717999458313,
"learning_rate": 2.6937831978377325e-05,
"loss": 0.0367,
"step": 20100
},
{
"grad_norm": 0.21794377267360687,
"learning_rate": 2.6888943325283485e-05,
"loss": 0.0327,
"step": 20110
},
{
"grad_norm": 0.2315199226140976,
"learning_rate": 2.6840082754302733e-05,
"loss": 0.0342,
"step": 20120
},
{
"grad_norm": 0.23786017298698425,
"learning_rate": 2.679125032480525e-05,
"loss": 0.0357,
"step": 20130
},
{
"grad_norm": 0.2549615204334259,
"learning_rate": 2.674244609612708e-05,
"loss": 0.0401,
"step": 20140
},
{
"grad_norm": 0.21104735136032104,
"learning_rate": 2.6693670127569958e-05,
"loss": 0.0363,
"step": 20150
},
{
"grad_norm": 0.23846204578876495,
"learning_rate": 2.6644922478401268e-05,
"loss": 0.0376,
"step": 20160
},
{
"grad_norm": 0.23106716573238373,
"learning_rate": 2.6596203207854004e-05,
"loss": 0.0322,
"step": 20170
},
{
"grad_norm": 0.26020514965057373,
"learning_rate": 2.654751237512666e-05,
"loss": 0.0339,
"step": 20180
},
{
"grad_norm": 0.29553329944610596,
"learning_rate": 2.649885003938323e-05,
"loss": 0.0318,
"step": 20190
},
{
"grad_norm": 0.2088714838027954,
"learning_rate": 2.6450216259753e-05,
"loss": 0.036,
"step": 20200
},
{
"grad_norm": 0.21375878155231476,
"learning_rate": 2.640161109533063e-05,
"loss": 0.0352,
"step": 20210
},
{
"grad_norm": 0.21708621084690094,
"learning_rate": 2.6353034605175936e-05,
"loss": 0.0365,
"step": 20220
},
{
"grad_norm": 0.2764700651168823,
"learning_rate": 2.630448684831398e-05,
"loss": 0.0396,
"step": 20230
},
{
"grad_norm": 0.29964324831962585,
"learning_rate": 2.625596788373482e-05,
"loss": 0.0389,
"step": 20240
},
{
"grad_norm": 0.26454266905784607,
"learning_rate": 2.620747777039363e-05,
"loss": 0.0366,
"step": 20250
},
{
"grad_norm": 0.24117180705070496,
"learning_rate": 2.6159016567210424e-05,
"loss": 0.0429,
"step": 20260
},
{
"grad_norm": 0.3040686845779419,
"learning_rate": 2.6110584333070153e-05,
"loss": 0.037,
"step": 20270
},
{
"grad_norm": 0.24788342416286469,
"learning_rate": 2.6062181126822537e-05,
"loss": 0.0377,
"step": 20280
},
{
"grad_norm": 0.18469078838825226,
"learning_rate": 2.601380700728203e-05,
"loss": 0.0348,
"step": 20290
},
{
"grad_norm": 0.22909514605998993,
"learning_rate": 2.596546203322777e-05,
"loss": 0.0347,
"step": 20300
},
{
"grad_norm": 0.22121627628803253,
"learning_rate": 2.5917146263403457e-05,
"loss": 0.0347,
"step": 20310
},
{
"grad_norm": 0.23250320553779602,
"learning_rate": 2.5868859756517296e-05,
"loss": 0.04,
"step": 20320
},
{
"grad_norm": 0.26366162300109863,
"learning_rate": 2.5820602571241947e-05,
"loss": 0.0387,
"step": 20330
},
{
"grad_norm": 0.26516279578208923,
"learning_rate": 2.577237476621442e-05,
"loss": 0.0437,
"step": 20340
},
{
"grad_norm": 0.18215222656726837,
"learning_rate": 2.572417640003609e-05,
"loss": 0.0368,
"step": 20350
},
{
"grad_norm": 0.19739647209644318,
"learning_rate": 2.5676007531272473e-05,
"loss": 0.0427,
"step": 20360
},
{
"grad_norm": 0.19717362523078918,
"learning_rate": 2.5627868218453332e-05,
"loss": 0.0389,
"step": 20370
},
{
"grad_norm": 0.18554255366325378,
"learning_rate": 2.5579758520072443e-05,
"loss": 0.0377,
"step": 20380
},
{
"grad_norm": 0.2145162969827652,
"learning_rate": 2.5531678494587615e-05,
"loss": 0.0399,
"step": 20390
},
{
"grad_norm": 0.24747347831726074,
"learning_rate": 2.5483628200420644e-05,
"loss": 0.0394,
"step": 20400
},
{
"grad_norm": 0.19843804836273193,
"learning_rate": 2.5435607695957154e-05,
"loss": 0.0422,
"step": 20410
},
{
"grad_norm": 0.25495854020118713,
"learning_rate": 2.5387617039546586e-05,
"loss": 0.0377,
"step": 20420
},
{
"grad_norm": 0.1701011061668396,
"learning_rate": 2.5339656289502106e-05,
"loss": 0.0371,
"step": 20430
},
{
"grad_norm": 0.23653551936149597,
"learning_rate": 2.529172550410056e-05,
"loss": 0.0426,
"step": 20440
},
{
"grad_norm": 0.1779911369085312,
"learning_rate": 2.524382474158234e-05,
"loss": 0.0382,
"step": 20450
},
{
"grad_norm": 0.22654680907726288,
"learning_rate": 2.5195954060151433e-05,
"loss": 0.0353,
"step": 20460
},
{
"grad_norm": 0.19598910212516785,
"learning_rate": 2.5148113517975212e-05,
"loss": 0.0399,
"step": 20470
},
{
"grad_norm": 0.15057498216629028,
"learning_rate": 2.5100303173184447e-05,
"loss": 0.0322,
"step": 20480
},
{
"grad_norm": 0.18491095304489136,
"learning_rate": 2.5052523083873196e-05,
"loss": 0.0349,
"step": 20490
},
{
"grad_norm": 0.2452964335680008,
"learning_rate": 2.500477330809881e-05,
"loss": 0.0359,
"step": 20500
},
{
"grad_norm": 0.2063591629266739,
"learning_rate": 2.4957053903881733e-05,
"loss": 0.04,
"step": 20510
},
{
"grad_norm": 0.2718079686164856,
"learning_rate": 2.4909364929205576e-05,
"loss": 0.0351,
"step": 20520
},
{
"grad_norm": 0.21857582032680511,
"learning_rate": 2.4861706442016923e-05,
"loss": 0.0371,
"step": 20530
},
{
"grad_norm": 0.27503538131713867,
"learning_rate": 2.481407850022533e-05,
"loss": 0.042,
"step": 20540
},
{
"grad_norm": 0.22487691044807434,
"learning_rate": 2.4766481161703213e-05,
"loss": 0.0359,
"step": 20550
},
{
"grad_norm": 0.2401731014251709,
"learning_rate": 2.4718914484285876e-05,
"loss": 0.0378,
"step": 20560
},
{
"grad_norm": 0.25331974029541016,
"learning_rate": 2.467137852577129e-05,
"loss": 0.0404,
"step": 20570
},
{
"grad_norm": 0.2122095674276352,
"learning_rate": 2.4623873343920123e-05,
"loss": 0.039,
"step": 20580
},
{
"grad_norm": 0.23126626014709473,
"learning_rate": 2.4576398996455656e-05,
"loss": 0.0371,
"step": 20590
},
{
"grad_norm": 0.23790284991264343,
"learning_rate": 2.452895554106368e-05,
"loss": 0.0387,
"step": 20600
},
{
"grad_norm": 0.2141856551170349,
"learning_rate": 2.4481543035392506e-05,
"loss": 0.0389,
"step": 20610
},
{
"grad_norm": 0.2521611154079437,
"learning_rate": 2.4434161537052774e-05,
"loss": 0.0356,
"step": 20620
},
{
"grad_norm": 0.18922841548919678,
"learning_rate": 2.4386811103617474e-05,
"loss": 0.0333,
"step": 20630
},
{
"grad_norm": 0.27379629015922546,
"learning_rate": 2.4339491792621833e-05,
"loss": 0.0342,
"step": 20640
},
{
"grad_norm": 0.17814895510673523,
"learning_rate": 2.4292203661563312e-05,
"loss": 0.0358,
"step": 20650
},
{
"grad_norm": 0.2952808141708374,
"learning_rate": 2.424494676790141e-05,
"loss": 0.036,
"step": 20660
},
{
"grad_norm": 0.19882376492023468,
"learning_rate": 2.4197721169057745e-05,
"loss": 0.0313,
"step": 20670
},
{
"grad_norm": 0.2552473247051239,
"learning_rate": 2.4150526922415857e-05,
"loss": 0.0347,
"step": 20680
},
{
"grad_norm": 0.22291389107704163,
"learning_rate": 2.41033640853212e-05,
"loss": 0.0376,
"step": 20690
},
{
"grad_norm": 0.25662899017333984,
"learning_rate": 2.405623271508108e-05,
"loss": 0.0417,
"step": 20700
},
{
"grad_norm": 0.2582562565803528,
"learning_rate": 2.4009132868964522e-05,
"loss": 0.0414,
"step": 20710
},
{
"grad_norm": 0.23722343146800995,
"learning_rate": 2.3962064604202327e-05,
"loss": 0.0361,
"step": 20720
},
{
"grad_norm": 0.22454728186130524,
"learning_rate": 2.3915027977986858e-05,
"loss": 0.0427,
"step": 20730
},
{
"grad_norm": 0.21354511380195618,
"learning_rate": 2.386802304747205e-05,
"loss": 0.0398,
"step": 20740
},
{
"grad_norm": 0.21074089407920837,
"learning_rate": 2.382104986977332e-05,
"loss": 0.0401,
"step": 20750
},
{
"grad_norm": 0.228531613945961,
"learning_rate": 2.377410850196749e-05,
"loss": 0.036,
"step": 20760
},
{
"grad_norm": 0.2090863585472107,
"learning_rate": 2.3727199001092797e-05,
"loss": 0.0401,
"step": 20770
},
{
"grad_norm": 0.2561509907245636,
"learning_rate": 2.3680321424148677e-05,
"loss": 0.0454,
"step": 20780
},
{
"grad_norm": 0.24620649218559265,
"learning_rate": 2.3633475828095793e-05,
"loss": 0.0403,
"step": 20790
},
{
"grad_norm": 0.19917407631874084,
"learning_rate": 2.358666226985599e-05,
"loss": 0.0393,
"step": 20800
},
{
"grad_norm": 0.2224746197462082,
"learning_rate": 2.353988080631213e-05,
"loss": 0.0359,
"step": 20810
},
{
"grad_norm": 0.17652179300785065,
"learning_rate": 2.3493131494308142e-05,
"loss": 0.0364,
"step": 20820
},
{
"grad_norm": 0.2611381411552429,
"learning_rate": 2.344641439064881e-05,
"loss": 0.0374,
"step": 20830
},
{
"grad_norm": 0.19176985323429108,
"learning_rate": 2.339972955209984e-05,
"loss": 0.0354,
"step": 20840
},
{
"grad_norm": 0.21573065221309662,
"learning_rate": 2.335307703538771e-05,
"loss": 0.0418,
"step": 20850
},
{
"grad_norm": 0.18042756617069244,
"learning_rate": 2.330645689719962e-05,
"loss": 0.0344,
"step": 20860
},
{
"grad_norm": 0.1971328854560852,
"learning_rate": 2.3259869194183414e-05,
"loss": 0.0352,
"step": 20870
},
{
"grad_norm": 0.23652096092700958,
"learning_rate": 2.321331398294759e-05,
"loss": 0.0382,
"step": 20880
},
{
"grad_norm": 0.21927939355373383,
"learning_rate": 2.3166791320061094e-05,
"loss": 0.0403,
"step": 20890
},
{
"grad_norm": 0.19436945021152496,
"learning_rate": 2.312030126205335e-05,
"loss": 0.0357,
"step": 20900
},
{
"grad_norm": 0.18774205446243286,
"learning_rate": 2.3073843865414162e-05,
"loss": 0.0362,
"step": 20910
},
{
"grad_norm": 0.18360066413879395,
"learning_rate": 2.302741918659363e-05,
"loss": 0.0401,
"step": 20920
},
{
"grad_norm": 0.321674644947052,
"learning_rate": 2.2981027282002154e-05,
"loss": 0.0443,
"step": 20930
},
{
"grad_norm": 0.17846287786960602,
"learning_rate": 2.2934668208010236e-05,
"loss": 0.0395,
"step": 20940
},
{
"grad_norm": 0.2013472616672516,
"learning_rate": 2.2888342020948555e-05,
"loss": 0.038,
"step": 20950
},
{
"grad_norm": 0.22748152911663055,
"learning_rate": 2.2842048777107783e-05,
"loss": 0.0381,
"step": 20960
},
{
"grad_norm": 0.2502395510673523,
"learning_rate": 2.2795788532738553e-05,
"loss": 0.041,
"step": 20970
},
{
"grad_norm": 0.18183571100234985,
"learning_rate": 2.2749561344051468e-05,
"loss": 0.0403,
"step": 20980
},
{
"grad_norm": 0.21390889585018158,
"learning_rate": 2.2703367267216896e-05,
"loss": 0.0427,
"step": 20990
},
{
"grad_norm": 0.3070128262042999,
"learning_rate": 2.2657206358364997e-05,
"loss": 0.0459,
"step": 21000
},
{
"grad_norm": 0.21548451483249664,
"learning_rate": 2.261107867358563e-05,
"loss": 0.0383,
"step": 21010
},
{
"grad_norm": 0.20574261248111725,
"learning_rate": 2.2564984268928263e-05,
"loss": 0.0371,
"step": 21020
},
{
"grad_norm": 0.20345918834209442,
"learning_rate": 2.251892320040198e-05,
"loss": 0.0389,
"step": 21030
},
{
"grad_norm": 0.20436224341392517,
"learning_rate": 2.2472895523975312e-05,
"loss": 0.0372,
"step": 21040
},
{
"grad_norm": 0.24329958856105804,
"learning_rate": 2.2426901295576212e-05,
"loss": 0.0406,
"step": 21050
},
{
"grad_norm": 0.22425851225852966,
"learning_rate": 2.2380940571092013e-05,
"loss": 0.0421,
"step": 21060
},
{
"grad_norm": 0.18522213399410248,
"learning_rate": 2.2335013406369302e-05,
"loss": 0.0374,
"step": 21070
},
{
"grad_norm": 0.2295093983411789,
"learning_rate": 2.228911985721397e-05,
"loss": 0.0451,
"step": 21080
},
{
"grad_norm": 0.23965975642204285,
"learning_rate": 2.224325997939095e-05,
"loss": 0.0408,
"step": 21090
},
{
"grad_norm": 0.23921312391757965,
"learning_rate": 2.219743382862437e-05,
"loss": 0.042,
"step": 21100
},
{
"grad_norm": 0.24748393893241882,
"learning_rate": 2.2151641460597294e-05,
"loss": 0.038,
"step": 21110
},
{
"grad_norm": 0.2995326817035675,
"learning_rate": 2.210588293095177e-05,
"loss": 0.0502,
"step": 21120
},
{
"grad_norm": 0.20778805017471313,
"learning_rate": 2.2060158295288714e-05,
"loss": 0.0441,
"step": 21130
},
{
"grad_norm": 0.21855351328849792,
"learning_rate": 2.2014467609167905e-05,
"loss": 0.0446,
"step": 21140
},
{
"grad_norm": 0.26528626680374146,
"learning_rate": 2.1968810928107806e-05,
"loss": 0.0462,
"step": 21150
},
{
"grad_norm": 0.2655385732650757,
"learning_rate": 2.1923188307585606e-05,
"loss": 0.0443,
"step": 21160
},
{
"grad_norm": 0.206245556473732,
"learning_rate": 2.187759980303708e-05,
"loss": 0.042,
"step": 21170
},
{
"grad_norm": 0.23579102754592896,
"learning_rate": 2.1832045469856545e-05,
"loss": 0.0434,
"step": 21180
},
{
"grad_norm": 0.21801422536373138,
"learning_rate": 2.1786525363396844e-05,
"loss": 0.0404,
"step": 21190
},
{
"grad_norm": 0.29967206716537476,
"learning_rate": 2.1741039538969183e-05,
"loss": 0.0431,
"step": 21200
},
{
"grad_norm": 0.20192112028598785,
"learning_rate": 2.169558805184313e-05,
"loss": 0.0396,
"step": 21210
},
{
"grad_norm": 0.282464861869812,
"learning_rate": 2.165017095724651e-05,
"loss": 0.0433,
"step": 21220
},
{
"grad_norm": 0.18897965550422668,
"learning_rate": 2.1604788310365405e-05,
"loss": 0.0425,
"step": 21230
},
{
"grad_norm": 0.2030584216117859,
"learning_rate": 2.155944016634398e-05,
"loss": 0.04,
"step": 21240
},
{
"grad_norm": 0.19546188414096832,
"learning_rate": 2.1514126580284538e-05,
"loss": 0.0385,
"step": 21250
},
{
"grad_norm": 0.1942000687122345,
"learning_rate": 2.1468847607247344e-05,
"loss": 0.0384,
"step": 21260
},
{
"grad_norm": 0.20162087678909302,
"learning_rate": 2.1423603302250623e-05,
"loss": 0.0388,
"step": 21270
},
{
"grad_norm": 0.19028404355049133,
"learning_rate": 2.1378393720270468e-05,
"loss": 0.0403,
"step": 21280
},
{
"grad_norm": 0.23134730756282806,
"learning_rate": 2.133321891624076e-05,
"loss": 0.0369,
"step": 21290
},
{
"grad_norm": 0.2570262849330902,
"learning_rate": 2.1288078945053193e-05,
"loss": 0.0377,
"step": 21300
},
{
"grad_norm": 0.19804701209068298,
"learning_rate": 2.1242973861557064e-05,
"loss": 0.0323,
"step": 21310
},
{
"grad_norm": 0.22698380053043365,
"learning_rate": 2.1197903720559304e-05,
"loss": 0.0357,
"step": 21320
},
{
"grad_norm": 0.284479558467865,
"learning_rate": 2.115286857682438e-05,
"loss": 0.0345,
"step": 21330
},
{
"grad_norm": 0.20733642578125,
"learning_rate": 2.1107868485074228e-05,
"loss": 0.0356,
"step": 21340
},
{
"grad_norm": 0.2057272493839264,
"learning_rate": 2.1062903499988234e-05,
"loss": 0.0366,
"step": 21350
},
{
"grad_norm": 0.24864676594734192,
"learning_rate": 2.101797367620308e-05,
"loss": 0.0397,
"step": 21360
},
{
"grad_norm": 0.20604996383190155,
"learning_rate": 2.0973079068312713e-05,
"loss": 0.0346,
"step": 21370
},
{
"grad_norm": 0.28327465057373047,
"learning_rate": 2.0928219730868358e-05,
"loss": 0.0338,
"step": 21380
},
{
"grad_norm": 0.20866821706295013,
"learning_rate": 2.0883395718378302e-05,
"loss": 0.0313,
"step": 21390
},
{
"grad_norm": 0.16616901755332947,
"learning_rate": 2.083860708530798e-05,
"loss": 0.0335,
"step": 21400
},
{
"grad_norm": 0.22390931844711304,
"learning_rate": 2.0793853886079796e-05,
"loss": 0.0355,
"step": 21410
},
{
"grad_norm": 0.148369699716568,
"learning_rate": 2.074913617507309e-05,
"loss": 0.0359,
"step": 21420
},
{
"grad_norm": 0.18207670748233795,
"learning_rate": 2.070445400662411e-05,
"loss": 0.0421,
"step": 21430
},
{
"grad_norm": 0.19708731770515442,
"learning_rate": 2.0659807435025907e-05,
"loss": 0.0493,
"step": 21440
},
{
"grad_norm": 0.18977856636047363,
"learning_rate": 2.061519651452825e-05,
"loss": 0.0444,
"step": 21450
},
{
"grad_norm": 0.23934000730514526,
"learning_rate": 2.0570621299337656e-05,
"loss": 0.0416,
"step": 21460
},
{
"grad_norm": 0.22606970369815826,
"learning_rate": 2.0526081843617183e-05,
"loss": 0.0436,
"step": 21470
},
{
"grad_norm": 0.22899135947227478,
"learning_rate": 2.0481578201486485e-05,
"loss": 0.0463,
"step": 21480
},
{
"grad_norm": 0.1853393018245697,
"learning_rate": 2.0437110427021677e-05,
"loss": 0.0413,
"step": 21490
},
{
"grad_norm": 0.2045687884092331,
"learning_rate": 2.0392678574255282e-05,
"loss": 0.0413,
"step": 21500
},
{
"grad_norm": 0.23020720481872559,
"learning_rate": 2.0348282697176217e-05,
"loss": 0.0402,
"step": 21510
},
{
"grad_norm": 0.15244992077350616,
"learning_rate": 2.030392284972964e-05,
"loss": 0.0421,
"step": 21520
},
{
"grad_norm": 0.20619356632232666,
"learning_rate": 2.0259599085816973e-05,
"loss": 0.0379,
"step": 21530
},
{
"grad_norm": 0.23374305665493011,
"learning_rate": 2.0215311459295755e-05,
"loss": 0.0368,
"step": 21540
},
{
"grad_norm": 0.2343476414680481,
"learning_rate": 2.0171060023979606e-05,
"loss": 0.0329,
"step": 21550
},
{
"grad_norm": 0.173087939620018,
"learning_rate": 2.012684483363823e-05,
"loss": 0.0368,
"step": 21560
},
{
"grad_norm": 0.23633894324302673,
"learning_rate": 2.0082665941997237e-05,
"loss": 0.0389,
"step": 21570
},
{
"grad_norm": 0.18118853867053986,
"learning_rate": 2.0038523402738146e-05,
"loss": 0.0402,
"step": 21580
},
{
"grad_norm": 0.20321035385131836,
"learning_rate": 1.99944172694983e-05,
"loss": 0.046,
"step": 21590
},
{
"grad_norm": 0.20035482943058014,
"learning_rate": 1.99503475958708e-05,
"loss": 0.0381,
"step": 21600
},
{
"grad_norm": 0.19728951156139374,
"learning_rate": 1.9906314435404484e-05,
"loss": 0.0451,
"step": 21610
},
{
"grad_norm": 0.22929500043392181,
"learning_rate": 1.986231784160378e-05,
"loss": 0.0391,
"step": 21620
},
{
"grad_norm": 0.19313189387321472,
"learning_rate": 1.9818357867928693e-05,
"loss": 0.0373,
"step": 21630
},
{
"grad_norm": 0.22061873972415924,
"learning_rate": 1.977443456779474e-05,
"loss": 0.0395,
"step": 21640
},
{
"grad_norm": 0.2101905792951584,
"learning_rate": 1.9730547994572858e-05,
"loss": 0.0376,
"step": 21650
},
{
"grad_norm": 0.21324148774147034,
"learning_rate": 1.9686698201589393e-05,
"loss": 0.0387,
"step": 21660
},
{
"grad_norm": 0.25892964005470276,
"learning_rate": 1.9642885242125964e-05,
"loss": 0.0424,
"step": 21670
},
{
"grad_norm": 0.21896907687187195,
"learning_rate": 1.9599109169419466e-05,
"loss": 0.0397,
"step": 21680
},
{
"grad_norm": 0.19612598419189453,
"learning_rate": 1.9555370036661948e-05,
"loss": 0.0435,
"step": 21690
},
{
"grad_norm": 0.21141347289085388,
"learning_rate": 1.9511667897000574e-05,
"loss": 0.0395,
"step": 21700
},
{
"grad_norm": 0.20380882918834686,
"learning_rate": 1.946800280353755e-05,
"loss": 0.0403,
"step": 21710
},
{
"grad_norm": 0.23413388431072235,
"learning_rate": 1.9424374809330114e-05,
"loss": 0.0401,
"step": 21720
},
{
"grad_norm": 0.2564195394515991,
"learning_rate": 1.9380783967390382e-05,
"loss": 0.0462,
"step": 21730
},
{
"grad_norm": 0.18020562827587128,
"learning_rate": 1.933723033068533e-05,
"loss": 0.0361,
"step": 21740
},
{
"grad_norm": 0.26402461528778076,
"learning_rate": 1.929371395213674e-05,
"loss": 0.0444,
"step": 21750
},
{
"grad_norm": 0.22266747057437897,
"learning_rate": 1.9250234884621092e-05,
"loss": 0.0354,
"step": 21760
},
{
"grad_norm": 0.17776279151439667,
"learning_rate": 1.920679318096959e-05,
"loss": 0.0345,
"step": 21770
},
{
"grad_norm": 0.21332547068595886,
"learning_rate": 1.9163388893967982e-05,
"loss": 0.0374,
"step": 21780
},
{
"grad_norm": 0.21355228126049042,
"learning_rate": 1.9120022076356575e-05,
"loss": 0.0408,
"step": 21790
},
{
"grad_norm": 0.2276710867881775,
"learning_rate": 1.9076692780830114e-05,
"loss": 0.0348,
"step": 21800
},
{
"grad_norm": 0.2492753118276596,
"learning_rate": 1.903340106003782e-05,
"loss": 0.0368,
"step": 21810
},
{
"grad_norm": 0.18199560046195984,
"learning_rate": 1.8990146966583183e-05,
"loss": 0.0356,
"step": 21820
},
{
"grad_norm": 0.1957540512084961,
"learning_rate": 1.8946930553024035e-05,
"loss": 0.0393,
"step": 21830
},
{
"grad_norm": 0.17549556493759155,
"learning_rate": 1.8903751871872378e-05,
"loss": 0.0403,
"step": 21840
},
{
"grad_norm": 0.3051057457923889,
"learning_rate": 1.8860610975594382e-05,
"loss": 0.039,
"step": 21850
},
{
"grad_norm": 0.21554066240787506,
"learning_rate": 1.8817507916610307e-05,
"loss": 0.0338,
"step": 21860
},
{
"grad_norm": 0.2683010995388031,
"learning_rate": 1.8774442747294407e-05,
"loss": 0.0414,
"step": 21870
},
{
"grad_norm": 0.20552904903888702,
"learning_rate": 1.8731415519974965e-05,
"loss": 0.0378,
"step": 21880
},
{
"grad_norm": 0.2367197424173355,
"learning_rate": 1.86884262869341e-05,
"loss": 0.0345,
"step": 21890
},
{
"grad_norm": 0.20127518475055695,
"learning_rate": 1.8645475100407788e-05,
"loss": 0.0365,
"step": 21900
},
{
"grad_norm": 0.20096677541732788,
"learning_rate": 1.8602562012585768e-05,
"loss": 0.0412,
"step": 21910
},
{
"grad_norm": 0.21187731623649597,
"learning_rate": 1.8559687075611466e-05,
"loss": 0.0363,
"step": 21920
},
{
"grad_norm": 0.17884133756160736,
"learning_rate": 1.8516850341582017e-05,
"loss": 0.0393,
"step": 21930
},
{
"grad_norm": 0.23608508706092834,
"learning_rate": 1.847405186254807e-05,
"loss": 0.0412,
"step": 21940
},
{
"grad_norm": 0.1819131225347519,
"learning_rate": 1.843129169051379e-05,
"loss": 0.0354,
"step": 21950
},
{
"grad_norm": 0.18228501081466675,
"learning_rate": 1.8388569877436863e-05,
"loss": 0.0368,
"step": 21960
},
{
"grad_norm": 0.18096892535686493,
"learning_rate": 1.834588647522828e-05,
"loss": 0.032,
"step": 21970
},
{
"grad_norm": 0.19546526670455933,
"learning_rate": 1.8303241535752435e-05,
"loss": 0.0452,
"step": 21980
},
{
"grad_norm": 0.16101308166980743,
"learning_rate": 1.8260635110826934e-05,
"loss": 0.0385,
"step": 21990
},
{
"grad_norm": 0.25127822160720825,
"learning_rate": 1.8218067252222597e-05,
"loss": 0.0337,
"step": 22000
},
{
"grad_norm": 0.1664080023765564,
"learning_rate": 1.817553801166339e-05,
"loss": 0.039,
"step": 22010
},
{
"grad_norm": 0.19500654935836792,
"learning_rate": 1.8133047440826333e-05,
"loss": 0.0359,
"step": 22020
},
{
"grad_norm": 0.2295587658882141,
"learning_rate": 1.809059559134151e-05,
"loss": 0.0379,
"step": 22030
},
{
"grad_norm": 0.2246081829071045,
"learning_rate": 1.80481825147919e-05,
"loss": 0.0408,
"step": 22040
},
{
"grad_norm": 0.23786970973014832,
"learning_rate": 1.8005808262713396e-05,
"loss": 0.0391,
"step": 22050
},
{
"grad_norm": 0.18641743063926697,
"learning_rate": 1.7963472886594714e-05,
"loss": 0.0341,
"step": 22060
},
{
"grad_norm": 0.24793487787246704,
"learning_rate": 1.7921176437877302e-05,
"loss": 0.0365,
"step": 22070
},
{
"grad_norm": 0.19699017703533173,
"learning_rate": 1.7878918967955366e-05,
"loss": 0.0351,
"step": 22080
},
{
"grad_norm": 0.2239817976951599,
"learning_rate": 1.7836700528175693e-05,
"loss": 0.0359,
"step": 22090
},
{
"grad_norm": 0.2340039312839508,
"learning_rate": 1.779452116983769e-05,
"loss": 0.0334,
"step": 22100
},
{
"grad_norm": 0.23447147011756897,
"learning_rate": 1.775238094419325e-05,
"loss": 0.0406,
"step": 22110
},
{
"grad_norm": 0.19273117184638977,
"learning_rate": 1.7710279902446708e-05,
"loss": 0.0355,
"step": 22120
},
{
"grad_norm": 0.2749791741371155,
"learning_rate": 1.7668218095754795e-05,
"loss": 0.0384,
"step": 22130
},
{
"grad_norm": 0.2161046415567398,
"learning_rate": 1.7626195575226594e-05,
"loss": 0.0404,
"step": 22140
},
{
"grad_norm": 0.2150353640317917,
"learning_rate": 1.758421239192343e-05,
"loss": 0.0393,
"step": 22150
},
{
"grad_norm": 0.2026449292898178,
"learning_rate": 1.7542268596858814e-05,
"loss": 0.0425,
"step": 22160
},
{
"grad_norm": 0.23117214441299438,
"learning_rate": 1.750036424099841e-05,
"loss": 0.0359,
"step": 22170
},
{
"grad_norm": 0.18065892159938812,
"learning_rate": 1.7458499375259956e-05,
"loss": 0.0386,
"step": 22180
},
{
"grad_norm": 0.25054895877838135,
"learning_rate": 1.7416674050513245e-05,
"loss": 0.038,
"step": 22190
},
{
"grad_norm": 0.19106796383857727,
"learning_rate": 1.737488831757997e-05,
"loss": 0.04,
"step": 22200
},
{
"grad_norm": 0.16423672437667847,
"learning_rate": 1.733314222723373e-05,
"loss": 0.0339,
"step": 22210
},
{
"grad_norm": 0.25283917784690857,
"learning_rate": 1.7291435830199954e-05,
"loss": 0.036,
"step": 22220
},
{
"grad_norm": 0.19551143050193787,
"learning_rate": 1.7249769177155878e-05,
"loss": 0.0357,
"step": 22230
},
{
"grad_norm": 0.3064842224121094,
"learning_rate": 1.720814231873038e-05,
"loss": 0.0389,
"step": 22240
},
{
"grad_norm": 0.18697325885295868,
"learning_rate": 1.716655530550405e-05,
"loss": 0.0368,
"step": 22250
},
{
"grad_norm": 0.2274559885263443,
"learning_rate": 1.7125008188009017e-05,
"loss": 0.0363,
"step": 22260
},
{
"grad_norm": 0.22496716678142548,
"learning_rate": 1.708350101672894e-05,
"loss": 0.0363,
"step": 22270
},
{
"grad_norm": 0.21342577040195465,
"learning_rate": 1.7042033842098957e-05,
"loss": 0.0341,
"step": 22280
},
{
"grad_norm": 0.23419596254825592,
"learning_rate": 1.7000606714505567e-05,
"loss": 0.0318,
"step": 22290
},
{
"grad_norm": 0.21617811918258667,
"learning_rate": 1.695921968428668e-05,
"loss": 0.0291,
"step": 22300
},
{
"grad_norm": 0.2533870339393616,
"learning_rate": 1.6917872801731416e-05,
"loss": 0.0343,
"step": 22310
},
{
"grad_norm": 0.2820796072483063,
"learning_rate": 1.687656611708014e-05,
"loss": 0.0352,
"step": 22320
},
{
"grad_norm": 0.22426827251911163,
"learning_rate": 1.6835299680524368e-05,
"loss": 0.0365,
"step": 22330
},
{
"grad_norm": 0.2956727147102356,
"learning_rate": 1.679407354220669e-05,
"loss": 0.0369,
"step": 22340
},
{
"grad_norm": 0.1969355344772339,
"learning_rate": 1.6752887752220793e-05,
"loss": 0.0377,
"step": 22350
},
{
"grad_norm": 0.22599226236343384,
"learning_rate": 1.6711742360611277e-05,
"loss": 0.0368,
"step": 22360
},
{
"grad_norm": 0.2024817019701004,
"learning_rate": 1.6670637417373652e-05,
"loss": 0.0394,
"step": 22370
},
{
"grad_norm": 0.16458505392074585,
"learning_rate": 1.6629572972454333e-05,
"loss": 0.0376,
"step": 22380
},
{
"grad_norm": 0.21054644882678986,
"learning_rate": 1.6588549075750464e-05,
"loss": 0.0373,
"step": 22390
},
{
"grad_norm": 0.22767816483974457,
"learning_rate": 1.6547565777109976e-05,
"loss": 0.0362,
"step": 22400
},
{
"grad_norm": 0.22495850920677185,
"learning_rate": 1.650662312633143e-05,
"loss": 0.0409,
"step": 22410
},
{
"grad_norm": 0.1712123602628708,
"learning_rate": 1.6465721173164002e-05,
"loss": 0.0339,
"step": 22420
},
{
"grad_norm": 0.24093368649482727,
"learning_rate": 1.6424859967307428e-05,
"loss": 0.0335,
"step": 22430
},
{
"grad_norm": 0.21538522839546204,
"learning_rate": 1.6384039558411903e-05,
"loss": 0.0326,
"step": 22440
},
{
"grad_norm": 0.20245462656021118,
"learning_rate": 1.634325999607811e-05,
"loss": 0.0398,
"step": 22450
},
{
"grad_norm": 0.24620772898197174,
"learning_rate": 1.6302521329857045e-05,
"loss": 0.0387,
"step": 22460
},
{
"grad_norm": 0.18312226235866547,
"learning_rate": 1.6261823609250027e-05,
"loss": 0.0381,
"step": 22470
},
{
"grad_norm": 0.20423147082328796,
"learning_rate": 1.6221166883708627e-05,
"loss": 0.0374,
"step": 22480
},
{
"grad_norm": 0.1582416296005249,
"learning_rate": 1.6180551202634603e-05,
"loss": 0.0367,
"step": 22490
},
{
"grad_norm": 0.1772347390651703,
"learning_rate": 1.613997661537981e-05,
"loss": 0.029,
"step": 22500
},
{
"grad_norm": 0.17890992760658264,
"learning_rate": 1.609944317124624e-05,
"loss": 0.0328,
"step": 22510
},
{
"grad_norm": 0.3024440109729767,
"learning_rate": 1.605895091948582e-05,
"loss": 0.0412,
"step": 22520
},
{
"grad_norm": 0.24228602647781372,
"learning_rate": 1.6018499909300477e-05,
"loss": 0.0367,
"step": 22530
},
{
"grad_norm": 0.17716093361377716,
"learning_rate": 1.597809018984199e-05,
"loss": 0.0349,
"step": 22540
},
{
"grad_norm": 0.18945686519145966,
"learning_rate": 1.593772181021196e-05,
"loss": 0.0303,
"step": 22550
},
{
"grad_norm": 0.1959318369626999,
"learning_rate": 1.5897394819461813e-05,
"loss": 0.0328,
"step": 22560
},
{
"grad_norm": 0.18893995881080627,
"learning_rate": 1.5857109266592625e-05,
"loss": 0.0317,
"step": 22570
},
{
"grad_norm": 0.18034042418003082,
"learning_rate": 1.581686520055514e-05,
"loss": 0.0303,
"step": 22580
},
{
"grad_norm": 0.16626392304897308,
"learning_rate": 1.5776662670249704e-05,
"loss": 0.0315,
"step": 22590
},
{
"grad_norm": 0.20742277801036835,
"learning_rate": 1.573650172452615e-05,
"loss": 0.0329,
"step": 22600
},
{
"grad_norm": 0.42917564511299133,
"learning_rate": 1.5696382412183852e-05,
"loss": 0.037,
"step": 22610
},
{
"grad_norm": 0.18621651828289032,
"learning_rate": 1.565630478197155e-05,
"loss": 0.0346,
"step": 22620
},
{
"grad_norm": 0.24330507218837738,
"learning_rate": 1.5616268882587332e-05,
"loss": 0.031,
"step": 22630
},
{
"grad_norm": 0.23923663794994354,
"learning_rate": 1.5576274762678594e-05,
"loss": 0.0351,
"step": 22640
},
{
"grad_norm": 0.19905850291252136,
"learning_rate": 1.5536322470841952e-05,
"loss": 0.0351,
"step": 22650
},
{
"grad_norm": 0.23712609708309174,
"learning_rate": 1.549641205562324e-05,
"loss": 0.0348,
"step": 22660
},
{
"grad_norm": 0.17332106828689575,
"learning_rate": 1.5456543565517334e-05,
"loss": 0.0309,
"step": 22670
},
{
"grad_norm": 0.19904932379722595,
"learning_rate": 1.541671704896825e-05,
"loss": 0.0356,
"step": 22680
},
{
"grad_norm": 0.26263731718063354,
"learning_rate": 1.5376932554368938e-05,
"loss": 0.036,
"step": 22690
},
{
"grad_norm": 0.24293218553066254,
"learning_rate": 1.5337190130061318e-05,
"loss": 0.036,
"step": 22700
},
{
"grad_norm": 0.20116055011749268,
"learning_rate": 1.5297489824336165e-05,
"loss": 0.0323,
"step": 22710
},
{
"grad_norm": 0.21575824916362762,
"learning_rate": 1.5257831685433121e-05,
"loss": 0.0334,
"step": 22720
},
{
"grad_norm": 0.24044173955917358,
"learning_rate": 1.521821576154055e-05,
"loss": 0.0366,
"step": 22730
},
{
"grad_norm": 0.21021106839179993,
"learning_rate": 1.5178642100795542e-05,
"loss": 0.0333,
"step": 22740
},
{
"grad_norm": 0.23764625191688538,
"learning_rate": 1.5139110751283819e-05,
"loss": 0.0355,
"step": 22750
},
{
"grad_norm": 0.2668387293815613,
"learning_rate": 1.5099621761039683e-05,
"loss": 0.0357,
"step": 22760
},
{
"grad_norm": 0.22924350202083588,
"learning_rate": 1.5060175178046016e-05,
"loss": 0.0305,
"step": 22770
},
{
"grad_norm": 0.18499214947223663,
"learning_rate": 1.5020771050234117e-05,
"loss": 0.033,
"step": 22780
},
{
"grad_norm": 0.2517329156398773,
"learning_rate": 1.4981409425483716e-05,
"loss": 0.033,
"step": 22790
},
{
"grad_norm": 0.2147057056427002,
"learning_rate": 1.4942090351622883e-05,
"loss": 0.0353,
"step": 22800
},
{
"grad_norm": 0.22826455533504486,
"learning_rate": 1.490281387642804e-05,
"loss": 0.0327,
"step": 22810
},
{
"grad_norm": 0.16469426453113556,
"learning_rate": 1.4863580047623765e-05,
"loss": 0.0282,
"step": 22820
},
{
"grad_norm": 0.21418945491313934,
"learning_rate": 1.4824388912882897e-05,
"loss": 0.031,
"step": 22830
},
{
"grad_norm": 0.2257077991962433,
"learning_rate": 1.4785240519826344e-05,
"loss": 0.0315,
"step": 22840
},
{
"grad_norm": 0.1837363839149475,
"learning_rate": 1.4746134916023097e-05,
"loss": 0.0367,
"step": 22850
},
{
"grad_norm": 0.1932159811258316,
"learning_rate": 1.4707072148990141e-05,
"loss": 0.0354,
"step": 22860
},
{
"grad_norm": 0.1930593103170395,
"learning_rate": 1.4668052266192423e-05,
"loss": 0.0334,
"step": 22870
},
{
"grad_norm": 0.18781617283821106,
"learning_rate": 1.4629075315042795e-05,
"loss": 0.0341,
"step": 22880
},
{
"grad_norm": 0.20860779285430908,
"learning_rate": 1.4590141342901925e-05,
"loss": 0.0434,
"step": 22890
},
{
"grad_norm": 0.2166423201560974,
"learning_rate": 1.4551250397078253e-05,
"loss": 0.0409,
"step": 22900
},
{
"grad_norm": 0.15797056257724762,
"learning_rate": 1.4512402524827945e-05,
"loss": 0.0385,
"step": 22910
},
{
"grad_norm": 0.19482962787151337,
"learning_rate": 1.447359777335482e-05,
"loss": 0.0427,
"step": 22920
},
{
"grad_norm": 0.26034629344940186,
"learning_rate": 1.4434836189810335e-05,
"loss": 0.0414,
"step": 22930
},
{
"grad_norm": 0.18671546876430511,
"learning_rate": 1.4396117821293453e-05,
"loss": 0.0361,
"step": 22940
},
{
"grad_norm": 0.2589699327945709,
"learning_rate": 1.4357442714850633e-05,
"loss": 0.0417,
"step": 22950
},
{
"grad_norm": 0.2040921300649643,
"learning_rate": 1.43188109174758e-05,
"loss": 0.0407,
"step": 22960
},
{
"grad_norm": 0.1595921516418457,
"learning_rate": 1.4280222476110206e-05,
"loss": 0.0382,
"step": 22970
},
{
"grad_norm": 0.1747111827135086,
"learning_rate": 1.4241677437642469e-05,
"loss": 0.0453,
"step": 22980
},
{
"grad_norm": 0.27932411432266235,
"learning_rate": 1.420317584890844e-05,
"loss": 0.0439,
"step": 22990
},
{
"grad_norm": 0.23738005757331848,
"learning_rate": 1.4164717756691176e-05,
"loss": 0.0456,
"step": 23000
},
{
"grad_norm": 0.22628755867481232,
"learning_rate": 1.4126303207720882e-05,
"loss": 0.0428,
"step": 23010
},
{
"grad_norm": 0.18822930753231049,
"learning_rate": 1.408793224867484e-05,
"loss": 0.0419,
"step": 23020
},
{
"grad_norm": 0.17955468595027924,
"learning_rate": 1.4049604926177423e-05,
"loss": 0.0447,
"step": 23030
},
{
"grad_norm": 0.23468677699565887,
"learning_rate": 1.4011321286799916e-05,
"loss": 0.0396,
"step": 23040
},
{
"grad_norm": 0.20670485496520996,
"learning_rate": 1.3973081377060565e-05,
"loss": 0.0407,
"step": 23050
},
{
"grad_norm": 0.19423407316207886,
"learning_rate": 1.393488524342445e-05,
"loss": 0.0355,
"step": 23060
},
{
"grad_norm": 0.2609492838382721,
"learning_rate": 1.3896732932303486e-05,
"loss": 0.0441,
"step": 23070
},
{
"grad_norm": 0.2226841300725937,
"learning_rate": 1.3858624490056304e-05,
"loss": 0.0369,
"step": 23080
},
{
"grad_norm": 0.25144368410110474,
"learning_rate": 1.3820559962988299e-05,
"loss": 0.0366,
"step": 23090
},
{
"grad_norm": 0.225273996591568,
"learning_rate": 1.3782539397351418e-05,
"loss": 0.0386,
"step": 23100
},
{
"grad_norm": 0.20086705684661865,
"learning_rate": 1.3744562839344266e-05,
"loss": 0.0378,
"step": 23110
},
{
"grad_norm": 0.18323999643325806,
"learning_rate": 1.370663033511193e-05,
"loss": 0.0348,
"step": 23120
},
{
"grad_norm": 0.23329462110996246,
"learning_rate": 1.3668741930745964e-05,
"loss": 0.0309,
"step": 23130
},
{
"grad_norm": 0.16277958452701569,
"learning_rate": 1.3630897672284382e-05,
"loss": 0.0319,
"step": 23140
},
{
"grad_norm": 0.23894961178302765,
"learning_rate": 1.3593097605711508e-05,
"loss": 0.0373,
"step": 23150
},
{
"grad_norm": 0.1615697294473648,
"learning_rate": 1.3555341776957992e-05,
"loss": 0.0344,
"step": 23160
},
{
"grad_norm": 0.22898811101913452,
"learning_rate": 1.3517630231900724e-05,
"loss": 0.0362,
"step": 23170
},
{
"grad_norm": 0.18154974281787872,
"learning_rate": 1.3479963016362767e-05,
"loss": 0.0376,
"step": 23180
},
{
"grad_norm": 0.27294838428497314,
"learning_rate": 1.3442340176113377e-05,
"loss": 0.0347,
"step": 23190
},
{
"grad_norm": 0.18982717394828796,
"learning_rate": 1.340476175686784e-05,
"loss": 0.0321,
"step": 23200
},
{
"grad_norm": 0.21996819972991943,
"learning_rate": 1.3367227804287469e-05,
"loss": 0.0416,
"step": 23210
},
{
"grad_norm": 0.1882348209619522,
"learning_rate": 1.332973836397956e-05,
"loss": 0.0341,
"step": 23220
},
{
"grad_norm": 0.20760047435760498,
"learning_rate": 1.329229348149731e-05,
"loss": 0.0408,
"step": 23230
},
{
"grad_norm": 0.17178040742874146,
"learning_rate": 1.3254893202339796e-05,
"loss": 0.0351,
"step": 23240
},
{
"grad_norm": 0.22256483137607574,
"learning_rate": 1.3217537571951872e-05,
"loss": 0.046,
"step": 23250
},
{
"grad_norm": 0.22335106134414673,
"learning_rate": 1.3180226635724169e-05,
"loss": 0.0441,
"step": 23260
},
{
"grad_norm": 0.20773710310459137,
"learning_rate": 1.314296043899298e-05,
"loss": 0.0377,
"step": 23270
},
{
"grad_norm": 0.239211767911911,
"learning_rate": 1.3105739027040248e-05,
"loss": 0.0378,
"step": 23280
},
{
"grad_norm": 0.22174808382987976,
"learning_rate": 1.3068562445093491e-05,
"loss": 0.0407,
"step": 23290
},
{
"grad_norm": 0.18679125607013702,
"learning_rate": 1.3031430738325778e-05,
"loss": 0.0378,
"step": 23300
},
{
"grad_norm": 0.20027878880500793,
"learning_rate": 1.299434395185563e-05,
"loss": 0.0402,
"step": 23310
},
{
"grad_norm": 0.2675543427467346,
"learning_rate": 1.295730213074699e-05,
"loss": 0.0385,
"step": 23320
},
{
"grad_norm": 0.17766377329826355,
"learning_rate": 1.2920305320009152e-05,
"loss": 0.0401,
"step": 23330
},
{
"grad_norm": 0.2139299511909485,
"learning_rate": 1.2883353564596729e-05,
"loss": 0.0347,
"step": 23340
},
{
"grad_norm": 0.24178346991539001,
"learning_rate": 1.2846446909409604e-05,
"loss": 0.0415,
"step": 23350
},
{
"grad_norm": 0.19027981162071228,
"learning_rate": 1.280958539929284e-05,
"loss": 0.0349,
"step": 23360
},
{
"grad_norm": 0.21986611187458038,
"learning_rate": 1.277276907903664e-05,
"loss": 0.038,
"step": 23370
},
{
"grad_norm": 0.2592710256576538,
"learning_rate": 1.2735997993376298e-05,
"loss": 0.0354,
"step": 23380
},
{
"grad_norm": 0.16752168536186218,
"learning_rate": 1.2699272186992166e-05,
"loss": 0.0345,
"step": 23390
},
{
"grad_norm": 0.25470098853111267,
"learning_rate": 1.2662591704509547e-05,
"loss": 0.0362,
"step": 23400
},
{
"grad_norm": 0.17857319116592407,
"learning_rate": 1.2625956590498711e-05,
"loss": 0.035,
"step": 23410
},
{
"grad_norm": 0.2320038080215454,
"learning_rate": 1.2589366889474757e-05,
"loss": 0.0397,
"step": 23420
},
{
"grad_norm": 0.20005041360855103,
"learning_rate": 1.2552822645897621e-05,
"loss": 0.0427,
"step": 23430
},
{
"grad_norm": 0.2114519625902176,
"learning_rate": 1.2516323904172e-05,
"loss": 0.0414,
"step": 23440
},
{
"grad_norm": 0.20445914566516876,
"learning_rate": 1.2479870708647324e-05,
"loss": 0.0411,
"step": 23450
},
{
"grad_norm": 0.18438376486301422,
"learning_rate": 1.2443463103617658e-05,
"loss": 0.0409,
"step": 23460
},
{
"grad_norm": 0.20348145067691803,
"learning_rate": 1.240710113332167e-05,
"loss": 0.0418,
"step": 23470
},
{
"grad_norm": 0.19719134271144867,
"learning_rate": 1.2370784841942584e-05,
"loss": 0.0405,
"step": 23480
},
{
"grad_norm": 0.21505559980869293,
"learning_rate": 1.2334514273608117e-05,
"loss": 0.0419,
"step": 23490
},
{
"grad_norm": 0.21380212903022766,
"learning_rate": 1.2298289472390417e-05,
"loss": 0.0426,
"step": 23500
},
{
"grad_norm": 0.2479470819234848,
"learning_rate": 1.226211048230606e-05,
"loss": 0.0419,
"step": 23510
},
{
"grad_norm": 0.1996820569038391,
"learning_rate": 1.2225977347315921e-05,
"loss": 0.0418,
"step": 23520
},
{
"grad_norm": 0.20600031316280365,
"learning_rate": 1.2189890111325148e-05,
"loss": 0.0348,
"step": 23530
},
{
"grad_norm": 0.19264976680278778,
"learning_rate": 1.2153848818183161e-05,
"loss": 0.0443,
"step": 23540
},
{
"grad_norm": 0.1800975203514099,
"learning_rate": 1.2117853511683507e-05,
"loss": 0.0346,
"step": 23550
},
{
"grad_norm": 0.15906058251857758,
"learning_rate": 1.2081904235563906e-05,
"loss": 0.0413,
"step": 23560
},
{
"grad_norm": 0.22102585434913635,
"learning_rate": 1.20460010335061e-05,
"loss": 0.04,
"step": 23570
},
{
"grad_norm": 0.23027312755584717,
"learning_rate": 1.2010143949135865e-05,
"loss": 0.0364,
"step": 23580
},
{
"grad_norm": 0.26344889402389526,
"learning_rate": 1.1974333026022938e-05,
"loss": 0.0405,
"step": 23590
},
{
"grad_norm": 0.238390251994133,
"learning_rate": 1.1938568307680964e-05,
"loss": 0.0386,
"step": 23600
},
{
"grad_norm": 0.2628254294395447,
"learning_rate": 1.1902849837567464e-05,
"loss": 0.0378,
"step": 23610
},
{
"grad_norm": 0.19207504391670227,
"learning_rate": 1.1867177659083739e-05,
"loss": 0.0383,
"step": 23620
},
{
"grad_norm": 0.2528882324695587,
"learning_rate": 1.1831551815574848e-05,
"loss": 0.0408,
"step": 23630
},
{
"grad_norm": 0.21754851937294006,
"learning_rate": 1.1795972350329554e-05,
"loss": 0.0428,
"step": 23640
},
{
"grad_norm": 0.20590808987617493,
"learning_rate": 1.176043930658024e-05,
"loss": 0.0389,
"step": 23650
},
{
"grad_norm": 0.18157683312892914,
"learning_rate": 1.1724952727502942e-05,
"loss": 0.0381,
"step": 23660
},
{
"grad_norm": 0.23386608064174652,
"learning_rate": 1.1689512656217178e-05,
"loss": 0.0421,
"step": 23670
},
{
"grad_norm": 0.20880161225795746,
"learning_rate": 1.1654119135785962e-05,
"loss": 0.0393,
"step": 23680
},
{
"grad_norm": 0.21291187405586243,
"learning_rate": 1.1618772209215794e-05,
"loss": 0.0398,
"step": 23690
},
{
"grad_norm": 0.21790309250354767,
"learning_rate": 1.1583471919456506e-05,
"loss": 0.0368,
"step": 23700
},
{
"grad_norm": 0.21689258515834808,
"learning_rate": 1.1548218309401266e-05,
"loss": 0.0353,
"step": 23710
},
{
"grad_norm": 0.1767474114894867,
"learning_rate": 1.1513011421886555e-05,
"loss": 0.0395,
"step": 23720
},
{
"grad_norm": 0.18343332409858704,
"learning_rate": 1.1477851299692056e-05,
"loss": 0.0394,
"step": 23730
},
{
"grad_norm": 0.18294543027877808,
"learning_rate": 1.144273798554063e-05,
"loss": 0.0328,
"step": 23740
},
{
"grad_norm": 0.20064060389995575,
"learning_rate": 1.1407671522098262e-05,
"loss": 0.0346,
"step": 23750
},
{
"grad_norm": 0.22689735889434814,
"learning_rate": 1.1372651951974e-05,
"loss": 0.0388,
"step": 23760
},
{
"grad_norm": 0.2066619098186493,
"learning_rate": 1.1337679317719952e-05,
"loss": 0.0374,
"step": 23770
},
{
"grad_norm": 0.18367694318294525,
"learning_rate": 1.1302753661831151e-05,
"loss": 0.0346,
"step": 23780
},
{
"grad_norm": 0.22273485362529755,
"learning_rate": 1.1267875026745561e-05,
"loss": 0.0356,
"step": 23790
},
{
"grad_norm": 0.18771523237228394,
"learning_rate": 1.1233043454844016e-05,
"loss": 0.0384,
"step": 23800
},
{
"grad_norm": 0.22385771572589874,
"learning_rate": 1.1198258988450144e-05,
"loss": 0.0385,
"step": 23810
},
{
"grad_norm": 0.13540440797805786,
"learning_rate": 1.1163521669830368e-05,
"loss": 0.0346,
"step": 23820
},
{
"grad_norm": 0.17024365067481995,
"learning_rate": 1.1128831541193818e-05,
"loss": 0.0416,
"step": 23830
},
{
"grad_norm": 0.21784012019634247,
"learning_rate": 1.1094188644692255e-05,
"loss": 0.0359,
"step": 23840
},
{
"grad_norm": 0.18060486018657684,
"learning_rate": 1.1059593022420067e-05,
"loss": 0.0438,
"step": 23850
},
{
"grad_norm": 0.17789499461650848,
"learning_rate": 1.1025044716414184e-05,
"loss": 0.0384,
"step": 23860
},
{
"grad_norm": 0.2596547603607178,
"learning_rate": 1.0990543768654082e-05,
"loss": 0.0398,
"step": 23870
},
{
"grad_norm": 0.24579423666000366,
"learning_rate": 1.0956090221061655e-05,
"loss": 0.0413,
"step": 23880
},
{
"grad_norm": 0.20934265851974487,
"learning_rate": 1.0921684115501208e-05,
"loss": 0.0372,
"step": 23890
},
{
"grad_norm": 0.15605300664901733,
"learning_rate": 1.0887325493779405e-05,
"loss": 0.0332,
"step": 23900
},
{
"grad_norm": 0.27476954460144043,
"learning_rate": 1.0853014397645207e-05,
"loss": 0.032,
"step": 23910
},
{
"grad_norm": 0.21390210092067719,
"learning_rate": 1.0818750868789829e-05,
"loss": 0.0324,
"step": 23920
},
{
"grad_norm": 0.18390192091464996,
"learning_rate": 1.0784534948846703e-05,
"loss": 0.0324,
"step": 23930
},
{
"grad_norm": 0.17887698113918304,
"learning_rate": 1.0750366679391393e-05,
"loss": 0.0279,
"step": 23940
},
{
"grad_norm": 0.19990065693855286,
"learning_rate": 1.0716246101941557e-05,
"loss": 0.0312,
"step": 23950
},
{
"grad_norm": 0.22170302271842957,
"learning_rate": 1.0682173257956934e-05,
"loss": 0.0336,
"step": 23960
},
{
"grad_norm": 0.17926593124866486,
"learning_rate": 1.064814818883922e-05,
"loss": 0.0306,
"step": 23970
},
{
"grad_norm": 0.18762008845806122,
"learning_rate": 1.06141709359321e-05,
"loss": 0.0291,
"step": 23980
},
{
"grad_norm": 0.19169163703918457,
"learning_rate": 1.058024154052114e-05,
"loss": 0.0351,
"step": 23990
},
{
"grad_norm": 0.14914429187774658,
"learning_rate": 1.0546360043833747e-05,
"loss": 0.0316,
"step": 24000
},
{
"grad_norm": 0.15681490302085876,
"learning_rate": 1.0512526487039138e-05,
"loss": 0.0333,
"step": 24010
},
{
"grad_norm": 0.17099972069263458,
"learning_rate": 1.0478740911248258e-05,
"loss": 0.0324,
"step": 24020
},
{
"grad_norm": 0.1534363329410553,
"learning_rate": 1.0445003357513793e-05,
"loss": 0.0291,
"step": 24030
},
{
"grad_norm": 0.19713850319385529,
"learning_rate": 1.0411313866830042e-05,
"loss": 0.0362,
"step": 24040
},
{
"grad_norm": 0.2533229887485504,
"learning_rate": 1.0377672480132915e-05,
"loss": 0.0386,
"step": 24050
},
{
"grad_norm": 0.17173029482364655,
"learning_rate": 1.0344079238299864e-05,
"loss": 0.0395,
"step": 24060
},
{
"grad_norm": 0.18817080557346344,
"learning_rate": 1.0310534182149833e-05,
"loss": 0.0355,
"step": 24070
},
{
"grad_norm": 0.15675614774227142,
"learning_rate": 1.0277037352443258e-05,
"loss": 0.0329,
"step": 24080
},
{
"grad_norm": 0.24644015729427338,
"learning_rate": 1.024358878988193e-05,
"loss": 0.0411,
"step": 24090
},
{
"grad_norm": 0.1980200558900833,
"learning_rate": 1.0210188535108995e-05,
"loss": 0.0348,
"step": 24100
},
{
"grad_norm": 0.1955983191728592,
"learning_rate": 1.0176836628708935e-05,
"loss": 0.0413,
"step": 24110
},
{
"grad_norm": 0.21426765620708466,
"learning_rate": 1.0143533111207454e-05,
"loss": 0.0397,
"step": 24120
},
{
"grad_norm": 0.21278069913387299,
"learning_rate": 1.0110278023071446e-05,
"loss": 0.0348,
"step": 24130
},
{
"grad_norm": 0.22771716117858887,
"learning_rate": 1.007707140470901e-05,
"loss": 0.0342,
"step": 24140
},
{
"grad_norm": 0.26802703738212585,
"learning_rate": 1.0043913296469298e-05,
"loss": 0.035,
"step": 24150
},
{
"grad_norm": 0.18416860699653625,
"learning_rate": 1.001080373864255e-05,
"loss": 0.0361,
"step": 24160
},
{
"grad_norm": 0.2285734862089157,
"learning_rate": 9.977742771459991e-06,
"loss": 0.0369,
"step": 24170
},
{
"grad_norm": 0.19544701278209686,
"learning_rate": 9.944730435093802e-06,
"loss": 0.036,
"step": 24180
},
{
"grad_norm": 0.2433275580406189,
"learning_rate": 9.911766769657115e-06,
"loss": 0.0409,
"step": 24190
},
{
"grad_norm": 0.21004357933998108,
"learning_rate": 9.878851815203882e-06,
"loss": 0.0343,
"step": 24200
},
{
"grad_norm": 0.21688443422317505,
"learning_rate": 9.845985611728864e-06,
"loss": 0.0348,
"step": 24210
},
{
"grad_norm": 0.22794075310230255,
"learning_rate": 9.813168199167606e-06,
"loss": 0.0401,
"step": 24220
},
{
"grad_norm": 0.20268043875694275,
"learning_rate": 9.780399617396341e-06,
"loss": 0.0342,
"step": 24230
},
{
"grad_norm": 0.2068203240633011,
"learning_rate": 9.747679906232016e-06,
"loss": 0.0387,
"step": 24240
},
{
"grad_norm": 0.17917129397392273,
"learning_rate": 9.715009105432138e-06,
"loss": 0.0337,
"step": 24250
},
{
"grad_norm": 0.26544103026390076,
"learning_rate": 9.682387254694835e-06,
"loss": 0.0411,
"step": 24260
},
{
"grad_norm": 0.16896457970142365,
"learning_rate": 9.649814393658723e-06,
"loss": 0.0363,
"step": 24270
},
{
"grad_norm": 0.1958674043416977,
"learning_rate": 9.617290561902881e-06,
"loss": 0.0312,
"step": 24280
},
{
"grad_norm": 0.26005280017852783,
"learning_rate": 9.584815798946861e-06,
"loss": 0.0389,
"step": 24290
},
{
"grad_norm": 0.2262255698442459,
"learning_rate": 9.55239014425055e-06,
"loss": 0.0329,
"step": 24300
},
{
"grad_norm": 0.19896724820137024,
"learning_rate": 9.520013637214176e-06,
"loss": 0.0364,
"step": 24310
},
{
"grad_norm": 0.21827764809131622,
"learning_rate": 9.48768631717824e-06,
"loss": 0.0393,
"step": 24320
},
{
"grad_norm": 0.2604480981826782,
"learning_rate": 9.455408223423496e-06,
"loss": 0.0381,
"step": 24330
},
{
"grad_norm": 0.24648134410381317,
"learning_rate": 9.423179395170845e-06,
"loss": 0.0364,
"step": 24340
},
{
"grad_norm": 0.18727527558803558,
"learning_rate": 9.390999871581391e-06,
"loss": 0.0338,
"step": 24350
},
{
"grad_norm": 0.2093559354543686,
"learning_rate": 9.358869691756273e-06,
"loss": 0.0367,
"step": 24360
},
{
"grad_norm": 0.21332038938999176,
"learning_rate": 9.326788894736688e-06,
"loss": 0.0362,
"step": 24370
},
{
"grad_norm": 0.27581986784935,
"learning_rate": 9.294757519503811e-06,
"loss": 0.0328,
"step": 24380
},
{
"grad_norm": 0.21681338548660278,
"learning_rate": 9.262775604978819e-06,
"loss": 0.0393,
"step": 24390
},
{
"grad_norm": 0.20855584740638733,
"learning_rate": 9.230843190022725e-06,
"loss": 0.0301,
"step": 24400
},
{
"grad_norm": 0.20897409319877625,
"learning_rate": 9.198960313436445e-06,
"loss": 0.0294,
"step": 24410
},
{
"grad_norm": 0.21171963214874268,
"learning_rate": 9.16712701396067e-06,
"loss": 0.0337,
"step": 24420
},
{
"grad_norm": 0.23461666703224182,
"learning_rate": 9.135343330275864e-06,
"loss": 0.029,
"step": 24430
},
{
"grad_norm": 0.2153220772743225,
"learning_rate": 9.10360930100218e-06,
"loss": 0.0325,
"step": 24440
},
{
"grad_norm": 0.23157557845115662,
"learning_rate": 9.07192496469949e-06,
"loss": 0.0337,
"step": 24450
},
{
"grad_norm": 0.1828492283821106,
"learning_rate": 9.040290359867231e-06,
"loss": 0.0323,
"step": 24460
},
{
"grad_norm": 0.1797216683626175,
"learning_rate": 9.008705524944438e-06,
"loss": 0.0305,
"step": 24470
},
{
"grad_norm": 0.15407446026802063,
"learning_rate": 8.977170498309651e-06,
"loss": 0.0335,
"step": 24480
},
{
"grad_norm": 0.17882442474365234,
"learning_rate": 8.945685318280916e-06,
"loss": 0.03,
"step": 24490
},
{
"grad_norm": 0.1426704376935959,
"learning_rate": 8.914250023115672e-06,
"loss": 0.0296,
"step": 24500
},
{
"grad_norm": 0.17062900960445404,
"learning_rate": 8.882864651010798e-06,
"loss": 0.0295,
"step": 24510
},
{
"grad_norm": 0.18100108206272125,
"learning_rate": 8.851529240102462e-06,
"loss": 0.0325,
"step": 24520
},
{
"grad_norm": 0.24238081276416779,
"learning_rate": 8.820243828466134e-06,
"loss": 0.0337,
"step": 24530
},
{
"grad_norm": 0.20483867824077606,
"learning_rate": 8.789008454116566e-06,
"loss": 0.0382,
"step": 24540
},
{
"grad_norm": 0.21139168739318848,
"learning_rate": 8.757823155007655e-06,
"loss": 0.0353,
"step": 24550
},
{
"grad_norm": 0.19928909838199615,
"learning_rate": 8.72668796903251e-06,
"loss": 0.0401,
"step": 24560
},
{
"grad_norm": 0.22222256660461426,
"learning_rate": 8.695602934023301e-06,
"loss": 0.0352,
"step": 24570
},
{
"grad_norm": 0.14943648874759674,
"learning_rate": 8.664568087751274e-06,
"loss": 0.0378,
"step": 24580
},
{
"grad_norm": 0.16387082636356354,
"learning_rate": 8.633583467926697e-06,
"loss": 0.0394,
"step": 24590
},
{
"grad_norm": 0.19438165426254272,
"learning_rate": 8.602649112198796e-06,
"loss": 0.0418,
"step": 24600
},
{
"grad_norm": 0.2061309665441513,
"learning_rate": 8.571765058155745e-06,
"loss": 0.0386,
"step": 24610
},
{
"grad_norm": 0.2485935389995575,
"learning_rate": 8.540931343324583e-06,
"loss": 0.0388,
"step": 24620
},
{
"grad_norm": 0.18815116584300995,
"learning_rate": 8.51014800517117e-06,
"loss": 0.0375,
"step": 24630
},
{
"grad_norm": 0.20986588299274445,
"learning_rate": 8.479415081100167e-06,
"loss": 0.0364,
"step": 24640
},
{
"grad_norm": 0.18513725697994232,
"learning_rate": 8.448732608454967e-06,
"loss": 0.0325,
"step": 24650
},
{
"grad_norm": 0.17452077567577362,
"learning_rate": 8.418100624517688e-06,
"loss": 0.0325,
"step": 24660
},
{
"grad_norm": 0.24269820749759674,
"learning_rate": 8.387519166509062e-06,
"loss": 0.0318,
"step": 24670
},
{
"grad_norm": 0.15794600546360016,
"learning_rate": 8.356988271588445e-06,
"loss": 0.031,
"step": 24680
},
{
"grad_norm": 0.283208966255188,
"learning_rate": 8.326507976853765e-06,
"loss": 0.039,
"step": 24690
},
{
"grad_norm": 0.19195295870304108,
"learning_rate": 8.296078319341443e-06,
"loss": 0.034,
"step": 24700
},
{
"grad_norm": 0.17826755344867706,
"learning_rate": 8.265699336026383e-06,
"loss": 0.0321,
"step": 24710
},
{
"grad_norm": 0.15649399161338806,
"learning_rate": 8.235371063821923e-06,
"loss": 0.0349,
"step": 24720
},
{
"grad_norm": 0.2310352623462677,
"learning_rate": 8.205093539579767e-06,
"loss": 0.0356,
"step": 24730
},
{
"grad_norm": 0.15615054965019226,
"learning_rate": 8.174866800089963e-06,
"loss": 0.0306,
"step": 24740
},
{
"grad_norm": 0.20134374499320984,
"learning_rate": 8.144690882080853e-06,
"loss": 0.0338,
"step": 24750
},
{
"grad_norm": 0.15863995254039764,
"learning_rate": 8.114565822219006e-06,
"loss": 0.0307,
"step": 24760
},
{
"grad_norm": 0.22674250602722168,
"learning_rate": 8.084491657109233e-06,
"loss": 0.0354,
"step": 24770
},
{
"grad_norm": 0.15616664290428162,
"learning_rate": 8.054468423294476e-06,
"loss": 0.035,
"step": 24780
},
{
"grad_norm": 0.19647470116615295,
"learning_rate": 8.024496157255785e-06,
"loss": 0.0369,
"step": 24790
},
{
"grad_norm": 0.1819557547569275,
"learning_rate": 7.994574895412293e-06,
"loss": 0.0359,
"step": 24800
},
{
"grad_norm": 0.20564055442810059,
"learning_rate": 7.964704674121149e-06,
"loss": 0.0349,
"step": 24810
},
{
"grad_norm": 0.18771442770957947,
"learning_rate": 7.934885529677505e-06,
"loss": 0.035,
"step": 24820
},
{
"grad_norm": 0.17739535868167877,
"learning_rate": 7.905117498314412e-06,
"loss": 0.0346,
"step": 24830
},
{
"grad_norm": 0.18636682629585266,
"learning_rate": 7.87540061620286e-06,
"loss": 0.0321,
"step": 24840
},
{
"grad_norm": 0.22322791814804077,
"learning_rate": 7.845734919451647e-06,
"loss": 0.0372,
"step": 24850
},
{
"grad_norm": 0.19257836043834686,
"learning_rate": 7.816120444107383e-06,
"loss": 0.036,
"step": 24860
},
{
"grad_norm": 0.19602902233600616,
"learning_rate": 7.786557226154473e-06,
"loss": 0.0319,
"step": 24870
},
{
"grad_norm": 0.18849128484725952,
"learning_rate": 7.757045301514998e-06,
"loss": 0.0354,
"step": 24880
},
{
"grad_norm": 0.20831118524074554,
"learning_rate": 7.727584706048735e-06,
"loss": 0.0385,
"step": 24890
},
{
"grad_norm": 0.1676318645477295,
"learning_rate": 7.698175475553077e-06,
"loss": 0.0392,
"step": 24900
},
{
"grad_norm": 0.22026681900024414,
"learning_rate": 7.668817645763021e-06,
"loss": 0.0372,
"step": 24910
},
{
"grad_norm": 0.18115563690662384,
"learning_rate": 7.639511252351088e-06,
"loss": 0.0367,
"step": 24920
},
{
"grad_norm": 0.21893097460269928,
"learning_rate": 7.610256330927324e-06,
"loss": 0.0386,
"step": 24930
},
{
"grad_norm": 0.19247709214687347,
"learning_rate": 7.581052917039211e-06,
"loss": 0.0453,
"step": 24940
},
{
"grad_norm": 0.25413358211517334,
"learning_rate": 7.551901046171645e-06,
"loss": 0.0405,
"step": 24950
},
{
"grad_norm": 0.19903209805488586,
"learning_rate": 7.522800753746895e-06,
"loss": 0.0404,
"step": 24960
},
{
"grad_norm": 0.21743354201316833,
"learning_rate": 7.493752075124577e-06,
"loss": 0.0428,
"step": 24970
},
{
"grad_norm": 0.17956329882144928,
"learning_rate": 7.464755045601557e-06,
"loss": 0.0381,
"step": 24980
},
{
"grad_norm": 0.23483504354953766,
"learning_rate": 7.435809700411972e-06,
"loss": 0.0388,
"step": 24990
},
{
"grad_norm": 0.1919252872467041,
"learning_rate": 7.40691607472715e-06,
"loss": 0.0391,
"step": 25000
},
{
"grad_norm": 0.18066823482513428,
"learning_rate": 7.3780742036555606e-06,
"loss": 0.0401,
"step": 25010
},
{
"grad_norm": 0.1938222050666809,
"learning_rate": 7.349284122242783e-06,
"loss": 0.0353,
"step": 25020
},
{
"grad_norm": 0.19637170433998108,
"learning_rate": 7.320545865471512e-06,
"loss": 0.0352,
"step": 25030
},
{
"grad_norm": 0.2064981460571289,
"learning_rate": 7.291859468261425e-06,
"loss": 0.0324,
"step": 25040
},
{
"grad_norm": 0.23065944015979767,
"learning_rate": 7.263224965469195e-06,
"loss": 0.0348,
"step": 25050
},
{
"grad_norm": 0.23484154045581818,
"learning_rate": 7.234642391888446e-06,
"loss": 0.0365,
"step": 25060
},
{
"grad_norm": 0.16796699166297913,
"learning_rate": 7.206111782249697e-06,
"loss": 0.0319,
"step": 25070
},
{
"grad_norm": 0.156756192445755,
"learning_rate": 7.1776331712203385e-06,
"loss": 0.0284,
"step": 25080
},
{
"grad_norm": 0.21304923295974731,
"learning_rate": 7.149206593404561e-06,
"loss": 0.0308,
"step": 25090
},
{
"grad_norm": 0.20190036296844482,
"learning_rate": 7.120832083343337e-06,
"loss": 0.0328,
"step": 25100
},
{
"grad_norm": 0.22565123438835144,
"learning_rate": 7.092509675514369e-06,
"loss": 0.0366,
"step": 25110
},
{
"grad_norm": 0.1947108954191208,
"learning_rate": 7.064239404332063e-06,
"loss": 0.0354,
"step": 25120
},
{
"grad_norm": 0.18892329931259155,
"learning_rate": 7.03602130414745e-06,
"loss": 0.0304,
"step": 25130
},
{
"grad_norm": 0.24032840132713318,
"learning_rate": 7.007855409248198e-06,
"loss": 0.0358,
"step": 25140
},
{
"grad_norm": 0.15574835240840912,
"learning_rate": 6.979741753858521e-06,
"loss": 0.0401,
"step": 25150
},
{
"grad_norm": 0.3073306679725647,
"learning_rate": 6.951680372139158e-06,
"loss": 0.0395,
"step": 25160
},
{
"grad_norm": 0.18491801619529724,
"learning_rate": 6.923671298187334e-06,
"loss": 0.0379,
"step": 25170
},
{
"grad_norm": 0.2675468325614929,
"learning_rate": 6.895714566036704e-06,
"loss": 0.0401,
"step": 25180
},
{
"grad_norm": 0.1535646617412567,
"learning_rate": 6.86781020965736e-06,
"loss": 0.037,
"step": 25190
},
{
"grad_norm": 0.2230897694826126,
"learning_rate": 6.839958262955709e-06,
"loss": 0.037,
"step": 25200
},
{
"grad_norm": 0.2286185771226883,
"learning_rate": 6.8121587597744894e-06,
"loss": 0.0351,
"step": 25210
},
{
"grad_norm": 0.22103099524974823,
"learning_rate": 6.784411733892732e-06,
"loss": 0.0328,
"step": 25220
},
{
"grad_norm": 0.15428020060062408,
"learning_rate": 6.756717219025666e-06,
"loss": 0.0324,
"step": 25230
},
{
"grad_norm": 0.2118782103061676,
"learning_rate": 6.729075248824762e-06,
"loss": 0.0371,
"step": 25240
},
{
"grad_norm": 0.20298250019550323,
"learning_rate": 6.701485856877615e-06,
"loss": 0.0364,
"step": 25250
},
{
"grad_norm": 0.2596721053123474,
"learning_rate": 6.673949076707925e-06,
"loss": 0.0312,
"step": 25260
},
{
"grad_norm": 0.20175741612911224,
"learning_rate": 6.646464941775499e-06,
"loss": 0.0378,
"step": 25270
},
{
"grad_norm": 0.2263554334640503,
"learning_rate": 6.619033485476128e-06,
"loss": 0.0362,
"step": 25280
},
{
"grad_norm": 0.2336198389530182,
"learning_rate": 6.59165474114164e-06,
"loss": 0.0342,
"step": 25290
},
{
"grad_norm": 0.27127358317375183,
"learning_rate": 6.564328742039783e-06,
"loss": 0.033,
"step": 25300
},
{
"grad_norm": 0.20276868343353271,
"learning_rate": 6.537055521374219e-06,
"loss": 0.031,
"step": 25310
},
{
"grad_norm": 0.20111382007598877,
"learning_rate": 6.509835112284485e-06,
"loss": 0.0295,
"step": 25320
},
{
"grad_norm": 0.20731808245182037,
"learning_rate": 6.482667547845944e-06,
"loss": 0.0314,
"step": 25330
},
{
"grad_norm": 0.18052595853805542,
"learning_rate": 6.455552861069736e-06,
"loss": 0.0287,
"step": 25340
},
{
"grad_norm": 0.17958711087703705,
"learning_rate": 6.428491084902788e-06,
"loss": 0.033,
"step": 25350
},
{
"grad_norm": 0.1632886528968811,
"learning_rate": 6.401482252227698e-06,
"loss": 0.0285,
"step": 25360
},
{
"grad_norm": 0.20821455121040344,
"learning_rate": 6.37452639586274e-06,
"loss": 0.0333,
"step": 25370
},
{
"grad_norm": 0.20519405603408813,
"learning_rate": 6.347623548561826e-06,
"loss": 0.038,
"step": 25380
},
{
"grad_norm": 0.20152124762535095,
"learning_rate": 6.320773743014441e-06,
"loss": 0.0348,
"step": 25390
},
{
"grad_norm": 0.16101081669330597,
"learning_rate": 6.293977011845648e-06,
"loss": 0.0304,
"step": 25400
},
{
"grad_norm": 0.1880352646112442,
"learning_rate": 6.2672333876159835e-06,
"loss": 0.0328,
"step": 25410
},
{
"grad_norm": 0.1922726333141327,
"learning_rate": 6.2405429028215e-06,
"loss": 0.0341,
"step": 25420
},
{
"grad_norm": 0.19048094749450684,
"learning_rate": 6.213905589893631e-06,
"loss": 0.0384,
"step": 25430
},
{
"grad_norm": 0.23592276871204376,
"learning_rate": 6.187321481199221e-06,
"loss": 0.0395,
"step": 25440
},
{
"grad_norm": 0.17596854269504547,
"learning_rate": 6.1607906090404895e-06,
"loss": 0.0382,
"step": 25450
},
{
"grad_norm": 0.22248589992523193,
"learning_rate": 6.134313005654929e-06,
"loss": 0.0369,
"step": 25460
},
{
"grad_norm": 0.2005740851163864,
"learning_rate": 6.107888703215337e-06,
"loss": 0.0346,
"step": 25470
},
{
"grad_norm": 0.20605795085430145,
"learning_rate": 6.081517733829722e-06,
"loss": 0.0322,
"step": 25480
},
{
"grad_norm": 0.2689594626426697,
"learning_rate": 6.055200129541294e-06,
"loss": 0.0373,
"step": 25490
},
{
"grad_norm": 0.20270395278930664,
"learning_rate": 6.0289359223284434e-06,
"loss": 0.0385,
"step": 25500
},
{
"grad_norm": 0.1983395665884018,
"learning_rate": 6.002725144104648e-06,
"loss": 0.0397,
"step": 25510
},
{
"grad_norm": 0.192221000790596,
"learning_rate": 5.976567826718477e-06,
"loss": 0.0366,
"step": 25520
},
{
"grad_norm": 0.2265564650297165,
"learning_rate": 5.9504640019535315e-06,
"loss": 0.044,
"step": 25530
},
{
"grad_norm": 0.23999322950839996,
"learning_rate": 5.924413701528414e-06,
"loss": 0.0342,
"step": 25540
},
{
"grad_norm": 0.2028564065694809,
"learning_rate": 5.898416957096703e-06,
"loss": 0.0313,
"step": 25550
},
{
"grad_norm": 0.1974341869354248,
"learning_rate": 5.872473800246913e-06,
"loss": 0.0339,
"step": 25560
},
{
"grad_norm": 0.1961073875427246,
"learning_rate": 5.846584262502402e-06,
"loss": 0.035,
"step": 25570
},
{
"grad_norm": 0.22909915447235107,
"learning_rate": 5.820748375321411e-06,
"loss": 0.0342,
"step": 25580
},
{
"grad_norm": 0.21388410031795502,
"learning_rate": 5.794966170096977e-06,
"loss": 0.0343,
"step": 25590
},
{
"grad_norm": 0.18144114315509796,
"learning_rate": 5.769237678156897e-06,
"loss": 0.0347,
"step": 25600
},
{
"grad_norm": 0.16462798416614532,
"learning_rate": 5.743562930763735e-06,
"loss": 0.0324,
"step": 25610
},
{
"grad_norm": 0.31449681520462036,
"learning_rate": 5.717941959114726e-06,
"loss": 0.0329,
"step": 25620
},
{
"grad_norm": 0.20670804381370544,
"learning_rate": 5.69237479434176e-06,
"loss": 0.0386,
"step": 25630
},
{
"grad_norm": 0.18057267367839813,
"learning_rate": 5.666861467511353e-06,
"loss": 0.0372,
"step": 25640
},
{
"grad_norm": 0.2072240114212036,
"learning_rate": 5.641402009624591e-06,
"loss": 0.0351,
"step": 25650
},
{
"grad_norm": 0.19102953374385834,
"learning_rate": 5.615996451617145e-06,
"loss": 0.0405,
"step": 25660
},
{
"grad_norm": 0.23088519275188446,
"learning_rate": 5.590644824359148e-06,
"loss": 0.0364,
"step": 25670
},
{
"grad_norm": 0.19517070055007935,
"learning_rate": 5.56534715865521e-06,
"loss": 0.0414,
"step": 25680
},
{
"grad_norm": 0.20782841742038727,
"learning_rate": 5.5401034852443965e-06,
"loss": 0.0398,
"step": 25690
},
{
"grad_norm": 0.24214865267276764,
"learning_rate": 5.514913834800134e-06,
"loss": 0.0461,
"step": 25700
},
{
"grad_norm": 0.21279871463775635,
"learning_rate": 5.489778237930238e-06,
"loss": 0.0367,
"step": 25710
},
{
"grad_norm": 0.22451657056808472,
"learning_rate": 5.464696725176827e-06,
"loss": 0.0426,
"step": 25720
},
{
"grad_norm": 0.16567926108837128,
"learning_rate": 5.439669327016294e-06,
"loss": 0.0433,
"step": 25730
},
{
"grad_norm": 0.2070925086736679,
"learning_rate": 5.4146960738592985e-06,
"loss": 0.0358,
"step": 25740
},
{
"grad_norm": 0.20479975640773773,
"learning_rate": 5.389776996050694e-06,
"loss": 0.0303,
"step": 25750
},
{
"grad_norm": 0.16662999987602234,
"learning_rate": 5.364912123869492e-06,
"loss": 0.0316,
"step": 25760
},
{
"grad_norm": 0.16534346342086792,
"learning_rate": 5.3401014875288864e-06,
"loss": 0.0344,
"step": 25770
},
{
"grad_norm": 0.21096844971179962,
"learning_rate": 5.315345117176129e-06,
"loss": 0.038,
"step": 25780
},
{
"grad_norm": 0.23314110934734344,
"learning_rate": 5.2906430428925405e-06,
"loss": 0.0318,
"step": 25790
},
{
"grad_norm": 0.1966601312160492,
"learning_rate": 5.265995294693471e-06,
"loss": 0.0314,
"step": 25800
},
{
"grad_norm": 0.19453799724578857,
"learning_rate": 5.241401902528254e-06,
"loss": 0.0308,
"step": 25810
},
{
"grad_norm": 0.17336805164813995,
"learning_rate": 5.216862896280189e-06,
"loss": 0.0272,
"step": 25820
},
{
"grad_norm": 0.20418553054332733,
"learning_rate": 5.192378305766471e-06,
"loss": 0.0283,
"step": 25830
},
{
"grad_norm": 0.16264602541923523,
"learning_rate": 5.1679481607382065e-06,
"loss": 0.0329,
"step": 25840
},
{
"grad_norm": 0.19209501147270203,
"learning_rate": 5.1435724908803105e-06,
"loss": 0.0314,
"step": 25850
},
{
"grad_norm": 0.171164408326149,
"learning_rate": 5.119251325811514e-06,
"loss": 0.0331,
"step": 25860
},
{
"grad_norm": 0.13701258599758148,
"learning_rate": 5.094984695084348e-06,
"loss": 0.0304,
"step": 25870
},
{
"grad_norm": 0.20167377591133118,
"learning_rate": 5.070772628185039e-06,
"loss": 0.0311,
"step": 25880
},
{
"grad_norm": 0.18954886496067047,
"learning_rate": 5.046615154533535e-06,
"loss": 0.0306,
"step": 25890
},
{
"grad_norm": 0.20127075910568237,
"learning_rate": 5.022512303483451e-06,
"loss": 0.0326,
"step": 25900
},
{
"grad_norm": 0.2184620499610901,
"learning_rate": 4.998464104322015e-06,
"loss": 0.0354,
"step": 25910
},
{
"grad_norm": 0.15516583621501923,
"learning_rate": 4.974470586270047e-06,
"loss": 0.0307,
"step": 25920
},
{
"grad_norm": 0.17588339745998383,
"learning_rate": 4.950531778481965e-06,
"loss": 0.0321,
"step": 25930
},
{
"grad_norm": 0.21232159435749054,
"learning_rate": 4.926647710045651e-06,
"loss": 0.0352,
"step": 25940
},
{
"grad_norm": 0.21604269742965698,
"learning_rate": 4.9028184099825125e-06,
"loss": 0.0342,
"step": 25950
},
{
"grad_norm": 0.22263774275779724,
"learning_rate": 4.879043907247382e-06,
"loss": 0.032,
"step": 25960
},
{
"grad_norm": 0.27130556106567383,
"learning_rate": 4.8553242307285414e-06,
"loss": 0.0359,
"step": 25970
},
{
"grad_norm": 0.1889198273420334,
"learning_rate": 4.8316594092476185e-06,
"loss": 0.0318,
"step": 25980
},
{
"grad_norm": 0.2060389220714569,
"learning_rate": 4.808049471559617e-06,
"loss": 0.0345,
"step": 25990
},
{
"grad_norm": 0.22062867879867554,
"learning_rate": 4.784494446352833e-06,
"loss": 0.0375,
"step": 26000
},
{
"grad_norm": 0.22988341748714447,
"learning_rate": 4.760994362248833e-06,
"loss": 0.0367,
"step": 26010
},
{
"grad_norm": 0.20314031839370728,
"learning_rate": 4.737549247802442e-06,
"loss": 0.0365,
"step": 26020
},
{
"grad_norm": 0.2126205414533615,
"learning_rate": 4.714159131501689e-06,
"loss": 0.0372,
"step": 26030
},
{
"grad_norm": 0.15630999207496643,
"learning_rate": 4.690824041767766e-06,
"loss": 0.0367,
"step": 26040
},
{
"grad_norm": 0.20806379616260529,
"learning_rate": 4.6675440069550105e-06,
"loss": 0.0359,
"step": 26050
},
{
"grad_norm": 0.2267577052116394,
"learning_rate": 4.64431905535086e-06,
"loss": 0.0381,
"step": 26060
},
{
"grad_norm": 0.25793811678886414,
"learning_rate": 4.621149215175808e-06,
"loss": 0.0371,
"step": 26070
},
{
"grad_norm": 0.1975090652704239,
"learning_rate": 4.5980345145834154e-06,
"loss": 0.0429,
"step": 26080
},
{
"grad_norm": 0.241678386926651,
"learning_rate": 4.574974981660213e-06,
"loss": 0.0387,
"step": 26090
},
{
"grad_norm": 0.19274018704891205,
"learning_rate": 4.551970644425707e-06,
"loss": 0.0413,
"step": 26100
},
{
"grad_norm": 0.2316385805606842,
"learning_rate": 4.529021530832328e-06,
"loss": 0.0384,
"step": 26110
},
{
"grad_norm": 0.20950907468795776,
"learning_rate": 4.5061276687654306e-06,
"loss": 0.0431,
"step": 26120
},
{
"grad_norm": 0.20928706228733063,
"learning_rate": 4.4832890860431965e-06,
"loss": 0.0367,
"step": 26130
},
{
"grad_norm": 0.21495896577835083,
"learning_rate": 4.460505810416682e-06,
"loss": 0.0349,
"step": 26140
},
{
"grad_norm": 0.17650671303272247,
"learning_rate": 4.4377778695696984e-06,
"loss": 0.0379,
"step": 26150
},
{
"grad_norm": 0.22892631590366364,
"learning_rate": 4.415105291118843e-06,
"loss": 0.0402,
"step": 26160
},
{
"grad_norm": 0.16991275548934937,
"learning_rate": 4.392488102613435e-06,
"loss": 0.037,
"step": 26170
},
{
"grad_norm": 0.23251581192016602,
"learning_rate": 4.369926331535473e-06,
"loss": 0.0408,
"step": 26180
},
{
"grad_norm": 0.15092848241329193,
"learning_rate": 4.347420005299668e-06,
"loss": 0.0342,
"step": 26190
},
{
"grad_norm": 0.2652461528778076,
"learning_rate": 4.324969151253317e-06,
"loss": 0.0349,
"step": 26200
},
{
"grad_norm": 0.21410644054412842,
"learning_rate": 4.302573796676312e-06,
"loss": 0.0378,
"step": 26210
},
{
"grad_norm": 0.21196499466896057,
"learning_rate": 4.2802339687811385e-06,
"loss": 0.0338,
"step": 26220
},
{
"grad_norm": 0.18901300430297852,
"learning_rate": 4.257949694712771e-06,
"loss": 0.0326,
"step": 26230
},
{
"grad_norm": 0.21811161935329437,
"learning_rate": 4.235721001548726e-06,
"loss": 0.033,
"step": 26240
},
{
"grad_norm": 0.1892784684896469,
"learning_rate": 4.21354791629896e-06,
"loss": 0.0339,
"step": 26250
},
{
"grad_norm": 0.18820306658744812,
"learning_rate": 4.191430465905843e-06,
"loss": 0.0349,
"step": 26260
},
{
"grad_norm": 0.2130112498998642,
"learning_rate": 4.169368677244184e-06,
"loss": 0.0333,
"step": 26270
},
{
"grad_norm": 0.21873033046722412,
"learning_rate": 4.147362577121122e-06,
"loss": 0.0353,
"step": 26280
},
{
"grad_norm": 0.23470018804073334,
"learning_rate": 4.125412192276157e-06,
"loss": 0.0354,
"step": 26290
},
{
"grad_norm": 0.18592198193073273,
"learning_rate": 4.10351754938107e-06,
"loss": 0.0324,
"step": 26300
},
{
"grad_norm": 0.15963780879974365,
"learning_rate": 4.081678675039913e-06,
"loss": 0.0275,
"step": 26310
},
{
"grad_norm": 0.1756141483783722,
"learning_rate": 4.0598955957889876e-06,
"loss": 0.0274,
"step": 26320
},
{
"grad_norm": 0.19199016690254211,
"learning_rate": 4.038168338096776e-06,
"loss": 0.0343,
"step": 26330
},
{
"grad_norm": 0.206266850233078,
"learning_rate": 4.016496928363944e-06,
"loss": 0.0285,
"step": 26340
},
{
"grad_norm": 0.19693589210510254,
"learning_rate": 3.994881392923317e-06,
"loss": 0.0314,
"step": 26350
},
{
"grad_norm": 0.1256352812051773,
"learning_rate": 3.9733217580397945e-06,
"loss": 0.0281,
"step": 26360
},
{
"grad_norm": 0.24775584042072296,
"learning_rate": 3.951818049910366e-06,
"loss": 0.0327,
"step": 26370
},
{
"grad_norm": 0.20777025818824768,
"learning_rate": 3.930370294664071e-06,
"loss": 0.032,
"step": 26380
},
{
"grad_norm": 0.14757362008094788,
"learning_rate": 3.908978518361938e-06,
"loss": 0.0279,
"step": 26390
},
{
"grad_norm": 0.1970275193452835,
"learning_rate": 3.887642746997017e-06,
"loss": 0.0302,
"step": 26400
},
{
"grad_norm": 0.20244228839874268,
"learning_rate": 3.866363006494255e-06,
"loss": 0.0303,
"step": 26410
},
{
"grad_norm": 0.3108057975769043,
"learning_rate": 3.8451393227105725e-06,
"loss": 0.0352,
"step": 26420
},
{
"grad_norm": 0.16092510521411896,
"learning_rate": 3.823971721434727e-06,
"loss": 0.0342,
"step": 26430
},
{
"grad_norm": 0.2555609941482544,
"learning_rate": 3.8028602283873505e-06,
"loss": 0.0366,
"step": 26440
},
{
"grad_norm": 0.17621466517448425,
"learning_rate": 3.781804869220912e-06,
"loss": 0.0379,
"step": 26450
},
{
"grad_norm": 0.2678952217102051,
"learning_rate": 3.760805669519646e-06,
"loss": 0.0351,
"step": 26460
},
{
"grad_norm": 0.1531263142824173,
"learning_rate": 3.7398626547995587e-06,
"loss": 0.033,
"step": 26470
},
{
"grad_norm": 0.21087443828582764,
"learning_rate": 3.7189758505083995e-06,
"loss": 0.0343,
"step": 26480
},
{
"grad_norm": 0.15880614519119263,
"learning_rate": 3.6981452820255837e-06,
"loss": 0.0311,
"step": 26490
},
{
"grad_norm": 0.20133022964000702,
"learning_rate": 3.6773709746622363e-06,
"loss": 0.034,
"step": 26500
},
{
"grad_norm": 0.17077286541461945,
"learning_rate": 3.6566529536610995e-06,
"loss": 0.0328,
"step": 26510
},
{
"grad_norm": 0.2948630154132843,
"learning_rate": 3.635991244196513e-06,
"loss": 0.0383,
"step": 26520
},
{
"grad_norm": 0.19197946786880493,
"learning_rate": 3.615385871374405e-06,
"loss": 0.0296,
"step": 26530
},
{
"grad_norm": 0.24558839201927185,
"learning_rate": 3.5948368602322367e-06,
"loss": 0.0382,
"step": 26540
},
{
"grad_norm": 0.24087628722190857,
"learning_rate": 3.5743442357390187e-06,
"loss": 0.0385,
"step": 26550
},
{
"grad_norm": 0.2480492889881134,
"learning_rate": 3.5539080227951937e-06,
"loss": 0.0338,
"step": 26560
},
{
"grad_norm": 0.21580128371715546,
"learning_rate": 3.533528246232709e-06,
"loss": 0.0349,
"step": 26570
},
{
"grad_norm": 0.2574542462825775,
"learning_rate": 3.5132049308149117e-06,
"loss": 0.0381,
"step": 26580
},
{
"grad_norm": 0.13560357689857483,
"learning_rate": 3.4929381012365425e-06,
"loss": 0.0338,
"step": 26590
},
{
"grad_norm": 0.27568548917770386,
"learning_rate": 3.472727782123697e-06,
"loss": 0.038,
"step": 26600
},
{
"grad_norm": 0.18689028918743134,
"learning_rate": 3.4525739980338425e-06,
"loss": 0.0354,
"step": 26610
},
{
"grad_norm": 0.1971084475517273,
"learning_rate": 3.432476773455712e-06,
"loss": 0.0357,
"step": 26620
},
{
"grad_norm": 0.1868872493505478,
"learning_rate": 3.4124361328093378e-06,
"loss": 0.0378,
"step": 26630
},
{
"grad_norm": 0.20983004570007324,
"learning_rate": 3.392452100445975e-06,
"loss": 0.0415,
"step": 26640
},
{
"grad_norm": 0.22753924131393433,
"learning_rate": 3.3725247006481163e-06,
"loss": 0.0389,
"step": 26650
},
{
"grad_norm": 0.2411780059337616,
"learning_rate": 3.3526539576294313e-06,
"loss": 0.0377,
"step": 26660
},
{
"grad_norm": 0.18486426770687103,
"learning_rate": 3.332839895534745e-06,
"loss": 0.034,
"step": 26670
},
{
"grad_norm": 0.243433877825737,
"learning_rate": 3.3130825384400155e-06,
"loss": 0.0325,
"step": 26680
},
{
"grad_norm": 0.2794117331504822,
"learning_rate": 3.293381910352278e-06,
"loss": 0.0315,
"step": 26690
},
{
"grad_norm": 0.17934608459472656,
"learning_rate": 3.273738035209678e-06,
"loss": 0.0325,
"step": 26700
},
{
"grad_norm": 0.18842627108097076,
"learning_rate": 3.254150936881356e-06,
"loss": 0.0354,
"step": 26710
},
{
"grad_norm": 0.1628698855638504,
"learning_rate": 3.234620639167496e-06,
"loss": 0.0345,
"step": 26720
},
{
"grad_norm": 0.24785351753234863,
"learning_rate": 3.2151471657992482e-06,
"loss": 0.0323,
"step": 26730
},
{
"grad_norm": 0.14857564866542816,
"learning_rate": 3.195730540438718e-06,
"loss": 0.0278,
"step": 26740
},
{
"grad_norm": 0.20546548068523407,
"learning_rate": 3.1763707866789336e-06,
"loss": 0.0291,
"step": 26750
},
{
"grad_norm": 0.15533821284770966,
"learning_rate": 3.1570679280438165e-06,
"loss": 0.0271,
"step": 26760
},
{
"grad_norm": 0.17334353923797607,
"learning_rate": 3.1378219879881767e-06,
"loss": 0.0296,
"step": 26770
},
{
"grad_norm": 0.22950275242328644,
"learning_rate": 3.11863298989763e-06,
"loss": 0.0323,
"step": 26780
},
{
"grad_norm": 0.19203022122383118,
"learning_rate": 3.0995009570886303e-06,
"loss": 0.0316,
"step": 26790
},
{
"grad_norm": 0.2657698392868042,
"learning_rate": 3.0804259128083936e-06,
"loss": 0.0347,
"step": 26800
},
{
"grad_norm": 0.2824636697769165,
"learning_rate": 3.06140788023489e-06,
"loss": 0.0324,
"step": 26810
},
{
"grad_norm": 0.21115641295909882,
"learning_rate": 3.042446882476846e-06,
"loss": 0.0369,
"step": 26820
},
{
"grad_norm": 0.37935954332351685,
"learning_rate": 3.023542942573643e-06,
"loss": 0.0373,
"step": 26830
},
{
"grad_norm": 0.21027018129825592,
"learning_rate": 3.004696083495351e-06,
"loss": 0.0393,
"step": 26840
},
{
"grad_norm": 0.27575919032096863,
"learning_rate": 2.98590632814269e-06,
"loss": 0.0395,
"step": 26850
},
{
"grad_norm": 0.25530073046684265,
"learning_rate": 2.9671736993469702e-06,
"loss": 0.0406,
"step": 26860
},
{
"grad_norm": 0.26740407943725586,
"learning_rate": 2.9484982198701216e-06,
"loss": 0.0378,
"step": 26870
},
{
"grad_norm": 0.2092030644416809,
"learning_rate": 2.929879912404604e-06,
"loss": 0.0399,
"step": 26880
},
{
"grad_norm": 0.22626225650310516,
"learning_rate": 2.9113187995734147e-06,
"loss": 0.0392,
"step": 26890
},
{
"grad_norm": 0.18050073087215424,
"learning_rate": 2.8928149039300524e-06,
"loss": 0.0369,
"step": 26900
},
{
"grad_norm": 0.2467006891965866,
"learning_rate": 2.8743682479584975e-06,
"loss": 0.0392,
"step": 26910
},
{
"grad_norm": 0.1885499209165573,
"learning_rate": 2.8559788540731824e-06,
"loss": 0.0384,
"step": 26920
},
{
"grad_norm": 0.19736219942569733,
"learning_rate": 2.8376467446189493e-06,
"loss": 0.0337,
"step": 26930
},
{
"grad_norm": 0.21519911289215088,
"learning_rate": 2.8193719418710406e-06,
"loss": 0.0362,
"step": 26940
},
{
"grad_norm": 0.1742061972618103,
"learning_rate": 2.8011544680350667e-06,
"loss": 0.0344,
"step": 26950
},
{
"grad_norm": 0.21678932011127472,
"learning_rate": 2.7829943452469753e-06,
"loss": 0.0333,
"step": 26960
},
{
"grad_norm": 0.20437230169773102,
"learning_rate": 2.7648915955730213e-06,
"loss": 0.0342,
"step": 26970
},
{
"grad_norm": 0.22159773111343384,
"learning_rate": 2.746846241009765e-06,
"loss": 0.0379,
"step": 26980
},
{
"grad_norm": 0.1510358601808548,
"learning_rate": 2.7288583034839945e-06,
"loss": 0.0365,
"step": 26990
},
{
"grad_norm": 0.23773546516895294,
"learning_rate": 2.7109278048527752e-06,
"loss": 0.0411,
"step": 27000
},
{
"grad_norm": 0.24329926073551178,
"learning_rate": 2.6930547669033413e-06,
"loss": 0.0341,
"step": 27010
},
{
"grad_norm": 0.2060578614473343,
"learning_rate": 2.675239211353109e-06,
"loss": 0.0318,
"step": 27020
},
{
"grad_norm": 0.20968450605869293,
"learning_rate": 2.6574811598496786e-06,
"loss": 0.0331,
"step": 27030
},
{
"grad_norm": 0.18646112084388733,
"learning_rate": 2.6397806339707456e-06,
"loss": 0.0365,
"step": 27040
},
{
"grad_norm": 0.1555396020412445,
"learning_rate": 2.6221376552241217e-06,
"loss": 0.0287,
"step": 27050
},
{
"grad_norm": 0.14469699561595917,
"learning_rate": 2.604552245047681e-06,
"loss": 0.0333,
"step": 27060
},
{
"grad_norm": 0.20040424168109894,
"learning_rate": 2.587024424809359e-06,
"loss": 0.0338,
"step": 27070
},
{
"grad_norm": 0.13866524398326874,
"learning_rate": 2.5695542158071186e-06,
"loss": 0.0334,
"step": 27080
},
{
"grad_norm": 0.21213102340698242,
"learning_rate": 2.5521416392689068e-06,
"loss": 0.0365,
"step": 27090
},
{
"grad_norm": 0.12232153117656708,
"learning_rate": 2.5347867163526384e-06,
"loss": 0.034,
"step": 27100
},
{
"grad_norm": 0.1966865211725235,
"learning_rate": 2.517489468146189e-06,
"loss": 0.0281,
"step": 27110
},
{
"grad_norm": 0.16872932016849518,
"learning_rate": 2.500249915667341e-06,
"loss": 0.0328,
"step": 27120
},
{
"grad_norm": 0.2221117615699768,
"learning_rate": 2.4830680798637817e-06,
"loss": 0.0314,
"step": 27130
},
{
"grad_norm": 0.1839471012353897,
"learning_rate": 2.465943981613056e-06,
"loss": 0.0291,
"step": 27140
},
{
"grad_norm": 0.22661961615085602,
"learning_rate": 2.448877641722569e-06,
"loss": 0.0295,
"step": 27150
},
{
"grad_norm": 0.16786102950572968,
"learning_rate": 2.431869080929522e-06,
"loss": 0.0294,
"step": 27160
},
{
"grad_norm": 0.2016066014766693,
"learning_rate": 2.4149183199009216e-06,
"loss": 0.0292,
"step": 27170
},
{
"grad_norm": 0.175208181142807,
"learning_rate": 2.3980253792335427e-06,
"loss": 0.0344,
"step": 27180
},
{
"grad_norm": 0.16838862001895905,
"learning_rate": 2.381190279453899e-06,
"loss": 0.0317,
"step": 27190
},
{
"grad_norm": 0.21897850930690765,
"learning_rate": 2.364413041018232e-06,
"loss": 0.0314,
"step": 27200
},
{
"grad_norm": 0.21111492812633514,
"learning_rate": 2.347693684312463e-06,
"loss": 0.032,
"step": 27210
},
{
"grad_norm": 0.20869970321655273,
"learning_rate": 2.331032229652186e-06,
"loss": 0.0321,
"step": 27220
},
{
"grad_norm": 0.1417779177427292,
"learning_rate": 2.314428697282628e-06,
"loss": 0.0328,
"step": 27230
},
{
"grad_norm": 0.2098666876554489,
"learning_rate": 2.297883107378673e-06,
"loss": 0.0307,
"step": 27240
},
{
"grad_norm": 0.14892929792404175,
"learning_rate": 2.2813954800447513e-06,
"loss": 0.0331,
"step": 27250
},
{
"grad_norm": 0.21149231493473053,
"learning_rate": 2.2649658353148974e-06,
"loss": 0.0347,
"step": 27260
},
{
"grad_norm": 0.2041424959897995,
"learning_rate": 2.2485941931526645e-06,
"loss": 0.0345,
"step": 27270
},
{
"grad_norm": 0.19133861362934113,
"learning_rate": 2.232280573451151e-06,
"loss": 0.0353,
"step": 27280
},
{
"grad_norm": 0.28061044216156006,
"learning_rate": 2.216024996032945e-06,
"loss": 0.0365,
"step": 27290
},
{
"grad_norm": 0.20739299058914185,
"learning_rate": 2.1998274806501074e-06,
"loss": 0.0362,
"step": 27300
},
{
"grad_norm": 0.23969867825508118,
"learning_rate": 2.183688046984139e-06,
"loss": 0.0379,
"step": 27310
},
{
"grad_norm": 0.18556906282901764,
"learning_rate": 2.167606714645981e-06,
"loss": 0.038,
"step": 27320
},
{
"grad_norm": 0.28039368987083435,
"learning_rate": 2.151583503175958e-06,
"loss": 0.0414,
"step": 27330
},
{
"grad_norm": 0.2219340205192566,
"learning_rate": 2.1356184320437957e-06,
"loss": 0.0367,
"step": 27340
},
{
"grad_norm": 0.23656632006168365,
"learning_rate": 2.119711520648554e-06,
"loss": 0.04,
"step": 27350
},
{
"grad_norm": 0.23218496143817902,
"learning_rate": 2.103862788318628e-06,
"loss": 0.0387,
"step": 27360
},
{
"grad_norm": 0.2153058648109436,
"learning_rate": 2.088072254311729e-06,
"loss": 0.0431,
"step": 27370
},
{
"grad_norm": 0.2309587150812149,
"learning_rate": 2.0723399378148435e-06,
"loss": 0.0325,
"step": 27380
},
{
"grad_norm": 0.2268764227628708,
"learning_rate": 2.0566658579442065e-06,
"loss": 0.0379,
"step": 27390
},
{
"grad_norm": 0.22586838901042938,
"learning_rate": 2.0410500337453176e-06,
"loss": 0.0413,
"step": 27400
},
{
"grad_norm": 0.1834147870540619,
"learning_rate": 2.0254924841928645e-06,
"loss": 0.0363,
"step": 27410
},
{
"grad_norm": 0.18279118835926056,
"learning_rate": 2.009993228190754e-06,
"loss": 0.0381,
"step": 27420
},
{
"grad_norm": 0.1710832566022873,
"learning_rate": 1.9945522845720323e-06,
"loss": 0.0319,
"step": 27430
},
{
"grad_norm": 0.17672070860862732,
"learning_rate": 1.9791696720988963e-06,
"loss": 0.0372,
"step": 27440
},
{
"grad_norm": 0.24893201887607574,
"learning_rate": 1.9638454094626833e-06,
"loss": 0.0373,
"step": 27450
},
{
"grad_norm": 0.15010783076286316,
"learning_rate": 1.9485795152838104e-06,
"loss": 0.0396,
"step": 27460
},
{
"grad_norm": 0.19107681512832642,
"learning_rate": 1.933372008111778e-06,
"loss": 0.0334,
"step": 27470
},
{
"grad_norm": 0.27231091260910034,
"learning_rate": 1.918222906425143e-06,
"loss": 0.0341,
"step": 27480
},
{
"grad_norm": 0.22653913497924805,
"learning_rate": 1.903132228631488e-06,
"loss": 0.034,
"step": 27490
},
{
"grad_norm": 0.300688773393631,
"learning_rate": 1.8880999930674215e-06,
"loss": 0.0418,
"step": 27500
},
{
"grad_norm": 0.15966728329658508,
"learning_rate": 1.8731262179985166e-06,
"loss": 0.035,
"step": 27510
},
{
"grad_norm": 0.2199893444776535,
"learning_rate": 1.8582109216193244e-06,
"loss": 0.0361,
"step": 27520
},
{
"grad_norm": 0.2506609559059143,
"learning_rate": 1.8433541220533368e-06,
"loss": 0.0311,
"step": 27530
},
{
"grad_norm": 0.21797513961791992,
"learning_rate": 1.8285558373529577e-06,
"loss": 0.035,
"step": 27540
},
{
"grad_norm": 0.3753775656223297,
"learning_rate": 1.8138160854995145e-06,
"loss": 0.0344,
"step": 27550
},
{
"grad_norm": 0.20994073152542114,
"learning_rate": 1.7991348844031864e-06,
"loss": 0.0332,
"step": 27560
},
{
"grad_norm": 0.21781712770462036,
"learning_rate": 1.78451225190302e-06,
"loss": 0.0319,
"step": 27570
},
{
"grad_norm": 0.22071443498134613,
"learning_rate": 1.7699482057668914e-06,
"loss": 0.0318,
"step": 27580
},
{
"grad_norm": 0.2253059297800064,
"learning_rate": 1.7554427636914895e-06,
"loss": 0.0297,
"step": 27590
},
{
"grad_norm": 0.18853643536567688,
"learning_rate": 1.7409959433022871e-06,
"loss": 0.0295,
"step": 27600
},
{
"grad_norm": 0.26162466406822205,
"learning_rate": 1.7266077621535481e-06,
"loss": 0.0282,
"step": 27610
},
{
"grad_norm": 0.21118517220020294,
"learning_rate": 1.7122782377282597e-06,
"loss": 0.0289,
"step": 27620
},
{
"grad_norm": 0.19400931894779205,
"learning_rate": 1.6980073874381497e-06,
"loss": 0.0276,
"step": 27630
},
{
"grad_norm": 0.22362148761749268,
"learning_rate": 1.6837952286236413e-06,
"loss": 0.0307,
"step": 27640
},
{
"grad_norm": 0.2128913551568985,
"learning_rate": 1.6696417785538488e-06,
"loss": 0.0312,
"step": 27650
},
{
"grad_norm": 0.2635255455970764,
"learning_rate": 1.6555470544265538e-06,
"loss": 0.0297,
"step": 27660
},
{
"grad_norm": 0.18413704633712769,
"learning_rate": 1.6415110733681738e-06,
"loss": 0.0307,
"step": 27670
},
{
"grad_norm": 0.2402702122926712,
"learning_rate": 1.6275338524337436e-06,
"loss": 0.0339,
"step": 27680
},
{
"grad_norm": 0.19961842894554138,
"learning_rate": 1.6136154086069056e-06,
"loss": 0.0283,
"step": 27690
},
{
"grad_norm": 0.21418826282024384,
"learning_rate": 1.5997557587998868e-06,
"loss": 0.0368,
"step": 27700
},
{
"grad_norm": 0.21191567182540894,
"learning_rate": 1.5859549198534606e-06,
"loss": 0.0317,
"step": 27710
},
{
"grad_norm": 0.19403713941574097,
"learning_rate": 1.572212908536963e-06,
"loss": 0.0304,
"step": 27720
},
{
"grad_norm": 0.15019190311431885,
"learning_rate": 1.5585297415482202e-06,
"loss": 0.0395,
"step": 27730
},
{
"grad_norm": 0.16130033135414124,
"learning_rate": 1.5449054355135717e-06,
"loss": 0.0324,
"step": 27740
},
{
"grad_norm": 0.19686198234558105,
"learning_rate": 1.5313400069878415e-06,
"loss": 0.0314,
"step": 27750
},
{
"grad_norm": 0.2224598079919815,
"learning_rate": 1.5178334724542887e-06,
"loss": 0.0325,
"step": 27760
},
{
"grad_norm": 0.18694256246089935,
"learning_rate": 1.504385848324641e-06,
"loss": 0.0377,
"step": 27770
},
{
"grad_norm": 0.2017853856086731,
"learning_rate": 1.490997150939033e-06,
"loss": 0.035,
"step": 27780
},
{
"grad_norm": 0.18776647746562958,
"learning_rate": 1.4776673965659793e-06,
"loss": 0.0355,
"step": 27790
},
{
"grad_norm": 0.15866783261299133,
"learning_rate": 1.4643966014023957e-06,
"loss": 0.0329,
"step": 27800
},
{
"grad_norm": 0.20602932572364807,
"learning_rate": 1.4511847815735503e-06,
"loss": 0.0388,
"step": 27810
},
{
"grad_norm": 0.17626391351222992,
"learning_rate": 1.438031953133051e-06,
"loss": 0.0355,
"step": 27820
},
{
"grad_norm": 0.187279611825943,
"learning_rate": 1.4249381320628197e-06,
"loss": 0.0339,
"step": 27830
},
{
"grad_norm": 0.1761024296283722,
"learning_rate": 1.4119033342730902e-06,
"loss": 0.034,
"step": 27840
},
{
"grad_norm": 0.20913295447826385,
"learning_rate": 1.3989275756023656e-06,
"loss": 0.0361,
"step": 27850
},
{
"grad_norm": 0.1689627766609192,
"learning_rate": 1.386010871817417e-06,
"loss": 0.03,
"step": 27860
},
{
"grad_norm": 0.18365339934825897,
"learning_rate": 1.3731532386132616e-06,
"loss": 0.0373,
"step": 27870
},
{
"grad_norm": 0.19457945227622986,
"learning_rate": 1.360354691613136e-06,
"loss": 0.0322,
"step": 27880
},
{
"grad_norm": 0.18660353124141693,
"learning_rate": 1.3476152463684776e-06,
"loss": 0.0398,
"step": 27890
},
{
"grad_norm": 0.2751769721508026,
"learning_rate": 1.3349349183589155e-06,
"loss": 0.0338,
"step": 27900
},
{
"grad_norm": 0.23868528008460999,
"learning_rate": 1.3223137229922356e-06,
"loss": 0.0374,
"step": 27910
},
{
"grad_norm": 0.20784129202365875,
"learning_rate": 1.3097516756043981e-06,
"loss": 0.0381,
"step": 27920
},
{
"grad_norm": 0.2453920990228653,
"learning_rate": 1.29724879145946e-06,
"loss": 0.0371,
"step": 27930
},
{
"grad_norm": 0.1552310287952423,
"learning_rate": 1.284805085749613e-06,
"loss": 0.0354,
"step": 27940
},
{
"grad_norm": 0.18091744184494019,
"learning_rate": 1.2724205735951288e-06,
"loss": 0.0406,
"step": 27950
},
{
"grad_norm": 0.14014825224876404,
"learning_rate": 1.2600952700443591e-06,
"loss": 0.0321,
"step": 27960
},
{
"grad_norm": 0.1746359020471573,
"learning_rate": 1.247829190073707e-06,
"loss": 0.0342,
"step": 27970
},
{
"grad_norm": 0.2595737874507904,
"learning_rate": 1.2356223485876173e-06,
"loss": 0.0346,
"step": 27980
},
{
"grad_norm": 0.18114635348320007,
"learning_rate": 1.2234747604185526e-06,
"loss": 0.0327,
"step": 27990
},
{
"grad_norm": 0.1816367208957672,
"learning_rate": 1.2113864403269836e-06,
"loss": 0.0324,
"step": 28000
},
{
"grad_norm": 0.21650968492031097,
"learning_rate": 1.1993574030013555e-06,
"loss": 0.0355,
"step": 28010
},
{
"grad_norm": 0.24884501099586487,
"learning_rate": 1.1873876630580816e-06,
"loss": 0.0387,
"step": 28020
},
{
"grad_norm": 0.165457084774971,
"learning_rate": 1.1754772350415278e-06,
"loss": 0.0326,
"step": 28030
},
{
"grad_norm": 0.20390920341014862,
"learning_rate": 1.163626133423984e-06,
"loss": 0.0352,
"step": 28040
},
{
"grad_norm": 0.23500585556030273,
"learning_rate": 1.1518343726056591e-06,
"loss": 0.0304,
"step": 28050
},
{
"grad_norm": 0.20029349625110626,
"learning_rate": 1.1401019669146474e-06,
"loss": 0.0307,
"step": 28060
},
{
"grad_norm": 0.1903732568025589,
"learning_rate": 1.128428930606934e-06,
"loss": 0.0323,
"step": 28070
},
{
"grad_norm": 0.1861230731010437,
"learning_rate": 1.1168152778663621e-06,
"loss": 0.034,
"step": 28080
},
{
"grad_norm": 0.20783047378063202,
"learning_rate": 1.10526102280461e-06,
"loss": 0.0332,
"step": 28090
},
{
"grad_norm": 0.18503524363040924,
"learning_rate": 1.0937661794611864e-06,
"loss": 0.0333,
"step": 28100
},
{
"grad_norm": 0.1878349334001541,
"learning_rate": 1.082330761803413e-06,
"loss": 0.033,
"step": 28110
},
{
"grad_norm": 0.20222771167755127,
"learning_rate": 1.0709547837263966e-06,
"loss": 0.0344,
"step": 28120
},
{
"grad_norm": 0.1854458749294281,
"learning_rate": 1.059638259053025e-06,
"loss": 0.0353,
"step": 28130
},
{
"grad_norm": 0.18822869658470154,
"learning_rate": 1.0483812015339434e-06,
"loss": 0.0288,
"step": 28140
},
{
"grad_norm": 0.17033694684505463,
"learning_rate": 1.037183624847543e-06,
"loss": 0.0345,
"step": 28150
},
{
"grad_norm": 0.1799994260072708,
"learning_rate": 1.0260455425999238e-06,
"loss": 0.0298,
"step": 28160
},
{
"grad_norm": 0.15347857773303986,
"learning_rate": 1.0149669683249208e-06,
"loss": 0.0323,
"step": 28170
},
{
"grad_norm": 0.29782533645629883,
"learning_rate": 1.003947915484027e-06,
"loss": 0.0345,
"step": 28180
},
{
"grad_norm": 0.17779317498207092,
"learning_rate": 9.92988397466449e-07,
"loss": 0.0321,
"step": 28190
},
{
"grad_norm": 0.16707853972911835,
"learning_rate": 9.820884275890286e-07,
"loss": 0.0344,
"step": 28200
},
{
"grad_norm": 0.18473440408706665,
"learning_rate": 9.712480190962546e-07,
"loss": 0.0337,
"step": 28210
},
{
"grad_norm": 0.2282411754131317,
"learning_rate": 9.604671851602464e-07,
"loss": 0.0346,
"step": 28220
},
{
"grad_norm": 0.17498600482940674,
"learning_rate": 9.497459388807306e-07,
"loss": 0.0256,
"step": 28230
},
{
"grad_norm": 0.16003865003585815,
"learning_rate": 9.390842932850364e-07,
"loss": 0.0263,
"step": 28240
},
{
"grad_norm": 0.1881251484155655,
"learning_rate": 9.28482261328073e-07,
"loss": 0.03,
"step": 28250
},
{
"grad_norm": 0.20345298945903778,
"learning_rate": 9.179398558923025e-07,
"loss": 0.0305,
"step": 28260
},
{
"grad_norm": 0.24660596251487732,
"learning_rate": 9.074570897877388e-07,
"loss": 0.0299,
"step": 28270
},
{
"grad_norm": 0.21088597178459167,
"learning_rate": 8.970339757519375e-07,
"loss": 0.0266,
"step": 28280
},
{
"grad_norm": 0.22768820822238922,
"learning_rate": 8.866705264499619e-07,
"loss": 0.0322,
"step": 28290
},
{
"grad_norm": 0.20227080583572388,
"learning_rate": 8.763667544743836e-07,
"loss": 0.0325,
"step": 28300
},
{
"grad_norm": 0.2542444169521332,
"learning_rate": 8.661226723452543e-07,
"loss": 0.032,
"step": 28310
},
{
"grad_norm": 0.18932412564754486,
"learning_rate": 8.559382925101001e-07,
"loss": 0.0303,
"step": 28320
},
{
"grad_norm": 0.28206169605255127,
"learning_rate": 8.458136273438943e-07,
"loss": 0.0344,
"step": 28330
},
{
"grad_norm": 0.20444028079509735,
"learning_rate": 8.357486891490795e-07,
"loss": 0.0284,
"step": 28340
},
{
"grad_norm": 0.18067412078380585,
"learning_rate": 8.257434901554895e-07,
"loss": 0.0294,
"step": 28350
},
{
"grad_norm": 0.25492680072784424,
"learning_rate": 8.157980425203937e-07,
"loss": 0.03,
"step": 28360
},
{
"grad_norm": 0.19316114485263824,
"learning_rate": 8.059123583284367e-07,
"loss": 0.0319,
"step": 28370
},
{
"grad_norm": 0.2045619785785675,
"learning_rate": 7.960864495916654e-07,
"loss": 0.0279,
"step": 28380
},
{
"grad_norm": 0.1863790899515152,
"learning_rate": 7.863203282494847e-07,
"loss": 0.0277,
"step": 28390
},
{
"grad_norm": 0.20044109225273132,
"learning_rate": 7.766140061686522e-07,
"loss": 0.0323,
"step": 28400
},
{
"grad_norm": 0.20885857939720154,
"learning_rate": 7.669674951432615e-07,
"loss": 0.0316,
"step": 28410
},
{
"grad_norm": 0.1598440408706665,
"learning_rate": 7.573808068947363e-07,
"loss": 0.0312,
"step": 28420
},
{
"grad_norm": 0.17936713993549347,
"learning_rate": 7.478539530718087e-07,
"loss": 0.032,
"step": 28430
},
{
"grad_norm": 0.18919233977794647,
"learning_rate": 7.383869452504965e-07,
"loss": 0.0336,
"step": 28440
},
{
"grad_norm": 0.1603318452835083,
"learning_rate": 7.289797949341204e-07,
"loss": 0.0303,
"step": 28450
},
{
"grad_norm": 0.1829591542482376,
"learning_rate": 7.196325135532423e-07,
"loss": 0.035,
"step": 28460
},
{
"grad_norm": 0.18491652607917786,
"learning_rate": 7.10345112465699e-07,
"loss": 0.0306,
"step": 28470
},
{
"grad_norm": 0.2358941286802292,
"learning_rate": 7.011176029565525e-07,
"loss": 0.0368,
"step": 28480
},
{
"grad_norm": 0.23379741609096527,
"learning_rate": 6.919499962381004e-07,
"loss": 0.04,
"step": 28490
},
{
"grad_norm": 0.2433539777994156,
"learning_rate": 6.828423034498488e-07,
"loss": 0.0347,
"step": 28500
},
{
"grad_norm": 0.2050427496433258,
"learning_rate": 6.737945356585007e-07,
"loss": 0.0376,
"step": 28510
},
{
"grad_norm": 0.1672799289226532,
"learning_rate": 6.648067038579508e-07,
"loss": 0.0358,
"step": 28520
},
{
"grad_norm": 0.1557297706604004,
"learning_rate": 6.558788189692578e-07,
"loss": 0.0346,
"step": 28530
},
{
"grad_norm": 0.2391682118177414,
"learning_rate": 6.470108918406492e-07,
"loss": 0.0318,
"step": 28540
},
{
"grad_norm": 0.23948504030704498,
"learning_rate": 6.382029332474893e-07,
"loss": 0.0326,
"step": 28550
},
{
"grad_norm": 0.19147975742816925,
"learning_rate": 6.294549538922778e-07,
"loss": 0.028,
"step": 28560
},
{
"grad_norm": 0.1760636568069458,
"learning_rate": 6.207669644046344e-07,
"loss": 0.029,
"step": 28570
},
{
"grad_norm": 0.1910676658153534,
"learning_rate": 6.121389753412865e-07,
"loss": 0.0285,
"step": 28580
},
{
"grad_norm": 0.15190190076828003,
"learning_rate": 6.035709971860592e-07,
"loss": 0.0288,
"step": 28590
},
{
"grad_norm": 0.19925153255462646,
"learning_rate": 5.950630403498469e-07,
"loss": 0.0297,
"step": 28600
},
{
"grad_norm": 0.15815521776676178,
"learning_rate": 5.866151151706189e-07,
"loss": 0.0302,
"step": 28610
},
{
"grad_norm": 0.21183480322360992,
"learning_rate": 5.782272319134086e-07,
"loss": 0.0362,
"step": 28620
},
{
"grad_norm": 0.20627401769161224,
"learning_rate": 5.698994007702796e-07,
"loss": 0.0303,
"step": 28630
},
{
"grad_norm": 0.20052334666252136,
"learning_rate": 5.616316318603321e-07,
"loss": 0.0326,
"step": 28640
},
{
"grad_norm": 0.2002720683813095,
"learning_rate": 5.534239352296799e-07,
"loss": 0.0314,
"step": 28650
},
{
"grad_norm": 0.15933628380298615,
"learning_rate": 5.452763208514621e-07,
"loss": 0.0301,
"step": 28660
},
{
"grad_norm": 0.17894089221954346,
"learning_rate": 5.371887986257873e-07,
"loss": 0.0286,
"step": 28670
},
{
"grad_norm": 0.19548213481903076,
"learning_rate": 5.291613783797611e-07,
"loss": 0.0322,
"step": 28680
},
{
"grad_norm": 0.19147153198719025,
"learning_rate": 5.211940698674534e-07,
"loss": 0.0292,
"step": 28690
},
{
"grad_norm": 0.1998213827610016,
"learning_rate": 5.132868827698978e-07,
"loss": 0.0288,
"step": 28700
},
{
"grad_norm": 0.14438310265541077,
"learning_rate": 5.054398266950755e-07,
"loss": 0.0262,
"step": 28710
},
{
"grad_norm": 0.18305571377277374,
"learning_rate": 4.976529111778872e-07,
"loss": 0.0296,
"step": 28720
},
{
"grad_norm": 0.18567264080047607,
"learning_rate": 4.899261456801862e-07,
"loss": 0.0245,
"step": 28730
},
{
"grad_norm": 0.2223803848028183,
"learning_rate": 4.822595395907126e-07,
"loss": 0.0266,
"step": 28740
},
{
"grad_norm": 0.18931105732917786,
"learning_rate": 4.7465310222510886e-07,
"loss": 0.0261,
"step": 28750
},
{
"grad_norm": 0.180213063955307,
"learning_rate": 4.6710684282593175e-07,
"loss": 0.0311,
"step": 28760
},
{
"grad_norm": 0.20747826993465424,
"learning_rate": 4.5962077056257993e-07,
"loss": 0.0306,
"step": 28770
},
{
"grad_norm": 0.22823959589004517,
"learning_rate": 4.521948945313492e-07,
"loss": 0.0337,
"step": 28780
},
{
"grad_norm": 0.17368264496326447,
"learning_rate": 4.4482922375537195e-07,
"loss": 0.0326,
"step": 28790
},
{
"grad_norm": 0.1953117996454239,
"learning_rate": 4.375237671846333e-07,
"loss": 0.0323,
"step": 28800
},
{
"grad_norm": 0.22263897955417633,
"learning_rate": 4.302785336959547e-07,
"loss": 0.0344,
"step": 28810
},
{
"grad_norm": 0.24437640607357025,
"learning_rate": 4.2309353209297744e-07,
"loss": 0.0295,
"step": 28820
},
{
"grad_norm": 0.18094122409820557,
"learning_rate": 4.159687711061566e-07,
"loss": 0.0302,
"step": 28830
},
{
"grad_norm": 0.19272062182426453,
"learning_rate": 4.0890425939275055e-07,
"loss": 0.0281,
"step": 28840
},
{
"grad_norm": 0.21651127934455872,
"learning_rate": 4.0190000553679827e-07,
"loss": 0.0354,
"step": 28850
},
{
"grad_norm": 0.2156757116317749,
"learning_rate": 3.9495601804913627e-07,
"loss": 0.0322,
"step": 28860
},
{
"grad_norm": 0.21865883469581604,
"learning_rate": 3.880723053673652e-07,
"loss": 0.0375,
"step": 28870
},
{
"grad_norm": 0.17753100395202637,
"learning_rate": 3.812488758558386e-07,
"loss": 0.0322,
"step": 28880
},
{
"grad_norm": 0.1433328241109848,
"learning_rate": 3.744857378056743e-07,
"loss": 0.031,
"step": 28890
},
{
"grad_norm": 0.17325064539909363,
"learning_rate": 3.677828994347154e-07,
"loss": 0.0328,
"step": 28900
},
{
"grad_norm": 0.207637220621109,
"learning_rate": 3.61140368887547e-07,
"loss": 0.0375,
"step": 28910
},
{
"grad_norm": 0.22459979355335236,
"learning_rate": 3.545581542354681e-07,
"loss": 0.0362,
"step": 28920
},
{
"grad_norm": 0.14907418191432953,
"learning_rate": 3.480362634764922e-07,
"loss": 0.0342,
"step": 28930
},
{
"grad_norm": 0.19613823294639587,
"learning_rate": 3.4157470453533014e-07,
"loss": 0.0308,
"step": 28940
},
{
"grad_norm": 0.1955443173646927,
"learning_rate": 3.3517348526339034e-07,
"loss": 0.0327,
"step": 28950
},
{
"grad_norm": 0.21833966672420502,
"learning_rate": 3.288326134387454e-07,
"loss": 0.0348,
"step": 28960
},
{
"grad_norm": 0.20331119000911713,
"learning_rate": 3.225520967661655e-07,
"loss": 0.0355,
"step": 28970
},
{
"grad_norm": 0.1821715235710144,
"learning_rate": 3.163319428770628e-07,
"loss": 0.033,
"step": 28980
},
{
"grad_norm": 0.2200372964143753,
"learning_rate": 3.1017215932951374e-07,
"loss": 0.0371,
"step": 28990
},
{
"grad_norm": 0.15630647540092468,
"learning_rate": 3.040727536082366e-07,
"loss": 0.0345,
"step": 29000
},
{
"grad_norm": 0.1526121348142624,
"learning_rate": 2.980337331245864e-07,
"loss": 0.0357,
"step": 29010
},
{
"grad_norm": 0.15940600633621216,
"learning_rate": 2.9205510521653213e-07,
"loss": 0.037,
"step": 29020
},
{
"grad_norm": 0.18748238682746887,
"learning_rate": 2.86136877148685e-07,
"loss": 0.03,
"step": 29030
},
{
"grad_norm": 0.2448747605085373,
"learning_rate": 2.8027905611223704e-07,
"loss": 0.036,
"step": 29040
},
{
"grad_norm": 0.19934611022472382,
"learning_rate": 2.7448164922500573e-07,
"loss": 0.0364,
"step": 29050
},
{
"grad_norm": 0.2184101790189743,
"learning_rate": 2.687446635313784e-07,
"loss": 0.041,
"step": 29060
},
{
"grad_norm": 0.16742043197155,
"learning_rate": 2.630681060023343e-07,
"loss": 0.0361,
"step": 29070
},
{
"grad_norm": 0.2069026231765747,
"learning_rate": 2.5745198353542833e-07,
"loss": 0.0369,
"step": 29080
},
{
"grad_norm": 0.19885164499282837,
"learning_rate": 2.518963029547794e-07,
"loss": 0.0403,
"step": 29090
},
{
"grad_norm": 0.1839226335287094,
"learning_rate": 2.464010710110598e-07,
"loss": 0.0333,
"step": 29100
},
{
"grad_norm": 0.28503116965293884,
"learning_rate": 2.4096629438150054e-07,
"loss": 0.0389,
"step": 29110
},
{
"grad_norm": 0.19080856442451477,
"learning_rate": 2.3559197966985802e-07,
"loss": 0.0374,
"step": 29120
},
{
"grad_norm": 0.22954346239566803,
"learning_rate": 2.3027813340644188e-07,
"loss": 0.036,
"step": 29130
},
{
"grad_norm": 0.24877487123012543,
"learning_rate": 2.2502476204807055e-07,
"loss": 0.0359,
"step": 29140
},
{
"grad_norm": 0.1800680309534073,
"learning_rate": 2.1983187197808786e-07,
"loss": 0.0381,
"step": 29150
},
{
"grad_norm": 0.18487083911895752,
"learning_rate": 2.1469946950634644e-07,
"loss": 0.0362,
"step": 29160
},
{
"grad_norm": 0.22050683200359344,
"learning_rate": 2.096275608691911e-07,
"loss": 0.0373,
"step": 29170
},
{
"grad_norm": 0.21246419847011566,
"learning_rate": 2.046161522294754e-07,
"loss": 0.0355,
"step": 29180
},
{
"grad_norm": 0.1887367069721222,
"learning_rate": 1.9966524967653944e-07,
"loss": 0.0358,
"step": 29190
},
{
"grad_norm": 0.17115497589111328,
"learning_rate": 1.9477485922618222e-07,
"loss": 0.0351,
"step": 29200
},
{
"grad_norm": 0.18320408463478088,
"learning_rate": 1.899449868207004e-07,
"loss": 0.0391,
"step": 29210
},
{
"grad_norm": 0.16281452775001526,
"learning_rate": 1.851756383288439e-07,
"loss": 0.0379,
"step": 29220
},
{
"grad_norm": 0.23046189546585083,
"learning_rate": 1.8046681954581035e-07,
"loss": 0.0386,
"step": 29230
},
{
"grad_norm": 0.1920708417892456,
"learning_rate": 1.7581853619327294e-07,
"loss": 0.0393,
"step": 29240
},
{
"grad_norm": 0.1852489858865738,
"learning_rate": 1.7123079391932472e-07,
"loss": 0.0326,
"step": 29250
},
{
"grad_norm": 0.2059720754623413,
"learning_rate": 1.6670359829850657e-07,
"loss": 0.0373,
"step": 29260
},
{
"grad_norm": 0.18409974873065948,
"learning_rate": 1.6223695483179037e-07,
"loss": 0.0385,
"step": 29270
},
{
"grad_norm": 0.20481891930103302,
"learning_rate": 1.5783086894656795e-07,
"loss": 0.0299,
"step": 29280
},
{
"grad_norm": 0.20854105055332184,
"learning_rate": 1.5348534599665121e-07,
"loss": 0.0369,
"step": 29290
},
{
"grad_norm": 0.1993677020072937,
"learning_rate": 1.4920039126225527e-07,
"loss": 0.0327,
"step": 29300
},
{
"grad_norm": 0.17689667642116547,
"learning_rate": 1.449760099500097e-07,
"loss": 0.0337,
"step": 29310
},
{
"grad_norm": 0.20842567086219788,
"learning_rate": 1.4081220719293075e-07,
"loss": 0.0323,
"step": 29320
},
{
"grad_norm": 0.20056729018688202,
"learning_rate": 1.3670898805043797e-07,
"loss": 0.0365,
"step": 29330
},
{
"grad_norm": 0.22962814569473267,
"learning_rate": 1.326663575083209e-07,
"loss": 0.0368,
"step": 29340
},
{
"grad_norm": 0.18686366081237793,
"learning_rate": 1.2868432047876688e-07,
"loss": 0.0332,
"step": 29350
},
{
"grad_norm": 0.2225220650434494,
"learning_rate": 1.2476288180032213e-07,
"loss": 0.0314,
"step": 29360
},
{
"grad_norm": 0.18584488332271576,
"learning_rate": 1.209020462379029e-07,
"loss": 0.0279,
"step": 29370
},
{
"grad_norm": 0.2154630571603775,
"learning_rate": 1.1710181848278435e-07,
"loss": 0.0376,
"step": 29380
},
{
"grad_norm": 0.25119099020957947,
"learning_rate": 1.133622031526116e-07,
"loss": 0.0352,
"step": 29390
},
{
"grad_norm": 0.27894043922424316,
"learning_rate": 1.0968320479136095e-07,
"loss": 0.0366,
"step": 29400
},
{
"grad_norm": 0.21918430924415588,
"learning_rate": 1.0606482786936767e-07,
"loss": 0.0352,
"step": 29410
},
{
"grad_norm": 0.2265995293855667,
"learning_rate": 1.0250707678329808e-07,
"loss": 0.0316,
"step": 29420
},
{
"grad_norm": 0.2396053820848465,
"learning_rate": 9.900995585615525e-08,
"loss": 0.0359,
"step": 29430
},
{
"grad_norm": 0.16162151098251343,
"learning_rate": 9.55734693372734e-08,
"loss": 0.0339,
"step": 29440
},
{
"grad_norm": 0.2057119607925415,
"learning_rate": 9.219762140231236e-08,
"loss": 0.0355,
"step": 29450
},
{
"grad_norm": 0.2744728922843933,
"learning_rate": 8.888241615322978e-08,
"loss": 0.0386,
"step": 29460
},
{
"grad_norm": 0.15655767917633057,
"learning_rate": 8.562785761833114e-08,
"loss": 0.0306,
"step": 29470
},
{
"grad_norm": 0.2418740838766098,
"learning_rate": 8.243394975219753e-08,
"loss": 0.0336,
"step": 29480
},
{
"grad_norm": 0.20268963277339935,
"learning_rate": 7.930069643573568e-08,
"loss": 0.0316,
"step": 29490
},
{
"grad_norm": 0.232268825173378,
"learning_rate": 7.622810147614456e-08,
"loss": 0.0361,
"step": 29500
},
{
"grad_norm": 0.2125527560710907,
"learning_rate": 7.321616860690995e-08,
"loss": 0.0374,
"step": 29510
},
{
"grad_norm": 0.19314096868038177,
"learning_rate": 7.026490148782095e-08,
"loss": 0.0354,
"step": 29520
},
{
"grad_norm": 0.22315093874931335,
"learning_rate": 6.737430370494236e-08,
"loss": 0.0353,
"step": 29530
},
{
"grad_norm": 0.2335837483406067,
"learning_rate": 6.454437877062569e-08,
"loss": 0.0346,
"step": 29540
},
{
"grad_norm": 0.24504299461841583,
"learning_rate": 6.177513012349256e-08,
"loss": 0.0342,
"step": 29550
},
{
"grad_norm": 0.22245462238788605,
"learning_rate": 5.9066561128445775e-08,
"loss": 0.035,
"step": 29560
},
{
"grad_norm": 0.1679915338754654,
"learning_rate": 5.6418675076641556e-08,
"loss": 0.0358,
"step": 29570
},
{
"grad_norm": 0.25395259261131287,
"learning_rate": 5.383147518552845e-08,
"loss": 0.0369,
"step": 29580
},
{
"grad_norm": 0.16364385187625885,
"learning_rate": 5.1304964598786196e-08,
"loss": 0.0361,
"step": 29590
},
{
"grad_norm": 0.20032702386379242,
"learning_rate": 4.883914638636467e-08,
"loss": 0.0331,
"step": 29600
},
{
"grad_norm": 0.16169103980064392,
"learning_rate": 4.643402354446713e-08,
"loss": 0.0345,
"step": 29610
},
{
"grad_norm": 0.1847226917743683,
"learning_rate": 4.4089598995544766e-08,
"loss": 0.0334,
"step": 29620
},
{
"grad_norm": 0.1536235362291336,
"learning_rate": 4.180587558829663e-08,
"loss": 0.0344,
"step": 29630
},
{
"grad_norm": 0.21013644337654114,
"learning_rate": 3.958285609765855e-08,
"loss": 0.0283,
"step": 29640
},
{
"grad_norm": 0.20534594357013702,
"learning_rate": 3.74205432248087e-08,
"loss": 0.0377,
"step": 29650
},
{
"grad_norm": 0.21907639503479004,
"learning_rate": 3.531893959716204e-08,
"loss": 0.0396,
"step": 29660
},
{
"grad_norm": 0.2088647335767746,
"learning_rate": 3.327804776837029e-08,
"loss": 0.0359,
"step": 29670
},
{
"grad_norm": 0.24326613545417786,
"learning_rate": 3.129787021829977e-08,
"loss": 0.0371,
"step": 29680
},
{
"grad_norm": 0.19639475643634796,
"learning_rate": 2.9378409353059133e-08,
"loss": 0.0328,
"step": 29690
},
{
"grad_norm": 0.21212221682071686,
"learning_rate": 2.7519667504971593e-08,
"loss": 0.0334,
"step": 29700
},
{
"grad_norm": 0.22771959006786346,
"learning_rate": 2.572164693258605e-08,
"loss": 0.0393,
"step": 29710
},
{
"grad_norm": 0.20089668035507202,
"learning_rate": 2.3984349820665996e-08,
"loss": 0.0357,
"step": 29720
},
{
"grad_norm": 0.20129364728927612,
"learning_rate": 2.2307778280189485e-08,
"loss": 0.0384,
"step": 29730
},
{
"grad_norm": 0.24074164032936096,
"learning_rate": 2.0691934348354704e-08,
"loss": 0.0403,
"step": 29740
},
{
"grad_norm": 0.2764643430709839,
"learning_rate": 1.9136819988557763e-08,
"loss": 0.0371,
"step": 29750
},
{
"grad_norm": 0.23757904767990112,
"learning_rate": 1.7642437090414908e-08,
"loss": 0.0388,
"step": 29760
},
{
"grad_norm": 0.2216019481420517,
"learning_rate": 1.6208787469734755e-08,
"loss": 0.0324,
"step": 29770
},
{
"grad_norm": 0.2253427952528,
"learning_rate": 1.483587286854604e-08,
"loss": 0.0319,
"step": 29780
},
{
"grad_norm": 0.24056778848171234,
"learning_rate": 1.3523694955064336e-08,
"loss": 0.0339,
"step": 29790
},
{
"grad_norm": 0.2266574651002884,
"learning_rate": 1.2272255323708681e-08,
"loss": 0.0323,
"step": 29800
},
{
"grad_norm": 0.21901734173297882,
"learning_rate": 1.1081555495096042e-08,
"loss": 0.0294,
"step": 29810
},
{
"grad_norm": 0.20699453353881836,
"learning_rate": 9.951596916041306e-09,
"loss": 0.0343,
"step": 29820
},
{
"grad_norm": 0.1703358143568039,
"learning_rate": 8.882380959551739e-09,
"loss": 0.0262,
"step": 29830
},
{
"grad_norm": 0.1927197277545929,
"learning_rate": 7.873908924821428e-09,
"loss": 0.0334,
"step": 29840
},
{
"grad_norm": 0.2077469676733017,
"learning_rate": 6.926182037242379e-09,
"loss": 0.0334,
"step": 29850
},
{
"grad_norm": 0.17766863107681274,
"learning_rate": 6.039201448387876e-09,
"loss": 0.032,
"step": 29860
},
{
"grad_norm": 0.1982751041650772,
"learning_rate": 5.212968236029125e-09,
"loss": 0.0363,
"step": 29870
},
{
"grad_norm": 0.20739732682704926,
"learning_rate": 4.447483404118602e-09,
"loss": 0.0344,
"step": 29880
},
{
"grad_norm": 0.20854595303535461,
"learning_rate": 3.742747882784503e-09,
"loss": 0.0327,
"step": 29890
},
{
"grad_norm": 0.18219459056854248,
"learning_rate": 3.0987625283473987e-09,
"loss": 0.0393,
"step": 29900
},
{
"grad_norm": 0.22119903564453125,
"learning_rate": 2.515528123320232e-09,
"loss": 0.0372,
"step": 29910
},
{
"grad_norm": 0.20093749463558197,
"learning_rate": 1.9930453763750133e-09,
"loss": 0.0354,
"step": 29920
},
{
"grad_norm": 0.20442412793636322,
"learning_rate": 1.5313149223872281e-09,
"loss": 0.035,
"step": 29930
},
{
"grad_norm": 0.22766539454460144,
"learning_rate": 1.1303373224025305e-09,
"loss": 0.04,
"step": 29940
},
{
"grad_norm": 0.25613829493522644,
"learning_rate": 7.901130636367438e-10,
"loss": 0.0386,
"step": 29950
},
{
"grad_norm": 0.25260430574417114,
"learning_rate": 5.106425595036158e-10,
"loss": 0.0419,
"step": 29960
},
{
"grad_norm": 0.22208313643932343,
"learning_rate": 2.9192614958706286e-10,
"loss": 0.0413,
"step": 29970
},
{
"grad_norm": 0.2517436146736145,
"learning_rate": 1.3396409964117064e-10,
"loss": 0.0377,
"step": 29980
},
{
"grad_norm": 0.23551464080810547,
"learning_rate": 3.6756601606846574e-11,
"loss": 0.0363,
"step": 29990
},
{
"grad_norm": 0.20905114710330963,
"learning_rate": 3.0377360626943076e-13,
"loss": 0.038,
"step": 30000
}
],
"logging_steps": 10,
"max_steps": 30000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}