upper_body_60-task_cls / trainer_state.json
gubam's picture
Upload folder using huggingface_hub
f460142 verified
Raw
History Blame Contribute Delete
420 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": 1.1163818836212158,
"learning_rate": 6.000000000000001e-07,
"loss": 1.3241,
"step": 10
},
{
"grad_norm": 1.1505200862884521,
"learning_rate": 1.2666666666666667e-06,
"loss": 1.3212,
"step": 20
},
{
"grad_norm": 1.0313507318496704,
"learning_rate": 1.9333333333333336e-06,
"loss": 1.3077,
"step": 30
},
{
"grad_norm": 0.8380591869354248,
"learning_rate": 2.6e-06,
"loss": 1.2963,
"step": 40
},
{
"grad_norm": 0.7542089819908142,
"learning_rate": 3.2666666666666666e-06,
"loss": 1.2766,
"step": 50
},
{
"grad_norm": 0.7316833138465881,
"learning_rate": 3.9333333333333335e-06,
"loss": 1.2423,
"step": 60
},
{
"grad_norm": 0.9064944982528687,
"learning_rate": 4.6e-06,
"loss": 1.1995,
"step": 70
},
{
"grad_norm": 0.6927876472473145,
"learning_rate": 5.266666666666667e-06,
"loss": 1.162,
"step": 80
},
{
"grad_norm": 0.8706201910972595,
"learning_rate": 5.933333333333334e-06,
"loss": 1.1388,
"step": 90
},
{
"grad_norm": 0.6237123012542725,
"learning_rate": 6.6e-06,
"loss": 1.1191,
"step": 100
},
{
"grad_norm": 0.6539652943611145,
"learning_rate": 7.266666666666668e-06,
"loss": 1.1177,
"step": 110
},
{
"grad_norm": 0.48477816581726074,
"learning_rate": 7.933333333333334e-06,
"loss": 1.1094,
"step": 120
},
{
"grad_norm": 0.4369802474975586,
"learning_rate": 8.599999999999999e-06,
"loss": 1.1107,
"step": 130
},
{
"grad_norm": 0.6129502058029175,
"learning_rate": 9.266666666666667e-06,
"loss": 1.0988,
"step": 140
},
{
"grad_norm": 0.5187477469444275,
"learning_rate": 9.933333333333334e-06,
"loss": 1.0965,
"step": 150
},
{
"grad_norm": 0.5889061689376831,
"learning_rate": 1.06e-05,
"loss": 1.0774,
"step": 160
},
{
"grad_norm": 0.48414167761802673,
"learning_rate": 1.1266666666666667e-05,
"loss": 1.0685,
"step": 170
},
{
"grad_norm": 0.5607510805130005,
"learning_rate": 1.1933333333333333e-05,
"loss": 1.0732,
"step": 180
},
{
"grad_norm": 0.7136213779449463,
"learning_rate": 1.2600000000000001e-05,
"loss": 1.0669,
"step": 190
},
{
"grad_norm": 0.5070462226867676,
"learning_rate": 1.3266666666666666e-05,
"loss": 1.0525,
"step": 200
},
{
"grad_norm": 0.6223626732826233,
"learning_rate": 1.3933333333333334e-05,
"loss": 1.056,
"step": 210
},
{
"grad_norm": 0.580681562423706,
"learning_rate": 1.4599999999999999e-05,
"loss": 1.0478,
"step": 220
},
{
"grad_norm": 0.6932505369186401,
"learning_rate": 1.5266666666666667e-05,
"loss": 1.0541,
"step": 230
},
{
"grad_norm": 0.5870380997657776,
"learning_rate": 1.5933333333333332e-05,
"loss": 1.0479,
"step": 240
},
{
"grad_norm": 0.6588231921195984,
"learning_rate": 1.66e-05,
"loss": 1.0432,
"step": 250
},
{
"grad_norm": 0.6166058778762817,
"learning_rate": 1.726666666666667e-05,
"loss": 1.0441,
"step": 260
},
{
"grad_norm": 0.49625009298324585,
"learning_rate": 1.7933333333333337e-05,
"loss": 1.0407,
"step": 270
},
{
"grad_norm": 0.6466280221939087,
"learning_rate": 1.86e-05,
"loss": 1.047,
"step": 280
},
{
"grad_norm": 0.5774076581001282,
"learning_rate": 1.926666666666667e-05,
"loss": 1.031,
"step": 290
},
{
"grad_norm": 0.6434250473976135,
"learning_rate": 1.9933333333333334e-05,
"loss": 1.0219,
"step": 300
},
{
"grad_norm": 0.5921059250831604,
"learning_rate": 2.06e-05,
"loss": 1.0242,
"step": 310
},
{
"grad_norm": 0.6857638955116272,
"learning_rate": 2.1266666666666667e-05,
"loss": 1.0319,
"step": 320
},
{
"grad_norm": 0.6548111438751221,
"learning_rate": 2.1933333333333332e-05,
"loss": 1.0167,
"step": 330
},
{
"grad_norm": 0.5840401649475098,
"learning_rate": 2.26e-05,
"loss": 1.0164,
"step": 340
},
{
"grad_norm": 0.6341369152069092,
"learning_rate": 2.326666666666667e-05,
"loss": 1.0088,
"step": 350
},
{
"grad_norm": 0.5580436587333679,
"learning_rate": 2.3933333333333337e-05,
"loss": 1.0021,
"step": 360
},
{
"grad_norm": 0.5720800161361694,
"learning_rate": 2.46e-05,
"loss": 0.9997,
"step": 370
},
{
"grad_norm": 0.6285766959190369,
"learning_rate": 2.5266666666666666e-05,
"loss": 0.9963,
"step": 380
},
{
"grad_norm": 0.7693972587585449,
"learning_rate": 2.5933333333333338e-05,
"loss": 0.9874,
"step": 390
},
{
"grad_norm": 0.6983421444892883,
"learning_rate": 2.6600000000000003e-05,
"loss": 0.9776,
"step": 400
},
{
"grad_norm": 0.608896791934967,
"learning_rate": 2.7266666666666668e-05,
"loss": 0.9744,
"step": 410
},
{
"grad_norm": 0.7339090704917908,
"learning_rate": 2.7933333333333332e-05,
"loss": 0.9611,
"step": 420
},
{
"grad_norm": 0.7828398942947388,
"learning_rate": 2.86e-05,
"loss": 0.9481,
"step": 430
},
{
"grad_norm": 0.7411054372787476,
"learning_rate": 2.926666666666667e-05,
"loss": 0.9318,
"step": 440
},
{
"grad_norm": 0.9737377762794495,
"learning_rate": 2.9933333333333337e-05,
"loss": 0.9144,
"step": 450
},
{
"grad_norm": 1.0443695783615112,
"learning_rate": 3.06e-05,
"loss": 0.912,
"step": 460
},
{
"grad_norm": 0.7744823694229126,
"learning_rate": 3.126666666666666e-05,
"loss": 0.8997,
"step": 470
},
{
"grad_norm": 0.7731150984764099,
"learning_rate": 3.1933333333333335e-05,
"loss": 0.8869,
"step": 480
},
{
"grad_norm": 0.8609164357185364,
"learning_rate": 3.26e-05,
"loss": 0.8594,
"step": 490
},
{
"grad_norm": 1.0301389694213867,
"learning_rate": 3.326666666666667e-05,
"loss": 0.8494,
"step": 500
},
{
"grad_norm": 0.9303619265556335,
"learning_rate": 3.3933333333333336e-05,
"loss": 0.8517,
"step": 510
},
{
"grad_norm": 0.8962922692298889,
"learning_rate": 3.46e-05,
"loss": 0.8341,
"step": 520
},
{
"grad_norm": 0.9600597620010376,
"learning_rate": 3.526666666666667e-05,
"loss": 0.8244,
"step": 530
},
{
"grad_norm": 0.8156377673149109,
"learning_rate": 3.593333333333334e-05,
"loss": 0.7997,
"step": 540
},
{
"grad_norm": 0.8684089779853821,
"learning_rate": 3.66e-05,
"loss": 0.8197,
"step": 550
},
{
"grad_norm": 0.984448254108429,
"learning_rate": 3.726666666666667e-05,
"loss": 0.7946,
"step": 560
},
{
"grad_norm": 0.8076623678207397,
"learning_rate": 3.793333333333334e-05,
"loss": 0.7742,
"step": 570
},
{
"grad_norm": 0.8329934477806091,
"learning_rate": 3.86e-05,
"loss": 0.766,
"step": 580
},
{
"grad_norm": 1.0042364597320557,
"learning_rate": 3.926666666666667e-05,
"loss": 0.7589,
"step": 590
},
{
"grad_norm": 0.9964665174484253,
"learning_rate": 3.993333333333333e-05,
"loss": 0.7454,
"step": 600
},
{
"grad_norm": 1.0122058391571045,
"learning_rate": 4.0600000000000004e-05,
"loss": 0.7255,
"step": 610
},
{
"grad_norm": 1.0226969718933105,
"learning_rate": 4.126666666666667e-05,
"loss": 0.7177,
"step": 620
},
{
"grad_norm": 1.0136826038360596,
"learning_rate": 4.1933333333333334e-05,
"loss": 0.7204,
"step": 630
},
{
"grad_norm": 1.0898646116256714,
"learning_rate": 4.26e-05,
"loss": 0.7003,
"step": 640
},
{
"grad_norm": 1.1982061862945557,
"learning_rate": 4.3266666666666664e-05,
"loss": 0.6937,
"step": 650
},
{
"grad_norm": 1.0605641603469849,
"learning_rate": 4.3933333333333335e-05,
"loss": 0.69,
"step": 660
},
{
"grad_norm": 0.8114312291145325,
"learning_rate": 4.46e-05,
"loss": 0.6644,
"step": 670
},
{
"grad_norm": 1.0191421508789062,
"learning_rate": 4.526666666666667e-05,
"loss": 0.6651,
"step": 680
},
{
"grad_norm": 0.9930680394172668,
"learning_rate": 4.5933333333333336e-05,
"loss": 0.6658,
"step": 690
},
{
"grad_norm": 1.050426721572876,
"learning_rate": 4.660000000000001e-05,
"loss": 0.6608,
"step": 700
},
{
"grad_norm": 1.185225248336792,
"learning_rate": 4.726666666666667e-05,
"loss": 0.6382,
"step": 710
},
{
"grad_norm": 1.0115975141525269,
"learning_rate": 4.793333333333334e-05,
"loss": 0.637,
"step": 720
},
{
"grad_norm": 1.0402801036834717,
"learning_rate": 4.86e-05,
"loss": 0.6254,
"step": 730
},
{
"grad_norm": 1.02639639377594,
"learning_rate": 4.926666666666667e-05,
"loss": 0.6076,
"step": 740
},
{
"grad_norm": 0.814791738986969,
"learning_rate": 4.993333333333334e-05,
"loss": 0.6152,
"step": 750
},
{
"grad_norm": 1.0697119235992432,
"learning_rate": 5.0600000000000003e-05,
"loss": 0.607,
"step": 760
},
{
"grad_norm": 0.96463942527771,
"learning_rate": 5.1266666666666675e-05,
"loss": 0.5999,
"step": 770
},
{
"grad_norm": 1.2382808923721313,
"learning_rate": 5.193333333333333e-05,
"loss": 0.6241,
"step": 780
},
{
"grad_norm": 0.8918554782867432,
"learning_rate": 5.2600000000000005e-05,
"loss": 0.5858,
"step": 790
},
{
"grad_norm": 1.0972034931182861,
"learning_rate": 5.326666666666666e-05,
"loss": 0.5805,
"step": 800
},
{
"grad_norm": 0.995185136795044,
"learning_rate": 5.3933333333333334e-05,
"loss": 0.58,
"step": 810
},
{
"grad_norm": 1.009131908416748,
"learning_rate": 5.4600000000000006e-05,
"loss": 0.5842,
"step": 820
},
{
"grad_norm": 1.1561076641082764,
"learning_rate": 5.5266666666666664e-05,
"loss": 0.5871,
"step": 830
},
{
"grad_norm": 1.06320321559906,
"learning_rate": 5.5933333333333335e-05,
"loss": 0.5816,
"step": 840
},
{
"grad_norm": 0.8474621772766113,
"learning_rate": 5.66e-05,
"loss": 0.5756,
"step": 850
},
{
"grad_norm": 0.9998260736465454,
"learning_rate": 5.726666666666667e-05,
"loss": 0.5594,
"step": 860
},
{
"grad_norm": 0.9477384686470032,
"learning_rate": 5.7933333333333337e-05,
"loss": 0.5739,
"step": 870
},
{
"grad_norm": 1.1020270586013794,
"learning_rate": 5.86e-05,
"loss": 0.5552,
"step": 880
},
{
"grad_norm": 0.9148961305618286,
"learning_rate": 5.926666666666667e-05,
"loss": 0.5642,
"step": 890
},
{
"grad_norm": 0.9651272296905518,
"learning_rate": 5.9933333333333345e-05,
"loss": 0.5524,
"step": 900
},
{
"grad_norm": 1.0408751964569092,
"learning_rate": 6.06e-05,
"loss": 0.5516,
"step": 910
},
{
"grad_norm": 0.9110714197158813,
"learning_rate": 6.126666666666667e-05,
"loss": 0.5412,
"step": 920
},
{
"grad_norm": 0.9199972152709961,
"learning_rate": 6.193333333333333e-05,
"loss": 0.5477,
"step": 930
},
{
"grad_norm": 1.0447932481765747,
"learning_rate": 6.26e-05,
"loss": 0.544,
"step": 940
},
{
"grad_norm": 0.9080489277839661,
"learning_rate": 6.326666666666667e-05,
"loss": 0.538,
"step": 950
},
{
"grad_norm": 0.8295878171920776,
"learning_rate": 6.393333333333333e-05,
"loss": 0.5371,
"step": 960
},
{
"grad_norm": 1.1033858060836792,
"learning_rate": 6.460000000000001e-05,
"loss": 0.5518,
"step": 970
},
{
"grad_norm": 0.8905893564224243,
"learning_rate": 6.526666666666666e-05,
"loss": 0.5209,
"step": 980
},
{
"grad_norm": 1.0580471754074097,
"learning_rate": 6.593333333333334e-05,
"loss": 0.5191,
"step": 990
},
{
"grad_norm": 1.088282823562622,
"learning_rate": 6.66e-05,
"loss": 0.5217,
"step": 1000
},
{
"grad_norm": 1.0257587432861328,
"learning_rate": 6.726666666666667e-05,
"loss": 0.5128,
"step": 1010
},
{
"grad_norm": 1.1624850034713745,
"learning_rate": 6.793333333333334e-05,
"loss": 0.5319,
"step": 1020
},
{
"grad_norm": 0.9272876977920532,
"learning_rate": 6.860000000000001e-05,
"loss": 0.5056,
"step": 1030
},
{
"grad_norm": 0.7771590352058411,
"learning_rate": 6.926666666666667e-05,
"loss": 0.501,
"step": 1040
},
{
"grad_norm": 1.2056543827056885,
"learning_rate": 6.993333333333334e-05,
"loss": 0.5172,
"step": 1050
},
{
"grad_norm": 1.0876524448394775,
"learning_rate": 7.06e-05,
"loss": 0.5121,
"step": 1060
},
{
"grad_norm": 0.9903247356414795,
"learning_rate": 7.126666666666667e-05,
"loss": 0.5098,
"step": 1070
},
{
"grad_norm": 1.0215466022491455,
"learning_rate": 7.193333333333334e-05,
"loss": 0.5191,
"step": 1080
},
{
"grad_norm": 0.9327561259269714,
"learning_rate": 7.26e-05,
"loss": 0.5087,
"step": 1090
},
{
"grad_norm": 0.9732714891433716,
"learning_rate": 7.326666666666667e-05,
"loss": 0.4915,
"step": 1100
},
{
"grad_norm": 1.0927966833114624,
"learning_rate": 7.393333333333333e-05,
"loss": 0.5201,
"step": 1110
},
{
"grad_norm": 0.9160197377204895,
"learning_rate": 7.46e-05,
"loss": 0.4912,
"step": 1120
},
{
"grad_norm": 0.885166347026825,
"learning_rate": 7.526666666666668e-05,
"loss": 0.5006,
"step": 1130
},
{
"grad_norm": 0.8786702752113342,
"learning_rate": 7.593333333333334e-05,
"loss": 0.5133,
"step": 1140
},
{
"grad_norm": 0.9320246577262878,
"learning_rate": 7.66e-05,
"loss": 0.5046,
"step": 1150
},
{
"grad_norm": 0.8129745125770569,
"learning_rate": 7.726666666666667e-05,
"loss": 0.4952,
"step": 1160
},
{
"grad_norm": 0.944871723651886,
"learning_rate": 7.793333333333333e-05,
"loss": 0.4882,
"step": 1170
},
{
"grad_norm": 0.8445220589637756,
"learning_rate": 7.860000000000001e-05,
"loss": 0.4774,
"step": 1180
},
{
"grad_norm": 0.9477908611297607,
"learning_rate": 7.926666666666666e-05,
"loss": 0.4872,
"step": 1190
},
{
"grad_norm": 0.9060473442077637,
"learning_rate": 7.993333333333334e-05,
"loss": 0.4903,
"step": 1200
},
{
"grad_norm": 1.2685356140136719,
"learning_rate": 8.060000000000001e-05,
"loss": 0.5047,
"step": 1210
},
{
"grad_norm": 0.8890970945358276,
"learning_rate": 8.126666666666667e-05,
"loss": 0.4808,
"step": 1220
},
{
"grad_norm": 0.8808901906013489,
"learning_rate": 8.193333333333334e-05,
"loss": 0.4799,
"step": 1230
},
{
"grad_norm": 0.8536569476127625,
"learning_rate": 8.26e-05,
"loss": 0.482,
"step": 1240
},
{
"grad_norm": 0.9718847870826721,
"learning_rate": 8.326666666666667e-05,
"loss": 0.4646,
"step": 1250
},
{
"grad_norm": 1.0003539323806763,
"learning_rate": 8.393333333333335e-05,
"loss": 0.4705,
"step": 1260
},
{
"grad_norm": 0.8586186766624451,
"learning_rate": 8.46e-05,
"loss": 0.4562,
"step": 1270
},
{
"grad_norm": 0.8693225383758545,
"learning_rate": 8.526666666666667e-05,
"loss": 0.4662,
"step": 1280
},
{
"grad_norm": 0.8493294715881348,
"learning_rate": 8.593333333333333e-05,
"loss": 0.4641,
"step": 1290
},
{
"grad_norm": 0.8591540455818176,
"learning_rate": 8.66e-05,
"loss": 0.4587,
"step": 1300
},
{
"grad_norm": 0.8848975896835327,
"learning_rate": 8.726666666666667e-05,
"loss": 0.4494,
"step": 1310
},
{
"grad_norm": 0.8884143233299255,
"learning_rate": 8.793333333333333e-05,
"loss": 0.452,
"step": 1320
},
{
"grad_norm": 0.7653201818466187,
"learning_rate": 8.86e-05,
"loss": 0.4426,
"step": 1330
},
{
"grad_norm": 0.9352204203605652,
"learning_rate": 8.926666666666668e-05,
"loss": 0.4326,
"step": 1340
},
{
"grad_norm": 0.960019588470459,
"learning_rate": 8.993333333333334e-05,
"loss": 0.4618,
"step": 1350
},
{
"grad_norm": 1.0280357599258423,
"learning_rate": 9.06e-05,
"loss": 0.4485,
"step": 1360
},
{
"grad_norm": 0.9692069292068481,
"learning_rate": 9.126666666666667e-05,
"loss": 0.4407,
"step": 1370
},
{
"grad_norm": 0.8127903342247009,
"learning_rate": 9.193333333333334e-05,
"loss": 0.4405,
"step": 1380
},
{
"grad_norm": 0.9468965530395508,
"learning_rate": 9.260000000000001e-05,
"loss": 0.4447,
"step": 1390
},
{
"grad_norm": 0.7770897150039673,
"learning_rate": 9.326666666666667e-05,
"loss": 0.4418,
"step": 1400
},
{
"grad_norm": 0.747828483581543,
"learning_rate": 9.393333333333334e-05,
"loss": 0.4463,
"step": 1410
},
{
"grad_norm": 0.96014004945755,
"learning_rate": 9.46e-05,
"loss": 0.4555,
"step": 1420
},
{
"grad_norm": 0.8238260746002197,
"learning_rate": 9.526666666666667e-05,
"loss": 0.4331,
"step": 1430
},
{
"grad_norm": 0.7926999926567078,
"learning_rate": 9.593333333333334e-05,
"loss": 0.4486,
"step": 1440
},
{
"grad_norm": 0.745404839515686,
"learning_rate": 9.66e-05,
"loss": 0.4541,
"step": 1450
},
{
"grad_norm": 0.8123156428337097,
"learning_rate": 9.726666666666667e-05,
"loss": 0.453,
"step": 1460
},
{
"grad_norm": 0.8227003812789917,
"learning_rate": 9.793333333333333e-05,
"loss": 0.4538,
"step": 1470
},
{
"grad_norm": 0.8354244828224182,
"learning_rate": 9.86e-05,
"loss": 0.4452,
"step": 1480
},
{
"grad_norm": 0.9338411688804626,
"learning_rate": 9.926666666666668e-05,
"loss": 0.4378,
"step": 1490
},
{
"grad_norm": 0.6889047026634216,
"learning_rate": 9.993333333333334e-05,
"loss": 0.4567,
"step": 1500
},
{
"grad_norm": 0.7608139514923096,
"learning_rate": 9.999997539434007e-05,
"loss": 0.4413,
"step": 1510
},
{
"grad_norm": 0.8127821087837219,
"learning_rate": 9.999989033776898e-05,
"loss": 0.4443,
"step": 1520
},
{
"grad_norm": 0.6642680168151855,
"learning_rate": 9.999974452661641e-05,
"loss": 0.4481,
"step": 1530
},
{
"grad_norm": 0.7917938828468323,
"learning_rate": 9.999953796105959e-05,
"loss": 0.439,
"step": 1540
},
{
"grad_norm": 0.7324179410934448,
"learning_rate": 9.999927064134949e-05,
"loss": 0.4345,
"step": 1550
},
{
"grad_norm": 0.7585936188697815,
"learning_rate": 9.999894256781095e-05,
"loss": 0.4523,
"step": 1560
},
{
"grad_norm": 0.870932400226593,
"learning_rate": 9.99985537408426e-05,
"loss": 0.4318,
"step": 1570
},
{
"grad_norm": 0.7509883642196655,
"learning_rate": 9.999810416091688e-05,
"loss": 0.4352,
"step": 1580
},
{
"grad_norm": 0.8793972134590149,
"learning_rate": 9.99975938285801e-05,
"loss": 0.4424,
"step": 1590
},
{
"grad_norm": 0.8379639387130737,
"learning_rate": 9.999702274445236e-05,
"loss": 0.4383,
"step": 1600
},
{
"grad_norm": 0.6467755436897278,
"learning_rate": 9.999639090922756e-05,
"loss": 0.4217,
"step": 1610
},
{
"grad_norm": 0.7293924689292908,
"learning_rate": 9.999569832367346e-05,
"loss": 0.4321,
"step": 1620
},
{
"grad_norm": 0.8434097766876221,
"learning_rate": 9.999494498863162e-05,
"loss": 0.4539,
"step": 1630
},
{
"grad_norm": 0.7917138338088989,
"learning_rate": 9.99941309050174e-05,
"loss": 0.436,
"step": 1640
},
{
"grad_norm": 0.7262437343597412,
"learning_rate": 9.999325607381999e-05,
"loss": 0.4353,
"step": 1650
},
{
"grad_norm": 0.7967081069946289,
"learning_rate": 9.999232049610238e-05,
"loss": 0.4354,
"step": 1660
},
{
"grad_norm": 0.7270679473876953,
"learning_rate": 9.999132417300142e-05,
"loss": 0.4141,
"step": 1670
},
{
"grad_norm": 0.8002856373786926,
"learning_rate": 9.99902671057277e-05,
"loss": 0.4269,
"step": 1680
},
{
"grad_norm": 0.667580783367157,
"learning_rate": 9.998914929556569e-05,
"loss": 0.4253,
"step": 1690
},
{
"grad_norm": 0.7109050154685974,
"learning_rate": 9.998797074387361e-05,
"loss": 0.4233,
"step": 1700
},
{
"grad_norm": 0.8006998300552368,
"learning_rate": 9.99867314520835e-05,
"loss": 0.4427,
"step": 1710
},
{
"grad_norm": 0.7001204490661621,
"learning_rate": 9.998543142170126e-05,
"loss": 0.4311,
"step": 1720
},
{
"grad_norm": 0.6047891974449158,
"learning_rate": 9.99840706543065e-05,
"loss": 0.4226,
"step": 1730
},
{
"grad_norm": 0.7993099093437195,
"learning_rate": 9.998264915155274e-05,
"loss": 0.4201,
"step": 1740
},
{
"grad_norm": 0.7204874753952026,
"learning_rate": 9.998116691516718e-05,
"loss": 0.4291,
"step": 1750
},
{
"grad_norm": 0.8226584196090698,
"learning_rate": 9.997962394695091e-05,
"loss": 0.432,
"step": 1760
},
{
"grad_norm": 0.783210039138794,
"learning_rate": 9.997802024877875e-05,
"loss": 0.4223,
"step": 1770
},
{
"grad_norm": 0.6845479607582092,
"learning_rate": 9.99763558225994e-05,
"loss": 0.431,
"step": 1780
},
{
"grad_norm": 0.69374680519104,
"learning_rate": 9.997463067043526e-05,
"loss": 0.4194,
"step": 1790
},
{
"grad_norm": 0.7323899865150452,
"learning_rate": 9.997284479438253e-05,
"loss": 0.4201,
"step": 1800
},
{
"grad_norm": 0.6646573543548584,
"learning_rate": 9.997099819661127e-05,
"loss": 0.4135,
"step": 1810
},
{
"grad_norm": 0.7762008905410767,
"learning_rate": 9.996909087936524e-05,
"loss": 0.4188,
"step": 1820
},
{
"grad_norm": 0.6358603239059448,
"learning_rate": 9.996712284496201e-05,
"loss": 0.4202,
"step": 1830
},
{
"grad_norm": 0.9141177535057068,
"learning_rate": 9.996509409579293e-05,
"loss": 0.4264,
"step": 1840
},
{
"grad_norm": 0.7645198106765747,
"learning_rate": 9.996300463432312e-05,
"loss": 0.4451,
"step": 1850
},
{
"grad_norm": 0.809575080871582,
"learning_rate": 9.996085446309148e-05,
"loss": 0.4329,
"step": 1860
},
{
"grad_norm": 0.7992050051689148,
"learning_rate": 9.995864358471066e-05,
"loss": 0.419,
"step": 1870
},
{
"grad_norm": 0.6914141178131104,
"learning_rate": 9.99563720018671e-05,
"loss": 0.4026,
"step": 1880
},
{
"grad_norm": 0.5682405233383179,
"learning_rate": 9.995403971732098e-05,
"loss": 0.4076,
"step": 1890
},
{
"grad_norm": 0.7482168674468994,
"learning_rate": 9.995164673390625e-05,
"loss": 0.4025,
"step": 1900
},
{
"grad_norm": 0.9186742305755615,
"learning_rate": 9.994919305453059e-05,
"loss": 0.4249,
"step": 1910
},
{
"grad_norm": 0.7261620759963989,
"learning_rate": 9.994667868217548e-05,
"loss": 0.4082,
"step": 1920
},
{
"grad_norm": 0.758364200592041,
"learning_rate": 9.99441036198961e-05,
"loss": 0.4024,
"step": 1930
},
{
"grad_norm": 0.8376240730285645,
"learning_rate": 9.99414678708214e-05,
"loss": 0.4055,
"step": 1940
},
{
"grad_norm": 0.683440089225769,
"learning_rate": 9.993877143815407e-05,
"loss": 0.3972,
"step": 1950
},
{
"grad_norm": 0.7009807825088501,
"learning_rate": 9.993601432517053e-05,
"loss": 0.4178,
"step": 1960
},
{
"grad_norm": 0.6763379573822021,
"learning_rate": 9.993319653522091e-05,
"loss": 0.3953,
"step": 1970
},
{
"grad_norm": 0.7284714579582214,
"learning_rate": 9.993031807172911e-05,
"loss": 0.4132,
"step": 1980
},
{
"grad_norm": 0.6284232139587402,
"learning_rate": 9.992737893819273e-05,
"loss": 0.4031,
"step": 1990
},
{
"grad_norm": 0.609225332736969,
"learning_rate": 9.992437913818312e-05,
"loss": 0.4083,
"step": 2000
},
{
"grad_norm": 0.7316564917564392,
"learning_rate": 9.992131867534526e-05,
"loss": 0.4248,
"step": 2010
},
{
"grad_norm": 0.7980116009712219,
"learning_rate": 9.991819755339796e-05,
"loss": 0.4064,
"step": 2020
},
{
"grad_norm": 0.7345552444458008,
"learning_rate": 9.991501577613365e-05,
"loss": 0.3936,
"step": 2030
},
{
"grad_norm": 0.8199358582496643,
"learning_rate": 9.99117733474185e-05,
"loss": 0.4014,
"step": 2040
},
{
"grad_norm": 0.6312980055809021,
"learning_rate": 9.990847027119234e-05,
"loss": 0.4161,
"step": 2050
},
{
"grad_norm": 0.6444507837295532,
"learning_rate": 9.990510655146877e-05,
"loss": 0.4001,
"step": 2060
},
{
"grad_norm": 0.6979199051856995,
"learning_rate": 9.990168219233496e-05,
"loss": 0.3991,
"step": 2070
},
{
"grad_norm": 0.6546664834022522,
"learning_rate": 9.989819719795188e-05,
"loss": 0.3996,
"step": 2080
},
{
"grad_norm": 0.9151248335838318,
"learning_rate": 9.989465157255412e-05,
"loss": 0.4001,
"step": 2090
},
{
"grad_norm": 0.7367625832557678,
"learning_rate": 9.989104532044994e-05,
"loss": 0.4021,
"step": 2100
},
{
"grad_norm": 0.8062440752983093,
"learning_rate": 9.988737844602128e-05,
"loss": 0.4177,
"step": 2110
},
{
"grad_norm": 0.66869056224823,
"learning_rate": 9.988365095372372e-05,
"loss": 0.3926,
"step": 2120
},
{
"grad_norm": 0.738856315612793,
"learning_rate": 9.987986284808654e-05,
"loss": 0.3979,
"step": 2130
},
{
"grad_norm": 0.7004558444023132,
"learning_rate": 9.987601413371264e-05,
"loss": 0.3805,
"step": 2140
},
{
"grad_norm": 0.836412250995636,
"learning_rate": 9.987210481527855e-05,
"loss": 0.4002,
"step": 2150
},
{
"grad_norm": 0.6521918177604675,
"learning_rate": 9.98681348975345e-05,
"loss": 0.3852,
"step": 2160
},
{
"grad_norm": 0.6960434913635254,
"learning_rate": 9.986410438530427e-05,
"loss": 0.3836,
"step": 2170
},
{
"grad_norm": 0.6714290380477905,
"learning_rate": 9.986001328348534e-05,
"loss": 0.3935,
"step": 2180
},
{
"grad_norm": 0.7229437232017517,
"learning_rate": 9.985586159704878e-05,
"loss": 0.3838,
"step": 2190
},
{
"grad_norm": 0.698377251625061,
"learning_rate": 9.985164933103929e-05,
"loss": 0.3793,
"step": 2200
},
{
"grad_norm": 0.7945483922958374,
"learning_rate": 9.984737649057513e-05,
"loss": 0.3855,
"step": 2210
},
{
"grad_norm": 0.6426589488983154,
"learning_rate": 9.984304308084827e-05,
"loss": 0.3712,
"step": 2220
},
{
"grad_norm": 0.6760197281837463,
"learning_rate": 9.983864910712416e-05,
"loss": 0.3846,
"step": 2230
},
{
"grad_norm": 0.8168231248855591,
"learning_rate": 9.98341945747419e-05,
"loss": 0.3758,
"step": 2240
},
{
"grad_norm": 0.6996021866798401,
"learning_rate": 9.98296794891142e-05,
"loss": 0.3866,
"step": 2250
},
{
"grad_norm": 0.5832933187484741,
"learning_rate": 9.982510385572725e-05,
"loss": 0.3733,
"step": 2260
},
{
"grad_norm": 0.6716319918632507,
"learning_rate": 9.982046768014094e-05,
"loss": 0.3649,
"step": 2270
},
{
"grad_norm": 0.7418034672737122,
"learning_rate": 9.981577096798863e-05,
"loss": 0.3675,
"step": 2280
},
{
"grad_norm": 0.727208137512207,
"learning_rate": 9.981101372497727e-05,
"loss": 0.3969,
"step": 2290
},
{
"grad_norm": 0.6294035315513611,
"learning_rate": 9.980619595688737e-05,
"loss": 0.3628,
"step": 2300
},
{
"grad_norm": 0.7221769094467163,
"learning_rate": 9.980131766957295e-05,
"loss": 0.375,
"step": 2310
},
{
"grad_norm": 0.7024942636489868,
"learning_rate": 9.979637886896163e-05,
"loss": 0.3676,
"step": 2320
},
{
"grad_norm": 0.7094406485557556,
"learning_rate": 9.979137956105447e-05,
"loss": 0.3747,
"step": 2330
},
{
"grad_norm": 0.6109972596168518,
"learning_rate": 9.978631975192613e-05,
"loss": 0.3704,
"step": 2340
},
{
"grad_norm": 0.6860019564628601,
"learning_rate": 9.978119944772475e-05,
"loss": 0.3688,
"step": 2350
},
{
"grad_norm": 0.6933342218399048,
"learning_rate": 9.977601865467197e-05,
"loss": 0.373,
"step": 2360
},
{
"grad_norm": 0.7283564209938049,
"learning_rate": 9.977077737906297e-05,
"loss": 0.3696,
"step": 2370
},
{
"grad_norm": 0.6756188273429871,
"learning_rate": 9.976547562726636e-05,
"loss": 0.3656,
"step": 2380
},
{
"grad_norm": 0.7115433812141418,
"learning_rate": 9.976011340572429e-05,
"loss": 0.3723,
"step": 2390
},
{
"grad_norm": 0.6298806667327881,
"learning_rate": 9.975469072095237e-05,
"loss": 0.3641,
"step": 2400
},
{
"grad_norm": 0.5875228643417358,
"learning_rate": 9.974920757953965e-05,
"loss": 0.3605,
"step": 2410
},
{
"grad_norm": 0.6618392467498779,
"learning_rate": 9.97436639881487e-05,
"loss": 0.3839,
"step": 2420
},
{
"grad_norm": 0.7790018320083618,
"learning_rate": 9.973805995351545e-05,
"loss": 0.3625,
"step": 2430
},
{
"grad_norm": 0.6831006407737732,
"learning_rate": 9.973239548244939e-05,
"loss": 0.3671,
"step": 2440
},
{
"grad_norm": 0.7966041564941406,
"learning_rate": 9.972667058183333e-05,
"loss": 0.3547,
"step": 2450
},
{
"grad_norm": 0.582952618598938,
"learning_rate": 9.972088525862362e-05,
"loss": 0.347,
"step": 2460
},
{
"grad_norm": 0.7932450771331787,
"learning_rate": 9.971503951984995e-05,
"loss": 0.3508,
"step": 2470
},
{
"grad_norm": 0.6803222298622131,
"learning_rate": 9.970913337261543e-05,
"loss": 0.3636,
"step": 2480
},
{
"grad_norm": 0.8180113434791565,
"learning_rate": 9.97031668240966e-05,
"loss": 0.3596,
"step": 2490
},
{
"grad_norm": 0.6797927618026733,
"learning_rate": 9.969713988154339e-05,
"loss": 0.3533,
"step": 2500
},
{
"grad_norm": 0.7941766977310181,
"learning_rate": 9.969105255227906e-05,
"loss": 0.3503,
"step": 2510
},
{
"grad_norm": 0.8432013988494873,
"learning_rate": 9.968490484370035e-05,
"loss": 0.352,
"step": 2520
},
{
"grad_norm": 0.7826078534126282,
"learning_rate": 9.967869676327726e-05,
"loss": 0.3409,
"step": 2530
},
{
"grad_norm": 0.618452250957489,
"learning_rate": 9.96724283185532e-05,
"loss": 0.3499,
"step": 2540
},
{
"grad_norm": 0.7181870937347412,
"learning_rate": 9.966609951714494e-05,
"loss": 0.3546,
"step": 2550
},
{
"grad_norm": 0.6708511114120483,
"learning_rate": 9.965971036674255e-05,
"loss": 0.356,
"step": 2560
},
{
"grad_norm": 0.691321611404419,
"learning_rate": 9.965326087510947e-05,
"loss": 0.3451,
"step": 2570
},
{
"grad_norm": 0.5737016797065735,
"learning_rate": 9.964675105008244e-05,
"loss": 0.35,
"step": 2580
},
{
"grad_norm": 0.767196774482727,
"learning_rate": 9.964018089957147e-05,
"loss": 0.3454,
"step": 2590
},
{
"grad_norm": 0.6091346740722656,
"learning_rate": 9.963355043155996e-05,
"loss": 0.3566,
"step": 2600
},
{
"grad_norm": 0.7480655312538147,
"learning_rate": 9.962685965410455e-05,
"loss": 0.3569,
"step": 2610
},
{
"grad_norm": 0.6881885528564453,
"learning_rate": 9.962010857533514e-05,
"loss": 0.3363,
"step": 2620
},
{
"grad_norm": 0.9496521353721619,
"learning_rate": 9.961329720345493e-05,
"loss": 0.3443,
"step": 2630
},
{
"grad_norm": 0.7808911204338074,
"learning_rate": 9.96064255467404e-05,
"loss": 0.352,
"step": 2640
},
{
"grad_norm": 0.6344329714775085,
"learning_rate": 9.959949361354126e-05,
"loss": 0.3439,
"step": 2650
},
{
"grad_norm": 0.7219773530960083,
"learning_rate": 9.959250141228045e-05,
"loss": 0.3487,
"step": 2660
},
{
"grad_norm": 0.6620672345161438,
"learning_rate": 9.958544895145414e-05,
"loss": 0.3493,
"step": 2670
},
{
"grad_norm": 0.7466215491294861,
"learning_rate": 9.957833623963177e-05,
"loss": 0.3643,
"step": 2680
},
{
"grad_norm": 0.6851657629013062,
"learning_rate": 9.957116328545593e-05,
"loss": 0.3388,
"step": 2690
},
{
"grad_norm": 0.6105020642280579,
"learning_rate": 9.956393009764244e-05,
"loss": 0.3298,
"step": 2700
},
{
"grad_norm": 0.6620052456855774,
"learning_rate": 9.955663668498032e-05,
"loss": 0.3334,
"step": 2710
},
{
"grad_norm": 0.5547780990600586,
"learning_rate": 9.954928305633173e-05,
"loss": 0.3498,
"step": 2720
},
{
"grad_norm": 0.684063732624054,
"learning_rate": 9.954186922063204e-05,
"loss": 0.3349,
"step": 2730
},
{
"grad_norm": 0.6296693086624146,
"learning_rate": 9.953439518688974e-05,
"loss": 0.3284,
"step": 2740
},
{
"grad_norm": 0.8140157461166382,
"learning_rate": 9.952686096418652e-05,
"loss": 0.3419,
"step": 2750
},
{
"grad_norm": 0.6329033374786377,
"learning_rate": 9.951926656167714e-05,
"loss": 0.3296,
"step": 2760
},
{
"grad_norm": 0.6835638880729675,
"learning_rate": 9.951161198858953e-05,
"loss": 0.3304,
"step": 2770
},
{
"grad_norm": 0.7543403506278992,
"learning_rate": 9.950389725422471e-05,
"loss": 0.3372,
"step": 2780
},
{
"grad_norm": 0.80125492811203,
"learning_rate": 9.949612236795682e-05,
"loss": 0.334,
"step": 2790
},
{
"grad_norm": 0.8037199378013611,
"learning_rate": 9.948828733923305e-05,
"loss": 0.322,
"step": 2800
},
{
"grad_norm": 0.6507173776626587,
"learning_rate": 9.948039217757374e-05,
"loss": 0.3518,
"step": 2810
},
{
"grad_norm": 0.6338552832603455,
"learning_rate": 9.947243689257225e-05,
"loss": 0.3286,
"step": 2820
},
{
"grad_norm": 0.6869152784347534,
"learning_rate": 9.946442149389497e-05,
"loss": 0.3166,
"step": 2830
},
{
"grad_norm": 0.5835918188095093,
"learning_rate": 9.945634599128139e-05,
"loss": 0.3316,
"step": 2840
},
{
"grad_norm": 0.6610328555107117,
"learning_rate": 9.944821039454402e-05,
"loss": 0.3319,
"step": 2850
},
{
"grad_norm": 0.6664662957191467,
"learning_rate": 9.944001471356835e-05,
"loss": 0.3191,
"step": 2860
},
{
"grad_norm": 0.5826230645179749,
"learning_rate": 9.94317589583129e-05,
"loss": 0.3119,
"step": 2870
},
{
"grad_norm": 0.6326906085014343,
"learning_rate": 9.942344313880922e-05,
"loss": 0.3086,
"step": 2880
},
{
"grad_norm": 0.7808578014373779,
"learning_rate": 9.941506726516179e-05,
"loss": 0.3235,
"step": 2890
},
{
"grad_norm": 0.6272144913673401,
"learning_rate": 9.94066313475481e-05,
"loss": 0.3218,
"step": 2900
},
{
"grad_norm": 0.5796061754226685,
"learning_rate": 9.939813539621857e-05,
"loss": 0.3116,
"step": 2910
},
{
"grad_norm": 0.6038811206817627,
"learning_rate": 9.93895794214966e-05,
"loss": 0.311,
"step": 2920
},
{
"grad_norm": 0.6317859292030334,
"learning_rate": 9.938096343377852e-05,
"loss": 0.2998,
"step": 2930
},
{
"grad_norm": 0.6989952325820923,
"learning_rate": 9.937228744353353e-05,
"loss": 0.3042,
"step": 2940
},
{
"grad_norm": 0.6888319849967957,
"learning_rate": 9.936355146130379e-05,
"loss": 0.3073,
"step": 2950
},
{
"grad_norm": 0.6472424268722534,
"learning_rate": 9.935475549770435e-05,
"loss": 0.3067,
"step": 2960
},
{
"grad_norm": 0.6531728506088257,
"learning_rate": 9.934589956342315e-05,
"loss": 0.2981,
"step": 2970
},
{
"grad_norm": 0.6818107962608337,
"learning_rate": 9.933698366922093e-05,
"loss": 0.3078,
"step": 2980
},
{
"grad_norm": 0.6490950584411621,
"learning_rate": 9.93280078259314e-05,
"loss": 0.3107,
"step": 2990
},
{
"grad_norm": 0.5881419777870178,
"learning_rate": 9.931897204446104e-05,
"loss": 0.3051,
"step": 3000
},
{
"grad_norm": 0.6761528849601746,
"learning_rate": 9.930987633578915e-05,
"loss": 0.2905,
"step": 3010
},
{
"grad_norm": 0.5939033031463623,
"learning_rate": 9.93007207109679e-05,
"loss": 0.2967,
"step": 3020
},
{
"grad_norm": 0.6588555574417114,
"learning_rate": 9.929150518112224e-05,
"loss": 0.2949,
"step": 3030
},
{
"grad_norm": 0.648625373840332,
"learning_rate": 9.928222975744991e-05,
"loss": 0.3014,
"step": 3040
},
{
"grad_norm": 0.6281964182853699,
"learning_rate": 9.92728944512214e-05,
"loss": 0.3028,
"step": 3050
},
{
"grad_norm": 0.5988143682479858,
"learning_rate": 9.926349927378001e-05,
"loss": 0.3029,
"step": 3060
},
{
"grad_norm": 0.6383551359176636,
"learning_rate": 9.925404423654174e-05,
"loss": 0.2868,
"step": 3070
},
{
"grad_norm": 0.7227466702461243,
"learning_rate": 9.924452935099537e-05,
"loss": 0.302,
"step": 3080
},
{
"grad_norm": 0.7182079553604126,
"learning_rate": 9.92349546287024e-05,
"loss": 0.3234,
"step": 3090
},
{
"grad_norm": 0.6264972686767578,
"learning_rate": 9.9225320081297e-05,
"loss": 0.305,
"step": 3100
},
{
"grad_norm": 0.7251865863800049,
"learning_rate": 9.921562572048606e-05,
"loss": 0.2921,
"step": 3110
},
{
"grad_norm": 0.7024000883102417,
"learning_rate": 9.920587155804913e-05,
"loss": 0.2892,
"step": 3120
},
{
"grad_norm": 0.6056602001190186,
"learning_rate": 9.919605760583845e-05,
"loss": 0.2919,
"step": 3130
},
{
"grad_norm": 0.7880511283874512,
"learning_rate": 9.91861838757789e-05,
"loss": 0.2932,
"step": 3140
},
{
"grad_norm": 0.6231911778450012,
"learning_rate": 9.917625037986798e-05,
"loss": 0.2879,
"step": 3150
},
{
"grad_norm": 0.6722307801246643,
"learning_rate": 9.916625713017583e-05,
"loss": 0.2963,
"step": 3160
},
{
"grad_norm": 0.6776823401451111,
"learning_rate": 9.915620413884519e-05,
"loss": 0.2867,
"step": 3170
},
{
"grad_norm": 0.6715995669364929,
"learning_rate": 9.914609141809139e-05,
"loss": 0.2993,
"step": 3180
},
{
"grad_norm": 0.6652569770812988,
"learning_rate": 9.913591898020235e-05,
"loss": 0.2992,
"step": 3190
},
{
"grad_norm": 0.7991640567779541,
"learning_rate": 9.912568683753853e-05,
"loss": 0.2936,
"step": 3200
},
{
"grad_norm": 0.5877777338027954,
"learning_rate": 9.911539500253295e-05,
"loss": 0.2801,
"step": 3210
},
{
"grad_norm": 0.7298980951309204,
"learning_rate": 9.910504348769118e-05,
"loss": 0.3111,
"step": 3220
},
{
"grad_norm": 0.6657949686050415,
"learning_rate": 9.909463230559127e-05,
"loss": 0.2965,
"step": 3230
},
{
"grad_norm": 0.5663632154464722,
"learning_rate": 9.908416146888376e-05,
"loss": 0.2793,
"step": 3240
},
{
"grad_norm": 0.6907248497009277,
"learning_rate": 9.907363099029175e-05,
"loss": 0.2835,
"step": 3250
},
{
"grad_norm": 0.8290233016014099,
"learning_rate": 9.906304088261073e-05,
"loss": 0.2907,
"step": 3260
},
{
"grad_norm": 0.7471662163734436,
"learning_rate": 9.905239115870872e-05,
"loss": 0.2887,
"step": 3270
},
{
"grad_norm": 0.6640713214874268,
"learning_rate": 9.90416818315261e-05,
"loss": 0.273,
"step": 3280
},
{
"grad_norm": 0.5558493733406067,
"learning_rate": 9.903091291407573e-05,
"loss": 0.2798,
"step": 3290
},
{
"grad_norm": 0.7830227613449097,
"learning_rate": 9.902008441944286e-05,
"loss": 0.2737,
"step": 3300
},
{
"grad_norm": 0.7315648794174194,
"learning_rate": 9.900919636078512e-05,
"loss": 0.2819,
"step": 3310
},
{
"grad_norm": 0.6222036480903625,
"learning_rate": 9.899824875133255e-05,
"loss": 0.3014,
"step": 3320
},
{
"grad_norm": 0.6810315847396851,
"learning_rate": 9.898724160438749e-05,
"loss": 0.2792,
"step": 3330
},
{
"grad_norm": 0.7829352617263794,
"learning_rate": 9.89761749333247e-05,
"loss": 0.2668,
"step": 3340
},
{
"grad_norm": 0.7544236779212952,
"learning_rate": 9.896504875159122e-05,
"loss": 0.2714,
"step": 3350
},
{
"grad_norm": 0.7376437187194824,
"learning_rate": 9.89538630727064e-05,
"loss": 0.2698,
"step": 3360
},
{
"grad_norm": 0.6954911351203918,
"learning_rate": 9.894261791026189e-05,
"loss": 0.2713,
"step": 3370
},
{
"grad_norm": 0.5811148881912231,
"learning_rate": 9.893131327792165e-05,
"loss": 0.3016,
"step": 3380
},
{
"grad_norm": 0.6499139666557312,
"learning_rate": 9.891994918942182e-05,
"loss": 0.2586,
"step": 3390
},
{
"grad_norm": 0.7431524395942688,
"learning_rate": 9.890852565857092e-05,
"loss": 0.2725,
"step": 3400
},
{
"grad_norm": 0.7640831470489502,
"learning_rate": 9.889704269924954e-05,
"loss": 0.2667,
"step": 3410
},
{
"grad_norm": 0.741356611251831,
"learning_rate": 9.888550032541059e-05,
"loss": 0.2672,
"step": 3420
},
{
"grad_norm": 0.6311560273170471,
"learning_rate": 9.887389855107916e-05,
"loss": 0.2693,
"step": 3430
},
{
"grad_norm": 0.7636248469352722,
"learning_rate": 9.886223739035248e-05,
"loss": 0.2744,
"step": 3440
},
{
"grad_norm": 0.6490204334259033,
"learning_rate": 9.885051685739997e-05,
"loss": 0.2633,
"step": 3450
},
{
"grad_norm": 0.7360467910766602,
"learning_rate": 9.883873696646316e-05,
"loss": 0.2584,
"step": 3460
},
{
"grad_norm": 0.6869045495986938,
"learning_rate": 9.882689773185575e-05,
"loss": 0.2702,
"step": 3470
},
{
"grad_norm": 0.8184191584587097,
"learning_rate": 9.881499916796353e-05,
"loss": 0.2718,
"step": 3480
},
{
"grad_norm": 0.6298326849937439,
"learning_rate": 9.880304128924434e-05,
"loss": 0.2658,
"step": 3490
},
{
"grad_norm": 0.6409152150154114,
"learning_rate": 9.879102411022817e-05,
"loss": 0.2509,
"step": 3500
},
{
"grad_norm": 0.6272318959236145,
"learning_rate": 9.877894764551703e-05,
"loss": 0.2587,
"step": 3510
},
{
"grad_norm": 0.7787696719169617,
"learning_rate": 9.876681190978494e-05,
"loss": 0.2628,
"step": 3520
},
{
"grad_norm": 0.9701261520385742,
"learning_rate": 9.875461691777797e-05,
"loss": 0.2664,
"step": 3530
},
{
"grad_norm": 0.7884742021560669,
"learning_rate": 9.874236268431417e-05,
"loss": 0.2582,
"step": 3540
},
{
"grad_norm": 0.865479588508606,
"learning_rate": 9.873004922428361e-05,
"loss": 0.2699,
"step": 3550
},
{
"grad_norm": 0.6470707654953003,
"learning_rate": 9.871767655264829e-05,
"loss": 0.2523,
"step": 3560
},
{
"grad_norm": 0.7726714015007019,
"learning_rate": 9.87052446844422e-05,
"loss": 0.2651,
"step": 3570
},
{
"grad_norm": 0.7063262462615967,
"learning_rate": 9.869275363477122e-05,
"loss": 0.2757,
"step": 3580
},
{
"grad_norm": 0.7393220663070679,
"learning_rate": 9.868020341881312e-05,
"loss": 0.2697,
"step": 3590
},
{
"grad_norm": 0.6041222810745239,
"learning_rate": 9.866759405181765e-05,
"loss": 0.2645,
"step": 3600
},
{
"grad_norm": 0.7207736968994141,
"learning_rate": 9.865492554910633e-05,
"loss": 0.2632,
"step": 3610
},
{
"grad_norm": 0.7190155386924744,
"learning_rate": 9.864219792607262e-05,
"loss": 0.2557,
"step": 3620
},
{
"grad_norm": 0.6632683277130127,
"learning_rate": 9.862941119818177e-05,
"loss": 0.2482,
"step": 3630
},
{
"grad_norm": 0.8325068354606628,
"learning_rate": 9.861656538097086e-05,
"loss": 0.2598,
"step": 3640
},
{
"grad_norm": 0.7415052652359009,
"learning_rate": 9.860366049004877e-05,
"loss": 0.2635,
"step": 3650
},
{
"grad_norm": 0.7114351987838745,
"learning_rate": 9.859069654109615e-05,
"loss": 0.2562,
"step": 3660
},
{
"grad_norm": 0.69629967212677,
"learning_rate": 9.857767354986545e-05,
"loss": 0.2667,
"step": 3670
},
{
"grad_norm": 0.6470484137535095,
"learning_rate": 9.856459153218078e-05,
"loss": 0.2583,
"step": 3680
},
{
"grad_norm": 0.6998385190963745,
"learning_rate": 9.855145050393808e-05,
"loss": 0.2442,
"step": 3690
},
{
"grad_norm": 0.5356374979019165,
"learning_rate": 9.85382504811049e-05,
"loss": 0.2444,
"step": 3700
},
{
"grad_norm": 0.8032253384590149,
"learning_rate": 9.852499147972054e-05,
"loss": 0.2361,
"step": 3710
},
{
"grad_norm": 0.8585163950920105,
"learning_rate": 9.851167351589592e-05,
"loss": 0.2402,
"step": 3720
},
{
"grad_norm": 0.8090064525604248,
"learning_rate": 9.849829660581363e-05,
"loss": 0.2272,
"step": 3730
},
{
"grad_norm": 0.7175264358520508,
"learning_rate": 9.848486076572787e-05,
"loss": 0.2345,
"step": 3740
},
{
"grad_norm": 0.6926993727684021,
"learning_rate": 9.847136601196446e-05,
"loss": 0.2406,
"step": 3750
},
{
"grad_norm": 0.6136336326599121,
"learning_rate": 9.845781236092078e-05,
"loss": 0.2218,
"step": 3760
},
{
"grad_norm": 0.9006650447845459,
"learning_rate": 9.844419982906583e-05,
"loss": 0.2358,
"step": 3770
},
{
"grad_norm": 0.7359917163848877,
"learning_rate": 9.843052843294008e-05,
"loss": 0.2309,
"step": 3780
},
{
"grad_norm": 0.7071955800056458,
"learning_rate": 9.841679818915559e-05,
"loss": 0.2203,
"step": 3790
},
{
"grad_norm": 0.7072486877441406,
"learning_rate": 9.840300911439591e-05,
"loss": 0.2468,
"step": 3800
},
{
"grad_norm": 0.8127968907356262,
"learning_rate": 9.838916122541603e-05,
"loss": 0.2171,
"step": 3810
},
{
"grad_norm": 0.6996515393257141,
"learning_rate": 9.837525453904246e-05,
"loss": 0.2308,
"step": 3820
},
{
"grad_norm": 0.7189144492149353,
"learning_rate": 9.836128907217314e-05,
"loss": 0.2145,
"step": 3830
},
{
"grad_norm": 0.6873388886451721,
"learning_rate": 9.834726484177743e-05,
"loss": 0.213,
"step": 3840
},
{
"grad_norm": 0.7843382358551025,
"learning_rate": 9.833318186489609e-05,
"loss": 0.2342,
"step": 3850
},
{
"grad_norm": 0.7304880619049072,
"learning_rate": 9.831904015864126e-05,
"loss": 0.2163,
"step": 3860
},
{
"grad_norm": 0.6932791471481323,
"learning_rate": 9.830483974019645e-05,
"loss": 0.2149,
"step": 3870
},
{
"grad_norm": 0.7439470291137695,
"learning_rate": 9.82905806268165e-05,
"loss": 0.2141,
"step": 3880
},
{
"grad_norm": 0.6758773326873779,
"learning_rate": 9.82762628358276e-05,
"loss": 0.2253,
"step": 3890
},
{
"grad_norm": 0.6273120045661926,
"learning_rate": 9.826188638462718e-05,
"loss": 0.2195,
"step": 3900
},
{
"grad_norm": 0.6467409133911133,
"learning_rate": 9.824745129068402e-05,
"loss": 0.2202,
"step": 3910
},
{
"grad_norm": 0.582391083240509,
"learning_rate": 9.82329575715381e-05,
"loss": 0.2294,
"step": 3920
},
{
"grad_norm": 0.6178143620491028,
"learning_rate": 9.821840524480066e-05,
"loss": 0.2135,
"step": 3930
},
{
"grad_norm": 0.5768554210662842,
"learning_rate": 9.820379432815414e-05,
"loss": 0.2295,
"step": 3940
},
{
"grad_norm": 0.740070104598999,
"learning_rate": 9.81891248393522e-05,
"loss": 0.2203,
"step": 3950
},
{
"grad_norm": 0.667249858379364,
"learning_rate": 9.817439679621963e-05,
"loss": 0.2249,
"step": 3960
},
{
"grad_norm": 0.8875001072883606,
"learning_rate": 9.815961021665243e-05,
"loss": 0.2121,
"step": 3970
},
{
"grad_norm": 0.7105627059936523,
"learning_rate": 9.814476511861763e-05,
"loss": 0.2121,
"step": 3980
},
{
"grad_norm": 0.7251412868499756,
"learning_rate": 9.812986152015348e-05,
"loss": 0.2132,
"step": 3990
},
{
"grad_norm": 0.7231500744819641,
"learning_rate": 9.811489943936922e-05,
"loss": 0.2142,
"step": 4000
},
{
"grad_norm": 0.7124753594398499,
"learning_rate": 9.809987889444522e-05,
"loss": 0.2069,
"step": 4010
},
{
"grad_norm": 0.7132613658905029,
"learning_rate": 9.808479990363282e-05,
"loss": 0.2087,
"step": 4020
},
{
"grad_norm": 0.7888582348823547,
"learning_rate": 9.806966248525445e-05,
"loss": 0.2103,
"step": 4030
},
{
"grad_norm": 0.6670511960983276,
"learning_rate": 9.805446665770348e-05,
"loss": 0.1908,
"step": 4040
},
{
"grad_norm": 0.6733547449111938,
"learning_rate": 9.803921243944429e-05,
"loss": 0.2074,
"step": 4050
},
{
"grad_norm": 0.6576470136642456,
"learning_rate": 9.802389984901218e-05,
"loss": 0.2028,
"step": 4060
},
{
"grad_norm": 0.645409882068634,
"learning_rate": 9.80085289050134e-05,
"loss": 0.1958,
"step": 4070
},
{
"grad_norm": 0.7557162642478943,
"learning_rate": 9.799309962612508e-05,
"loss": 0.2132,
"step": 4080
},
{
"grad_norm": 0.7431875467300415,
"learning_rate": 9.797761203109527e-05,
"loss": 0.2032,
"step": 4090
},
{
"grad_norm": 0.6959190964698792,
"learning_rate": 9.796206613874283e-05,
"loss": 0.1987,
"step": 4100
},
{
"grad_norm": 0.7518453001976013,
"learning_rate": 9.794646196795754e-05,
"loss": 0.2081,
"step": 4110
},
{
"grad_norm": 0.6748153567314148,
"learning_rate": 9.793079953769987e-05,
"loss": 0.2058,
"step": 4120
},
{
"grad_norm": 0.6845585703849792,
"learning_rate": 9.79150788670012e-05,
"loss": 0.1953,
"step": 4130
},
{
"grad_norm": 0.899891197681427,
"learning_rate": 9.78992999749636e-05,
"loss": 0.2028,
"step": 4140
},
{
"grad_norm": 0.7274357676506042,
"learning_rate": 9.788346288075994e-05,
"loss": 0.2114,
"step": 4150
},
{
"grad_norm": 0.7095952033996582,
"learning_rate": 9.786756760363373e-05,
"loss": 0.1954,
"step": 4160
},
{
"grad_norm": 0.6773634552955627,
"learning_rate": 9.78516141628993e-05,
"loss": 0.1959,
"step": 4170
},
{
"grad_norm": 0.874722957611084,
"learning_rate": 9.783560257794154e-05,
"loss": 0.1869,
"step": 4180
},
{
"grad_norm": 0.7582895159721375,
"learning_rate": 9.781953286821603e-05,
"loss": 0.193,
"step": 4190
},
{
"grad_norm": 0.7728500962257385,
"learning_rate": 9.780340505324901e-05,
"loss": 0.1897,
"step": 4200
},
{
"grad_norm": 0.6582096815109253,
"learning_rate": 9.778721915263727e-05,
"loss": 0.1881,
"step": 4210
},
{
"grad_norm": 0.6903952360153198,
"learning_rate": 9.777097518604824e-05,
"loss": 0.1981,
"step": 4220
},
{
"grad_norm": 0.6549860239028931,
"learning_rate": 9.775467317321984e-05,
"loss": 0.2055,
"step": 4230
},
{
"grad_norm": 0.6893287301063538,
"learning_rate": 9.773831313396055e-05,
"loss": 0.1983,
"step": 4240
},
{
"grad_norm": 0.74331134557724,
"learning_rate": 9.77218950881494e-05,
"loss": 0.1957,
"step": 4250
},
{
"grad_norm": 0.7214449048042297,
"learning_rate": 9.770541905573583e-05,
"loss": 0.1961,
"step": 4260
},
{
"grad_norm": 0.7280714511871338,
"learning_rate": 9.768888505673976e-05,
"loss": 0.1945,
"step": 4270
},
{
"grad_norm": 0.7373641133308411,
"learning_rate": 9.767229311125162e-05,
"loss": 0.1832,
"step": 4280
},
{
"grad_norm": 0.7736567258834839,
"learning_rate": 9.765564323943211e-05,
"loss": 0.185,
"step": 4290
},
{
"grad_norm": 0.6676798462867737,
"learning_rate": 9.763893546151244e-05,
"loss": 0.177,
"step": 4300
},
{
"grad_norm": 0.7867143750190735,
"learning_rate": 9.762216979779412e-05,
"loss": 0.1849,
"step": 4310
},
{
"grad_norm": 0.7856062054634094,
"learning_rate": 9.760534626864902e-05,
"loss": 0.2078,
"step": 4320
},
{
"grad_norm": 0.7747724652290344,
"learning_rate": 9.758846489451931e-05,
"loss": 0.1781,
"step": 4330
},
{
"grad_norm": 0.5981408953666687,
"learning_rate": 9.757152569591748e-05,
"loss": 0.1764,
"step": 4340
},
{
"grad_norm": 0.5866038203239441,
"learning_rate": 9.75545286934262e-05,
"loss": 0.1699,
"step": 4350
},
{
"grad_norm": 0.7155736088752747,
"learning_rate": 9.753747390769847e-05,
"loss": 0.1705,
"step": 4360
},
{
"grad_norm": 0.8444572687149048,
"learning_rate": 9.752036135945744e-05,
"loss": 0.1804,
"step": 4370
},
{
"grad_norm": 0.7528351545333862,
"learning_rate": 9.750319106949649e-05,
"loss": 0.1781,
"step": 4380
},
{
"grad_norm": 0.6849138736724854,
"learning_rate": 9.748596305867913e-05,
"loss": 0.1638,
"step": 4390
},
{
"grad_norm": 0.6855719089508057,
"learning_rate": 9.746867734793903e-05,
"loss": 0.1791,
"step": 4400
},
{
"grad_norm": 0.7258173823356628,
"learning_rate": 9.745133395827993e-05,
"loss": 0.175,
"step": 4410
},
{
"grad_norm": 0.5329680442810059,
"learning_rate": 9.743393291077572e-05,
"loss": 0.1871,
"step": 4420
},
{
"grad_norm": 0.686873733997345,
"learning_rate": 9.741647422657028e-05,
"loss": 0.1802,
"step": 4430
},
{
"grad_norm": 0.706271767616272,
"learning_rate": 9.739895792687758e-05,
"loss": 0.1722,
"step": 4440
},
{
"grad_norm": 0.7103954553604126,
"learning_rate": 9.738138403298157e-05,
"loss": 0.1814,
"step": 4450
},
{
"grad_norm": 0.6127023100852966,
"learning_rate": 9.736375256623619e-05,
"loss": 0.1664,
"step": 4460
},
{
"grad_norm": 0.6501713991165161,
"learning_rate": 9.734606354806533e-05,
"loss": 0.1621,
"step": 4470
},
{
"grad_norm": 0.6074185371398926,
"learning_rate": 9.73283169999628e-05,
"loss": 0.161,
"step": 4480
},
{
"grad_norm": 0.9301014542579651,
"learning_rate": 9.731051294349238e-05,
"loss": 0.1638,
"step": 4490
},
{
"grad_norm": 0.7157642841339111,
"learning_rate": 9.729265140028762e-05,
"loss": 0.1581,
"step": 4500
},
{
"grad_norm": 0.7186063528060913,
"learning_rate": 9.727473239205201e-05,
"loss": 0.1852,
"step": 4510
},
{
"grad_norm": 0.690217137336731,
"learning_rate": 9.725675594055883e-05,
"loss": 0.1572,
"step": 4520
},
{
"grad_norm": 0.7358818650245667,
"learning_rate": 9.723872206765116e-05,
"loss": 0.1545,
"step": 4530
},
{
"grad_norm": 0.8105913400650024,
"learning_rate": 9.722063079524185e-05,
"loss": 0.1615,
"step": 4540
},
{
"grad_norm": 0.7895976901054382,
"learning_rate": 9.720248214531351e-05,
"loss": 0.1613,
"step": 4550
},
{
"grad_norm": 0.8814677000045776,
"learning_rate": 9.718427613991848e-05,
"loss": 0.1475,
"step": 4560
},
{
"grad_norm": 0.7542166113853455,
"learning_rate": 9.716601280117873e-05,
"loss": 0.1562,
"step": 4570
},
{
"grad_norm": 0.844075620174408,
"learning_rate": 9.714769215128596e-05,
"loss": 0.1455,
"step": 4580
},
{
"grad_norm": 0.743806004524231,
"learning_rate": 9.712931421250152e-05,
"loss": 0.1517,
"step": 4590
},
{
"grad_norm": 0.6393651962280273,
"learning_rate": 9.711087900715627e-05,
"loss": 0.1487,
"step": 4600
},
{
"grad_norm": 0.8024899959564209,
"learning_rate": 9.709238655765078e-05,
"loss": 0.1469,
"step": 4610
},
{
"grad_norm": 0.588691771030426,
"learning_rate": 9.707383688645511e-05,
"loss": 0.1335,
"step": 4620
},
{
"grad_norm": 0.7212942838668823,
"learning_rate": 9.705523001610883e-05,
"loss": 0.1437,
"step": 4630
},
{
"grad_norm": 0.6818627715110779,
"learning_rate": 9.703656596922107e-05,
"loss": 0.1401,
"step": 4640
},
{
"grad_norm": 0.7202953100204468,
"learning_rate": 9.70178447684704e-05,
"loss": 0.1261,
"step": 4650
},
{
"grad_norm": 0.7244252562522888,
"learning_rate": 9.699906643660483e-05,
"loss": 0.1387,
"step": 4660
},
{
"grad_norm": 0.7823927402496338,
"learning_rate": 9.698023099644185e-05,
"loss": 0.1286,
"step": 4670
},
{
"grad_norm": 0.8115151524543762,
"learning_rate": 9.696133847086823e-05,
"loss": 0.1289,
"step": 4680
},
{
"grad_norm": 0.8499891757965088,
"learning_rate": 9.694238888284022e-05,
"loss": 0.1421,
"step": 4690
},
{
"grad_norm": 0.7232232689857483,
"learning_rate": 9.692338225538333e-05,
"loss": 0.1268,
"step": 4700
},
{
"grad_norm": 0.7501139044761658,
"learning_rate": 9.690431861159241e-05,
"loss": 0.1249,
"step": 4710
},
{
"grad_norm": 0.6424766778945923,
"learning_rate": 9.688519797463161e-05,
"loss": 0.1199,
"step": 4720
},
{
"grad_norm": 0.5778738260269165,
"learning_rate": 9.686602036773426e-05,
"loss": 0.1388,
"step": 4730
},
{
"grad_norm": 0.7467142343521118,
"learning_rate": 9.684678581420302e-05,
"loss": 0.1239,
"step": 4740
},
{
"grad_norm": 0.6017611026763916,
"learning_rate": 9.682749433740962e-05,
"loss": 0.114,
"step": 4750
},
{
"grad_norm": 0.7175586819648743,
"learning_rate": 9.680814596079507e-05,
"loss": 0.1225,
"step": 4760
},
{
"grad_norm": 0.7513552308082581,
"learning_rate": 9.678874070786945e-05,
"loss": 0.1284,
"step": 4770
},
{
"grad_norm": 0.858982503414154,
"learning_rate": 9.676927860221199e-05,
"loss": 0.1391,
"step": 4780
},
{
"grad_norm": 0.7743598222732544,
"learning_rate": 9.674975966747097e-05,
"loss": 0.1124,
"step": 4790
},
{
"grad_norm": 0.7889891266822815,
"learning_rate": 9.673018392736374e-05,
"loss": 0.1189,
"step": 4800
},
{
"grad_norm": 0.6704162359237671,
"learning_rate": 9.671055140567667e-05,
"loss": 0.1173,
"step": 4810
},
{
"grad_norm": 0.7439952492713928,
"learning_rate": 9.669086212626511e-05,
"loss": 0.1181,
"step": 4820
},
{
"grad_norm": 1.075419306755066,
"learning_rate": 9.667111611305341e-05,
"loss": 0.1158,
"step": 4830
},
{
"grad_norm": 0.8188729286193848,
"learning_rate": 9.665131339003486e-05,
"loss": 0.1167,
"step": 4840
},
{
"grad_norm": 0.8400049209594727,
"learning_rate": 9.663145398127158e-05,
"loss": 0.115,
"step": 4850
},
{
"grad_norm": 0.6420927047729492,
"learning_rate": 9.661153791089467e-05,
"loss": 0.1013,
"step": 4860
},
{
"grad_norm": 0.6801747679710388,
"learning_rate": 9.659156520310402e-05,
"loss": 0.1051,
"step": 4870
},
{
"grad_norm": 0.6923588514328003,
"learning_rate": 9.657153588216835e-05,
"loss": 0.0991,
"step": 4880
},
{
"grad_norm": 0.7888909578323364,
"learning_rate": 9.655144997242516e-05,
"loss": 0.1205,
"step": 4890
},
{
"grad_norm": 0.6130096912384033,
"learning_rate": 9.653130749828075e-05,
"loss": 0.0925,
"step": 4900
},
{
"grad_norm": 0.7016632556915283,
"learning_rate": 9.65111084842101e-05,
"loss": 0.1031,
"step": 4910
},
{
"grad_norm": 0.7130147814750671,
"learning_rate": 9.649085295475695e-05,
"loss": 0.1045,
"step": 4920
},
{
"grad_norm": 0.7954491972923279,
"learning_rate": 9.647054093453365e-05,
"loss": 0.1062,
"step": 4930
},
{
"grad_norm": 0.6415190100669861,
"learning_rate": 9.645017244822123e-05,
"loss": 0.0934,
"step": 4940
},
{
"grad_norm": 0.7961637377738953,
"learning_rate": 9.642974752056931e-05,
"loss": 0.106,
"step": 4950
},
{
"grad_norm": 0.675407350063324,
"learning_rate": 9.640926617639613e-05,
"loss": 0.1008,
"step": 4960
},
{
"grad_norm": 0.7687718272209167,
"learning_rate": 9.638872844058843e-05,
"loss": 0.1089,
"step": 4970
},
{
"grad_norm": 0.7104296088218689,
"learning_rate": 9.63681343381015e-05,
"loss": 0.1057,
"step": 4980
},
{
"grad_norm": 0.6876987218856812,
"learning_rate": 9.634748389395914e-05,
"loss": 0.0971,
"step": 4990
},
{
"grad_norm": 0.8525477647781372,
"learning_rate": 9.632677713325353e-05,
"loss": 0.1037,
"step": 5000
},
{
"grad_norm": 1.067914605140686,
"learning_rate": 9.63060140811454e-05,
"loss": 0.0991,
"step": 5010
},
{
"grad_norm": 0.7823187708854675,
"learning_rate": 9.628519476286379e-05,
"loss": 0.0955,
"step": 5020
},
{
"grad_norm": 0.6563372611999512,
"learning_rate": 9.626431920370612e-05,
"loss": 0.0821,
"step": 5030
},
{
"grad_norm": 0.6931690573692322,
"learning_rate": 9.624338742903819e-05,
"loss": 0.0846,
"step": 5040
},
{
"grad_norm": 0.8801899552345276,
"learning_rate": 9.622239946429406e-05,
"loss": 0.1038,
"step": 5050
},
{
"grad_norm": 0.7647538781166077,
"learning_rate": 9.620135533497609e-05,
"loss": 0.0768,
"step": 5060
},
{
"grad_norm": 0.8259631395339966,
"learning_rate": 9.61802550666549e-05,
"loss": 0.1143,
"step": 5070
},
{
"grad_norm": 0.8379261493682861,
"learning_rate": 9.615909868496928e-05,
"loss": 0.097,
"step": 5080
},
{
"grad_norm": 0.7736333608627319,
"learning_rate": 9.613788621562622e-05,
"loss": 0.0807,
"step": 5090
},
{
"grad_norm": 0.6134459376335144,
"learning_rate": 9.611661768440093e-05,
"loss": 0.0806,
"step": 5100
},
{
"grad_norm": 0.8735434412956238,
"learning_rate": 9.609529311713661e-05,
"loss": 0.0905,
"step": 5110
},
{
"grad_norm": 0.742935061454773,
"learning_rate": 9.607391253974466e-05,
"loss": 0.1067,
"step": 5120
},
{
"grad_norm": 0.660607099533081,
"learning_rate": 9.605247597820448e-05,
"loss": 0.1004,
"step": 5130
},
{
"grad_norm": 0.6215053796768188,
"learning_rate": 9.603098345856354e-05,
"loss": 0.0858,
"step": 5140
},
{
"grad_norm": 0.7790359258651733,
"learning_rate": 9.600943500693725e-05,
"loss": 0.0774,
"step": 5150
},
{
"grad_norm": 0.6415055394172668,
"learning_rate": 9.598783064950902e-05,
"loss": 0.1048,
"step": 5160
},
{
"grad_norm": 0.7468861937522888,
"learning_rate": 9.596617041253018e-05,
"loss": 0.0875,
"step": 5170
},
{
"grad_norm": 0.6383912563323975,
"learning_rate": 9.594445432231996e-05,
"loss": 0.0784,
"step": 5180
},
{
"grad_norm": 0.847959041595459,
"learning_rate": 9.592268240526547e-05,
"loss": 0.1072,
"step": 5190
},
{
"grad_norm": 0.7248074412345886,
"learning_rate": 9.590085468782162e-05,
"loss": 0.0738,
"step": 5200
},
{
"grad_norm": 0.7571699023246765,
"learning_rate": 9.587897119651116e-05,
"loss": 0.0844,
"step": 5210
},
{
"grad_norm": 0.7206233143806458,
"learning_rate": 9.585703195792459e-05,
"loss": 0.0813,
"step": 5220
},
{
"grad_norm": 0.8579068183898926,
"learning_rate": 9.583503699872016e-05,
"loss": 0.0801,
"step": 5230
},
{
"grad_norm": 0.8239566087722778,
"learning_rate": 9.581298634562381e-05,
"loss": 0.0851,
"step": 5240
},
{
"grad_norm": 0.7569472789764404,
"learning_rate": 9.579088002542917e-05,
"loss": 0.1023,
"step": 5250
},
{
"grad_norm": 0.8508474230766296,
"learning_rate": 9.57687180649975e-05,
"loss": 0.0826,
"step": 5260
},
{
"grad_norm": 0.7740755677223206,
"learning_rate": 9.574650049125768e-05,
"loss": 0.0821,
"step": 5270
},
{
"grad_norm": 0.7932382225990295,
"learning_rate": 9.572422733120614e-05,
"loss": 0.0937,
"step": 5280
},
{
"grad_norm": 0.717134952545166,
"learning_rate": 9.570189861190689e-05,
"loss": 0.094,
"step": 5290
},
{
"grad_norm": 0.6538495421409607,
"learning_rate": 9.56795143604914e-05,
"loss": 0.0785,
"step": 5300
},
{
"grad_norm": 0.7206899523735046,
"learning_rate": 9.565707460415869e-05,
"loss": 0.0711,
"step": 5310
},
{
"grad_norm": 0.7034664750099182,
"learning_rate": 9.563457937017515e-05,
"loss": 0.0782,
"step": 5320
},
{
"grad_norm": 0.7813044190406799,
"learning_rate": 9.56120286858746e-05,
"loss": 0.0851,
"step": 5330
},
{
"grad_norm": 0.7272282242774963,
"learning_rate": 9.558942257865829e-05,
"loss": 0.0751,
"step": 5340
},
{
"grad_norm": 0.9297890067100525,
"learning_rate": 9.556676107599472e-05,
"loss": 0.0677,
"step": 5350
},
{
"grad_norm": 0.635347843170166,
"learning_rate": 9.554404420541978e-05,
"loss": 0.0713,
"step": 5360
},
{
"grad_norm": 0.6273652911186218,
"learning_rate": 9.55212719945366e-05,
"loss": 0.0579,
"step": 5370
},
{
"grad_norm": 0.6494834423065186,
"learning_rate": 9.549844447101559e-05,
"loss": 0.0592,
"step": 5380
},
{
"grad_norm": 0.7968246936798096,
"learning_rate": 9.547556166259433e-05,
"loss": 0.0608,
"step": 5390
},
{
"grad_norm": 0.8095596432685852,
"learning_rate": 9.545262359707756e-05,
"loss": 0.0714,
"step": 5400
},
{
"grad_norm": 0.8394306302070618,
"learning_rate": 9.542963030233724e-05,
"loss": 0.0679,
"step": 5410
},
{
"grad_norm": 0.8043311834335327,
"learning_rate": 9.540658180631237e-05,
"loss": 0.0617,
"step": 5420
},
{
"grad_norm": 0.8117454648017883,
"learning_rate": 9.538347813700904e-05,
"loss": 0.0602,
"step": 5430
},
{
"grad_norm": 0.7123355865478516,
"learning_rate": 9.536031932250036e-05,
"loss": 0.0793,
"step": 5440
},
{
"grad_norm": 0.8153714537620544,
"learning_rate": 9.533710539092653e-05,
"loss": 0.0489,
"step": 5450
},
{
"grad_norm": 0.9228285551071167,
"learning_rate": 9.531383637049464e-05,
"loss": 0.0564,
"step": 5460
},
{
"grad_norm": 0.7224162220954895,
"learning_rate": 9.529051228947875e-05,
"loss": 0.0558,
"step": 5470
},
{
"grad_norm": 0.8940768241882324,
"learning_rate": 9.52671331762198e-05,
"loss": 0.0604,
"step": 5480
},
{
"grad_norm": 0.8763455152511597,
"learning_rate": 9.524369905912565e-05,
"loss": 0.0658,
"step": 5490
},
{
"grad_norm": 0.8506941795349121,
"learning_rate": 9.522020996667092e-05,
"loss": 0.067,
"step": 5500
},
{
"grad_norm": 0.6657439470291138,
"learning_rate": 9.519666592739709e-05,
"loss": 0.0683,
"step": 5510
},
{
"grad_norm": 0.7618071436882019,
"learning_rate": 9.517306696991241e-05,
"loss": 0.0574,
"step": 5520
},
{
"grad_norm": 0.6481887102127075,
"learning_rate": 9.51494131228918e-05,
"loss": 0.0582,
"step": 5530
},
{
"grad_norm": 0.6721943616867065,
"learning_rate": 9.512570441507695e-05,
"loss": 0.0498,
"step": 5540
},
{
"grad_norm": 0.7679053544998169,
"learning_rate": 9.510194087527615e-05,
"loss": 0.054,
"step": 5550
},
{
"grad_norm": 0.7131364941596985,
"learning_rate": 9.507812253236435e-05,
"loss": 0.0552,
"step": 5560
},
{
"grad_norm": 0.7968480587005615,
"learning_rate": 9.505424941528309e-05,
"loss": 0.0647,
"step": 5570
},
{
"grad_norm": 0.7074773907661438,
"learning_rate": 9.503032155304046e-05,
"loss": 0.0767,
"step": 5580
},
{
"grad_norm": 0.68541020154953,
"learning_rate": 9.500633897471106e-05,
"loss": 0.0574,
"step": 5590
},
{
"grad_norm": 0.8278937339782715,
"learning_rate": 9.498230170943596e-05,
"loss": 0.0717,
"step": 5600
},
{
"grad_norm": 0.7747924327850342,
"learning_rate": 9.495820978642275e-05,
"loss": 0.0599,
"step": 5610
},
{
"grad_norm": 0.774636447429657,
"learning_rate": 9.493406323494535e-05,
"loss": 0.0529,
"step": 5620
},
{
"grad_norm": 0.4860631227493286,
"learning_rate": 9.490986208434413e-05,
"loss": 0.0467,
"step": 5630
},
{
"grad_norm": 0.7017577290534973,
"learning_rate": 9.488560636402577e-05,
"loss": 0.0523,
"step": 5640
},
{
"grad_norm": 0.8965169787406921,
"learning_rate": 9.486129610346321e-05,
"loss": 0.0673,
"step": 5650
},
{
"grad_norm": 0.7559218406677246,
"learning_rate": 9.483693133219576e-05,
"loss": 0.0586,
"step": 5660
},
{
"grad_norm": 0.7744764685630798,
"learning_rate": 9.481251207982888e-05,
"loss": 0.0582,
"step": 5670
},
{
"grad_norm": 0.8262003064155579,
"learning_rate": 9.47880383760343e-05,
"loss": 0.0693,
"step": 5680
},
{
"grad_norm": 0.6478685140609741,
"learning_rate": 9.476351025054983e-05,
"loss": 0.0588,
"step": 5690
},
{
"grad_norm": 0.6692949533462524,
"learning_rate": 9.473892773317952e-05,
"loss": 0.0561,
"step": 5700
},
{
"grad_norm": 0.7504095435142517,
"learning_rate": 9.471429085379338e-05,
"loss": 0.0689,
"step": 5710
},
{
"grad_norm": 0.6733090877532959,
"learning_rate": 9.468959964232757e-05,
"loss": 0.0528,
"step": 5720
},
{
"grad_norm": 0.7132554054260254,
"learning_rate": 9.466485412878425e-05,
"loss": 0.055,
"step": 5730
},
{
"grad_norm": 0.544713020324707,
"learning_rate": 9.464005434323154e-05,
"loss": 0.0634,
"step": 5740
},
{
"grad_norm": 0.6640408039093018,
"learning_rate": 9.461520031580352e-05,
"loss": 0.0502,
"step": 5750
},
{
"grad_norm": 0.6543788313865662,
"learning_rate": 9.459029207670019e-05,
"loss": 0.0579,
"step": 5760
},
{
"grad_norm": 0.705933690071106,
"learning_rate": 9.456532965618737e-05,
"loss": 0.0574,
"step": 5770
},
{
"grad_norm": 0.7639266848564148,
"learning_rate": 9.454031308459681e-05,
"loss": 0.0681,
"step": 5780
},
{
"grad_norm": 0.7469704151153564,
"learning_rate": 9.451524239232595e-05,
"loss": 0.0547,
"step": 5790
},
{
"grad_norm": 0.8652675747871399,
"learning_rate": 9.449011760983809e-05,
"loss": 0.0685,
"step": 5800
},
{
"grad_norm": 0.6738976240158081,
"learning_rate": 9.446493876766218e-05,
"loss": 0.0588,
"step": 5810
},
{
"grad_norm": 0.6966020464897156,
"learning_rate": 9.44397058963929e-05,
"loss": 0.0587,
"step": 5820
},
{
"grad_norm": 0.5429330468177795,
"learning_rate": 9.441441902669056e-05,
"loss": 0.0621,
"step": 5830
},
{
"grad_norm": 0.6567779779434204,
"learning_rate": 9.43890781892811e-05,
"loss": 0.052,
"step": 5840
},
{
"grad_norm": 0.7521270513534546,
"learning_rate": 9.436368341495603e-05,
"loss": 0.0428,
"step": 5850
},
{
"grad_norm": 0.6410270929336548,
"learning_rate": 9.43382347345724e-05,
"loss": 0.0624,
"step": 5860
},
{
"grad_norm": 0.7417052388191223,
"learning_rate": 9.431273217905273e-05,
"loss": 0.0562,
"step": 5870
},
{
"grad_norm": 0.8881665468215942,
"learning_rate": 9.428717577938504e-05,
"loss": 0.0623,
"step": 5880
},
{
"grad_norm": 0.5850077867507935,
"learning_rate": 9.426156556662276e-05,
"loss": 0.0582,
"step": 5890
},
{
"grad_norm": 0.4818542003631592,
"learning_rate": 9.423590157188474e-05,
"loss": 0.0741,
"step": 5900
},
{
"grad_norm": 0.7062960863113403,
"learning_rate": 9.421018382635513e-05,
"loss": 0.0638,
"step": 5910
},
{
"grad_norm": 0.7811799645423889,
"learning_rate": 9.418441236128343e-05,
"loss": 0.0525,
"step": 5920
},
{
"grad_norm": 0.6753416061401367,
"learning_rate": 9.41585872079844e-05,
"loss": 0.0481,
"step": 5930
},
{
"grad_norm": 0.8710509538650513,
"learning_rate": 9.413270839783802e-05,
"loss": 0.0549,
"step": 5940
},
{
"grad_norm": 0.7185320854187012,
"learning_rate": 9.41067759622895e-05,
"loss": 0.0553,
"step": 5950
},
{
"grad_norm": 0.6032108664512634,
"learning_rate": 9.408078993284917e-05,
"loss": 0.0501,
"step": 5960
},
{
"grad_norm": 0.8627631068229675,
"learning_rate": 9.405475034109254e-05,
"loss": 0.06,
"step": 5970
},
{
"grad_norm": 0.8705095052719116,
"learning_rate": 9.402865721866015e-05,
"loss": 0.0557,
"step": 5980
},
{
"grad_norm": 0.692966878414154,
"learning_rate": 9.400251059725762e-05,
"loss": 0.066,
"step": 5990
},
{
"grad_norm": 0.6884616017341614,
"learning_rate": 9.397631050865554e-05,
"loss": 0.0741,
"step": 6000
},
{
"grad_norm": 0.62348473072052,
"learning_rate": 9.395005698468949e-05,
"loss": 0.0533,
"step": 6010
},
{
"grad_norm": 0.6259242296218872,
"learning_rate": 9.392375005726e-05,
"loss": 0.0548,
"step": 6020
},
{
"grad_norm": 0.5877556204795837,
"learning_rate": 9.389738975833242e-05,
"loss": 0.0516,
"step": 6030
},
{
"grad_norm": 0.7416625618934631,
"learning_rate": 9.387097611993707e-05,
"loss": 0.0481,
"step": 6040
},
{
"grad_norm": 0.9062529802322388,
"learning_rate": 9.384450917416894e-05,
"loss": 0.0576,
"step": 6050
},
{
"grad_norm": 0.7620342373847961,
"learning_rate": 9.381798895318792e-05,
"loss": 0.0494,
"step": 6060
},
{
"grad_norm": 0.6981661319732666,
"learning_rate": 9.379141548921853e-05,
"loss": 0.0536,
"step": 6070
},
{
"grad_norm": 0.6538633108139038,
"learning_rate": 9.376478881455009e-05,
"loss": 0.0582,
"step": 6080
},
{
"grad_norm": 0.6782255172729492,
"learning_rate": 9.373810896153647e-05,
"loss": 0.0563,
"step": 6090
},
{
"grad_norm": 0.7048159241676331,
"learning_rate": 9.371137596259623e-05,
"loss": 0.0627,
"step": 6100
},
{
"grad_norm": 0.8252936601638794,
"learning_rate": 9.368458985021248e-05,
"loss": 0.0411,
"step": 6110
},
{
"grad_norm": 0.5946969985961914,
"learning_rate": 9.365775065693287e-05,
"loss": 0.0517,
"step": 6120
},
{
"grad_norm": 0.7631013989448547,
"learning_rate": 9.363085841536957e-05,
"loss": 0.0413,
"step": 6130
},
{
"grad_norm": 0.5735177993774414,
"learning_rate": 9.360391315819917e-05,
"loss": 0.0493,
"step": 6140
},
{
"grad_norm": 0.7660733461380005,
"learning_rate": 9.357691491816269e-05,
"loss": 0.0674,
"step": 6150
},
{
"grad_norm": 0.7514684796333313,
"learning_rate": 9.354986372806557e-05,
"loss": 0.0714,
"step": 6160
},
{
"grad_norm": 0.6969489455223083,
"learning_rate": 9.352275962077752e-05,
"loss": 0.0577,
"step": 6170
},
{
"grad_norm": 0.563296914100647,
"learning_rate": 9.349560262923262e-05,
"loss": 0.046,
"step": 6180
},
{
"grad_norm": 0.6139219403266907,
"learning_rate": 9.346839278642913e-05,
"loss": 0.0617,
"step": 6190
},
{
"grad_norm": 0.6578940749168396,
"learning_rate": 9.344113012542963e-05,
"loss": 0.0486,
"step": 6200
},
{
"grad_norm": 0.748124361038208,
"learning_rate": 9.341381467936079e-05,
"loss": 0.0548,
"step": 6210
},
{
"grad_norm": 0.8145461082458496,
"learning_rate": 9.338644648141346e-05,
"loss": 0.0441,
"step": 6220
},
{
"grad_norm": 0.7148629426956177,
"learning_rate": 9.335902556484257e-05,
"loss": 0.0491,
"step": 6230
},
{
"grad_norm": 0.6139039397239685,
"learning_rate": 9.333155196296712e-05,
"loss": 0.0536,
"step": 6240
},
{
"grad_norm": 0.729573130607605,
"learning_rate": 9.330402570917016e-05,
"loss": 0.0565,
"step": 6250
},
{
"grad_norm": 0.6581088304519653,
"learning_rate": 9.327644683689865e-05,
"loss": 0.0387,
"step": 6260
},
{
"grad_norm": 0.6218430995941162,
"learning_rate": 9.324881537966354e-05,
"loss": 0.0492,
"step": 6270
},
{
"grad_norm": 0.7406831979751587,
"learning_rate": 9.322113137103964e-05,
"loss": 0.0437,
"step": 6280
},
{
"grad_norm": 0.7978484034538269,
"learning_rate": 9.319339484466564e-05,
"loss": 0.0473,
"step": 6290
},
{
"grad_norm": 0.5489411950111389,
"learning_rate": 9.316560583424403e-05,
"loss": 0.0488,
"step": 6300
},
{
"grad_norm": 0.6149821877479553,
"learning_rate": 9.313776437354108e-05,
"loss": 0.0412,
"step": 6310
},
{
"grad_norm": 0.6608788967132568,
"learning_rate": 9.31098704963868e-05,
"loss": 0.0457,
"step": 6320
},
{
"grad_norm": 0.5572587251663208,
"learning_rate": 9.308192423667485e-05,
"loss": 0.041,
"step": 6330
},
{
"grad_norm": 0.7157959938049316,
"learning_rate": 9.305392562836262e-05,
"loss": 0.0567,
"step": 6340
},
{
"grad_norm": 0.5528120398521423,
"learning_rate": 9.3025874705471e-05,
"loss": 0.044,
"step": 6350
},
{
"grad_norm": 0.7448693513870239,
"learning_rate": 9.299777150208457e-05,
"loss": 0.0483,
"step": 6360
},
{
"grad_norm": 0.7564453482627869,
"learning_rate": 9.296961605235133e-05,
"loss": 0.0429,
"step": 6370
},
{
"grad_norm": 0.6976820230484009,
"learning_rate": 9.29414083904828e-05,
"loss": 0.0437,
"step": 6380
},
{
"grad_norm": 0.7408413887023926,
"learning_rate": 9.291314855075397e-05,
"loss": 0.0554,
"step": 6390
},
{
"grad_norm": 0.6670358777046204,
"learning_rate": 9.288483656750322e-05,
"loss": 0.0568,
"step": 6400
},
{
"grad_norm": 0.5593525171279907,
"learning_rate": 9.285647247513225e-05,
"loss": 0.0379,
"step": 6410
},
{
"grad_norm": 0.6091060042381287,
"learning_rate": 9.282805630810614e-05,
"loss": 0.0506,
"step": 6420
},
{
"grad_norm": 0.7222121953964233,
"learning_rate": 9.279958810095317e-05,
"loss": 0.0483,
"step": 6430
},
{
"grad_norm": 0.6954823732376099,
"learning_rate": 9.277106788826494e-05,
"loss": 0.0465,
"step": 6440
},
{
"grad_norm": 0.6750673651695251,
"learning_rate": 9.274249570469618e-05,
"loss": 0.0435,
"step": 6450
},
{
"grad_norm": 0.7022892832756042,
"learning_rate": 9.271387158496476e-05,
"loss": 0.0436,
"step": 6460
},
{
"grad_norm": 0.6073474287986755,
"learning_rate": 9.268519556385173e-05,
"loss": 0.0433,
"step": 6470
},
{
"grad_norm": 0.7318218350410461,
"learning_rate": 9.265646767620113e-05,
"loss": 0.0373,
"step": 6480
},
{
"grad_norm": 0.6596046090126038,
"learning_rate": 9.262768795692006e-05,
"loss": 0.0506,
"step": 6490
},
{
"grad_norm": 0.6574918031692505,
"learning_rate": 9.259885644097861e-05,
"loss": 0.0366,
"step": 6500
},
{
"grad_norm": 0.572862982749939,
"learning_rate": 9.256997316340976e-05,
"loss": 0.0371,
"step": 6510
},
{
"grad_norm": 0.6681982278823853,
"learning_rate": 9.254103815930943e-05,
"loss": 0.047,
"step": 6520
},
{
"grad_norm": 0.6769469380378723,
"learning_rate": 9.251205146383636e-05,
"loss": 0.0408,
"step": 6530
},
{
"grad_norm": 0.6793182492256165,
"learning_rate": 9.248301311221216e-05,
"loss": 0.0517,
"step": 6540
},
{
"grad_norm": 0.7073116898536682,
"learning_rate": 9.245392313972115e-05,
"loss": 0.0392,
"step": 6550
},
{
"grad_norm": 0.7455390095710754,
"learning_rate": 9.242478158171037e-05,
"loss": 0.0481,
"step": 6560
},
{
"grad_norm": 0.6631140112876892,
"learning_rate": 9.239558847358957e-05,
"loss": 0.0527,
"step": 6570
},
{
"grad_norm": 0.6302623748779297,
"learning_rate": 9.236634385083114e-05,
"loss": 0.0372,
"step": 6580
},
{
"grad_norm": 0.7308129072189331,
"learning_rate": 9.233704774897005e-05,
"loss": 0.0593,
"step": 6590
},
{
"grad_norm": 0.6928110718727112,
"learning_rate": 9.230770020360382e-05,
"loss": 0.0547,
"step": 6600
},
{
"grad_norm": 0.6381212472915649,
"learning_rate": 9.227830125039247e-05,
"loss": 0.0404,
"step": 6610
},
{
"grad_norm": 0.5553435683250427,
"learning_rate": 9.224885092505853e-05,
"loss": 0.0391,
"step": 6620
},
{
"grad_norm": 0.6339812278747559,
"learning_rate": 9.221934926338691e-05,
"loss": 0.0387,
"step": 6630
},
{
"grad_norm": 0.588132917881012,
"learning_rate": 9.21897963012249e-05,
"loss": 0.0349,
"step": 6640
},
{
"grad_norm": 0.6249549388885498,
"learning_rate": 9.216019207448217e-05,
"loss": 0.0535,
"step": 6650
},
{
"grad_norm": 0.49740129709243774,
"learning_rate": 9.21305366191306e-05,
"loss": 0.035,
"step": 6660
},
{
"grad_norm": 0.6083484292030334,
"learning_rate": 9.21008299712044e-05,
"loss": 0.0342,
"step": 6670
},
{
"grad_norm": 0.5702409148216248,
"learning_rate": 9.207107216679993e-05,
"loss": 0.0478,
"step": 6680
},
{
"grad_norm": 0.7977477312088013,
"learning_rate": 9.204126324207574e-05,
"loss": 0.0493,
"step": 6690
},
{
"grad_norm": 0.8262686729431152,
"learning_rate": 9.201140323325248e-05,
"loss": 0.0409,
"step": 6700
},
{
"grad_norm": 0.536731481552124,
"learning_rate": 9.198149217661287e-05,
"loss": 0.0495,
"step": 6710
},
{
"grad_norm": 0.5465314984321594,
"learning_rate": 9.195153010850167e-05,
"loss": 0.0535,
"step": 6720
},
{
"grad_norm": 0.6884210705757141,
"learning_rate": 9.192151706532562e-05,
"loss": 0.0442,
"step": 6730
},
{
"grad_norm": 0.6468873023986816,
"learning_rate": 9.189145308355337e-05,
"loss": 0.0702,
"step": 6740
},
{
"grad_norm": 0.5701888799667358,
"learning_rate": 9.186133819971555e-05,
"loss": 0.041,
"step": 6750
},
{
"grad_norm": 0.7664920687675476,
"learning_rate": 9.183117245040455e-05,
"loss": 0.0401,
"step": 6760
},
{
"grad_norm": 0.6721729636192322,
"learning_rate": 9.180095587227458e-05,
"loss": 0.0388,
"step": 6770
},
{
"grad_norm": 0.7036067247390747,
"learning_rate": 9.177068850204167e-05,
"loss": 0.0454,
"step": 6780
},
{
"grad_norm": 0.5693458914756775,
"learning_rate": 9.174037037648351e-05,
"loss": 0.0369,
"step": 6790
},
{
"grad_norm": 0.5763115882873535,
"learning_rate": 9.171000153243947e-05,
"loss": 0.0487,
"step": 6800
},
{
"grad_norm": 0.6329843401908875,
"learning_rate": 9.167958200681057e-05,
"loss": 0.0329,
"step": 6810
},
{
"grad_norm": 0.8023676872253418,
"learning_rate": 9.164911183655943e-05,
"loss": 0.0494,
"step": 6820
},
{
"grad_norm": 0.6421040892601013,
"learning_rate": 9.161859105871013e-05,
"loss": 0.0428,
"step": 6830
},
{
"grad_norm": 0.7556735873222351,
"learning_rate": 9.158801971034832e-05,
"loss": 0.0433,
"step": 6840
},
{
"grad_norm": 0.6411566734313965,
"learning_rate": 9.155739782862107e-05,
"loss": 0.042,
"step": 6850
},
{
"grad_norm": 0.7058485150337219,
"learning_rate": 9.152672545073687e-05,
"loss": 0.0392,
"step": 6860
},
{
"grad_norm": 0.6122462153434753,
"learning_rate": 9.149600261396551e-05,
"loss": 0.0605,
"step": 6870
},
{
"grad_norm": 0.5858967900276184,
"learning_rate": 9.146522935563817e-05,
"loss": 0.0483,
"step": 6880
},
{
"grad_norm": 0.6201536655426025,
"learning_rate": 9.143440571314723e-05,
"loss": 0.0467,
"step": 6890
},
{
"grad_norm": 0.6456612348556519,
"learning_rate": 9.140353172394636e-05,
"loss": 0.0489,
"step": 6900
},
{
"grad_norm": 0.7748300433158875,
"learning_rate": 9.137260742555032e-05,
"loss": 0.0406,
"step": 6910
},
{
"grad_norm": 0.6247785687446594,
"learning_rate": 9.13416328555351e-05,
"loss": 0.0465,
"step": 6920
},
{
"grad_norm": 0.6419253945350647,
"learning_rate": 9.131060805153769e-05,
"loss": 0.05,
"step": 6930
},
{
"grad_norm": 0.7388933897018433,
"learning_rate": 9.127953305125618e-05,
"loss": 0.0362,
"step": 6940
},
{
"grad_norm": 0.666292667388916,
"learning_rate": 9.124840789244958e-05,
"loss": 0.0354,
"step": 6950
},
{
"grad_norm": 0.6128909587860107,
"learning_rate": 9.121723261293794e-05,
"loss": 0.0387,
"step": 6960
},
{
"grad_norm": 0.6145889759063721,
"learning_rate": 9.118600725060214e-05,
"loss": 0.0441,
"step": 6970
},
{
"grad_norm": 0.6859312653541565,
"learning_rate": 9.115473184338392e-05,
"loss": 0.0514,
"step": 6980
},
{
"grad_norm": 0.7586023211479187,
"learning_rate": 9.11234064292859e-05,
"loss": 0.0344,
"step": 6990
},
{
"grad_norm": 0.6913263201713562,
"learning_rate": 9.109203104637137e-05,
"loss": 0.0448,
"step": 7000
},
{
"grad_norm": 0.5343066453933716,
"learning_rate": 9.10606057327644e-05,
"loss": 0.0486,
"step": 7010
},
{
"grad_norm": 0.6005744338035583,
"learning_rate": 9.10291305266497e-05,
"loss": 0.0362,
"step": 7020
},
{
"grad_norm": 0.858198881149292,
"learning_rate": 9.099760546627261e-05,
"loss": 0.0615,
"step": 7030
},
{
"grad_norm": 0.6950360536575317,
"learning_rate": 9.096603058993907e-05,
"loss": 0.0406,
"step": 7040
},
{
"grad_norm": 0.6369425058364868,
"learning_rate": 9.093440593601552e-05,
"loss": 0.0363,
"step": 7050
},
{
"grad_norm": 0.6743416786193848,
"learning_rate": 9.090273154292889e-05,
"loss": 0.0354,
"step": 7060
},
{
"grad_norm": 0.702789306640625,
"learning_rate": 9.087100744916656e-05,
"loss": 0.0428,
"step": 7070
},
{
"grad_norm": 0.6569333672523499,
"learning_rate": 9.08392336932763e-05,
"loss": 0.0619,
"step": 7080
},
{
"grad_norm": 0.7255575060844421,
"learning_rate": 9.080741031386619e-05,
"loss": 0.045,
"step": 7090
},
{
"grad_norm": 0.5746154189109802,
"learning_rate": 9.077553734960468e-05,
"loss": 0.0376,
"step": 7100
},
{
"grad_norm": 0.5880788564682007,
"learning_rate": 9.074361483922041e-05,
"loss": 0.0507,
"step": 7110
},
{
"grad_norm": 0.6927939653396606,
"learning_rate": 9.071164282150224e-05,
"loss": 0.0381,
"step": 7120
},
{
"grad_norm": 0.5855864882469177,
"learning_rate": 9.067962133529919e-05,
"loss": 0.0376,
"step": 7130
},
{
"grad_norm": 0.5436041951179504,
"learning_rate": 9.064755041952036e-05,
"loss": 0.0619,
"step": 7140
},
{
"grad_norm": 0.6878156065940857,
"learning_rate": 9.061543011313497e-05,
"loss": 0.0559,
"step": 7150
},
{
"grad_norm": 0.6747798919677734,
"learning_rate": 9.05832604551722e-05,
"loss": 0.0545,
"step": 7160
},
{
"grad_norm": 0.622610867023468,
"learning_rate": 9.055104148472124e-05,
"loss": 0.051,
"step": 7170
},
{
"grad_norm": 0.7226181626319885,
"learning_rate": 9.051877324093114e-05,
"loss": 0.0423,
"step": 7180
},
{
"grad_norm": 0.5505620837211609,
"learning_rate": 9.04864557630109e-05,
"loss": 0.0505,
"step": 7190
},
{
"grad_norm": 0.7394242286682129,
"learning_rate": 9.045408909022928e-05,
"loss": 0.0411,
"step": 7200
},
{
"grad_norm": 0.5183990597724915,
"learning_rate": 9.042167326191484e-05,
"loss": 0.0437,
"step": 7210
},
{
"grad_norm": 0.5491803884506226,
"learning_rate": 9.038920831745587e-05,
"loss": 0.0428,
"step": 7220
},
{
"grad_norm": 0.6325652599334717,
"learning_rate": 9.035669429630036e-05,
"loss": 0.0403,
"step": 7230
},
{
"grad_norm": 0.6306131482124329,
"learning_rate": 9.032413123795588e-05,
"loss": 0.0424,
"step": 7240
},
{
"grad_norm": 0.7279338240623474,
"learning_rate": 9.029151918198962e-05,
"loss": 0.0431,
"step": 7250
},
{
"grad_norm": 0.5826907157897949,
"learning_rate": 9.025885816802833e-05,
"loss": 0.0379,
"step": 7260
},
{
"grad_norm": 0.5880635976791382,
"learning_rate": 9.02261482357582e-05,
"loss": 0.0385,
"step": 7270
},
{
"grad_norm": 0.6185583472251892,
"learning_rate": 9.019338942492485e-05,
"loss": 0.0361,
"step": 7280
},
{
"grad_norm": 0.6939125657081604,
"learning_rate": 9.01605817753334e-05,
"loss": 0.0391,
"step": 7290
},
{
"grad_norm": 0.5710512399673462,
"learning_rate": 9.012772532684818e-05,
"loss": 0.0353,
"step": 7300
},
{
"grad_norm": 0.530907154083252,
"learning_rate": 9.00948201193929e-05,
"loss": 0.0349,
"step": 7310
},
{
"grad_norm": 0.4984493553638458,
"learning_rate": 9.006186619295047e-05,
"loss": 0.0347,
"step": 7320
},
{
"grad_norm": 0.5582161545753479,
"learning_rate": 9.002886358756305e-05,
"loss": 0.0386,
"step": 7330
},
{
"grad_norm": 0.5856560468673706,
"learning_rate": 8.999581234333188e-05,
"loss": 0.0421,
"step": 7340
},
{
"grad_norm": 0.5844770669937134,
"learning_rate": 8.996271250041734e-05,
"loss": 0.0336,
"step": 7350
},
{
"grad_norm": 0.6166563630104065,
"learning_rate": 8.99295640990389e-05,
"loss": 0.0375,
"step": 7360
},
{
"grad_norm": 0.5409370064735413,
"learning_rate": 8.989636717947494e-05,
"loss": 0.0391,
"step": 7370
},
{
"grad_norm": 0.5650860071182251,
"learning_rate": 8.986312178206291e-05,
"loss": 0.0358,
"step": 7380
},
{
"grad_norm": 0.5486690402030945,
"learning_rate": 8.982982794719904e-05,
"loss": 0.0401,
"step": 7390
},
{
"grad_norm": 0.561327338218689,
"learning_rate": 8.979648571533852e-05,
"loss": 0.0362,
"step": 7400
},
{
"grad_norm": 0.6369312405586243,
"learning_rate": 8.976309512699528e-05,
"loss": 0.0405,
"step": 7410
},
{
"grad_norm": 0.6592939496040344,
"learning_rate": 8.972965622274205e-05,
"loss": 0.0462,
"step": 7420
},
{
"grad_norm": 0.7104017734527588,
"learning_rate": 8.969616904321026e-05,
"loss": 0.044,
"step": 7430
},
{
"grad_norm": 0.5764422416687012,
"learning_rate": 8.966263362908998e-05,
"loss": 0.0412,
"step": 7440
},
{
"grad_norm": 0.5725396871566772,
"learning_rate": 8.962905002112989e-05,
"loss": 0.0364,
"step": 7450
},
{
"grad_norm": 0.6328506469726562,
"learning_rate": 8.959541826013724e-05,
"loss": 0.0415,
"step": 7460
},
{
"grad_norm": 0.7069430947303772,
"learning_rate": 8.956173838697781e-05,
"loss": 0.0457,
"step": 7470
},
{
"grad_norm": 0.674612820148468,
"learning_rate": 8.952801044257581e-05,
"loss": 0.0513,
"step": 7480
},
{
"grad_norm": 0.6343667507171631,
"learning_rate": 8.949423446791387e-05,
"loss": 0.0406,
"step": 7490
},
{
"grad_norm": 0.6388131380081177,
"learning_rate": 8.9460410504033e-05,
"loss": 0.0359,
"step": 7500
},
{
"grad_norm": 0.569694995880127,
"learning_rate": 8.942653859203248e-05,
"loss": 0.0342,
"step": 7510
},
{
"grad_norm": 0.5964517593383789,
"learning_rate": 8.93926187730699e-05,
"loss": 0.0335,
"step": 7520
},
{
"grad_norm": 0.5617465376853943,
"learning_rate": 8.935865108836102e-05,
"loss": 0.0307,
"step": 7530
},
{
"grad_norm": 0.69667649269104,
"learning_rate": 8.932463557917981e-05,
"loss": 0.0389,
"step": 7540
},
{
"grad_norm": 0.6280410885810852,
"learning_rate": 8.929057228685829e-05,
"loss": 0.0377,
"step": 7550
},
{
"grad_norm": 0.6203128099441528,
"learning_rate": 8.925646125278658e-05,
"loss": 0.0376,
"step": 7560
},
{
"grad_norm": 0.6904357671737671,
"learning_rate": 8.92223025184128e-05,
"loss": 0.0468,
"step": 7570
},
{
"grad_norm": 0.561978816986084,
"learning_rate": 8.918809612524303e-05,
"loss": 0.0471,
"step": 7580
},
{
"grad_norm": 0.6063469052314758,
"learning_rate": 8.91538421148413e-05,
"loss": 0.0458,
"step": 7590
},
{
"grad_norm": 0.6514226198196411,
"learning_rate": 8.91195405288294e-05,
"loss": 0.0554,
"step": 7600
},
{
"grad_norm": 0.5756860375404358,
"learning_rate": 8.908519140888705e-05,
"loss": 0.0462,
"step": 7610
},
{
"grad_norm": 0.5900998711585999,
"learning_rate": 8.905079479675164e-05,
"loss": 0.0434,
"step": 7620
},
{
"grad_norm": 0.6459131240844727,
"learning_rate": 8.901635073421831e-05,
"loss": 0.0431,
"step": 7630
},
{
"grad_norm": 0.640886127948761,
"learning_rate": 8.898185926313983e-05,
"loss": 0.0381,
"step": 7640
},
{
"grad_norm": 0.6099423766136169,
"learning_rate": 8.894732042542659e-05,
"loss": 0.0377,
"step": 7650
},
{
"grad_norm": 0.602096676826477,
"learning_rate": 8.891273426304654e-05,
"loss": 0.032,
"step": 7660
},
{
"grad_norm": 0.5653091669082642,
"learning_rate": 8.887810081802515e-05,
"loss": 0.0408,
"step": 7670
},
{
"grad_norm": 0.6741763949394226,
"learning_rate": 8.884342013244529e-05,
"loss": 0.0428,
"step": 7680
},
{
"grad_norm": 0.5302419066429138,
"learning_rate": 8.880869224844725e-05,
"loss": 0.0402,
"step": 7690
},
{
"grad_norm": 0.5603381395339966,
"learning_rate": 8.877391720822875e-05,
"loss": 0.0317,
"step": 7700
},
{
"grad_norm": 0.5764555931091309,
"learning_rate": 8.873909505404468e-05,
"loss": 0.0474,
"step": 7710
},
{
"grad_norm": 0.6855901479721069,
"learning_rate": 8.870422582820726e-05,
"loss": 0.0466,
"step": 7720
},
{
"grad_norm": 0.5991953611373901,
"learning_rate": 8.866930957308588e-05,
"loss": 0.0453,
"step": 7730
},
{
"grad_norm": 0.6506611704826355,
"learning_rate": 8.86343463311071e-05,
"loss": 0.0441,
"step": 7740
},
{
"grad_norm": 0.5911173224449158,
"learning_rate": 8.859933614475452e-05,
"loss": 0.0325,
"step": 7750
},
{
"grad_norm": 0.7128370404243469,
"learning_rate": 8.85642790565689e-05,
"loss": 0.0427,
"step": 7760
},
{
"grad_norm": 0.5593247413635254,
"learning_rate": 8.852917510914782e-05,
"loss": 0.041,
"step": 7770
},
{
"grad_norm": 0.5970184206962585,
"learning_rate": 8.84940243451459e-05,
"loss": 0.0382,
"step": 7780
},
{
"grad_norm": 0.5335718393325806,
"learning_rate": 8.845882680727469e-05,
"loss": 0.0353,
"step": 7790
},
{
"grad_norm": 0.5941627025604248,
"learning_rate": 8.842358253830246e-05,
"loss": 0.0373,
"step": 7800
},
{
"grad_norm": 0.6231170296669006,
"learning_rate": 8.838829158105434e-05,
"loss": 0.0372,
"step": 7810
},
{
"grad_norm": 0.685299813747406,
"learning_rate": 8.835295397841217e-05,
"loss": 0.0298,
"step": 7820
},
{
"grad_norm": 0.6888481974601746,
"learning_rate": 8.831756977331447e-05,
"loss": 0.0331,
"step": 7830
},
{
"grad_norm": 0.5252216458320618,
"learning_rate": 8.828213900875638e-05,
"loss": 0.0334,
"step": 7840
},
{
"grad_norm": 0.5339848399162292,
"learning_rate": 8.824666172778964e-05,
"loss": 0.0347,
"step": 7850
},
{
"grad_norm": 0.5873726606369019,
"learning_rate": 8.821113797352245e-05,
"loss": 0.0446,
"step": 7860
},
{
"grad_norm": 0.6486777067184448,
"learning_rate": 8.817556778911956e-05,
"loss": 0.0418,
"step": 7870
},
{
"grad_norm": 0.6314468383789062,
"learning_rate": 8.81399512178021e-05,
"loss": 0.0407,
"step": 7880
},
{
"grad_norm": 0.5927188396453857,
"learning_rate": 8.810428830284752e-05,
"loss": 0.0313,
"step": 7890
},
{
"grad_norm": 0.6265239715576172,
"learning_rate": 8.806857908758967e-05,
"loss": 0.05,
"step": 7900
},
{
"grad_norm": 0.5644485354423523,
"learning_rate": 8.80328236154186e-05,
"loss": 0.0416,
"step": 7910
},
{
"grad_norm": 0.6644365191459656,
"learning_rate": 8.799702192978056e-05,
"loss": 0.0593,
"step": 7920
},
{
"grad_norm": 0.6032542586326599,
"learning_rate": 8.7961174074178e-05,
"loss": 0.0391,
"step": 7930
},
{
"grad_norm": 0.6656998991966248,
"learning_rate": 8.792528009216942e-05,
"loss": 0.0447,
"step": 7940
},
{
"grad_norm": 0.5895067453384399,
"learning_rate": 8.788934002736943e-05,
"loss": 0.0346,
"step": 7950
},
{
"grad_norm": 0.6622956395149231,
"learning_rate": 8.785335392344857e-05,
"loss": 0.034,
"step": 7960
},
{
"grad_norm": 0.4752400517463684,
"learning_rate": 8.781732182413336e-05,
"loss": 0.032,
"step": 7970
},
{
"grad_norm": 0.5423645377159119,
"learning_rate": 8.778124377320618e-05,
"loss": 0.0364,
"step": 7980
},
{
"grad_norm": 0.691588819026947,
"learning_rate": 8.774511981450529e-05,
"loss": 0.0354,
"step": 7990
},
{
"grad_norm": 0.6103001236915588,
"learning_rate": 8.770894999192467e-05,
"loss": 0.0308,
"step": 8000
},
{
"grad_norm": 0.4986715018749237,
"learning_rate": 8.767273434941412e-05,
"loss": 0.0277,
"step": 8010
},
{
"grad_norm": 0.5719414353370667,
"learning_rate": 8.7636472930979e-05,
"loss": 0.0337,
"step": 8020
},
{
"grad_norm": 0.5236717462539673,
"learning_rate": 8.76001657806804e-05,
"loss": 0.0327,
"step": 8030
},
{
"grad_norm": 0.6565148234367371,
"learning_rate": 8.756381294263491e-05,
"loss": 0.032,
"step": 8040
},
{
"grad_norm": 0.5476356744766235,
"learning_rate": 8.752741446101464e-05,
"loss": 0.0381,
"step": 8050
},
{
"grad_norm": 0.5435053706169128,
"learning_rate": 8.749097038004722e-05,
"loss": 0.0291,
"step": 8060
},
{
"grad_norm": 0.5327540636062622,
"learning_rate": 8.745448074401561e-05,
"loss": 0.0319,
"step": 8070
},
{
"grad_norm": 0.6264498829841614,
"learning_rate": 8.741794559725818e-05,
"loss": 0.0498,
"step": 8080
},
{
"grad_norm": 0.5779382586479187,
"learning_rate": 8.738136498416856e-05,
"loss": 0.0394,
"step": 8090
},
{
"grad_norm": 0.5851184725761414,
"learning_rate": 8.734473894919564e-05,
"loss": 0.0439,
"step": 8100
},
{
"grad_norm": 0.5970649123191833,
"learning_rate": 8.730806753684353e-05,
"loss": 0.0459,
"step": 8110
},
{
"grad_norm": 0.5445625185966492,
"learning_rate": 8.727135079167144e-05,
"loss": 0.0346,
"step": 8120
},
{
"grad_norm": 0.5895832777023315,
"learning_rate": 8.723458875829368e-05,
"loss": 0.0484,
"step": 8130
},
{
"grad_norm": 0.6479496955871582,
"learning_rate": 8.719778148137958e-05,
"loss": 0.0385,
"step": 8140
},
{
"grad_norm": 0.5875799059867859,
"learning_rate": 8.716092900565346e-05,
"loss": 0.0336,
"step": 8150
},
{
"grad_norm": 0.6689788103103638,
"learning_rate": 8.712403137589456e-05,
"loss": 0.0452,
"step": 8160
},
{
"grad_norm": 0.5614681243896484,
"learning_rate": 8.708708863693697e-05,
"loss": 0.0405,
"step": 8170
},
{
"grad_norm": 0.6117330193519592,
"learning_rate": 8.70501008336696e-05,
"loss": 0.0347,
"step": 8180
},
{
"grad_norm": 0.5564815998077393,
"learning_rate": 8.701306801103611e-05,
"loss": 0.0368,
"step": 8190
},
{
"grad_norm": 0.6981409192085266,
"learning_rate": 8.697599021403489e-05,
"loss": 0.0362,
"step": 8200
},
{
"grad_norm": 0.5376554727554321,
"learning_rate": 8.693886748771896e-05,
"loss": 0.0303,
"step": 8210
},
{
"grad_norm": 0.7426163554191589,
"learning_rate": 8.690169987719593e-05,
"loss": 0.0369,
"step": 8220
},
{
"grad_norm": 0.6097552180290222,
"learning_rate": 8.686448742762791e-05,
"loss": 0.0328,
"step": 8230
},
{
"grad_norm": 0.6823076009750366,
"learning_rate": 8.682723018423161e-05,
"loss": 0.0387,
"step": 8240
},
{
"grad_norm": 0.6721537113189697,
"learning_rate": 8.678992819227804e-05,
"loss": 0.0345,
"step": 8250
},
{
"grad_norm": 0.5575999021530151,
"learning_rate": 8.675258149709265e-05,
"loss": 0.0284,
"step": 8260
},
{
"grad_norm": 0.6113318204879761,
"learning_rate": 8.67151901440552e-05,
"loss": 0.0309,
"step": 8270
},
{
"grad_norm": 0.6697525382041931,
"learning_rate": 8.667775417859971e-05,
"loss": 0.0469,
"step": 8280
},
{
"grad_norm": 0.6293131113052368,
"learning_rate": 8.664027364621441e-05,
"loss": 0.0295,
"step": 8290
},
{
"grad_norm": 0.7675439715385437,
"learning_rate": 8.660274859244168e-05,
"loss": 0.0383,
"step": 8300
},
{
"grad_norm": 0.6316038966178894,
"learning_rate": 8.656517906287799e-05,
"loss": 0.0332,
"step": 8310
},
{
"grad_norm": 0.539495050907135,
"learning_rate": 8.652756510317387e-05,
"loss": 0.0279,
"step": 8320
},
{
"grad_norm": 0.6087834239006042,
"learning_rate": 8.648990675903382e-05,
"loss": 0.0322,
"step": 8330
},
{
"grad_norm": 0.6240699291229248,
"learning_rate": 8.64522040762163e-05,
"loss": 0.0279,
"step": 8340
},
{
"grad_norm": 0.6250036954879761,
"learning_rate": 8.64144571005336e-05,
"loss": 0.0357,
"step": 8350
},
{
"grad_norm": 0.6521341800689697,
"learning_rate": 8.637666587785184e-05,
"loss": 0.0288,
"step": 8360
},
{
"grad_norm": 0.5294894576072693,
"learning_rate": 8.633883045409096e-05,
"loss": 0.0323,
"step": 8370
},
{
"grad_norm": 0.6120601892471313,
"learning_rate": 8.630095087522457e-05,
"loss": 0.0364,
"step": 8380
},
{
"grad_norm": 0.44608214497566223,
"learning_rate": 8.62630271872799e-05,
"loss": 0.0295,
"step": 8390
},
{
"grad_norm": 0.45103901624679565,
"learning_rate": 8.62250594363378e-05,
"loss": 0.0321,
"step": 8400
},
{
"grad_norm": 0.5689276456832886,
"learning_rate": 8.61870476685327e-05,
"loss": 0.0324,
"step": 8410
},
{
"grad_norm": 0.5684999227523804,
"learning_rate": 8.614899193005248e-05,
"loss": 0.0301,
"step": 8420
},
{
"grad_norm": 0.6280667781829834,
"learning_rate": 8.611089226713842e-05,
"loss": 0.0298,
"step": 8430
},
{
"grad_norm": 0.6484363079071045,
"learning_rate": 8.607274872608521e-05,
"loss": 0.0328,
"step": 8440
},
{
"grad_norm": 0.5543585419654846,
"learning_rate": 8.603456135324088e-05,
"loss": 0.0305,
"step": 8450
},
{
"grad_norm": 0.7686638236045837,
"learning_rate": 8.599633019500665e-05,
"loss": 0.0304,
"step": 8460
},
{
"grad_norm": 0.5276733040809631,
"learning_rate": 8.595805529783702e-05,
"loss": 0.0281,
"step": 8470
},
{
"grad_norm": 0.6841892004013062,
"learning_rate": 8.591973670823958e-05,
"loss": 0.0322,
"step": 8480
},
{
"grad_norm": 0.7059366703033447,
"learning_rate": 8.588137447277503e-05,
"loss": 0.0355,
"step": 8490
},
{
"grad_norm": 0.7033182978630066,
"learning_rate": 8.584296863805709e-05,
"loss": 0.0387,
"step": 8500
},
{
"grad_norm": 0.6526648998260498,
"learning_rate": 8.580451925075248e-05,
"loss": 0.0329,
"step": 8510
},
{
"grad_norm": 0.5498099327087402,
"learning_rate": 8.576602635758085e-05,
"loss": 0.0259,
"step": 8520
},
{
"grad_norm": 0.6413003206253052,
"learning_rate": 8.572749000531469e-05,
"loss": 0.0323,
"step": 8530
},
{
"grad_norm": 0.6807985901832581,
"learning_rate": 8.568891024077926e-05,
"loss": 0.036,
"step": 8540
},
{
"grad_norm": 0.6169032454490662,
"learning_rate": 8.565028711085265e-05,
"loss": 0.0354,
"step": 8550
},
{
"grad_norm": 0.5226214528083801,
"learning_rate": 8.561162066246561e-05,
"loss": 0.037,
"step": 8560
},
{
"grad_norm": 0.6389352083206177,
"learning_rate": 8.557291094260152e-05,
"loss": 0.0593,
"step": 8570
},
{
"grad_norm": 0.7867082953453064,
"learning_rate": 8.553415799829629e-05,
"loss": 0.0396,
"step": 8580
},
{
"grad_norm": 0.5618152618408203,
"learning_rate": 8.549536187663847e-05,
"loss": 0.0345,
"step": 8590
},
{
"grad_norm": 0.5839462280273438,
"learning_rate": 8.545652262476897e-05,
"loss": 0.0265,
"step": 8600
},
{
"grad_norm": 0.6376568675041199,
"learning_rate": 8.541764028988115e-05,
"loss": 0.0282,
"step": 8610
},
{
"grad_norm": 0.7402397394180298,
"learning_rate": 8.537871491922071e-05,
"loss": 0.0424,
"step": 8620
},
{
"grad_norm": 0.6091654896736145,
"learning_rate": 8.533974656008566e-05,
"loss": 0.0287,
"step": 8630
},
{
"grad_norm": 0.6619576215744019,
"learning_rate": 8.530073525982621e-05,
"loss": 0.0374,
"step": 8640
},
{
"grad_norm": 0.6471283435821533,
"learning_rate": 8.526168106584476e-05,
"loss": 0.0414,
"step": 8650
},
{
"grad_norm": 0.5171893239021301,
"learning_rate": 8.522258402559586e-05,
"loss": 0.0338,
"step": 8660
},
{
"grad_norm": 0.5404852628707886,
"learning_rate": 8.518344418658613e-05,
"loss": 0.0299,
"step": 8670
},
{
"grad_norm": 0.6301115155220032,
"learning_rate": 8.51442615963741e-05,
"loss": 0.0273,
"step": 8680
},
{
"grad_norm": 0.4700194001197815,
"learning_rate": 8.510503630257034e-05,
"loss": 0.0299,
"step": 8690
},
{
"grad_norm": 0.5803355574607849,
"learning_rate": 8.50657683528373e-05,
"loss": 0.0288,
"step": 8700
},
{
"grad_norm": 0.6717302203178406,
"learning_rate": 8.502645779488922e-05,
"loss": 0.0383,
"step": 8710
},
{
"grad_norm": 0.6512241959571838,
"learning_rate": 8.498710467649213e-05,
"loss": 0.0478,
"step": 8720
},
{
"grad_norm": 0.5688778758049011,
"learning_rate": 8.494770904546381e-05,
"loss": 0.031,
"step": 8730
},
{
"grad_norm": 0.5589701533317566,
"learning_rate": 8.490827094967363e-05,
"loss": 0.0298,
"step": 8740
},
{
"grad_norm": 0.5661013722419739,
"learning_rate": 8.486879043704263e-05,
"loss": 0.0387,
"step": 8750
},
{
"grad_norm": 0.569941520690918,
"learning_rate": 8.482926755554332e-05,
"loss": 0.0295,
"step": 8760
},
{
"grad_norm": 0.6192848682403564,
"learning_rate": 8.478970235319975e-05,
"loss": 0.0362,
"step": 8770
},
{
"grad_norm": 0.6098713874816895,
"learning_rate": 8.475009487808737e-05,
"loss": 0.0432,
"step": 8780
},
{
"grad_norm": 0.7493323087692261,
"learning_rate": 8.471044517833298e-05,
"loss": 0.0325,
"step": 8790
},
{
"grad_norm": 0.5910696983337402,
"learning_rate": 8.467075330211474e-05,
"loss": 0.0291,
"step": 8800
},
{
"grad_norm": 0.5423613786697388,
"learning_rate": 8.463101929766198e-05,
"loss": 0.0328,
"step": 8810
},
{
"grad_norm": 0.6670126914978027,
"learning_rate": 8.459124321325528e-05,
"loss": 0.0357,
"step": 8820
},
{
"grad_norm": 0.6839043498039246,
"learning_rate": 8.455142509722634e-05,
"loss": 0.0359,
"step": 8830
},
{
"grad_norm": 0.5896878838539124,
"learning_rate": 8.451156499795792e-05,
"loss": 0.0322,
"step": 8840
},
{
"grad_norm": 0.5875719785690308,
"learning_rate": 8.447166296388381e-05,
"loss": 0.029,
"step": 8850
},
{
"grad_norm": 0.630513072013855,
"learning_rate": 8.443171904348872e-05,
"loss": 0.0317,
"step": 8860
},
{
"grad_norm": 0.6915355324745178,
"learning_rate": 8.43917332853083e-05,
"loss": 0.0385,
"step": 8870
},
{
"grad_norm": 0.5546975135803223,
"learning_rate": 8.435170573792901e-05,
"loss": 0.036,
"step": 8880
},
{
"grad_norm": 0.49335524439811707,
"learning_rate": 8.431163644998808e-05,
"loss": 0.0366,
"step": 8890
},
{
"grad_norm": 0.5926335453987122,
"learning_rate": 8.42715254701735e-05,
"loss": 0.0349,
"step": 8900
},
{
"grad_norm": 0.4683159291744232,
"learning_rate": 8.42313728472239e-05,
"loss": 0.0314,
"step": 8910
},
{
"grad_norm": 0.6267257332801819,
"learning_rate": 8.419117862992844e-05,
"loss": 0.0421,
"step": 8920
},
{
"grad_norm": 0.6858824491500854,
"learning_rate": 8.415094286712694e-05,
"loss": 0.0352,
"step": 8930
},
{
"grad_norm": 0.6632905602455139,
"learning_rate": 8.411066560770965e-05,
"loss": 0.0376,
"step": 8940
},
{
"grad_norm": 0.47960349917411804,
"learning_rate": 8.407034690061721e-05,
"loss": 0.0296,
"step": 8950
},
{
"grad_norm": 0.4371381402015686,
"learning_rate": 8.402998679484067e-05,
"loss": 0.0251,
"step": 8960
},
{
"grad_norm": 0.5468704104423523,
"learning_rate": 8.398958533942135e-05,
"loss": 0.0272,
"step": 8970
},
{
"grad_norm": 0.584854006767273,
"learning_rate": 8.394914258345083e-05,
"loss": 0.0329,
"step": 8980
},
{
"grad_norm": 0.6885668039321899,
"learning_rate": 8.390865857607089e-05,
"loss": 0.0514,
"step": 8990
},
{
"grad_norm": 0.6651278734207153,
"learning_rate": 8.386813336647339e-05,
"loss": 0.0325,
"step": 9000
},
{
"grad_norm": 0.5674067139625549,
"learning_rate": 8.38275670039003e-05,
"loss": 0.0294,
"step": 9010
},
{
"grad_norm": 0.5740431547164917,
"learning_rate": 8.378695953764356e-05,
"loss": 0.0326,
"step": 9020
},
{
"grad_norm": 0.6283910274505615,
"learning_rate": 8.374631101704509e-05,
"loss": 0.0451,
"step": 9030
},
{
"grad_norm": 0.6813831329345703,
"learning_rate": 8.370562149149665e-05,
"loss": 0.0375,
"step": 9040
},
{
"grad_norm": 0.42199113965034485,
"learning_rate": 8.366489101043989e-05,
"loss": 0.0292,
"step": 9050
},
{
"grad_norm": 0.5412783026695251,
"learning_rate": 8.362411962336614e-05,
"loss": 0.04,
"step": 9060
},
{
"grad_norm": 0.6151837110519409,
"learning_rate": 8.358330737981651e-05,
"loss": 0.0253,
"step": 9070
},
{
"grad_norm": 0.5114996433258057,
"learning_rate": 8.35424543293817e-05,
"loss": 0.0287,
"step": 9080
},
{
"grad_norm": 0.5295445919036865,
"learning_rate": 8.350156052170206e-05,
"loss": 0.0382,
"step": 9090
},
{
"grad_norm": 0.5946828126907349,
"learning_rate": 8.346062600646738e-05,
"loss": 0.0408,
"step": 9100
},
{
"grad_norm": 0.5739308595657349,
"learning_rate": 8.341965083341697e-05,
"loss": 0.0325,
"step": 9110
},
{
"grad_norm": 0.5593684315681458,
"learning_rate": 8.337863505233953e-05,
"loss": 0.0371,
"step": 9120
},
{
"grad_norm": 0.5210955739021301,
"learning_rate": 8.33375787130731e-05,
"loss": 0.0407,
"step": 9130
},
{
"grad_norm": 0.5242553949356079,
"learning_rate": 8.3296481865505e-05,
"loss": 0.0254,
"step": 9140
},
{
"grad_norm": 0.5588939785957336,
"learning_rate": 8.325534455957181e-05,
"loss": 0.0373,
"step": 9150
},
{
"grad_norm": 0.5350152254104614,
"learning_rate": 8.321416684525917e-05,
"loss": 0.0256,
"step": 9160
},
{
"grad_norm": 0.49569496512413025,
"learning_rate": 8.317294877260193e-05,
"loss": 0.0295,
"step": 9170
},
{
"grad_norm": 0.42370620369911194,
"learning_rate": 8.313169039168394e-05,
"loss": 0.0257,
"step": 9180
},
{
"grad_norm": 0.6131086349487305,
"learning_rate": 8.3090391752638e-05,
"loss": 0.0236,
"step": 9190
},
{
"grad_norm": 0.6043539047241211,
"learning_rate": 8.304905290564586e-05,
"loss": 0.031,
"step": 9200
},
{
"grad_norm": 0.470587819814682,
"learning_rate": 8.300767390093813e-05,
"loss": 0.0416,
"step": 9210
},
{
"grad_norm": 0.4834061861038208,
"learning_rate": 8.296625478879417e-05,
"loss": 0.0333,
"step": 9220
},
{
"grad_norm": 0.5486321449279785,
"learning_rate": 8.292479561954214e-05,
"loss": 0.0314,
"step": 9230
},
{
"grad_norm": 0.6123573184013367,
"learning_rate": 8.288329644355883e-05,
"loss": 0.0413,
"step": 9240
},
{
"grad_norm": 0.5733028650283813,
"learning_rate": 8.284175731126965e-05,
"loss": 0.0292,
"step": 9250
},
{
"grad_norm": 0.5841543674468994,
"learning_rate": 8.280017827314854e-05,
"loss": 0.0257,
"step": 9260
},
{
"grad_norm": 0.5915876030921936,
"learning_rate": 8.2758559379718e-05,
"loss": 0.033,
"step": 9270
},
{
"grad_norm": 0.6267388463020325,
"learning_rate": 8.271690068154887e-05,
"loss": 0.0295,
"step": 9280
},
{
"grad_norm": 0.788142740726471,
"learning_rate": 8.26752022292604e-05,
"loss": 0.0426,
"step": 9290
},
{
"grad_norm": 0.6792725920677185,
"learning_rate": 8.263346407352018e-05,
"loss": 0.0405,
"step": 9300
},
{
"grad_norm": 0.5279889702796936,
"learning_rate": 8.259168626504395e-05,
"loss": 0.0313,
"step": 9310
},
{
"grad_norm": 0.5980766415596008,
"learning_rate": 8.254986885459572e-05,
"loss": 0.0478,
"step": 9320
},
{
"grad_norm": 0.5487632155418396,
"learning_rate": 8.250801189298757e-05,
"loss": 0.0317,
"step": 9330
},
{
"grad_norm": 0.580762505531311,
"learning_rate": 8.246611543107967e-05,
"loss": 0.0342,
"step": 9340
},
{
"grad_norm": 0.5436245203018188,
"learning_rate": 8.242417951978018e-05,
"loss": 0.0312,
"step": 9350
},
{
"grad_norm": 0.5834146738052368,
"learning_rate": 8.238220421004517e-05,
"loss": 0.0266,
"step": 9360
},
{
"grad_norm": 0.6223360896110535,
"learning_rate": 8.234018955287859e-05,
"loss": 0.0436,
"step": 9370
},
{
"grad_norm": 0.5101218223571777,
"learning_rate": 8.229813559933224e-05,
"loss": 0.0305,
"step": 9380
},
{
"grad_norm": 0.5484694242477417,
"learning_rate": 8.22560424005056e-05,
"loss": 0.0306,
"step": 9390
},
{
"grad_norm": 0.6884644627571106,
"learning_rate": 8.22139100075459e-05,
"loss": 0.0421,
"step": 9400
},
{
"grad_norm": 0.5865536332130432,
"learning_rate": 8.217173847164799e-05,
"loss": 0.0378,
"step": 9410
},
{
"grad_norm": 0.5574385523796082,
"learning_rate": 8.212952784405422e-05,
"loss": 0.0313,
"step": 9420
},
{
"grad_norm": 0.6152393817901611,
"learning_rate": 8.208727817605451e-05,
"loss": 0.0269,
"step": 9430
},
{
"grad_norm": 0.5821828246116638,
"learning_rate": 8.204498951898618e-05,
"loss": 0.0296,
"step": 9440
},
{
"grad_norm": 0.5566200613975525,
"learning_rate": 8.200266192423395e-05,
"loss": 0.0265,
"step": 9450
},
{
"grad_norm": 0.6039608120918274,
"learning_rate": 8.196029544322981e-05,
"loss": 0.0246,
"step": 9460
},
{
"grad_norm": 0.5446297526359558,
"learning_rate": 8.191789012745305e-05,
"loss": 0.0449,
"step": 9470
},
{
"grad_norm": 0.5751950740814209,
"learning_rate": 8.187544602843014e-05,
"loss": 0.0262,
"step": 9480
},
{
"grad_norm": 0.6154511570930481,
"learning_rate": 8.183296319773466e-05,
"loss": 0.0282,
"step": 9490
},
{
"grad_norm": 0.5663873553276062,
"learning_rate": 8.179044168698721e-05,
"loss": 0.0252,
"step": 9500
},
{
"grad_norm": 0.5509154200553894,
"learning_rate": 8.174788154785548e-05,
"loss": 0.0265,
"step": 9510
},
{
"grad_norm": 0.5833673477172852,
"learning_rate": 8.170528283205403e-05,
"loss": 0.0343,
"step": 9520
},
{
"grad_norm": 0.5563690066337585,
"learning_rate": 8.166264559134435e-05,
"loss": 0.0286,
"step": 9530
},
{
"grad_norm": 0.7141296863555908,
"learning_rate": 8.161996987753467e-05,
"loss": 0.0381,
"step": 9540
},
{
"grad_norm": 0.6595776677131653,
"learning_rate": 8.157725574248e-05,
"loss": 0.0319,
"step": 9550
},
{
"grad_norm": 0.6489679217338562,
"learning_rate": 8.153450323808209e-05,
"loss": 0.0402,
"step": 9560
},
{
"grad_norm": 0.5790324807167053,
"learning_rate": 8.149171241628923e-05,
"loss": 0.0482,
"step": 9570
},
{
"grad_norm": 0.7813078165054321,
"learning_rate": 8.144888332909631e-05,
"loss": 0.0398,
"step": 9580
},
{
"grad_norm": 0.5433303117752075,
"learning_rate": 8.14060160285447e-05,
"loss": 0.0313,
"step": 9590
},
{
"grad_norm": 0.6028977632522583,
"learning_rate": 8.136311056672223e-05,
"loss": 0.0297,
"step": 9600
},
{
"grad_norm": 0.6106552481651306,
"learning_rate": 8.132016699576308e-05,
"loss": 0.0288,
"step": 9610
},
{
"grad_norm": 0.5814914107322693,
"learning_rate": 8.127718536784771e-05,
"loss": 0.0322,
"step": 9620
},
{
"grad_norm": 0.6080571413040161,
"learning_rate": 8.12341657352029e-05,
"loss": 0.0292,
"step": 9630
},
{
"grad_norm": 0.5781335234642029,
"learning_rate": 8.119110815010151e-05,
"loss": 0.0271,
"step": 9640
},
{
"grad_norm": 0.5819571614265442,
"learning_rate": 8.114801266486261e-05,
"loss": 0.0318,
"step": 9650
},
{
"grad_norm": 0.575080394744873,
"learning_rate": 8.110487933185122e-05,
"loss": 0.0308,
"step": 9660
},
{
"grad_norm": 0.5886939167976379,
"learning_rate": 8.106170820347847e-05,
"loss": 0.0342,
"step": 9670
},
{
"grad_norm": 0.44043856859207153,
"learning_rate": 8.101849933220132e-05,
"loss": 0.0258,
"step": 9680
},
{
"grad_norm": 0.4897483289241791,
"learning_rate": 8.097525277052264e-05,
"loss": 0.028,
"step": 9690
},
{
"grad_norm": 0.6758052706718445,
"learning_rate": 8.093196857099106e-05,
"loss": 0.026,
"step": 9700
},
{
"grad_norm": 0.5501713156700134,
"learning_rate": 8.088864678620096e-05,
"loss": 0.0246,
"step": 9710
},
{
"grad_norm": 0.5333536863327026,
"learning_rate": 8.084528746879243e-05,
"loss": 0.0321,
"step": 9720
},
{
"grad_norm": 0.6263291239738464,
"learning_rate": 8.080189067145108e-05,
"loss": 0.0352,
"step": 9730
},
{
"grad_norm": 0.5594651699066162,
"learning_rate": 8.075845644690813e-05,
"loss": 0.0289,
"step": 9740
},
{
"grad_norm": 0.6350171566009521,
"learning_rate": 8.07149848479403e-05,
"loss": 0.0301,
"step": 9750
},
{
"grad_norm": 0.5894553065299988,
"learning_rate": 8.067147592736962e-05,
"loss": 0.0313,
"step": 9760
},
{
"grad_norm": 0.7067338824272156,
"learning_rate": 8.062792973806357e-05,
"loss": 0.0312,
"step": 9770
},
{
"grad_norm": 0.5011563897132874,
"learning_rate": 8.058434633293485e-05,
"loss": 0.0322,
"step": 9780
},
{
"grad_norm": 0.5685155987739563,
"learning_rate": 8.054072576494143e-05,
"loss": 0.0308,
"step": 9790
},
{
"grad_norm": 0.7401490807533264,
"learning_rate": 8.04970680870864e-05,
"loss": 0.0257,
"step": 9800
},
{
"grad_norm": 0.5961191654205322,
"learning_rate": 8.045337335241793e-05,
"loss": 0.0297,
"step": 9810
},
{
"grad_norm": 0.7336726784706116,
"learning_rate": 8.040964161402932e-05,
"loss": 0.0352,
"step": 9820
},
{
"grad_norm": 0.6852769255638123,
"learning_rate": 8.036587292505869e-05,
"loss": 0.0324,
"step": 9830
},
{
"grad_norm": 0.5286868214607239,
"learning_rate": 8.032206733868912e-05,
"loss": 0.0378,
"step": 9840
},
{
"grad_norm": 0.5463317036628723,
"learning_rate": 8.027822490814858e-05,
"loss": 0.0379,
"step": 9850
},
{
"grad_norm": 0.5204600095748901,
"learning_rate": 8.023434568670972e-05,
"loss": 0.0265,
"step": 9860
},
{
"grad_norm": 0.5505666136741638,
"learning_rate": 8.019042972768992e-05,
"loss": 0.0274,
"step": 9870
},
{
"grad_norm": 0.5658701658248901,
"learning_rate": 8.014647708445124e-05,
"loss": 0.0348,
"step": 9880
},
{
"grad_norm": 0.5198262333869934,
"learning_rate": 8.010248781040027e-05,
"loss": 0.025,
"step": 9890
},
{
"grad_norm": 0.549225926399231,
"learning_rate": 8.005846195898814e-05,
"loss": 0.0297,
"step": 9900
},
{
"grad_norm": 0.5910757780075073,
"learning_rate": 8.00143995837104e-05,
"loss": 0.029,
"step": 9910
},
{
"grad_norm": 0.5789921283721924,
"learning_rate": 7.997030073810698e-05,
"loss": 0.0358,
"step": 9920
},
{
"grad_norm": 0.6346859931945801,
"learning_rate": 7.992616547576217e-05,
"loss": 0.027,
"step": 9930
},
{
"grad_norm": 0.6420550346374512,
"learning_rate": 7.988199385030445e-05,
"loss": 0.0286,
"step": 9940
},
{
"grad_norm": 0.6194259524345398,
"learning_rate": 7.98377859154065e-05,
"loss": 0.0269,
"step": 9950
},
{
"grad_norm": 0.569584310054779,
"learning_rate": 7.979354172478516e-05,
"loss": 0.0244,
"step": 9960
},
{
"grad_norm": 0.5307677388191223,
"learning_rate": 7.974926133220127e-05,
"loss": 0.0356,
"step": 9970
},
{
"grad_norm": 0.5357755422592163,
"learning_rate": 7.970494479145967e-05,
"loss": 0.0308,
"step": 9980
},
{
"grad_norm": 0.5300595164299011,
"learning_rate": 7.966059215640919e-05,
"loss": 0.0337,
"step": 9990
},
{
"grad_norm": 0.5895503163337708,
"learning_rate": 7.961620348094241e-05,
"loss": 0.036,
"step": 10000
},
{
"grad_norm": 0.6603275537490845,
"learning_rate": 7.957177881899577e-05,
"loss": 0.0325,
"step": 10010
},
{
"grad_norm": 0.48111557960510254,
"learning_rate": 7.952731822454944e-05,
"loss": 0.0333,
"step": 10020
},
{
"grad_norm": 0.6827055215835571,
"learning_rate": 7.948282175162722e-05,
"loss": 0.0242,
"step": 10030
},
{
"grad_norm": 0.5445454716682434,
"learning_rate": 7.943828945429652e-05,
"loss": 0.0372,
"step": 10040
},
{
"grad_norm": 0.5205521583557129,
"learning_rate": 7.939372138666827e-05,
"loss": 0.0305,
"step": 10050
},
{
"grad_norm": 0.5438241362571716,
"learning_rate": 7.934911760289692e-05,
"loss": 0.0351,
"step": 10060
},
{
"grad_norm": 0.6112908124923706,
"learning_rate": 7.930447815718022e-05,
"loss": 0.0237,
"step": 10070
},
{
"grad_norm": 0.4439740478992462,
"learning_rate": 7.925980310375934e-05,
"loss": 0.0273,
"step": 10080
},
{
"grad_norm": 0.5059341788291931,
"learning_rate": 7.921509249691865e-05,
"loss": 0.0283,
"step": 10090
},
{
"grad_norm": 0.6745112538337708,
"learning_rate": 7.917034639098579e-05,
"loss": 0.0284,
"step": 10100
},
{
"grad_norm": 0.4964512586593628,
"learning_rate": 7.912556484033146e-05,
"loss": 0.0222,
"step": 10110
},
{
"grad_norm": 0.5891860127449036,
"learning_rate": 7.908074789936952e-05,
"loss": 0.022,
"step": 10120
},
{
"grad_norm": 0.5039228796958923,
"learning_rate": 7.903589562255673e-05,
"loss": 0.0327,
"step": 10130
},
{
"grad_norm": 0.4523487687110901,
"learning_rate": 7.899100806439285e-05,
"loss": 0.0256,
"step": 10140
},
{
"grad_norm": 0.5442401170730591,
"learning_rate": 7.894608527942049e-05,
"loss": 0.0355,
"step": 10150
},
{
"grad_norm": 0.5850509405136108,
"learning_rate": 7.89011273222251e-05,
"loss": 0.0289,
"step": 10160
},
{
"grad_norm": 0.5431426763534546,
"learning_rate": 7.885613424743482e-05,
"loss": 0.032,
"step": 10170
},
{
"grad_norm": 0.47547927498817444,
"learning_rate": 7.881110610972044e-05,
"loss": 0.0385,
"step": 10180
},
{
"grad_norm": 0.6267944574356079,
"learning_rate": 7.876604296379544e-05,
"loss": 0.0361,
"step": 10190
},
{
"grad_norm": 0.546815812587738,
"learning_rate": 7.872094486441579e-05,
"loss": 0.0343,
"step": 10200
},
{
"grad_norm": 0.5742200016975403,
"learning_rate": 7.867581186637991e-05,
"loss": 0.0304,
"step": 10210
},
{
"grad_norm": 0.5584553480148315,
"learning_rate": 7.863064402452866e-05,
"loss": 0.0251,
"step": 10220
},
{
"grad_norm": 0.6229097247123718,
"learning_rate": 7.858544139374524e-05,
"loss": 0.0375,
"step": 10230
},
{
"grad_norm": 0.48090094327926636,
"learning_rate": 7.854020402895508e-05,
"loss": 0.0279,
"step": 10240
},
{
"grad_norm": 0.49592095613479614,
"learning_rate": 7.849493198512587e-05,
"loss": 0.0295,
"step": 10250
},
{
"grad_norm": 0.5183885097503662,
"learning_rate": 7.84496253172674e-05,
"loss": 0.0229,
"step": 10260
},
{
"grad_norm": 0.5413907766342163,
"learning_rate": 7.840428408043155e-05,
"loss": 0.0289,
"step": 10270
},
{
"grad_norm": 0.5817140936851501,
"learning_rate": 7.835890832971218e-05,
"loss": 0.0324,
"step": 10280
},
{
"grad_norm": 0.5629503726959229,
"learning_rate": 7.831349812024512e-05,
"loss": 0.0417,
"step": 10290
},
{
"grad_norm": 0.5762734413146973,
"learning_rate": 7.826805350720807e-05,
"loss": 0.0286,
"step": 10300
},
{
"grad_norm": 0.5128576755523682,
"learning_rate": 7.822257454582049e-05,
"loss": 0.0268,
"step": 10310
},
{
"grad_norm": 0.500552773475647,
"learning_rate": 7.817706129134363e-05,
"loss": 0.0293,
"step": 10320
},
{
"grad_norm": 0.5837981104850769,
"learning_rate": 7.813151379908036e-05,
"loss": 0.0297,
"step": 10330
},
{
"grad_norm": 0.6267739534378052,
"learning_rate": 7.808593212437523e-05,
"loss": 0.0327,
"step": 10340
},
{
"grad_norm": 0.510150134563446,
"learning_rate": 7.804031632261421e-05,
"loss": 0.0363,
"step": 10350
},
{
"grad_norm": 0.5368266105651855,
"learning_rate": 7.799466644922484e-05,
"loss": 0.0262,
"step": 10360
},
{
"grad_norm": 0.6090547442436218,
"learning_rate": 7.794898255967602e-05,
"loss": 0.0246,
"step": 10370
},
{
"grad_norm": 0.5788031816482544,
"learning_rate": 7.790326470947795e-05,
"loss": 0.029,
"step": 10380
},
{
"grad_norm": 0.578019380569458,
"learning_rate": 7.785751295418217e-05,
"loss": 0.0241,
"step": 10390
},
{
"grad_norm": 0.5326537489891052,
"learning_rate": 7.781172734938136e-05,
"loss": 0.0313,
"step": 10400
},
{
"grad_norm": 0.6269048452377319,
"learning_rate": 7.776590795070933e-05,
"loss": 0.0284,
"step": 10410
},
{
"grad_norm": 0.581630527973175,
"learning_rate": 7.772005481384099e-05,
"loss": 0.0339,
"step": 10420
},
{
"grad_norm": 0.653351366519928,
"learning_rate": 7.767416799449223e-05,
"loss": 0.0219,
"step": 10430
},
{
"grad_norm": 0.5283663272857666,
"learning_rate": 7.762824754841984e-05,
"loss": 0.0227,
"step": 10440
},
{
"grad_norm": 0.5754411816596985,
"learning_rate": 7.758229353142152e-05,
"loss": 0.0341,
"step": 10450
},
{
"grad_norm": 0.5917637348175049,
"learning_rate": 7.753630599933572e-05,
"loss": 0.03,
"step": 10460
},
{
"grad_norm": 0.5900212526321411,
"learning_rate": 7.749028500804161e-05,
"loss": 0.0275,
"step": 10470
},
{
"grad_norm": 0.5084112882614136,
"learning_rate": 7.744423061345906e-05,
"loss": 0.0257,
"step": 10480
},
{
"grad_norm": 0.5565437078475952,
"learning_rate": 7.73981428715485e-05,
"loss": 0.0235,
"step": 10490
},
{
"grad_norm": 0.5583534240722656,
"learning_rate": 7.735202183831085e-05,
"loss": 0.0202,
"step": 10500
},
{
"grad_norm": 0.6069707870483398,
"learning_rate": 7.730586756978757e-05,
"loss": 0.0319,
"step": 10510
},
{
"grad_norm": 0.5969206690788269,
"learning_rate": 7.72596801220604e-05,
"loss": 0.0217,
"step": 10520
},
{
"grad_norm": 0.5718361735343933,
"learning_rate": 7.721345955125147e-05,
"loss": 0.0304,
"step": 10530
},
{
"grad_norm": 0.42534777522087097,
"learning_rate": 7.71672059135231e-05,
"loss": 0.0276,
"step": 10540
},
{
"grad_norm": 0.528182864189148,
"learning_rate": 7.712091926507787e-05,
"loss": 0.0216,
"step": 10550
},
{
"grad_norm": 0.5216792225837708,
"learning_rate": 7.70745996621584e-05,
"loss": 0.0217,
"step": 10560
},
{
"grad_norm": 0.6057245135307312,
"learning_rate": 7.702824716104735e-05,
"loss": 0.0228,
"step": 10570
},
{
"grad_norm": 0.5128807425498962,
"learning_rate": 7.698186181806743e-05,
"loss": 0.0294,
"step": 10580
},
{
"grad_norm": 0.570521891117096,
"learning_rate": 7.693544368958116e-05,
"loss": 0.0297,
"step": 10590
},
{
"grad_norm": 0.5803155303001404,
"learning_rate": 7.688899283199096e-05,
"loss": 0.0378,
"step": 10600
},
{
"grad_norm": 0.5385972857475281,
"learning_rate": 7.684250930173901e-05,
"loss": 0.0478,
"step": 10610
},
{
"grad_norm": 0.5489588379859924,
"learning_rate": 7.679599315530718e-05,
"loss": 0.0479,
"step": 10620
},
{
"grad_norm": 0.6402607560157776,
"learning_rate": 7.674944444921695e-05,
"loss": 0.0241,
"step": 10630
},
{
"grad_norm": 0.610360324382782,
"learning_rate": 7.670286324002944e-05,
"loss": 0.0338,
"step": 10640
},
{
"grad_norm": 0.6462026238441467,
"learning_rate": 7.665624958434514e-05,
"loss": 0.038,
"step": 10650
},
{
"grad_norm": 0.5430339574813843,
"learning_rate": 7.66096035388041e-05,
"loss": 0.0376,
"step": 10660
},
{
"grad_norm": 0.6038399338722229,
"learning_rate": 7.656292516008563e-05,
"loss": 0.0327,
"step": 10670
},
{
"grad_norm": 0.4891054332256317,
"learning_rate": 7.651621450490837e-05,
"loss": 0.0235,
"step": 10680
},
{
"grad_norm": 0.7008992433547974,
"learning_rate": 7.646947163003017e-05,
"loss": 0.0231,
"step": 10690
},
{
"grad_norm": 0.5642795562744141,
"learning_rate": 7.642269659224803e-05,
"loss": 0.0521,
"step": 10700
},
{
"grad_norm": 0.608784556388855,
"learning_rate": 7.637588944839803e-05,
"loss": 0.0313,
"step": 10710
},
{
"grad_norm": 0.6123632192611694,
"learning_rate": 7.632905025535528e-05,
"loss": 0.0288,
"step": 10720
},
{
"grad_norm": 0.5852841138839722,
"learning_rate": 7.628217907003378e-05,
"loss": 0.0234,
"step": 10730
},
{
"grad_norm": 0.46447455883026123,
"learning_rate": 7.623527594938649e-05,
"loss": 0.0239,
"step": 10740
},
{
"grad_norm": 0.6004592776298523,
"learning_rate": 7.618834095040509e-05,
"loss": 0.0307,
"step": 10750
},
{
"grad_norm": 0.6370054483413696,
"learning_rate": 7.614137413012e-05,
"loss": 0.0356,
"step": 10760
},
{
"grad_norm": 0.5060777068138123,
"learning_rate": 7.609437554560042e-05,
"loss": 0.0359,
"step": 10770
},
{
"grad_norm": 0.5299996733665466,
"learning_rate": 7.604734525395398e-05,
"loss": 0.0323,
"step": 10780
},
{
"grad_norm": 0.5680559277534485,
"learning_rate": 7.600028331232699e-05,
"loss": 0.0261,
"step": 10790
},
{
"grad_norm": 0.6162432432174683,
"learning_rate": 7.595318977790408e-05,
"loss": 0.0399,
"step": 10800
},
{
"grad_norm": 0.5902808308601379,
"learning_rate": 7.590606470790836e-05,
"loss": 0.0265,
"step": 10810
},
{
"grad_norm": 0.5633955001831055,
"learning_rate": 7.585890815960125e-05,
"loss": 0.0289,
"step": 10820
},
{
"grad_norm": 0.6329830288887024,
"learning_rate": 7.581172019028238e-05,
"loss": 0.03,
"step": 10830
},
{
"grad_norm": 0.6195060610771179,
"learning_rate": 7.576450085728959e-05,
"loss": 0.0362,
"step": 10840
},
{
"grad_norm": 0.6009583473205566,
"learning_rate": 7.571725021799885e-05,
"loss": 0.028,
"step": 10850
},
{
"grad_norm": 0.5903528332710266,
"learning_rate": 7.566996832982409e-05,
"loss": 0.0246,
"step": 10860
},
{
"grad_norm": 0.5060582160949707,
"learning_rate": 7.56226552502173e-05,
"loss": 0.0315,
"step": 10870
},
{
"grad_norm": 0.6085615158081055,
"learning_rate": 7.557531103666833e-05,
"loss": 0.0323,
"step": 10880
},
{
"grad_norm": 0.5560912489891052,
"learning_rate": 7.552793574670485e-05,
"loss": 0.0311,
"step": 10890
},
{
"grad_norm": 0.631277859210968,
"learning_rate": 7.54805294378923e-05,
"loss": 0.0319,
"step": 10900
},
{
"grad_norm": 0.6949539184570312,
"learning_rate": 7.543309216783384e-05,
"loss": 0.0282,
"step": 10910
},
{
"grad_norm": 0.5869392156600952,
"learning_rate": 7.53856239941702e-05,
"loss": 0.0288,
"step": 10920
},
{
"grad_norm": 0.5116824507713318,
"learning_rate": 7.533812497457972e-05,
"loss": 0.0371,
"step": 10930
},
{
"grad_norm": 0.48919740319252014,
"learning_rate": 7.529059516677814e-05,
"loss": 0.0329,
"step": 10940
},
{
"grad_norm": 0.5898220539093018,
"learning_rate": 7.524303462851872e-05,
"loss": 0.0265,
"step": 10950
},
{
"grad_norm": 0.5486130714416504,
"learning_rate": 7.519544341759192e-05,
"loss": 0.0245,
"step": 10960
},
{
"grad_norm": 0.4380483329296112,
"learning_rate": 7.514782159182562e-05,
"loss": 0.0238,
"step": 10970
},
{
"grad_norm": 0.5393425822257996,
"learning_rate": 7.510016920908481e-05,
"loss": 0.0243,
"step": 10980
},
{
"grad_norm": 0.3718966841697693,
"learning_rate": 7.505248632727158e-05,
"loss": 0.0244,
"step": 10990
},
{
"grad_norm": 0.48129647970199585,
"learning_rate": 7.50047730043252e-05,
"loss": 0.0248,
"step": 11000
},
{
"grad_norm": 0.44795945286750793,
"learning_rate": 7.495702929822183e-05,
"loss": 0.0272,
"step": 11010
},
{
"grad_norm": 0.5548344850540161,
"learning_rate": 7.490925526697455e-05,
"loss": 0.0285,
"step": 11020
},
{
"grad_norm": 0.829288125038147,
"learning_rate": 7.486145096863334e-05,
"loss": 0.0276,
"step": 11030
},
{
"grad_norm": 0.5439048409461975,
"learning_rate": 7.481361646128491e-05,
"loss": 0.0294,
"step": 11040
},
{
"grad_norm": 0.6631721258163452,
"learning_rate": 7.476575180305271e-05,
"loss": 0.0249,
"step": 11050
},
{
"grad_norm": 0.6400963068008423,
"learning_rate": 7.471785705209682e-05,
"loss": 0.033,
"step": 11060
},
{
"grad_norm": 0.5484209656715393,
"learning_rate": 7.466993226661387e-05,
"loss": 0.0247,
"step": 11070
},
{
"grad_norm": 0.5449013113975525,
"learning_rate": 7.462197750483699e-05,
"loss": 0.0331,
"step": 11080
},
{
"grad_norm": 0.5215168595314026,
"learning_rate": 7.457399282503574e-05,
"loss": 0.0279,
"step": 11090
},
{
"grad_norm": 0.6095676422119141,
"learning_rate": 7.452597828551604e-05,
"loss": 0.033,
"step": 11100
},
{
"grad_norm": 0.6611076593399048,
"learning_rate": 7.447793394462006e-05,
"loss": 0.0309,
"step": 11110
},
{
"grad_norm": 0.6780716180801392,
"learning_rate": 7.442985986072624e-05,
"loss": 0.0255,
"step": 11120
},
{
"grad_norm": 0.47774091362953186,
"learning_rate": 7.438175609224908e-05,
"loss": 0.0333,
"step": 11130
},
{
"grad_norm": 0.5286259651184082,
"learning_rate": 7.433362269763924e-05,
"loss": 0.0229,
"step": 11140
},
{
"grad_norm": 0.5117261409759521,
"learning_rate": 7.428545973538329e-05,
"loss": 0.0185,
"step": 11150
},
{
"grad_norm": 0.5176396369934082,
"learning_rate": 7.42372672640038e-05,
"loss": 0.0249,
"step": 11160
},
{
"grad_norm": 0.6360921263694763,
"learning_rate": 7.418904534205917e-05,
"loss": 0.0232,
"step": 11170
},
{
"grad_norm": 0.5273647904396057,
"learning_rate": 7.414079402814356e-05,
"loss": 0.0268,
"step": 11180
},
{
"grad_norm": 0.4677692949771881,
"learning_rate": 7.409251338088687e-05,
"loss": 0.027,
"step": 11190
},
{
"grad_norm": 0.6061916351318359,
"learning_rate": 7.404420345895467e-05,
"loss": 0.0296,
"step": 11200
},
{
"grad_norm": 0.515214204788208,
"learning_rate": 7.399586432104804e-05,
"loss": 0.0284,
"step": 11210
},
{
"grad_norm": 0.49113893508911133,
"learning_rate": 7.394749602590358e-05,
"loss": 0.0242,
"step": 11220
},
{
"grad_norm": 0.6719169616699219,
"learning_rate": 7.389909863229336e-05,
"loss": 0.0395,
"step": 11230
},
{
"grad_norm": 0.5344063639640808,
"learning_rate": 7.385067219902477e-05,
"loss": 0.0279,
"step": 11240
},
{
"grad_norm": 0.4761667251586914,
"learning_rate": 7.380221678494049e-05,
"loss": 0.0214,
"step": 11250
},
{
"grad_norm": 0.5625593662261963,
"learning_rate": 7.37537324489184e-05,
"loss": 0.02,
"step": 11260
},
{
"grad_norm": 0.4344480335712433,
"learning_rate": 7.370521924987155e-05,
"loss": 0.0247,
"step": 11270
},
{
"grad_norm": 0.43648964166641235,
"learning_rate": 7.365667724674806e-05,
"loss": 0.0371,
"step": 11280
},
{
"grad_norm": 0.6037659645080566,
"learning_rate": 7.360810649853105e-05,
"loss": 0.0241,
"step": 11290
},
{
"grad_norm": 0.6314641237258911,
"learning_rate": 7.355950706423853e-05,
"loss": 0.0289,
"step": 11300
},
{
"grad_norm": 0.6193861961364746,
"learning_rate": 7.351087900292342e-05,
"loss": 0.0286,
"step": 11310
},
{
"grad_norm": 0.47861620783805847,
"learning_rate": 7.346222237367339e-05,
"loss": 0.0249,
"step": 11320
},
{
"grad_norm": 0.485915869474411,
"learning_rate": 7.341353723561081e-05,
"loss": 0.0239,
"step": 11330
},
{
"grad_norm": 0.5347712635993958,
"learning_rate": 7.336482364789277e-05,
"loss": 0.021,
"step": 11340
},
{
"grad_norm": 0.5797995924949646,
"learning_rate": 7.331608166971082e-05,
"loss": 0.0421,
"step": 11350
},
{
"grad_norm": 0.5185192227363586,
"learning_rate": 7.326731136029108e-05,
"loss": 0.0255,
"step": 11360
},
{
"grad_norm": 0.4394061863422394,
"learning_rate": 7.321851277889408e-05,
"loss": 0.0187,
"step": 11370
},
{
"grad_norm": 0.5574398040771484,
"learning_rate": 7.316968598481469e-05,
"loss": 0.0232,
"step": 11380
},
{
"grad_norm": 0.459883451461792,
"learning_rate": 7.312083103738206e-05,
"loss": 0.0309,
"step": 11390
},
{
"grad_norm": 0.507987380027771,
"learning_rate": 7.307194799595958e-05,
"loss": 0.0324,
"step": 11400
},
{
"grad_norm": 0.6104898452758789,
"learning_rate": 7.302303691994473e-05,
"loss": 0.0238,
"step": 11410
},
{
"grad_norm": 0.6142218112945557,
"learning_rate": 7.29740978687691e-05,
"loss": 0.0392,
"step": 11420
},
{
"grad_norm": 0.49763092398643494,
"learning_rate": 7.292513090189825e-05,
"loss": 0.0238,
"step": 11430
},
{
"grad_norm": 0.524719774723053,
"learning_rate": 7.287613607883163e-05,
"loss": 0.0225,
"step": 11440
},
{
"grad_norm": 0.4774492084980011,
"learning_rate": 7.282711345910263e-05,
"loss": 0.0233,
"step": 11450
},
{
"grad_norm": 0.5299280285835266,
"learning_rate": 7.27780631022783e-05,
"loss": 0.0302,
"step": 11460
},
{
"grad_norm": 0.5470672249794006,
"learning_rate": 7.272898506795948e-05,
"loss": 0.0204,
"step": 11470
},
{
"grad_norm": 0.5868731141090393,
"learning_rate": 7.267987941578058e-05,
"loss": 0.0213,
"step": 11480
},
{
"grad_norm": 0.7234907150268555,
"learning_rate": 7.263074620540962e-05,
"loss": 0.0284,
"step": 11490
},
{
"grad_norm": 0.5165469646453857,
"learning_rate": 7.25815854965481e-05,
"loss": 0.0227,
"step": 11500
},
{
"grad_norm": 0.47002914547920227,
"learning_rate": 7.253239734893089e-05,
"loss": 0.0201,
"step": 11510
},
{
"grad_norm": 0.5490991473197937,
"learning_rate": 7.248318182232623e-05,
"loss": 0.023,
"step": 11520
},
{
"grad_norm": 0.6184241771697998,
"learning_rate": 7.243393897653565e-05,
"loss": 0.0204,
"step": 11530
},
{
"grad_norm": 0.4693434238433838,
"learning_rate": 7.23846688713938e-05,
"loss": 0.0239,
"step": 11540
},
{
"grad_norm": 0.5601281523704529,
"learning_rate": 7.233537156676854e-05,
"loss": 0.0314,
"step": 11550
},
{
"grad_norm": 0.4588971436023712,
"learning_rate": 7.228604712256076e-05,
"loss": 0.0239,
"step": 11560
},
{
"grad_norm": 0.4428764581680298,
"learning_rate": 7.223669559870426e-05,
"loss": 0.0226,
"step": 11570
},
{
"grad_norm": 0.5917168259620667,
"learning_rate": 7.218731705516585e-05,
"loss": 0.0293,
"step": 11580
},
{
"grad_norm": 0.6205174922943115,
"learning_rate": 7.21379115519451e-05,
"loss": 0.0289,
"step": 11590
},
{
"grad_norm": 0.5595042109489441,
"learning_rate": 7.20884791490743e-05,
"loss": 0.0246,
"step": 11600
},
{
"grad_norm": 0.5059493780136108,
"learning_rate": 7.203901990661857e-05,
"loss": 0.0374,
"step": 11610
},
{
"grad_norm": 0.5916591286659241,
"learning_rate": 7.198953388467549e-05,
"loss": 0.0359,
"step": 11620
},
{
"grad_norm": 0.5272249579429626,
"learning_rate": 7.194002114337526e-05,
"loss": 0.045,
"step": 11630
},
{
"grad_norm": 0.54767906665802,
"learning_rate": 7.189048174288054e-05,
"loss": 0.0257,
"step": 11640
},
{
"grad_norm": 0.6054946780204773,
"learning_rate": 7.184091574338636e-05,
"loss": 0.0263,
"step": 11650
},
{
"grad_norm": 0.4777096211910248,
"learning_rate": 7.179132320512009e-05,
"loss": 0.0284,
"step": 11660
},
{
"grad_norm": 0.40717241168022156,
"learning_rate": 7.174170418834134e-05,
"loss": 0.0192,
"step": 11670
},
{
"grad_norm": 0.5232993960380554,
"learning_rate": 7.169205875334189e-05,
"loss": 0.0271,
"step": 11680
},
{
"grad_norm": 0.43350353837013245,
"learning_rate": 7.164238696044562e-05,
"loss": 0.0247,
"step": 11690
},
{
"grad_norm": 0.4589349031448364,
"learning_rate": 7.159268887000846e-05,
"loss": 0.0263,
"step": 11700
},
{
"grad_norm": 0.5655088424682617,
"learning_rate": 7.154296454241827e-05,
"loss": 0.0273,
"step": 11710
},
{
"grad_norm": 0.5438809394836426,
"learning_rate": 7.149321403809479e-05,
"loss": 0.0284,
"step": 11720
},
{
"grad_norm": 0.40988877415657043,
"learning_rate": 7.144343741748956e-05,
"loss": 0.0234,
"step": 11730
},
{
"grad_norm": 0.5398318767547607,
"learning_rate": 7.139363474108589e-05,
"loss": 0.025,
"step": 11740
},
{
"grad_norm": 0.5652521848678589,
"learning_rate": 7.134380606939874e-05,
"loss": 0.0333,
"step": 11750
},
{
"grad_norm": 0.4497848153114319,
"learning_rate": 7.12939514629746e-05,
"loss": 0.0268,
"step": 11760
},
{
"grad_norm": 0.44516339898109436,
"learning_rate": 7.124407098239155e-05,
"loss": 0.026,
"step": 11770
},
{
"grad_norm": 0.5375936627388,
"learning_rate": 7.119416468825908e-05,
"loss": 0.0233,
"step": 11780
},
{
"grad_norm": 0.5366640686988831,
"learning_rate": 7.114423264121804e-05,
"loss": 0.0222,
"step": 11790
},
{
"grad_norm": 0.5288453698158264,
"learning_rate": 7.109427490194056e-05,
"loss": 0.0272,
"step": 11800
},
{
"grad_norm": 0.6016205549240112,
"learning_rate": 7.104429153113001e-05,
"loss": 0.0327,
"step": 11810
},
{
"grad_norm": 0.660162627696991,
"learning_rate": 7.099428258952092e-05,
"loss": 0.0279,
"step": 11820
},
{
"grad_norm": 0.6235839128494263,
"learning_rate": 7.094424813787883e-05,
"loss": 0.0368,
"step": 11830
},
{
"grad_norm": 0.6197534203529358,
"learning_rate": 7.089418823700034e-05,
"loss": 0.0251,
"step": 11840
},
{
"grad_norm": 0.4736495614051819,
"learning_rate": 7.084410294771298e-05,
"loss": 0.0246,
"step": 11850
},
{
"grad_norm": 0.466431200504303,
"learning_rate": 7.079399233087504e-05,
"loss": 0.0296,
"step": 11860
},
{
"grad_norm": 0.5524370670318604,
"learning_rate": 7.074385644737568e-05,
"loss": 0.0233,
"step": 11870
},
{
"grad_norm": 0.4001089930534363,
"learning_rate": 7.069369535813473e-05,
"loss": 0.0372,
"step": 11880
},
{
"grad_norm": 0.5084970593452454,
"learning_rate": 7.06435091241026e-05,
"loss": 0.0212,
"step": 11890
},
{
"grad_norm": 0.5023282170295715,
"learning_rate": 7.059329780626034e-05,
"loss": 0.0247,
"step": 11900
},
{
"grad_norm": 0.5728763937950134,
"learning_rate": 7.054306146561944e-05,
"loss": 0.0289,
"step": 11910
},
{
"grad_norm": 0.5095998048782349,
"learning_rate": 7.049280016322178e-05,
"loss": 0.024,
"step": 11920
},
{
"grad_norm": 0.49976977705955505,
"learning_rate": 7.044251396013957e-05,
"loss": 0.0298,
"step": 11930
},
{
"grad_norm": 0.5828304290771484,
"learning_rate": 7.039220291747528e-05,
"loss": 0.0254,
"step": 11940
},
{
"grad_norm": 0.6110705733299255,
"learning_rate": 7.034186709636159e-05,
"loss": 0.0199,
"step": 11950
},
{
"grad_norm": 0.5116322040557861,
"learning_rate": 7.029150655796129e-05,
"loss": 0.0224,
"step": 11960
},
{
"grad_norm": 0.6065073609352112,
"learning_rate": 7.024112136346712e-05,
"loss": 0.0263,
"step": 11970
},
{
"grad_norm": 0.4956058859825134,
"learning_rate": 7.01907115741019e-05,
"loss": 0.0283,
"step": 11980
},
{
"grad_norm": 0.6078999042510986,
"learning_rate": 7.014027725111825e-05,
"loss": 0.0282,
"step": 11990
},
{
"grad_norm": 0.648478627204895,
"learning_rate": 7.008981845579864e-05,
"loss": 0.0396,
"step": 12000
},
{
"grad_norm": 0.47652679681777954,
"learning_rate": 7.003933524945528e-05,
"loss": 0.0254,
"step": 12010
},
{
"grad_norm": 0.6057568192481995,
"learning_rate": 6.998882769342998e-05,
"loss": 0.0247,
"step": 12020
},
{
"grad_norm": 0.5547711253166199,
"learning_rate": 6.993829584909423e-05,
"loss": 0.0227,
"step": 12030
},
{
"grad_norm": 0.5529880523681641,
"learning_rate": 6.988773977784895e-05,
"loss": 0.0292,
"step": 12040
},
{
"grad_norm": 0.6028531789779663,
"learning_rate": 6.983715954112454e-05,
"loss": 0.0326,
"step": 12050
},
{
"grad_norm": 0.4881529211997986,
"learning_rate": 6.978655520038079e-05,
"loss": 0.0284,
"step": 12060
},
{
"grad_norm": 0.5525237321853638,
"learning_rate": 6.97359268171067e-05,
"loss": 0.0397,
"step": 12070
},
{
"grad_norm": 0.6869008541107178,
"learning_rate": 6.968527445282056e-05,
"loss": 0.026,
"step": 12080
},
{
"grad_norm": 0.5513829588890076,
"learning_rate": 6.963459816906977e-05,
"loss": 0.0205,
"step": 12090
},
{
"grad_norm": 0.5451334118843079,
"learning_rate": 6.958389802743077e-05,
"loss": 0.0302,
"step": 12100
},
{
"grad_norm": 0.6114900708198547,
"learning_rate": 6.953317408950903e-05,
"loss": 0.0214,
"step": 12110
},
{
"grad_norm": 0.5863832831382751,
"learning_rate": 6.94824264169389e-05,
"loss": 0.022,
"step": 12120
},
{
"grad_norm": 0.558210015296936,
"learning_rate": 6.94316550713836e-05,
"loss": 0.0244,
"step": 12130
},
{
"grad_norm": 0.5351990461349487,
"learning_rate": 6.938086011453513e-05,
"loss": 0.0218,
"step": 12140
},
{
"grad_norm": 0.5699197053909302,
"learning_rate": 6.93300416081141e-05,
"loss": 0.0222,
"step": 12150
},
{
"grad_norm": 0.6497247815132141,
"learning_rate": 6.927919961386984e-05,
"loss": 0.0217,
"step": 12160
},
{
"grad_norm": 0.5265652537345886,
"learning_rate": 6.922833419358013e-05,
"loss": 0.0222,
"step": 12170
},
{
"grad_norm": 0.5426461100578308,
"learning_rate": 6.917744540905124e-05,
"loss": 0.0342,
"step": 12180
},
{
"grad_norm": 0.49767646193504333,
"learning_rate": 6.912653332211787e-05,
"loss": 0.0196,
"step": 12190
},
{
"grad_norm": 0.3577786087989807,
"learning_rate": 6.9075597994643e-05,
"loss": 0.0318,
"step": 12200
},
{
"grad_norm": 0.5193069577217102,
"learning_rate": 6.902463948851787e-05,
"loss": 0.0453,
"step": 12210
},
{
"grad_norm": 0.5483071208000183,
"learning_rate": 6.897365786566184e-05,
"loss": 0.0248,
"step": 12220
},
{
"grad_norm": 0.4992721974849701,
"learning_rate": 6.892265318802242e-05,
"loss": 0.0274,
"step": 12230
},
{
"grad_norm": 0.5099933743476868,
"learning_rate": 6.887162551757507e-05,
"loss": 0.0252,
"step": 12240
},
{
"grad_norm": 0.48819467425346375,
"learning_rate": 6.882057491632326e-05,
"loss": 0.0322,
"step": 12250
},
{
"grad_norm": 0.570290207862854,
"learning_rate": 6.876950144629824e-05,
"loss": 0.036,
"step": 12260
},
{
"grad_norm": 0.4806978106498718,
"learning_rate": 6.87184051695591e-05,
"loss": 0.0244,
"step": 12270
},
{
"grad_norm": 0.5586807727813721,
"learning_rate": 6.866728614819268e-05,
"loss": 0.0238,
"step": 12280
},
{
"grad_norm": 0.49823352694511414,
"learning_rate": 6.861614444431337e-05,
"loss": 0.0295,
"step": 12290
},
{
"grad_norm": 0.5393345952033997,
"learning_rate": 6.856498012006318e-05,
"loss": 0.0195,
"step": 12300
},
{
"grad_norm": 0.565272867679596,
"learning_rate": 6.851379323761157e-05,
"loss": 0.0252,
"step": 12310
},
{
"grad_norm": 0.6653050780296326,
"learning_rate": 6.846258385915545e-05,
"loss": 0.0297,
"step": 12320
},
{
"grad_norm": 0.49976181983947754,
"learning_rate": 6.841135204691902e-05,
"loss": 0.0218,
"step": 12330
},
{
"grad_norm": 0.6112423539161682,
"learning_rate": 6.836009786315377e-05,
"loss": 0.0278,
"step": 12340
},
{
"grad_norm": 0.49948421120643616,
"learning_rate": 6.830882137013839e-05,
"loss": 0.0232,
"step": 12350
},
{
"grad_norm": 0.5131206512451172,
"learning_rate": 6.825752263017863e-05,
"loss": 0.0307,
"step": 12360
},
{
"grad_norm": 0.5961295962333679,
"learning_rate": 6.820620170560731e-05,
"loss": 0.0222,
"step": 12370
},
{
"grad_norm": 0.6142684817314148,
"learning_rate": 6.815485865878418e-05,
"loss": 0.0259,
"step": 12380
},
{
"grad_norm": 0.5474580526351929,
"learning_rate": 6.810349355209588e-05,
"loss": 0.0291,
"step": 12390
},
{
"grad_norm": 0.5429353713989258,
"learning_rate": 6.805210644795588e-05,
"loss": 0.0309,
"step": 12400
},
{
"grad_norm": 0.597928524017334,
"learning_rate": 6.800069740880433e-05,
"loss": 0.0294,
"step": 12410
},
{
"grad_norm": 0.48671743273735046,
"learning_rate": 6.794926649710807e-05,
"loss": 0.0213,
"step": 12420
},
{
"grad_norm": 0.7151568531990051,
"learning_rate": 6.789781377536052e-05,
"loss": 0.034,
"step": 12430
},
{
"grad_norm": 0.640270471572876,
"learning_rate": 6.784633930608158e-05,
"loss": 0.0357,
"step": 12440
},
{
"grad_norm": 0.4663013517856598,
"learning_rate": 6.779484315181759e-05,
"loss": 0.0242,
"step": 12450
},
{
"grad_norm": 0.5653468370437622,
"learning_rate": 6.774332537514122e-05,
"loss": 0.0308,
"step": 12460
},
{
"grad_norm": 0.3421177566051483,
"learning_rate": 6.769178603865143e-05,
"loss": 0.0212,
"step": 12470
},
{
"grad_norm": 0.500761091709137,
"learning_rate": 6.764022520497337e-05,
"loss": 0.0181,
"step": 12480
},
{
"grad_norm": 0.43413594365119934,
"learning_rate": 6.758864293675833e-05,
"loss": 0.0233,
"step": 12490
},
{
"grad_norm": 0.45528745651245117,
"learning_rate": 6.753703929668362e-05,
"loss": 0.0287,
"step": 12500
},
{
"grad_norm": 0.6552013754844666,
"learning_rate": 6.748541434745253e-05,
"loss": 0.0396,
"step": 12510
},
{
"grad_norm": 0.7702287435531616,
"learning_rate": 6.743376815179424e-05,
"loss": 0.0272,
"step": 12520
},
{
"grad_norm": 0.6352901458740234,
"learning_rate": 6.738210077246376e-05,
"loss": 0.0262,
"step": 12530
},
{
"grad_norm": 0.5837974548339844,
"learning_rate": 6.733041227224181e-05,
"loss": 0.0354,
"step": 12540
},
{
"grad_norm": 0.5869095921516418,
"learning_rate": 6.72787027139348e-05,
"loss": 0.032,
"step": 12550
},
{
"grad_norm": 0.5836209058761597,
"learning_rate": 6.72269721603747e-05,
"loss": 0.0235,
"step": 12560
},
{
"grad_norm": 0.5046476721763611,
"learning_rate": 6.717522067441904e-05,
"loss": 0.0229,
"step": 12570
},
{
"grad_norm": 0.5972120761871338,
"learning_rate": 6.712344831895074e-05,
"loss": 0.0216,
"step": 12580
},
{
"grad_norm": 0.5265899300575256,
"learning_rate": 6.707165515687811e-05,
"loss": 0.0332,
"step": 12590
},
{
"grad_norm": 0.4666118621826172,
"learning_rate": 6.70198412511347e-05,
"loss": 0.0242,
"step": 12600
},
{
"grad_norm": 0.6096286773681641,
"learning_rate": 6.69680066646793e-05,
"loss": 0.0202,
"step": 12610
},
{
"grad_norm": 0.5215181708335876,
"learning_rate": 6.691615146049583e-05,
"loss": 0.0217,
"step": 12620
},
{
"grad_norm": 0.4468609690666199,
"learning_rate": 6.686427570159324e-05,
"loss": 0.0168,
"step": 12630
},
{
"grad_norm": 0.5390323996543884,
"learning_rate": 6.681237945100548e-05,
"loss": 0.0303,
"step": 12640
},
{
"grad_norm": 0.46825534105300903,
"learning_rate": 6.676046277179139e-05,
"loss": 0.0208,
"step": 12650
},
{
"grad_norm": 0.5392065644264221,
"learning_rate": 6.670852572703462e-05,
"loss": 0.0292,
"step": 12660
},
{
"grad_norm": 0.5165770053863525,
"learning_rate": 6.665656837984359e-05,
"loss": 0.0223,
"step": 12670
},
{
"grad_norm": 0.43299975991249084,
"learning_rate": 6.660459079335135e-05,
"loss": 0.0318,
"step": 12680
},
{
"grad_norm": 0.4496845602989197,
"learning_rate": 6.655259303071558e-05,
"loss": 0.0234,
"step": 12690
},
{
"grad_norm": 0.4688109755516052,
"learning_rate": 6.650057515511848e-05,
"loss": 0.0226,
"step": 12700
},
{
"grad_norm": 0.48314177989959717,
"learning_rate": 6.644853722976667e-05,
"loss": 0.0202,
"step": 12710
},
{
"grad_norm": 0.6051582098007202,
"learning_rate": 6.639647931789113e-05,
"loss": 0.0245,
"step": 12720
},
{
"grad_norm": 0.4984167814254761,
"learning_rate": 6.634440148274713e-05,
"loss": 0.0206,
"step": 12730
},
{
"grad_norm": 0.6376867294311523,
"learning_rate": 6.629230378761415e-05,
"loss": 0.0236,
"step": 12740
},
{
"grad_norm": 0.5019478797912598,
"learning_rate": 6.624018629579581e-05,
"loss": 0.0194,
"step": 12750
},
{
"grad_norm": 0.5625728368759155,
"learning_rate": 6.618804907061976e-05,
"loss": 0.0335,
"step": 12760
},
{
"grad_norm": 0.5341185331344604,
"learning_rate": 6.613589217543766e-05,
"loss": 0.0358,
"step": 12770
},
{
"grad_norm": 0.6405519843101501,
"learning_rate": 6.608371567362504e-05,
"loss": 0.0291,
"step": 12780
},
{
"grad_norm": 0.5221056342124939,
"learning_rate": 6.60315196285813e-05,
"loss": 0.0311,
"step": 12790
},
{
"grad_norm": 0.5446773171424866,
"learning_rate": 6.597930410372954e-05,
"loss": 0.0249,
"step": 12800
},
{
"grad_norm": 0.6241329312324524,
"learning_rate": 6.592706916251653e-05,
"loss": 0.0246,
"step": 12810
},
{
"grad_norm": 0.4280039370059967,
"learning_rate": 6.587481486841267e-05,
"loss": 0.0269,
"step": 12820
},
{
"grad_norm": 0.5963847041130066,
"learning_rate": 6.582254128491184e-05,
"loss": 0.0253,
"step": 12830
},
{
"grad_norm": 0.5306430459022522,
"learning_rate": 6.577024847553139e-05,
"loss": 0.0216,
"step": 12840
},
{
"grad_norm": 0.566694974899292,
"learning_rate": 6.571793650381201e-05,
"loss": 0.022,
"step": 12850
},
{
"grad_norm": 0.505228579044342,
"learning_rate": 6.566560543331766e-05,
"loss": 0.0255,
"step": 12860
},
{
"grad_norm": 0.4366820752620697,
"learning_rate": 6.561325532763554e-05,
"loss": 0.0209,
"step": 12870
},
{
"grad_norm": 0.5246875882148743,
"learning_rate": 6.556088625037597e-05,
"loss": 0.0179,
"step": 12880
},
{
"grad_norm": 0.42681029438972473,
"learning_rate": 6.55084982651723e-05,
"loss": 0.0215,
"step": 12890
},
{
"grad_norm": 0.4990365207195282,
"learning_rate": 6.54560914356809e-05,
"loss": 0.0296,
"step": 12900
},
{
"grad_norm": 0.44131505489349365,
"learning_rate": 6.540366582558097e-05,
"loss": 0.0211,
"step": 12910
},
{
"grad_norm": 0.589377224445343,
"learning_rate": 6.53512214985746e-05,
"loss": 0.0159,
"step": 12920
},
{
"grad_norm": 0.48958897590637207,
"learning_rate": 6.529875851838658e-05,
"loss": 0.0227,
"step": 12930
},
{
"grad_norm": 0.5186077356338501,
"learning_rate": 6.52462769487644e-05,
"loss": 0.0204,
"step": 12940
},
{
"grad_norm": 0.48851266503334045,
"learning_rate": 6.519377685347807e-05,
"loss": 0.0186,
"step": 12950
},
{
"grad_norm": 0.5329164862632751,
"learning_rate": 6.514125829632021e-05,
"loss": 0.0178,
"step": 12960
},
{
"grad_norm": 0.5639731287956238,
"learning_rate": 6.508872134110577e-05,
"loss": 0.0188,
"step": 12970
},
{
"grad_norm": 0.4259587228298187,
"learning_rate": 6.503616605167213e-05,
"loss": 0.0172,
"step": 12980
},
{
"grad_norm": 0.4810214042663574,
"learning_rate": 6.498359249187893e-05,
"loss": 0.0209,
"step": 12990
},
{
"grad_norm": 0.4611412286758423,
"learning_rate": 6.493100072560799e-05,
"loss": 0.0238,
"step": 13000
},
{
"grad_norm": 0.46649855375289917,
"learning_rate": 6.487839081676327e-05,
"loss": 0.0394,
"step": 13010
},
{
"grad_norm": 0.4901805520057678,
"learning_rate": 6.482576282927076e-05,
"loss": 0.0288,
"step": 13020
},
{
"grad_norm": 0.5271385312080383,
"learning_rate": 6.477311682707845e-05,
"loss": 0.0295,
"step": 13030
},
{
"grad_norm": 0.6626272201538086,
"learning_rate": 6.472045287415616e-05,
"loss": 0.0219,
"step": 13040
},
{
"grad_norm": 0.5435581207275391,
"learning_rate": 6.466777103449558e-05,
"loss": 0.0173,
"step": 13050
},
{
"grad_norm": 0.5784201622009277,
"learning_rate": 6.461507137211012e-05,
"loss": 0.0203,
"step": 13060
},
{
"grad_norm": 0.5615996718406677,
"learning_rate": 6.456235395103483e-05,
"loss": 0.0342,
"step": 13070
},
{
"grad_norm": 0.5899002552032471,
"learning_rate": 6.450961883532634e-05,
"loss": 0.0229,
"step": 13080
},
{
"grad_norm": 0.555974006652832,
"learning_rate": 6.445686608906283e-05,
"loss": 0.0233,
"step": 13090
},
{
"grad_norm": 0.44332408905029297,
"learning_rate": 6.44040957763438e-05,
"loss": 0.0238,
"step": 13100
},
{
"grad_norm": 0.46654751896858215,
"learning_rate": 6.435130796129018e-05,
"loss": 0.0183,
"step": 13110
},
{
"grad_norm": 0.567701518535614,
"learning_rate": 6.429850270804416e-05,
"loss": 0.0168,
"step": 13120
},
{
"grad_norm": 0.4689022898674011,
"learning_rate": 6.424568008076909e-05,
"loss": 0.0281,
"step": 13130
},
{
"grad_norm": 0.546794056892395,
"learning_rate": 6.419284014364943e-05,
"loss": 0.0253,
"step": 13140
},
{
"grad_norm": 0.5685749650001526,
"learning_rate": 6.41399829608907e-05,
"loss": 0.0228,
"step": 13150
},
{
"grad_norm": 0.433464914560318,
"learning_rate": 6.408710859671938e-05,
"loss": 0.0166,
"step": 13160
},
{
"grad_norm": 0.562315821647644,
"learning_rate": 6.403421711538278e-05,
"loss": 0.0193,
"step": 13170
},
{
"grad_norm": 0.409152626991272,
"learning_rate": 6.398130858114903e-05,
"loss": 0.0244,
"step": 13180
},
{
"grad_norm": 0.44981682300567627,
"learning_rate": 6.392838305830701e-05,
"loss": 0.0173,
"step": 13190
},
{
"grad_norm": 0.6406846046447754,
"learning_rate": 6.387544061116621e-05,
"loss": 0.0258,
"step": 13200
},
{
"grad_norm": 0.45574530959129333,
"learning_rate": 6.382248130405671e-05,
"loss": 0.0169,
"step": 13210
},
{
"grad_norm": 0.3962901532649994,
"learning_rate": 6.376950520132905e-05,
"loss": 0.0189,
"step": 13220
},
{
"grad_norm": 0.4806446433067322,
"learning_rate": 6.371651236735418e-05,
"loss": 0.0254,
"step": 13230
},
{
"grad_norm": 0.6413816213607788,
"learning_rate": 6.366350286652341e-05,
"loss": 0.031,
"step": 13240
},
{
"grad_norm": 0.4670848548412323,
"learning_rate": 6.361047676324827e-05,
"loss": 0.0201,
"step": 13250
},
{
"grad_norm": 0.6498979330062866,
"learning_rate": 6.355743412196047e-05,
"loss": 0.0249,
"step": 13260
},
{
"grad_norm": 0.5692505836486816,
"learning_rate": 6.350437500711184e-05,
"loss": 0.0228,
"step": 13270
},
{
"grad_norm": 0.5747967958450317,
"learning_rate": 6.345129948317418e-05,
"loss": 0.0213,
"step": 13280
},
{
"grad_norm": 0.5064454078674316,
"learning_rate": 6.33982076146393e-05,
"loss": 0.0213,
"step": 13290
},
{
"grad_norm": 0.5429168939590454,
"learning_rate": 6.334509946601879e-05,
"loss": 0.0193,
"step": 13300
},
{
"grad_norm": 0.5899654030799866,
"learning_rate": 6.329197510184406e-05,
"loss": 0.0261,
"step": 13310
},
{
"grad_norm": 0.5409785509109497,
"learning_rate": 6.323883458666624e-05,
"loss": 0.0272,
"step": 13320
},
{
"grad_norm": 0.4772477447986603,
"learning_rate": 6.318567798505605e-05,
"loss": 0.027,
"step": 13330
},
{
"grad_norm": 0.42332810163497925,
"learning_rate": 6.313250536160378e-05,
"loss": 0.0221,
"step": 13340
},
{
"grad_norm": 0.5681222677230835,
"learning_rate": 6.307931678091917e-05,
"loss": 0.0182,
"step": 13350
},
{
"grad_norm": 0.4789891839027405,
"learning_rate": 6.302611230763138e-05,
"loss": 0.0211,
"step": 13360
},
{
"grad_norm": 0.5313396453857422,
"learning_rate": 6.297289200638887e-05,
"loss": 0.0288,
"step": 13370
},
{
"grad_norm": 0.5089273452758789,
"learning_rate": 6.29196559418593e-05,
"loss": 0.0263,
"step": 13380
},
{
"grad_norm": 0.5141650438308716,
"learning_rate": 6.286640417872951e-05,
"loss": 0.0249,
"step": 13390
},
{
"grad_norm": 0.5642871856689453,
"learning_rate": 6.281313678170542e-05,
"loss": 0.0316,
"step": 13400
},
{
"grad_norm": 0.5751014947891235,
"learning_rate": 6.275985381551194e-05,
"loss": 0.0206,
"step": 13410
},
{
"grad_norm": 0.45068448781967163,
"learning_rate": 6.270655534489292e-05,
"loss": 0.0237,
"step": 13420
},
{
"grad_norm": 0.5279017686843872,
"learning_rate": 6.265324143461098e-05,
"loss": 0.0179,
"step": 13430
},
{
"grad_norm": 0.5911248326301575,
"learning_rate": 6.259991214944758e-05,
"loss": 0.0209,
"step": 13440
},
{
"grad_norm": 0.6161231398582458,
"learning_rate": 6.254656755420283e-05,
"loss": 0.0306,
"step": 13450
},
{
"grad_norm": 0.49795427918434143,
"learning_rate": 6.249320771369545e-05,
"loss": 0.0329,
"step": 13460
},
{
"grad_norm": 0.589379608631134,
"learning_rate": 6.243983269276263e-05,
"loss": 0.0345,
"step": 13470
},
{
"grad_norm": 0.4845764636993408,
"learning_rate": 6.238644255626012e-05,
"loss": 0.0266,
"step": 13480
},
{
"grad_norm": 0.4284992814064026,
"learning_rate": 6.233303736906193e-05,
"loss": 0.0241,
"step": 13490
},
{
"grad_norm": 0.6017924547195435,
"learning_rate": 6.227961719606039e-05,
"loss": 0.0161,
"step": 13500
},
{
"grad_norm": 0.470271497964859,
"learning_rate": 6.222618210216608e-05,
"loss": 0.028,
"step": 13510
},
{
"grad_norm": 0.5801962018013,
"learning_rate": 6.217273215230767e-05,
"loss": 0.0323,
"step": 13520
},
{
"grad_norm": 0.49787667393684387,
"learning_rate": 6.211926741143188e-05,
"loss": 0.0276,
"step": 13530
},
{
"grad_norm": 0.5178689360618591,
"learning_rate": 6.20657879445034e-05,
"loss": 0.0275,
"step": 13540
},
{
"grad_norm": 0.5051644444465637,
"learning_rate": 6.201229381650485e-05,
"loss": 0.0181,
"step": 13550
},
{
"grad_norm": 0.5673476457595825,
"learning_rate": 6.195878509243661e-05,
"loss": 0.0248,
"step": 13560
},
{
"grad_norm": 0.5351276397705078,
"learning_rate": 6.190526183731685e-05,
"loss": 0.0198,
"step": 13570
},
{
"grad_norm": 0.47061482071876526,
"learning_rate": 6.185172411618138e-05,
"loss": 0.0208,
"step": 13580
},
{
"grad_norm": 0.5516050457954407,
"learning_rate": 6.179817199408355e-05,
"loss": 0.03,
"step": 13590
},
{
"grad_norm": 0.4483397901058197,
"learning_rate": 6.174460553609426e-05,
"loss": 0.0404,
"step": 13600
},
{
"grad_norm": 0.47059401869773865,
"learning_rate": 6.16910248073018e-05,
"loss": 0.02,
"step": 13610
},
{
"grad_norm": 0.4948214888572693,
"learning_rate": 6.16374298728118e-05,
"loss": 0.0343,
"step": 13620
},
{
"grad_norm": 0.5488273501396179,
"learning_rate": 6.158382079774716e-05,
"loss": 0.0202,
"step": 13630
},
{
"grad_norm": 0.5804873704910278,
"learning_rate": 6.153019764724799e-05,
"loss": 0.0201,
"step": 13640
},
{
"grad_norm": 0.5161739587783813,
"learning_rate": 6.147656048647144e-05,
"loss": 0.0274,
"step": 13650
},
{
"grad_norm": 0.5509073734283447,
"learning_rate": 6.142290938059173e-05,
"loss": 0.0268,
"step": 13660
},
{
"grad_norm": 0.6063939332962036,
"learning_rate": 6.136924439480001e-05,
"loss": 0.0218,
"step": 13670
},
{
"grad_norm": 0.578978419303894,
"learning_rate": 6.13155655943043e-05,
"loss": 0.0193,
"step": 13680
},
{
"grad_norm": 0.6555829644203186,
"learning_rate": 6.126187304432941e-05,
"loss": 0.0241,
"step": 13690
},
{
"grad_norm": 0.5103971362113953,
"learning_rate": 6.120816681011684e-05,
"loss": 0.0237,
"step": 13700
},
{
"grad_norm": 0.5035702586174011,
"learning_rate": 6.115444695692474e-05,
"loss": 0.0196,
"step": 13710
},
{
"grad_norm": 0.5113937854766846,
"learning_rate": 6.110071355002779e-05,
"loss": 0.0184,
"step": 13720
},
{
"grad_norm": 0.49035438895225525,
"learning_rate": 6.104696665471714e-05,
"loss": 0.0193,
"step": 13730
},
{
"grad_norm": 0.4406224489212036,
"learning_rate": 6.099320633630036e-05,
"loss": 0.0191,
"step": 13740
},
{
"grad_norm": 0.43845370411872864,
"learning_rate": 6.093943266010128e-05,
"loss": 0.0221,
"step": 13750
},
{
"grad_norm": 0.5333425402641296,
"learning_rate": 6.088564569146e-05,
"loss": 0.0386,
"step": 13760
},
{
"grad_norm": 0.4694617688655853,
"learning_rate": 6.083184549573274e-05,
"loss": 0.0197,
"step": 13770
},
{
"grad_norm": 0.5224243402481079,
"learning_rate": 6.077803213829184e-05,
"loss": 0.019,
"step": 13780
},
{
"grad_norm": 0.5152093172073364,
"learning_rate": 6.0724205684525594e-05,
"loss": 0.0203,
"step": 13790
},
{
"grad_norm": 0.5144279599189758,
"learning_rate": 6.0670366199838215e-05,
"loss": 0.0186,
"step": 13800
},
{
"grad_norm": 0.5832288861274719,
"learning_rate": 6.061651374964974e-05,
"loss": 0.0289,
"step": 13810
},
{
"grad_norm": 0.4512421190738678,
"learning_rate": 6.0562648399396004e-05,
"loss": 0.0228,
"step": 13820
},
{
"grad_norm": 0.4812946915626526,
"learning_rate": 6.050877021452845e-05,
"loss": 0.0242,
"step": 13830
},
{
"grad_norm": 0.5242236256599426,
"learning_rate": 6.04548792605142e-05,
"loss": 0.0253,
"step": 13840
},
{
"grad_norm": 0.506477415561676,
"learning_rate": 6.04009756028358e-05,
"loss": 0.0171,
"step": 13850
},
{
"grad_norm": 0.5122888088226318,
"learning_rate": 6.0347059306991294e-05,
"loss": 0.02,
"step": 13860
},
{
"grad_norm": 0.43320560455322266,
"learning_rate": 6.029313043849407e-05,
"loss": 0.0228,
"step": 13870
},
{
"grad_norm": 0.3890727460384369,
"learning_rate": 6.023918906287279e-05,
"loss": 0.0146,
"step": 13880
},
{
"grad_norm": 0.49916714429855347,
"learning_rate": 6.01852352456713e-05,
"loss": 0.0268,
"step": 13890
},
{
"grad_norm": 0.49414682388305664,
"learning_rate": 6.013126905244858e-05,
"loss": 0.0155,
"step": 13900
},
{
"grad_norm": 0.4326798617839813,
"learning_rate": 6.007729054877864e-05,
"loss": 0.0206,
"step": 13910
},
{
"grad_norm": 0.5811951756477356,
"learning_rate": 6.0023299800250466e-05,
"loss": 0.0228,
"step": 13920
},
{
"grad_norm": 0.6308193206787109,
"learning_rate": 5.9969296872467894e-05,
"loss": 0.0205,
"step": 13930
},
{
"grad_norm": 0.495025098323822,
"learning_rate": 5.9915281831049585e-05,
"loss": 0.0208,
"step": 13940
},
{
"grad_norm": 0.5016384720802307,
"learning_rate": 5.98612547416289e-05,
"loss": 0.0201,
"step": 13950
},
{
"grad_norm": 0.4948132634162903,
"learning_rate": 5.9807215669853855e-05,
"loss": 0.0221,
"step": 13960
},
{
"grad_norm": 0.5061405897140503,
"learning_rate": 5.9753164681387e-05,
"loss": 0.0425,
"step": 13970
},
{
"grad_norm": 0.6127529740333557,
"learning_rate": 5.969910184190539e-05,
"loss": 0.0275,
"step": 13980
},
{
"grad_norm": 0.5329326391220093,
"learning_rate": 5.9645027217100477e-05,
"loss": 0.0211,
"step": 13990
},
{
"grad_norm": 0.5093228816986084,
"learning_rate": 5.959094087267804e-05,
"loss": 0.023,
"step": 14000
},
{
"grad_norm": 0.4908728003501892,
"learning_rate": 5.9536842874358065e-05,
"loss": 0.0185,
"step": 14010
},
{
"grad_norm": 0.48747098445892334,
"learning_rate": 5.9482733287874734e-05,
"loss": 0.0289,
"step": 14020
},
{
"grad_norm": 0.4782416522502899,
"learning_rate": 5.942861217897631e-05,
"loss": 0.0246,
"step": 14030
},
{
"grad_norm": 0.5104605555534363,
"learning_rate": 5.9374479613425e-05,
"loss": 0.0174,
"step": 14040
},
{
"grad_norm": 0.47266533970832825,
"learning_rate": 5.932033565699704e-05,
"loss": 0.0178,
"step": 14050
},
{
"grad_norm": 0.466538667678833,
"learning_rate": 5.926618037548237e-05,
"loss": 0.0246,
"step": 14060
},
{
"grad_norm": 0.3909836709499359,
"learning_rate": 5.9212013834684824e-05,
"loss": 0.022,
"step": 14070
},
{
"grad_norm": 0.46042880415916443,
"learning_rate": 5.9157836100421824e-05,
"loss": 0.0202,
"step": 14080
},
{
"grad_norm": 0.4964950978755951,
"learning_rate": 5.910364723852444e-05,
"loss": 0.0177,
"step": 14090
},
{
"grad_norm": 0.4591788053512573,
"learning_rate": 5.904944731483723e-05,
"loss": 0.0252,
"step": 14100
},
{
"grad_norm": 0.47894972562789917,
"learning_rate": 5.899523639521825e-05,
"loss": 0.0247,
"step": 14110
},
{
"grad_norm": 0.43118035793304443,
"learning_rate": 5.894101454553883e-05,
"loss": 0.0187,
"step": 14120
},
{
"grad_norm": 0.5279966592788696,
"learning_rate": 5.888678183168368e-05,
"loss": 0.0298,
"step": 14130
},
{
"grad_norm": 0.5553144216537476,
"learning_rate": 5.883253831955061e-05,
"loss": 0.0242,
"step": 14140
},
{
"grad_norm": 0.43117618560791016,
"learning_rate": 5.8778284075050625e-05,
"loss": 0.0186,
"step": 14150
},
{
"grad_norm": 0.6525658369064331,
"learning_rate": 5.872401916410777e-05,
"loss": 0.0375,
"step": 14160
},
{
"grad_norm": 0.5280880928039551,
"learning_rate": 5.8669743652659014e-05,
"loss": 0.0251,
"step": 14170
},
{
"grad_norm": 0.4885375201702118,
"learning_rate": 5.86154576066542e-05,
"loss": 0.0209,
"step": 14180
},
{
"grad_norm": 0.5092771053314209,
"learning_rate": 5.856116109205602e-05,
"loss": 0.0255,
"step": 14190
},
{
"grad_norm": 0.45492205023765564,
"learning_rate": 5.850685417483983e-05,
"loss": 0.0213,
"step": 14200
},
{
"grad_norm": 0.6266351342201233,
"learning_rate": 5.8452536920993684e-05,
"loss": 0.0219,
"step": 14210
},
{
"grad_norm": 0.5126073360443115,
"learning_rate": 5.839820939651817e-05,
"loss": 0.0297,
"step": 14220
},
{
"grad_norm": 0.47290733456611633,
"learning_rate": 5.834387166742632e-05,
"loss": 0.0216,
"step": 14230
},
{
"grad_norm": 0.5791200995445251,
"learning_rate": 5.828952379974364e-05,
"loss": 0.0216,
"step": 14240
},
{
"grad_norm": 0.468591570854187,
"learning_rate": 5.8235165859507864e-05,
"loss": 0.029,
"step": 14250
},
{
"grad_norm": 0.5315958857536316,
"learning_rate": 5.818079791276907e-05,
"loss": 0.0244,
"step": 14260
},
{
"grad_norm": 0.49462300539016724,
"learning_rate": 5.8126420025589415e-05,
"loss": 0.0265,
"step": 14270
},
{
"grad_norm": 0.5535178780555725,
"learning_rate": 5.807203226404313e-05,
"loss": 0.0228,
"step": 14280
},
{
"grad_norm": 0.4999871253967285,
"learning_rate": 5.801763469421652e-05,
"loss": 0.0184,
"step": 14290
},
{
"grad_norm": 0.4495685398578644,
"learning_rate": 5.796322738220774e-05,
"loss": 0.0254,
"step": 14300
},
{
"grad_norm": 0.5187375545501709,
"learning_rate": 5.79088103941268e-05,
"loss": 0.0189,
"step": 14310
},
{
"grad_norm": 0.48773083090782166,
"learning_rate": 5.785438379609549e-05,
"loss": 0.0325,
"step": 14320
},
{
"grad_norm": 0.5059367418289185,
"learning_rate": 5.779994765424724e-05,
"loss": 0.0183,
"step": 14330
},
{
"grad_norm": 0.5208343863487244,
"learning_rate": 5.7745502034727114e-05,
"loss": 0.0216,
"step": 14340
},
{
"grad_norm": 0.4834510087966919,
"learning_rate": 5.769104700369164e-05,
"loss": 0.0239,
"step": 14350
},
{
"grad_norm": 0.3189760148525238,
"learning_rate": 5.7636582627308854e-05,
"loss": 0.017,
"step": 14360
},
{
"grad_norm": 0.536418080329895,
"learning_rate": 5.7582108971758095e-05,
"loss": 0.0218,
"step": 14370
},
{
"grad_norm": 0.45111915469169617,
"learning_rate": 5.752762610322999e-05,
"loss": 0.0205,
"step": 14380
},
{
"grad_norm": 0.45027846097946167,
"learning_rate": 5.747313408792636e-05,
"loss": 0.0294,
"step": 14390
},
{
"grad_norm": 0.479526549577713,
"learning_rate": 5.741863299206014e-05,
"loss": 0.0217,
"step": 14400
},
{
"grad_norm": 0.5127474069595337,
"learning_rate": 5.73641228818553e-05,
"loss": 0.0169,
"step": 14410
},
{
"grad_norm": 0.5346008539199829,
"learning_rate": 5.730960382354676e-05,
"loss": 0.0198,
"step": 14420
},
{
"grad_norm": 0.4975365102291107,
"learning_rate": 5.725507588338035e-05,
"loss": 0.0194,
"step": 14430
},
{
"grad_norm": 0.5815244913101196,
"learning_rate": 5.7200539127612604e-05,
"loss": 0.0175,
"step": 14440
},
{
"grad_norm": 0.4840431809425354,
"learning_rate": 5.714599362251087e-05,
"loss": 0.0204,
"step": 14450
},
{
"grad_norm": 0.47391101717948914,
"learning_rate": 5.709143943435307e-05,
"loss": 0.0186,
"step": 14460
},
{
"grad_norm": 0.46357056498527527,
"learning_rate": 5.7036876629427646e-05,
"loss": 0.0258,
"step": 14470
},
{
"grad_norm": 0.44239541888237,
"learning_rate": 5.698230527403361e-05,
"loss": 0.0194,
"step": 14480
},
{
"grad_norm": 0.5034361481666565,
"learning_rate": 5.6927725434480274e-05,
"loss": 0.0194,
"step": 14490
},
{
"grad_norm": 0.31999504566192627,
"learning_rate": 5.687313717708728e-05,
"loss": 0.0338,
"step": 14500
},
{
"grad_norm": 0.49579980969429016,
"learning_rate": 5.681854056818453e-05,
"loss": 0.0242,
"step": 14510
},
{
"grad_norm": 0.5044465065002441,
"learning_rate": 5.676393567411205e-05,
"loss": 0.0165,
"step": 14520
},
{
"grad_norm": 0.5233392715454102,
"learning_rate": 5.670932256121991e-05,
"loss": 0.0371,
"step": 14530
},
{
"grad_norm": 0.46637219190597534,
"learning_rate": 5.6654701295868216e-05,
"loss": 0.0174,
"step": 14540
},
{
"grad_norm": 0.5456653833389282,
"learning_rate": 5.660007194442697e-05,
"loss": 0.0181,
"step": 14550
},
{
"grad_norm": 0.5302630066871643,
"learning_rate": 5.6545434573275946e-05,
"loss": 0.0193,
"step": 14560
},
{
"grad_norm": 0.4971660375595093,
"learning_rate": 5.649078924880472e-05,
"loss": 0.0162,
"step": 14570
},
{
"grad_norm": 0.5420553088188171,
"learning_rate": 5.643613603741252e-05,
"loss": 0.0248,
"step": 14580
},
{
"grad_norm": 0.4389548897743225,
"learning_rate": 5.6381475005508154e-05,
"loss": 0.0311,
"step": 14590
},
{
"grad_norm": 0.487795352935791,
"learning_rate": 5.632680621950992e-05,
"loss": 0.0272,
"step": 14600
},
{
"grad_norm": 0.4966619610786438,
"learning_rate": 5.627212974584555e-05,
"loss": 0.0175,
"step": 14610
},
{
"grad_norm": 0.5126991271972656,
"learning_rate": 5.62174456509521e-05,
"loss": 0.0199,
"step": 14620
},
{
"grad_norm": 0.42528289556503296,
"learning_rate": 5.616275400127594e-05,
"loss": 0.019,
"step": 14630
},
{
"grad_norm": 0.4664388597011566,
"learning_rate": 5.610805486327254e-05,
"loss": 0.0176,
"step": 14640
},
{
"grad_norm": 0.4523860812187195,
"learning_rate": 5.6053348303406536e-05,
"loss": 0.0198,
"step": 14650
},
{
"grad_norm": 0.5383732318878174,
"learning_rate": 5.599863438815156e-05,
"loss": 0.02,
"step": 14660
},
{
"grad_norm": 0.4415532648563385,
"learning_rate": 5.594391318399017e-05,
"loss": 0.0252,
"step": 14670
},
{
"grad_norm": 0.4967097342014313,
"learning_rate": 5.588918475741377e-05,
"loss": 0.0182,
"step": 14680
},
{
"grad_norm": 0.4659595787525177,
"learning_rate": 5.5834449174922586e-05,
"loss": 0.0237,
"step": 14690
},
{
"grad_norm": 0.43765997886657715,
"learning_rate": 5.57797065030255e-05,
"loss": 0.0247,
"step": 14700
},
{
"grad_norm": 0.4963107705116272,
"learning_rate": 5.5724956808240016e-05,
"loss": 0.0275,
"step": 14710
},
{
"grad_norm": 0.4939739406108856,
"learning_rate": 5.567020015709219e-05,
"loss": 0.0224,
"step": 14720
},
{
"grad_norm": 0.5268587470054626,
"learning_rate": 5.561543661611649e-05,
"loss": 0.023,
"step": 14730
},
{
"grad_norm": 0.5268352031707764,
"learning_rate": 5.556066625185583e-05,
"loss": 0.0257,
"step": 14740
},
{
"grad_norm": 0.4926978647708893,
"learning_rate": 5.550588913086131e-05,
"loss": 0.0229,
"step": 14750
},
{
"grad_norm": 0.5762534141540527,
"learning_rate": 5.545110531969234e-05,
"loss": 0.0231,
"step": 14760
},
{
"grad_norm": 0.5706159472465515,
"learning_rate": 5.539631488491641e-05,
"loss": 0.0228,
"step": 14770
},
{
"grad_norm": 0.5364542007446289,
"learning_rate": 5.534151789310904e-05,
"loss": 0.0219,
"step": 14780
},
{
"grad_norm": 0.48187071084976196,
"learning_rate": 5.528671441085376e-05,
"loss": 0.018,
"step": 14790
},
{
"grad_norm": 0.4640049338340759,
"learning_rate": 5.523190450474197e-05,
"loss": 0.0247,
"step": 14800
},
{
"grad_norm": 0.3426426351070404,
"learning_rate": 5.5177088241372865e-05,
"loss": 0.0154,
"step": 14810
},
{
"grad_norm": 0.4653962552547455,
"learning_rate": 5.512226568735338e-05,
"loss": 0.0206,
"step": 14820
},
{
"grad_norm": 0.46503666043281555,
"learning_rate": 5.506743690929809e-05,
"loss": 0.0295,
"step": 14830
},
{
"grad_norm": 0.4653470516204834,
"learning_rate": 5.501260197382912e-05,
"loss": 0.0358,
"step": 14840
},
{
"grad_norm": 0.5856786966323853,
"learning_rate": 5.49577609475761e-05,
"loss": 0.0367,
"step": 14850
},
{
"grad_norm": 0.4468640387058258,
"learning_rate": 5.490291389717603e-05,
"loss": 0.0187,
"step": 14860
},
{
"grad_norm": 0.5366262793540955,
"learning_rate": 5.484806088927329e-05,
"loss": 0.018,
"step": 14870
},
{
"grad_norm": 0.6121026873588562,
"learning_rate": 5.479320199051942e-05,
"loss": 0.0224,
"step": 14880
},
{
"grad_norm": 0.6029980182647705,
"learning_rate": 5.473833726757314e-05,
"loss": 0.0197,
"step": 14890
},
{
"grad_norm": 0.5525507926940918,
"learning_rate": 5.4683466787100304e-05,
"loss": 0.0183,
"step": 14900
},
{
"grad_norm": 0.5265474915504456,
"learning_rate": 5.462859061577369e-05,
"loss": 0.0217,
"step": 14910
},
{
"grad_norm": 0.41470423340797424,
"learning_rate": 5.4573708820273026e-05,
"loss": 0.0177,
"step": 14920
},
{
"grad_norm": 0.4511779844760895,
"learning_rate": 5.451882146728488e-05,
"loss": 0.0189,
"step": 14930
},
{
"grad_norm": 0.4308796525001526,
"learning_rate": 5.446392862350255e-05,
"loss": 0.0248,
"step": 14940
},
{
"grad_norm": 0.4632108509540558,
"learning_rate": 5.440903035562603e-05,
"loss": 0.0357,
"step": 14950
},
{
"grad_norm": 0.4534340500831604,
"learning_rate": 5.435412673036188e-05,
"loss": 0.0231,
"step": 14960
},
{
"grad_norm": 0.5301945209503174,
"learning_rate": 5.429921781442318e-05,
"loss": 0.0184,
"step": 14970
},
{
"grad_norm": 0.3546154201030731,
"learning_rate": 5.424430367452945e-05,
"loss": 0.0196,
"step": 14980
},
{
"grad_norm": 0.49920833110809326,
"learning_rate": 5.418938437740655e-05,
"loss": 0.0201,
"step": 14990
},
{
"grad_norm": 0.6312521696090698,
"learning_rate": 5.413445998978658e-05,
"loss": 0.0268,
"step": 15000
},
{
"grad_norm": 0.44467583298683167,
"learning_rate": 5.407953057840789e-05,
"loss": 0.016,
"step": 15010
},
{
"grad_norm": 0.5252653360366821,
"learning_rate": 5.4024596210014853e-05,
"loss": 0.0172,
"step": 15020
},
{
"grad_norm": 0.5147327184677124,
"learning_rate": 5.3969656951357935e-05,
"loss": 0.0187,
"step": 15030
},
{
"grad_norm": 0.3574521541595459,
"learning_rate": 5.3914712869193504e-05,
"loss": 0.0314,
"step": 15040
},
{
"grad_norm": 0.6194702386856079,
"learning_rate": 5.385976403028381e-05,
"loss": 0.0327,
"step": 15050
},
{
"grad_norm": 0.5916879177093506,
"learning_rate": 5.3804810501396875e-05,
"loss": 0.0282,
"step": 15060
},
{
"grad_norm": 0.35817068815231323,
"learning_rate": 5.374985234930639e-05,
"loss": 0.0216,
"step": 15070
},
{
"grad_norm": 0.4748634099960327,
"learning_rate": 5.3694889640791724e-05,
"loss": 0.0215,
"step": 15080
},
{
"grad_norm": 0.5741844773292542,
"learning_rate": 5.3639922442637736e-05,
"loss": 0.0236,
"step": 15090
},
{
"grad_norm": 0.4340136647224426,
"learning_rate": 5.358495082163476e-05,
"loss": 0.0208,
"step": 15100
},
{
"grad_norm": 0.4166621267795563,
"learning_rate": 5.35299748445785e-05,
"loss": 0.0213,
"step": 15110
},
{
"grad_norm": 0.5046685338020325,
"learning_rate": 5.3474994578269944e-05,
"loss": 0.0169,
"step": 15120
},
{
"grad_norm": 0.4489562511444092,
"learning_rate": 5.34200100895153e-05,
"loss": 0.0172,
"step": 15130
},
{
"grad_norm": 0.4482318162918091,
"learning_rate": 5.3365021445125916e-05,
"loss": 0.0143,
"step": 15140
},
{
"grad_norm": 0.5153161883354187,
"learning_rate": 5.331002871191817e-05,
"loss": 0.0226,
"step": 15150
},
{
"grad_norm": 0.4701783061027527,
"learning_rate": 5.3255031956713443e-05,
"loss": 0.0217,
"step": 15160
},
{
"grad_norm": 0.5924737453460693,
"learning_rate": 5.320003124633795e-05,
"loss": 0.0273,
"step": 15170
},
{
"grad_norm": 0.491658091545105,
"learning_rate": 5.314502664762274e-05,
"loss": 0.0205,
"step": 15180
},
{
"grad_norm": 0.47612354159355164,
"learning_rate": 5.3090018227403604e-05,
"loss": 0.0191,
"step": 15190
},
{
"grad_norm": 0.5122098326683044,
"learning_rate": 5.3035006052520955e-05,
"loss": 0.0269,
"step": 15200
},
{
"grad_norm": 0.3714747130870819,
"learning_rate": 5.297999018981976e-05,
"loss": 0.021,
"step": 15210
},
{
"grad_norm": 0.5548262596130371,
"learning_rate": 5.2924970706149505e-05,
"loss": 0.0184,
"step": 15220
},
{
"grad_norm": 0.4201054275035858,
"learning_rate": 5.2869947668364014e-05,
"loss": 0.013,
"step": 15230
},
{
"grad_norm": 0.3828338384628296,
"learning_rate": 5.2814921143321504e-05,
"loss": 0.0196,
"step": 15240
},
{
"grad_norm": 0.4574180543422699,
"learning_rate": 5.275989119788436e-05,
"loss": 0.0318,
"step": 15250
},
{
"grad_norm": 0.4920981526374817,
"learning_rate": 5.270485789891919e-05,
"loss": 0.0237,
"step": 15260
},
{
"grad_norm": 0.5518547892570496,
"learning_rate": 5.264982131329661e-05,
"loss": 0.0172,
"step": 15270
},
{
"grad_norm": 0.4956014156341553,
"learning_rate": 5.259478150789128e-05,
"loss": 0.0194,
"step": 15280
},
{
"grad_norm": 0.4764946401119232,
"learning_rate": 5.253973854958173e-05,
"loss": 0.0249,
"step": 15290
},
{
"grad_norm": 0.44023093581199646,
"learning_rate": 5.2484692505250376e-05,
"loss": 0.0211,
"step": 15300
},
{
"grad_norm": 0.4305736720561981,
"learning_rate": 5.242964344178333e-05,
"loss": 0.0275,
"step": 15310
},
{
"grad_norm": 0.37998518347740173,
"learning_rate": 5.237459142607041e-05,
"loss": 0.0151,
"step": 15320
},
{
"grad_norm": 0.5239267349243164,
"learning_rate": 5.231953652500498e-05,
"loss": 0.0165,
"step": 15330
},
{
"grad_norm": 0.4800921678543091,
"learning_rate": 5.2264478805483976e-05,
"loss": 0.0199,
"step": 15340
},
{
"grad_norm": 0.5239253640174866,
"learning_rate": 5.220941833440768e-05,
"loss": 0.0278,
"step": 15350
},
{
"grad_norm": 0.5484493970870972,
"learning_rate": 5.215435517867977e-05,
"loss": 0.0196,
"step": 15360
},
{
"grad_norm": 0.45815566182136536,
"learning_rate": 5.209928940520719e-05,
"loss": 0.0207,
"step": 15370
},
{
"grad_norm": 0.4087637662887573,
"learning_rate": 5.204422108090004e-05,
"loss": 0.0206,
"step": 15380
},
{
"grad_norm": 0.5465999245643616,
"learning_rate": 5.19891502726715e-05,
"loss": 0.0203,
"step": 15390
},
{
"grad_norm": 0.539454996585846,
"learning_rate": 5.193407704743782e-05,
"loss": 0.02,
"step": 15400
},
{
"grad_norm": 0.37997952103614807,
"learning_rate": 5.187900147211815e-05,
"loss": 0.0151,
"step": 15410
},
{
"grad_norm": 0.3879922926425934,
"learning_rate": 5.182392361363453e-05,
"loss": 0.0225,
"step": 15420
},
{
"grad_norm": 0.5663083791732788,
"learning_rate": 5.176884353891172e-05,
"loss": 0.0371,
"step": 15430
},
{
"grad_norm": 0.5053148865699768,
"learning_rate": 5.171376131487722e-05,
"loss": 0.02,
"step": 15440
},
{
"grad_norm": 0.538038969039917,
"learning_rate": 5.165867700846113e-05,
"loss": 0.0291,
"step": 15450
},
{
"grad_norm": 0.6184370517730713,
"learning_rate": 5.160359068659606e-05,
"loss": 0.0195,
"step": 15460
},
{
"grad_norm": 0.4717623293399811,
"learning_rate": 5.154850241621712e-05,
"loss": 0.0228,
"step": 15470
},
{
"grad_norm": 0.5252931118011475,
"learning_rate": 5.149341226426172e-05,
"loss": 0.0223,
"step": 15480
},
{
"grad_norm": 0.45526760816574097,
"learning_rate": 5.14383202976696e-05,
"loss": 0.0229,
"step": 15490
},
{
"grad_norm": 0.4614332914352417,
"learning_rate": 5.1383226583382684e-05,
"loss": 0.0195,
"step": 15500
},
{
"grad_norm": 0.5906754732131958,
"learning_rate": 5.132813118834504e-05,
"loss": 0.0237,
"step": 15510
},
{
"grad_norm": 0.40946319699287415,
"learning_rate": 5.1273034179502777e-05,
"loss": 0.014,
"step": 15520
},
{
"grad_norm": 0.4353945255279541,
"learning_rate": 5.1217935623803945e-05,
"loss": 0.0166,
"step": 15530
},
{
"grad_norm": 0.4610033929347992,
"learning_rate": 5.1162835588198476e-05,
"loss": 0.017,
"step": 15540
},
{
"grad_norm": 0.39268046617507935,
"learning_rate": 5.110773413963813e-05,
"loss": 0.016,
"step": 15550
},
{
"grad_norm": 0.47983095049858093,
"learning_rate": 5.105263134507636e-05,
"loss": 0.0203,
"step": 15560
},
{
"grad_norm": 0.554555356502533,
"learning_rate": 5.099752727146824e-05,
"loss": 0.0226,
"step": 15570
},
{
"grad_norm": 0.4779583215713501,
"learning_rate": 5.094242198577042e-05,
"loss": 0.0257,
"step": 15580
},
{
"grad_norm": 0.43633565306663513,
"learning_rate": 5.088731555494102e-05,
"loss": 0.016,
"step": 15590
},
{
"grad_norm": 0.347919762134552,
"learning_rate": 5.0832208045939556e-05,
"loss": 0.0166,
"step": 15600
},
{
"grad_norm": 0.4609246253967285,
"learning_rate": 5.0777099525726844e-05,
"loss": 0.0227,
"step": 15610
},
{
"grad_norm": 0.4452038109302521,
"learning_rate": 5.072199006126494e-05,
"loss": 0.0151,
"step": 15620
},
{
"grad_norm": 0.4674360752105713,
"learning_rate": 5.0666879719517026e-05,
"loss": 0.0175,
"step": 15630
},
{
"grad_norm": 0.4725985527038574,
"learning_rate": 5.061176856744737e-05,
"loss": 0.0174,
"step": 15640
},
{
"grad_norm": 0.4138374626636505,
"learning_rate": 5.0556656672021205e-05,
"loss": 0.0172,
"step": 15650
},
{
"grad_norm": 0.4620417356491089,
"learning_rate": 5.050154410020472e-05,
"loss": 0.0213,
"step": 15660
},
{
"grad_norm": 0.4797702133655548,
"learning_rate": 5.044643091896485e-05,
"loss": 0.0183,
"step": 15670
},
{
"grad_norm": 0.4944913387298584,
"learning_rate": 5.039131719526932e-05,
"loss": 0.0132,
"step": 15680
},
{
"grad_norm": 0.4205285906791687,
"learning_rate": 5.03362029960865e-05,
"loss": 0.0183,
"step": 15690
},
{
"grad_norm": 0.4850546419620514,
"learning_rate": 5.028108838838533e-05,
"loss": 0.023,
"step": 15700
},
{
"grad_norm": 0.45207902789115906,
"learning_rate": 5.0225973439135276e-05,
"loss": 0.0235,
"step": 15710
},
{
"grad_norm": 0.5000430345535278,
"learning_rate": 5.017085821530617e-05,
"loss": 0.0144,
"step": 15720
},
{
"grad_norm": 0.5730302929878235,
"learning_rate": 5.011574278386822e-05,
"loss": 0.0257,
"step": 15730
},
{
"grad_norm": 0.5539917945861816,
"learning_rate": 5.006062721179189e-05,
"loss": 0.0164,
"step": 15740
},
{
"grad_norm": 0.5315592885017395,
"learning_rate": 5.000551156604777e-05,
"loss": 0.0235,
"step": 15750
},
{
"grad_norm": 0.5321292877197266,
"learning_rate": 4.9950395913606594e-05,
"loss": 0.0202,
"step": 15760
},
{
"grad_norm": 0.5589334964752197,
"learning_rate": 4.989528032143903e-05,
"loss": 0.0206,
"step": 15770
},
{
"grad_norm": 0.40672385692596436,
"learning_rate": 4.984016485651578e-05,
"loss": 0.0177,
"step": 15780
},
{
"grad_norm": 0.5040760040283203,
"learning_rate": 4.9785049585807274e-05,
"loss": 0.0247,
"step": 15790
},
{
"grad_norm": 0.48457029461860657,
"learning_rate": 4.972993457628381e-05,
"loss": 0.0198,
"step": 15800
},
{
"grad_norm": 0.5615524053573608,
"learning_rate": 4.9674819894915306e-05,
"loss": 0.018,
"step": 15810
},
{
"grad_norm": 0.45237308740615845,
"learning_rate": 4.9619705608671265e-05,
"loss": 0.0185,
"step": 15820
},
{
"grad_norm": 0.5135700106620789,
"learning_rate": 4.956459178452079e-05,
"loss": 0.0173,
"step": 15830
},
{
"grad_norm": 0.5004363656044006,
"learning_rate": 4.950947848943235e-05,
"loss": 0.02,
"step": 15840
},
{
"grad_norm": 0.40345948934555054,
"learning_rate": 4.9454365790373805e-05,
"loss": 0.0191,
"step": 15850
},
{
"grad_norm": 0.44631311297416687,
"learning_rate": 4.939925375431226e-05,
"loss": 0.0187,
"step": 15860
},
{
"grad_norm": 0.5806166529655457,
"learning_rate": 4.9344142448214046e-05,
"loss": 0.023,
"step": 15870
},
{
"grad_norm": 0.5453324913978577,
"learning_rate": 4.928903193904461e-05,
"loss": 0.0235,
"step": 15880
},
{
"grad_norm": 0.5464229583740234,
"learning_rate": 4.923392229376842e-05,
"loss": 0.0191,
"step": 15890
},
{
"grad_norm": 0.41491982340812683,
"learning_rate": 4.9178813579348865e-05,
"loss": 0.014,
"step": 15900
},
{
"grad_norm": 0.47486695647239685,
"learning_rate": 4.9123705862748253e-05,
"loss": 0.0173,
"step": 15910
},
{
"grad_norm": 0.44494035840034485,
"learning_rate": 4.9068599210927627e-05,
"loss": 0.0197,
"step": 15920
},
{
"grad_norm": 0.4410898685455322,
"learning_rate": 4.901349369084681e-05,
"loss": 0.0198,
"step": 15930
},
{
"grad_norm": 0.514324963092804,
"learning_rate": 4.8958389369464156e-05,
"loss": 0.0157,
"step": 15940
},
{
"grad_norm": 0.4100477397441864,
"learning_rate": 4.890328631373666e-05,
"loss": 0.0148,
"step": 15950
},
{
"grad_norm": 0.3905644118785858,
"learning_rate": 4.88481845906197e-05,
"loss": 0.017,
"step": 15960
},
{
"grad_norm": 0.513152539730072,
"learning_rate": 4.879308426706707e-05,
"loss": 0.0204,
"step": 15970
},
{
"grad_norm": 0.4624914526939392,
"learning_rate": 4.873798541003084e-05,
"loss": 0.0151,
"step": 15980
},
{
"grad_norm": 0.5217762589454651,
"learning_rate": 4.868288808646136e-05,
"loss": 0.0207,
"step": 15990
},
{
"grad_norm": 0.4895175099372864,
"learning_rate": 4.862779236330705e-05,
"loss": 0.0181,
"step": 16000
},
{
"grad_norm": 0.4334188997745514,
"learning_rate": 4.8572698307514395e-05,
"loss": 0.0177,
"step": 16010
},
{
"grad_norm": 0.48886898159980774,
"learning_rate": 4.85176059860279e-05,
"loss": 0.0161,
"step": 16020
},
{
"grad_norm": 0.37678587436676025,
"learning_rate": 4.846251546578989e-05,
"loss": 0.02,
"step": 16030
},
{
"grad_norm": 0.4332294762134552,
"learning_rate": 4.840742681374058e-05,
"loss": 0.0168,
"step": 16040
},
{
"grad_norm": 0.5297786593437195,
"learning_rate": 4.835234009681787e-05,
"loss": 0.0175,
"step": 16050
},
{
"grad_norm": 0.45227107405662537,
"learning_rate": 4.8297255381957296e-05,
"loss": 0.0184,
"step": 16060
},
{
"grad_norm": 0.5056413412094116,
"learning_rate": 4.824217273609198e-05,
"loss": 0.0172,
"step": 16070
},
{
"grad_norm": 0.42688995599746704,
"learning_rate": 4.8187092226152555e-05,
"loss": 0.0189,
"step": 16080
},
{
"grad_norm": 0.4457613527774811,
"learning_rate": 4.8132013919067016e-05,
"loss": 0.0195,
"step": 16090
},
{
"grad_norm": 0.4380905330181122,
"learning_rate": 4.807693788176071e-05,
"loss": 0.021,
"step": 16100
},
{
"grad_norm": 0.4490067660808563,
"learning_rate": 4.8021864181156214e-05,
"loss": 0.0198,
"step": 16110
},
{
"grad_norm": 0.45158886909484863,
"learning_rate": 4.796679288417326e-05,
"loss": 0.0257,
"step": 16120
},
{
"grad_norm": 0.5157788395881653,
"learning_rate": 4.791172405772866e-05,
"loss": 0.0211,
"step": 16130
},
{
"grad_norm": 0.5067635178565979,
"learning_rate": 4.785665776873625e-05,
"loss": 0.0176,
"step": 16140
},
{
"grad_norm": 0.34866783022880554,
"learning_rate": 4.7801594084106763e-05,
"loss": 0.0227,
"step": 16150
},
{
"grad_norm": 0.44572070240974426,
"learning_rate": 4.774653307074775e-05,
"loss": 0.0224,
"step": 16160
},
{
"grad_norm": 0.5130992531776428,
"learning_rate": 4.769147479556355e-05,
"loss": 0.0171,
"step": 16170
},
{
"grad_norm": 0.41665777564048767,
"learning_rate": 4.763641932545515e-05,
"loss": 0.0159,
"step": 16180
},
{
"grad_norm": 0.4755811393260956,
"learning_rate": 4.758136672732013e-05,
"loss": 0.0179,
"step": 16190
},
{
"grad_norm": 0.472990483045578,
"learning_rate": 4.752631706805261e-05,
"loss": 0.018,
"step": 16200
},
{
"grad_norm": 0.538840651512146,
"learning_rate": 4.7471270414543114e-05,
"loss": 0.0256,
"step": 16210
},
{
"grad_norm": 0.37763509154319763,
"learning_rate": 4.741622683367849e-05,
"loss": 0.0153,
"step": 16220
},
{
"grad_norm": 0.45274749398231506,
"learning_rate": 4.736118639234191e-05,
"loss": 0.0289,
"step": 16230
},
{
"grad_norm": 0.4707541763782501,
"learning_rate": 4.730614915741267e-05,
"loss": 0.0237,
"step": 16240
},
{
"grad_norm": 0.5046401619911194,
"learning_rate": 4.7251115195766234e-05,
"loss": 0.0182,
"step": 16250
},
{
"grad_norm": 0.5368759036064148,
"learning_rate": 4.719608457427404e-05,
"loss": 0.0213,
"step": 16260
},
{
"grad_norm": 0.3866722881793976,
"learning_rate": 4.714105735980347e-05,
"loss": 0.0173,
"step": 16270
},
{
"grad_norm": 0.49749884009361267,
"learning_rate": 4.70860336192178e-05,
"loss": 0.023,
"step": 16280
},
{
"grad_norm": 0.28592053055763245,
"learning_rate": 4.7031013419376034e-05,
"loss": 0.0175,
"step": 16290
},
{
"grad_norm": 0.4941922724246979,
"learning_rate": 4.6975996827132926e-05,
"loss": 0.0139,
"step": 16300
},
{
"grad_norm": 0.43765896558761597,
"learning_rate": 4.692098390933883e-05,
"loss": 0.0224,
"step": 16310
},
{
"grad_norm": 0.5004605054855347,
"learning_rate": 4.6865974732839615e-05,
"loss": 0.0154,
"step": 16320
},
{
"grad_norm": 0.3991715908050537,
"learning_rate": 4.6810969364476615e-05,
"loss": 0.0226,
"step": 16330
},
{
"grad_norm": 0.5033056139945984,
"learning_rate": 4.675596787108653e-05,
"loss": 0.0149,
"step": 16340
},
{
"grad_norm": 0.4356118440628052,
"learning_rate": 4.670097031950139e-05,
"loss": 0.0127,
"step": 16350
},
{
"grad_norm": 0.48664993047714233,
"learning_rate": 4.664597677654838e-05,
"loss": 0.0193,
"step": 16360
},
{
"grad_norm": 0.5403888821601868,
"learning_rate": 4.659098730904986e-05,
"loss": 0.0246,
"step": 16370
},
{
"grad_norm": 0.4761137366294861,
"learning_rate": 4.6536001983823204e-05,
"loss": 0.0178,
"step": 16380
},
{
"grad_norm": 0.564595103263855,
"learning_rate": 4.648102086768078e-05,
"loss": 0.0237,
"step": 16390
},
{
"grad_norm": 0.4206317961215973,
"learning_rate": 4.6426044027429786e-05,
"loss": 0.0184,
"step": 16400
},
{
"grad_norm": 0.4059642255306244,
"learning_rate": 4.637107152987234e-05,
"loss": 0.0204,
"step": 16410
},
{
"grad_norm": 0.4529334008693695,
"learning_rate": 4.631610344180515e-05,
"loss": 0.0189,
"step": 16420
},
{
"grad_norm": 0.5059530138969421,
"learning_rate": 4.626113983001965e-05,
"loss": 0.023,
"step": 16430
},
{
"grad_norm": 0.4450704753398895,
"learning_rate": 4.620618076130182e-05,
"loss": 0.0198,
"step": 16440
},
{
"grad_norm": 0.47567087411880493,
"learning_rate": 4.6151226302432074e-05,
"loss": 0.0155,
"step": 16450
},
{
"grad_norm": 0.42478153109550476,
"learning_rate": 4.6096276520185304e-05,
"loss": 0.0181,
"step": 16460
},
{
"grad_norm": 0.37771472334861755,
"learning_rate": 4.604133148133066e-05,
"loss": 0.0154,
"step": 16470
},
{
"grad_norm": 0.45491811633110046,
"learning_rate": 4.598639125263155e-05,
"loss": 0.024,
"step": 16480
},
{
"grad_norm": 0.5239196419715881,
"learning_rate": 4.593145590084553e-05,
"loss": 0.023,
"step": 16490
},
{
"grad_norm": 0.44439220428466797,
"learning_rate": 4.58765254927242e-05,
"loss": 0.0149,
"step": 16500
},
{
"grad_norm": 0.46345189213752747,
"learning_rate": 4.582160009501323e-05,
"loss": 0.0158,
"step": 16510
},
{
"grad_norm": 0.38452160358428955,
"learning_rate": 4.5766679774452143e-05,
"loss": 0.0186,
"step": 16520
},
{
"grad_norm": 0.5093140602111816,
"learning_rate": 4.571176459777431e-05,
"loss": 0.0205,
"step": 16530
},
{
"grad_norm": 0.5206828117370605,
"learning_rate": 4.5656854631706846e-05,
"loss": 0.0166,
"step": 16540
},
{
"grad_norm": 0.5434250235557556,
"learning_rate": 4.560194994297054e-05,
"loss": 0.0204,
"step": 16550
},
{
"grad_norm": 0.480487197637558,
"learning_rate": 4.554705059827974e-05,
"loss": 0.0239,
"step": 16560
},
{
"grad_norm": 0.3538348078727722,
"learning_rate": 4.5492156664342365e-05,
"loss": 0.0189,
"step": 16570
},
{
"grad_norm": 0.42979592084884644,
"learning_rate": 4.5437268207859695e-05,
"loss": 0.0254,
"step": 16580
},
{
"grad_norm": 0.42062249779701233,
"learning_rate": 4.538238529552641e-05,
"loss": 0.0143,
"step": 16590
},
{
"grad_norm": 0.30756622552871704,
"learning_rate": 4.5327507994030394e-05,
"loss": 0.0127,
"step": 16600
},
{
"grad_norm": 0.42589271068573,
"learning_rate": 4.527263637005273e-05,
"loss": 0.0132,
"step": 16610
},
{
"grad_norm": 0.4414200484752655,
"learning_rate": 4.521777049026766e-05,
"loss": 0.0129,
"step": 16620
},
{
"grad_norm": 0.5160260796546936,
"learning_rate": 4.5162910421342386e-05,
"loss": 0.0156,
"step": 16630
},
{
"grad_norm": 0.5050916075706482,
"learning_rate": 4.510805622993706e-05,
"loss": 0.0229,
"step": 16640
},
{
"grad_norm": 0.4574771821498871,
"learning_rate": 4.505320798270467e-05,
"loss": 0.0214,
"step": 16650
},
{
"grad_norm": 0.4224357604980469,
"learning_rate": 4.499836574629105e-05,
"loss": 0.015,
"step": 16660
},
{
"grad_norm": 0.4318616986274719,
"learning_rate": 4.494352958733465e-05,
"loss": 0.0328,
"step": 16670
},
{
"grad_norm": 0.49073171615600586,
"learning_rate": 4.488869957246663e-05,
"loss": 0.0151,
"step": 16680
},
{
"grad_norm": 0.5508942008018494,
"learning_rate": 4.4833875768310575e-05,
"loss": 0.0185,
"step": 16690
},
{
"grad_norm": 0.4843858480453491,
"learning_rate": 4.47790582414826e-05,
"loss": 0.0184,
"step": 16700
},
{
"grad_norm": 0.49556872248649597,
"learning_rate": 4.472424705859114e-05,
"loss": 0.0212,
"step": 16710
},
{
"grad_norm": 0.47751182317733765,
"learning_rate": 4.466944228623701e-05,
"loss": 0.0335,
"step": 16720
},
{
"grad_norm": 0.5937556624412537,
"learning_rate": 4.461464399101312e-05,
"loss": 0.0271,
"step": 16730
},
{
"grad_norm": 0.4294058680534363,
"learning_rate": 4.45598522395046e-05,
"loss": 0.018,
"step": 16740
},
{
"grad_norm": 0.4545292258262634,
"learning_rate": 4.4505067098288577e-05,
"loss": 0.0242,
"step": 16750
},
{
"grad_norm": 0.5502038598060608,
"learning_rate": 4.445028863393417e-05,
"loss": 0.0181,
"step": 16760
},
{
"grad_norm": 0.48465967178344727,
"learning_rate": 4.439551691300236e-05,
"loss": 0.016,
"step": 16770
},
{
"grad_norm": 0.4531119167804718,
"learning_rate": 4.434075200204599e-05,
"loss": 0.0182,
"step": 16780
},
{
"grad_norm": 0.29305294156074524,
"learning_rate": 4.428599396760956e-05,
"loss": 0.013,
"step": 16790
},
{
"grad_norm": 0.4283970296382904,
"learning_rate": 4.423124287622926e-05,
"loss": 0.016,
"step": 16800
},
{
"grad_norm": 0.430448055267334,
"learning_rate": 4.417649879443282e-05,
"loss": 0.0128,
"step": 16810
},
{
"grad_norm": 0.44931647181510925,
"learning_rate": 4.4121761788739444e-05,
"loss": 0.0164,
"step": 16820
},
{
"grad_norm": 0.3733663558959961,
"learning_rate": 4.4067031925659805e-05,
"loss": 0.017,
"step": 16830
},
{
"grad_norm": 0.400789737701416,
"learning_rate": 4.4012309271695816e-05,
"loss": 0.0144,
"step": 16840
},
{
"grad_norm": 0.4875842332839966,
"learning_rate": 4.395759389334067e-05,
"loss": 0.0151,
"step": 16850
},
{
"grad_norm": 0.4128034710884094,
"learning_rate": 4.3902885857078685e-05,
"loss": 0.0218,
"step": 16860
},
{
"grad_norm": 0.41802412271499634,
"learning_rate": 4.384818522938531e-05,
"loss": 0.013,
"step": 16870
},
{
"grad_norm": 0.6621541380882263,
"learning_rate": 4.379349207672695e-05,
"loss": 0.0157,
"step": 16880
},
{
"grad_norm": 0.4162161350250244,
"learning_rate": 4.373880646556098e-05,
"loss": 0.0187,
"step": 16890
},
{
"grad_norm": 0.4720764756202698,
"learning_rate": 4.3684128462335533e-05,
"loss": 0.0163,
"step": 16900
},
{
"grad_norm": 0.5169637799263,
"learning_rate": 4.362945813348955e-05,
"loss": 0.0153,
"step": 16910
},
{
"grad_norm": 0.4132520258426666,
"learning_rate": 4.357479554545262e-05,
"loss": 0.0176,
"step": 16920
},
{
"grad_norm": 0.4332652688026428,
"learning_rate": 4.352014076464499e-05,
"loss": 0.0202,
"step": 16930
},
{
"grad_norm": 0.46775418519973755,
"learning_rate": 4.346549385747734e-05,
"loss": 0.0167,
"step": 16940
},
{
"grad_norm": 0.511955976486206,
"learning_rate": 4.34108548903508e-05,
"loss": 0.0243,
"step": 16950
},
{
"grad_norm": 0.584647536277771,
"learning_rate": 4.335622392965689e-05,
"loss": 0.0206,
"step": 16960
},
{
"grad_norm": 0.44684383273124695,
"learning_rate": 4.330160104177738e-05,
"loss": 0.0194,
"step": 16970
},
{
"grad_norm": 0.36751842498779297,
"learning_rate": 4.32469862930842e-05,
"loss": 0.0182,
"step": 16980
},
{
"grad_norm": 0.5229809880256653,
"learning_rate": 4.319237974993946e-05,
"loss": 0.0223,
"step": 16990
},
{
"grad_norm": 0.6980265378952026,
"learning_rate": 4.3137781478695236e-05,
"loss": 0.0195,
"step": 17000
},
{
"grad_norm": 0.4035058915615082,
"learning_rate": 4.308319154569358e-05,
"loss": 0.0299,
"step": 17010
},
{
"grad_norm": 0.6207200288772583,
"learning_rate": 4.302861001726641e-05,
"loss": 0.0248,
"step": 17020
},
{
"grad_norm": 0.444389283657074,
"learning_rate": 4.2974036959735415e-05,
"loss": 0.0154,
"step": 17030
},
{
"grad_norm": 0.4511665999889374,
"learning_rate": 4.2919472439412034e-05,
"loss": 0.0127,
"step": 17040
},
{
"grad_norm": 0.5390390157699585,
"learning_rate": 4.286491652259729e-05,
"loss": 0.0145,
"step": 17050
},
{
"grad_norm": 0.625353217124939,
"learning_rate": 4.281036927558178e-05,
"loss": 0.0276,
"step": 17060
},
{
"grad_norm": 0.4137900769710541,
"learning_rate": 4.275583076464552e-05,
"loss": 0.0156,
"step": 17070
},
{
"grad_norm": 0.49314960837364197,
"learning_rate": 4.270130105605794e-05,
"loss": 0.0148,
"step": 17080
},
{
"grad_norm": 0.44711920619010925,
"learning_rate": 4.264678021607782e-05,
"loss": 0.0193,
"step": 17090
},
{
"grad_norm": 0.3853570520877838,
"learning_rate": 4.25922683109531e-05,
"loss": 0.0142,
"step": 17100
},
{
"grad_norm": 0.5626659393310547,
"learning_rate": 4.25377654069209e-05,
"loss": 0.0182,
"step": 17110
},
{
"grad_norm": 0.5244747996330261,
"learning_rate": 4.248327157020737e-05,
"loss": 0.0153,
"step": 17120
},
{
"grad_norm": 0.5560617446899414,
"learning_rate": 4.242878686702763e-05,
"loss": 0.0146,
"step": 17130
},
{
"grad_norm": 0.4835559129714966,
"learning_rate": 4.23743113635858e-05,
"loss": 0.0122,
"step": 17140
},
{
"grad_norm": 0.43113401532173157,
"learning_rate": 4.2319845126074705e-05,
"loss": 0.0196,
"step": 17150
},
{
"grad_norm": 0.38143399357795715,
"learning_rate": 4.226538822067598e-05,
"loss": 0.0188,
"step": 17160
},
{
"grad_norm": 0.4144134223461151,
"learning_rate": 4.2210940713559896e-05,
"loss": 0.0174,
"step": 17170
},
{
"grad_norm": 0.49670249223709106,
"learning_rate": 4.21565026708853e-05,
"loss": 0.02,
"step": 17180
},
{
"grad_norm": 0.5124985575675964,
"learning_rate": 4.2102074158799535e-05,
"loss": 0.014,
"step": 17190
},
{
"grad_norm": 0.4831800162792206,
"learning_rate": 4.204765524343841e-05,
"loss": 0.0197,
"step": 17200
},
{
"grad_norm": 0.5432153344154358,
"learning_rate": 4.1993245990926036e-05,
"loss": 0.0174,
"step": 17210
},
{
"grad_norm": 0.40717804431915283,
"learning_rate": 4.193884646737474e-05,
"loss": 0.0174,
"step": 17220
},
{
"grad_norm": 0.46120816469192505,
"learning_rate": 4.1884456738885124e-05,
"loss": 0.0153,
"step": 17230
},
{
"grad_norm": 0.5277372598648071,
"learning_rate": 4.1830076871545795e-05,
"loss": 0.015,
"step": 17240
},
{
"grad_norm": 0.48084837198257446,
"learning_rate": 4.177570693143346e-05,
"loss": 0.0296,
"step": 17250
},
{
"grad_norm": 0.4313606321811676,
"learning_rate": 4.172134698461271e-05,
"loss": 0.0136,
"step": 17260
},
{
"grad_norm": 0.37397533655166626,
"learning_rate": 4.166699709713599e-05,
"loss": 0.0139,
"step": 17270
},
{
"grad_norm": 0.3367810547351837,
"learning_rate": 4.1612657335043555e-05,
"loss": 0.0123,
"step": 17280
},
{
"grad_norm": 0.4260478913784027,
"learning_rate": 4.155832776436331e-05,
"loss": 0.0174,
"step": 17290
},
{
"grad_norm": 0.3650907576084137,
"learning_rate": 4.1504008451110845e-05,
"loss": 0.0143,
"step": 17300
},
{
"grad_norm": 0.47170013189315796,
"learning_rate": 4.1449699461289225e-05,
"loss": 0.016,
"step": 17310
},
{
"grad_norm": 0.4058210849761963,
"learning_rate": 4.139540086088901e-05,
"loss": 0.0112,
"step": 17320
},
{
"grad_norm": 0.4133184254169464,
"learning_rate": 4.1341112715888106e-05,
"loss": 0.0126,
"step": 17330
},
{
"grad_norm": 0.4448685944080353,
"learning_rate": 4.128683509225172e-05,
"loss": 0.0174,
"step": 17340
},
{
"grad_norm": 0.39847812056541443,
"learning_rate": 4.123256805593231e-05,
"loss": 0.0136,
"step": 17350
},
{
"grad_norm": 0.4507867097854614,
"learning_rate": 4.117831167286943e-05,
"loss": 0.0209,
"step": 17360
},
{
"grad_norm": 0.5129208564758301,
"learning_rate": 4.1124066008989685e-05,
"loss": 0.0171,
"step": 17370
},
{
"grad_norm": 0.48708876967430115,
"learning_rate": 4.106983113020669e-05,
"loss": 0.0154,
"step": 17380
},
{
"grad_norm": 0.40911203622817993,
"learning_rate": 4.101560710242094e-05,
"loss": 0.0142,
"step": 17390
},
{
"grad_norm": 0.4711911976337433,
"learning_rate": 4.096139399151971e-05,
"loss": 0.0183,
"step": 17400
},
{
"grad_norm": 0.41986173391342163,
"learning_rate": 4.090719186337709e-05,
"loss": 0.0124,
"step": 17410
},
{
"grad_norm": 0.4648745357990265,
"learning_rate": 4.0853000783853746e-05,
"loss": 0.0181,
"step": 17420
},
{
"grad_norm": 0.42445722222328186,
"learning_rate": 4.0798820818796956e-05,
"loss": 0.0129,
"step": 17430
},
{
"grad_norm": 0.6091474294662476,
"learning_rate": 4.074465203404048e-05,
"loss": 0.0205,
"step": 17440
},
{
"grad_norm": 0.5341174006462097,
"learning_rate": 4.0690494495404495e-05,
"loss": 0.0193,
"step": 17450
},
{
"grad_norm": 0.5151488184928894,
"learning_rate": 4.0636348268695535e-05,
"loss": 0.0149,
"step": 17460
},
{
"grad_norm": 0.36311253905296326,
"learning_rate": 4.058221341970637e-05,
"loss": 0.013,
"step": 17470
},
{
"grad_norm": 0.4540540874004364,
"learning_rate": 4.0528090014215945e-05,
"loss": 0.0172,
"step": 17480
},
{
"grad_norm": 0.3352636396884918,
"learning_rate": 4.047397811798929e-05,
"loss": 0.0128,
"step": 17490
},
{
"grad_norm": 0.44019952416419983,
"learning_rate": 4.041987779677745e-05,
"loss": 0.0139,
"step": 17500
},
{
"grad_norm": 0.5217735767364502,
"learning_rate": 4.036578911631745e-05,
"loss": 0.0175,
"step": 17510
},
{
"grad_norm": 0.3993825316429138,
"learning_rate": 4.031171214233211e-05,
"loss": 0.016,
"step": 17520
},
{
"grad_norm": 0.43255704641342163,
"learning_rate": 4.025764694053008e-05,
"loss": 0.0153,
"step": 17530
},
{
"grad_norm": 0.4293983578681946,
"learning_rate": 4.020359357660566e-05,
"loss": 0.022,
"step": 17540
},
{
"grad_norm": 0.4519442021846771,
"learning_rate": 4.014955211623875e-05,
"loss": 0.0138,
"step": 17550
},
{
"grad_norm": 0.43478626012802124,
"learning_rate": 4.0095522625094895e-05,
"loss": 0.03,
"step": 17560
},
{
"grad_norm": 0.5119835138320923,
"learning_rate": 4.004150516882497e-05,
"loss": 0.021,
"step": 17570
},
{
"grad_norm": 0.5171064138412476,
"learning_rate": 3.99874998130653e-05,
"loss": 0.0157,
"step": 17580
},
{
"grad_norm": 0.3860531151294708,
"learning_rate": 3.9933506623437455e-05,
"loss": 0.0207,
"step": 17590
},
{
"grad_norm": 0.4299333989620209,
"learning_rate": 3.9879525665548276e-05,
"loss": 0.0242,
"step": 17600
},
{
"grad_norm": 0.5709996223449707,
"learning_rate": 3.9825557004989704e-05,
"loss": 0.0221,
"step": 17610
},
{
"grad_norm": 0.4006538689136505,
"learning_rate": 3.977160070733877e-05,
"loss": 0.0153,
"step": 17620
},
{
"grad_norm": 0.4045625925064087,
"learning_rate": 3.971765683815746e-05,
"loss": 0.012,
"step": 17630
},
{
"grad_norm": 0.3872022330760956,
"learning_rate": 3.966372546299265e-05,
"loss": 0.017,
"step": 17640
},
{
"grad_norm": 0.3713976740837097,
"learning_rate": 3.960980664737604e-05,
"loss": 0.0151,
"step": 17650
},
{
"grad_norm": 0.48039600253105164,
"learning_rate": 3.955590045682408e-05,
"loss": 0.0265,
"step": 17660
},
{
"grad_norm": 0.4872661530971527,
"learning_rate": 3.950200695683788e-05,
"loss": 0.0176,
"step": 17670
},
{
"grad_norm": 0.5464871525764465,
"learning_rate": 3.944812621290314e-05,
"loss": 0.0178,
"step": 17680
},
{
"grad_norm": 0.3874129056930542,
"learning_rate": 3.939425829049002e-05,
"loss": 0.0187,
"step": 17690
},
{
"grad_norm": 0.42265158891677856,
"learning_rate": 3.9340403255053135e-05,
"loss": 0.0224,
"step": 17700
},
{
"grad_norm": 0.4619930386543274,
"learning_rate": 3.9286561172031407e-05,
"loss": 0.018,
"step": 17710
},
{
"grad_norm": 0.46306854486465454,
"learning_rate": 3.923273210684809e-05,
"loss": 0.0161,
"step": 17720
},
{
"grad_norm": 0.3531588017940521,
"learning_rate": 3.9178916124910555e-05,
"loss": 0.0123,
"step": 17730
},
{
"grad_norm": 0.4237513840198517,
"learning_rate": 3.9125113291610274e-05,
"loss": 0.0235,
"step": 17740
},
{
"grad_norm": 0.4847213923931122,
"learning_rate": 3.907132367232278e-05,
"loss": 0.0309,
"step": 17750
},
{
"grad_norm": 0.44765254855155945,
"learning_rate": 3.901754733240752e-05,
"loss": 0.0208,
"step": 17760
},
{
"grad_norm": 0.4562167227268219,
"learning_rate": 3.896378433720785e-05,
"loss": 0.0136,
"step": 17770
},
{
"grad_norm": 0.4212327301502228,
"learning_rate": 3.891003475205086e-05,
"loss": 0.0279,
"step": 17780
},
{
"grad_norm": 0.3467431366443634,
"learning_rate": 3.885629864224736e-05,
"loss": 0.0086,
"step": 17790
},
{
"grad_norm": 0.4508669674396515,
"learning_rate": 3.880257607309178e-05,
"loss": 0.0135,
"step": 17800
},
{
"grad_norm": 0.5035721659660339,
"learning_rate": 3.874886710986213e-05,
"loss": 0.0189,
"step": 17810
},
{
"grad_norm": 0.45131590962409973,
"learning_rate": 3.8695171817819834e-05,
"loss": 0.0194,
"step": 17820
},
{
"grad_norm": 0.4182703197002411,
"learning_rate": 3.8641490262209774e-05,
"loss": 0.0149,
"step": 17830
},
{
"grad_norm": 0.5047512054443359,
"learning_rate": 3.8587822508260084e-05,
"loss": 0.0158,
"step": 17840
},
{
"grad_norm": 0.46264535188674927,
"learning_rate": 3.853416862118214e-05,
"loss": 0.0164,
"step": 17850
},
{
"grad_norm": 0.4324040710926056,
"learning_rate": 3.848052866617049e-05,
"loss": 0.015,
"step": 17860
},
{
"grad_norm": 0.3868952691555023,
"learning_rate": 3.84269027084027e-05,
"loss": 0.0156,
"step": 17870
},
{
"grad_norm": 0.4729918837547302,
"learning_rate": 3.83732908130394e-05,
"loss": 0.0153,
"step": 17880
},
{
"grad_norm": 0.4984741806983948,
"learning_rate": 3.831969304522409e-05,
"loss": 0.0155,
"step": 17890
},
{
"grad_norm": 0.5425384044647217,
"learning_rate": 3.826610947008312e-05,
"loss": 0.014,
"step": 17900
},
{
"grad_norm": 0.47965776920318604,
"learning_rate": 3.821254015272559e-05,
"loss": 0.0172,
"step": 17910
},
{
"grad_norm": 0.3790384531021118,
"learning_rate": 3.8158985158243214e-05,
"loss": 0.0148,
"step": 17920
},
{
"grad_norm": 0.415213018655777,
"learning_rate": 3.810544455171044e-05,
"loss": 0.0153,
"step": 17930
},
{
"grad_norm": 0.3903871476650238,
"learning_rate": 3.8051918398184114e-05,
"loss": 0.0143,
"step": 17940
},
{
"grad_norm": 0.4505022466182709,
"learning_rate": 3.799840676270356e-05,
"loss": 0.0207,
"step": 17950
},
{
"grad_norm": 0.4937152862548828,
"learning_rate": 3.794490971029048e-05,
"loss": 0.0202,
"step": 17960
},
{
"grad_norm": 0.380430668592453,
"learning_rate": 3.789142730594881e-05,
"loss": 0.0134,
"step": 17970
},
{
"grad_norm": 0.45218154788017273,
"learning_rate": 3.783795961466471e-05,
"loss": 0.0245,
"step": 17980
},
{
"grad_norm": 0.4809904098510742,
"learning_rate": 3.778450670140651e-05,
"loss": 0.0126,
"step": 17990
},
{
"grad_norm": 0.37509477138519287,
"learning_rate": 3.773106863112451e-05,
"loss": 0.0177,
"step": 18000
},
{
"grad_norm": 0.35761046409606934,
"learning_rate": 3.7677645468751e-05,
"loss": 0.025,
"step": 18010
},
{
"grad_norm": 0.4375065863132477,
"learning_rate": 3.762423727920018e-05,
"loss": 0.0165,
"step": 18020
},
{
"grad_norm": 0.3524833917617798,
"learning_rate": 3.7570844127367995e-05,
"loss": 0.0219,
"step": 18030
},
{
"grad_norm": 0.47392725944519043,
"learning_rate": 3.751746607813221e-05,
"loss": 0.0189,
"step": 18040
},
{
"grad_norm": 0.44055894017219543,
"learning_rate": 3.746410319635217e-05,
"loss": 0.0175,
"step": 18050
},
{
"grad_norm": 0.48054662346839905,
"learning_rate": 3.7410755546868806e-05,
"loss": 0.0133,
"step": 18060
},
{
"grad_norm": 0.398880273103714,
"learning_rate": 3.735742319450454e-05,
"loss": 0.017,
"step": 18070
},
{
"grad_norm": 0.3487880527973175,
"learning_rate": 3.7304106204063184e-05,
"loss": 0.0139,
"step": 18080
},
{
"grad_norm": 0.47175294160842896,
"learning_rate": 3.725080464032996e-05,
"loss": 0.015,
"step": 18090
},
{
"grad_norm": 0.42627474665641785,
"learning_rate": 3.719751856807125e-05,
"loss": 0.0197,
"step": 18100
},
{
"grad_norm": 0.4918067753314972,
"learning_rate": 3.71442480520347e-05,
"loss": 0.0137,
"step": 18110
},
{
"grad_norm": 0.40713074803352356,
"learning_rate": 3.709099315694897e-05,
"loss": 0.0154,
"step": 18120
},
{
"grad_norm": 0.4796452522277832,
"learning_rate": 3.703775394752378e-05,
"loss": 0.0168,
"step": 18130
},
{
"grad_norm": 0.3878817558288574,
"learning_rate": 3.698453048844983e-05,
"loss": 0.0112,
"step": 18140
},
{
"grad_norm": 0.6022557020187378,
"learning_rate": 3.693132284439861e-05,
"loss": 0.0154,
"step": 18150
},
{
"grad_norm": 0.3724147081375122,
"learning_rate": 3.6878131080022414e-05,
"loss": 0.0155,
"step": 18160
},
{
"grad_norm": 0.4457415044307709,
"learning_rate": 3.682495525995429e-05,
"loss": 0.0155,
"step": 18170
},
{
"grad_norm": 0.41124987602233887,
"learning_rate": 3.6771795448807846e-05,
"loss": 0.0121,
"step": 18180
},
{
"grad_norm": 0.35011178255081177,
"learning_rate": 3.671865171117724e-05,
"loss": 0.0147,
"step": 18190
},
{
"grad_norm": 0.41456279158592224,
"learning_rate": 3.666552411163718e-05,
"loss": 0.0106,
"step": 18200
},
{
"grad_norm": 0.3883676826953888,
"learning_rate": 3.661241271474269e-05,
"loss": 0.014,
"step": 18210
},
{
"grad_norm": 0.3940272331237793,
"learning_rate": 3.6559317585029116e-05,
"loss": 0.012,
"step": 18220
},
{
"grad_norm": 0.47107991576194763,
"learning_rate": 3.650623878701204e-05,
"loss": 0.0142,
"step": 18230
},
{
"grad_norm": 0.5025423169136047,
"learning_rate": 3.645317638518721e-05,
"loss": 0.0148,
"step": 18240
},
{
"grad_norm": 0.48330673575401306,
"learning_rate": 3.640013044403046e-05,
"loss": 0.0197,
"step": 18250
},
{
"grad_norm": 0.38045749068260193,
"learning_rate": 3.634710102799761e-05,
"loss": 0.0129,
"step": 18260
},
{
"grad_norm": 0.44960033893585205,
"learning_rate": 3.6294088201524394e-05,
"loss": 0.0172,
"step": 18270
},
{
"grad_norm": 0.4522477686405182,
"learning_rate": 3.62410920290264e-05,
"loss": 0.0119,
"step": 18280
},
{
"grad_norm": 0.4287990629673004,
"learning_rate": 3.6188112574898954e-05,
"loss": 0.0154,
"step": 18290
},
{
"grad_norm": 0.42192602157592773,
"learning_rate": 3.6135149903517115e-05,
"loss": 0.014,
"step": 18300
},
{
"grad_norm": 0.31292644143104553,
"learning_rate": 3.608220407923552e-05,
"loss": 0.0159,
"step": 18310
},
{
"grad_norm": 0.44275712966918945,
"learning_rate": 3.602927516638833e-05,
"loss": 0.0155,
"step": 18320
},
{
"grad_norm": 0.40655505657196045,
"learning_rate": 3.5976363229289165e-05,
"loss": 0.0153,
"step": 18330
},
{
"grad_norm": 0.5514722466468811,
"learning_rate": 3.5923468332231e-05,
"loss": 0.0162,
"step": 18340
},
{
"grad_norm": 0.39454907178878784,
"learning_rate": 3.587059053948616e-05,
"loss": 0.0142,
"step": 18350
},
{
"grad_norm": 0.4014619290828705,
"learning_rate": 3.5817729915306134e-05,
"loss": 0.0131,
"step": 18360
},
{
"grad_norm": 0.42949196696281433,
"learning_rate": 3.5764886523921566e-05,
"loss": 0.0129,
"step": 18370
},
{
"grad_norm": 0.5265833139419556,
"learning_rate": 3.571206042954214e-05,
"loss": 0.0135,
"step": 18380
},
{
"grad_norm": 0.3725685179233551,
"learning_rate": 3.5659251696356566e-05,
"loss": 0.014,
"step": 18390
},
{
"grad_norm": 0.4337877035140991,
"learning_rate": 3.56064603885324e-05,
"loss": 0.0231,
"step": 18400
},
{
"grad_norm": 0.5028528571128845,
"learning_rate": 3.5553686570216115e-05,
"loss": 0.0203,
"step": 18410
},
{
"grad_norm": 0.39936956763267517,
"learning_rate": 3.550093030553284e-05,
"loss": 0.0194,
"step": 18420
},
{
"grad_norm": 0.4264374375343323,
"learning_rate": 3.544819165858642e-05,
"loss": 0.0153,
"step": 18430
},
{
"grad_norm": 0.42669013142585754,
"learning_rate": 3.539547069345926e-05,
"loss": 0.0126,
"step": 18440
},
{
"grad_norm": 0.46507567167282104,
"learning_rate": 3.5342767474212344e-05,
"loss": 0.0219,
"step": 18450
},
{
"grad_norm": 0.485779345035553,
"learning_rate": 3.529008206488502e-05,
"loss": 0.0159,
"step": 18460
},
{
"grad_norm": 0.46321389079093933,
"learning_rate": 3.5237414529495055e-05,
"loss": 0.0235,
"step": 18470
},
{
"grad_norm": 0.37275978922843933,
"learning_rate": 3.5184764932038454e-05,
"loss": 0.0139,
"step": 18480
},
{
"grad_norm": 0.4192523658275604,
"learning_rate": 3.513213333648945e-05,
"loss": 0.012,
"step": 18490
},
{
"grad_norm": 0.34555143117904663,
"learning_rate": 3.5079519806800374e-05,
"loss": 0.0229,
"step": 18500
},
{
"grad_norm": 0.36041679978370667,
"learning_rate": 3.502692440690165e-05,
"loss": 0.0189,
"step": 18510
},
{
"grad_norm": 0.4773518741130829,
"learning_rate": 3.497434720070165e-05,
"loss": 0.0137,
"step": 18520
},
{
"grad_norm": 0.4039384722709656,
"learning_rate": 3.4921788252086616e-05,
"loss": 0.0161,
"step": 18530
},
{
"grad_norm": 0.4013559818267822,
"learning_rate": 3.486924762492065e-05,
"loss": 0.0105,
"step": 18540
},
{
"grad_norm": 0.3607863485813141,
"learning_rate": 3.4816725383045534e-05,
"loss": 0.015,
"step": 18550
},
{
"grad_norm": 0.3584037721157074,
"learning_rate": 3.476422159028079e-05,
"loss": 0.0273,
"step": 18560
},
{
"grad_norm": 0.44395920634269714,
"learning_rate": 3.471173631042345e-05,
"loss": 0.0162,
"step": 18570
},
{
"grad_norm": 0.472136914730072,
"learning_rate": 3.465926960724808e-05,
"loss": 0.0127,
"step": 18580
},
{
"grad_norm": 0.5398361086845398,
"learning_rate": 3.460682154450666e-05,
"loss": 0.0151,
"step": 18590
},
{
"grad_norm": 0.47547459602355957,
"learning_rate": 3.4554392185928564e-05,
"loss": 0.0254,
"step": 18600
},
{
"grad_norm": 0.5101744532585144,
"learning_rate": 3.450198159522037e-05,
"loss": 0.0132,
"step": 18610
},
{
"grad_norm": 0.38037049770355225,
"learning_rate": 3.444958983606592e-05,
"loss": 0.025,
"step": 18620
},
{
"grad_norm": 0.5962969660758972,
"learning_rate": 3.439721697212612e-05,
"loss": 0.0375,
"step": 18630
},
{
"grad_norm": 0.564574122428894,
"learning_rate": 3.4344863067038954e-05,
"loss": 0.0174,
"step": 18640
},
{
"grad_norm": 0.41285887360572815,
"learning_rate": 3.429252818441935e-05,
"loss": 0.0149,
"step": 18650
},
{
"grad_norm": 0.5477291941642761,
"learning_rate": 3.4240212387859094e-05,
"loss": 0.0173,
"step": 18660
},
{
"grad_norm": 0.5224332213401794,
"learning_rate": 3.418791574092686e-05,
"loss": 0.0152,
"step": 18670
},
{
"grad_norm": 0.38541659712791443,
"learning_rate": 3.413563830716796e-05,
"loss": 0.0151,
"step": 18680
},
{
"grad_norm": 0.37198877334594727,
"learning_rate": 3.408338015010445e-05,
"loss": 0.0121,
"step": 18690
},
{
"grad_norm": 0.4484640657901764,
"learning_rate": 3.40311413332349e-05,
"loss": 0.0105,
"step": 18700
},
{
"grad_norm": 0.29716673493385315,
"learning_rate": 3.3978921920034365e-05,
"loss": 0.0125,
"step": 18710
},
{
"grad_norm": 0.2875872850418091,
"learning_rate": 3.392672197395441e-05,
"loss": 0.0103,
"step": 18720
},
{
"grad_norm": 0.47655946016311646,
"learning_rate": 3.387454155842287e-05,
"loss": 0.0404,
"step": 18730
},
{
"grad_norm": 0.38600826263427734,
"learning_rate": 3.382238073684386e-05,
"loss": 0.0119,
"step": 18740
},
{
"grad_norm": 0.36877620220184326,
"learning_rate": 3.377023957259771e-05,
"loss": 0.0103,
"step": 18750
},
{
"grad_norm": 0.3797818720340729,
"learning_rate": 3.3718118129040835e-05,
"loss": 0.0145,
"step": 18760
},
{
"grad_norm": 0.40962105989456177,
"learning_rate": 3.3666016469505724e-05,
"loss": 0.0299,
"step": 18770
},
{
"grad_norm": 0.4484851658344269,
"learning_rate": 3.361393465730079e-05,
"loss": 0.0246,
"step": 18780
},
{
"grad_norm": 0.5320003628730774,
"learning_rate": 3.3561872755710366e-05,
"loss": 0.0145,
"step": 18790
},
{
"grad_norm": 0.44000014662742615,
"learning_rate": 3.350983082799456e-05,
"loss": 0.0119,
"step": 18800
},
{
"grad_norm": 0.3139103055000305,
"learning_rate": 3.34578089373892e-05,
"loss": 0.0143,
"step": 18810
},
{
"grad_norm": 0.45011618733406067,
"learning_rate": 3.340580714710581e-05,
"loss": 0.0181,
"step": 18820
},
{
"grad_norm": 0.4089092016220093,
"learning_rate": 3.3353825520331466e-05,
"loss": 0.014,
"step": 18830
},
{
"grad_norm": 0.3470788598060608,
"learning_rate": 3.330186412022876e-05,
"loss": 0.0149,
"step": 18840
},
{
"grad_norm": 0.49339786171913147,
"learning_rate": 3.324992300993568e-05,
"loss": 0.0179,
"step": 18850
},
{
"grad_norm": 0.5080187320709229,
"learning_rate": 3.319800225256556e-05,
"loss": 0.0188,
"step": 18860
},
{
"grad_norm": 0.6082074642181396,
"learning_rate": 3.314610191120702e-05,
"loss": 0.0155,
"step": 18870
},
{
"grad_norm": 0.30084535479545593,
"learning_rate": 3.30942220489239e-05,
"loss": 0.0134,
"step": 18880
},
{
"grad_norm": 0.48842066526412964,
"learning_rate": 3.3042362728755086e-05,
"loss": 0.0143,
"step": 18890
},
{
"grad_norm": 0.39984115958213806,
"learning_rate": 3.299052401371456e-05,
"loss": 0.0276,
"step": 18900
},
{
"grad_norm": 0.43989676237106323,
"learning_rate": 3.293870596679125e-05,
"loss": 0.012,
"step": 18910
},
{
"grad_norm": 0.40966060757637024,
"learning_rate": 3.288690865094895e-05,
"loss": 0.0122,
"step": 18920
},
{
"grad_norm": 0.48424264788627625,
"learning_rate": 3.283513212912632e-05,
"loss": 0.0169,
"step": 18930
},
{
"grad_norm": 0.4137212634086609,
"learning_rate": 3.2783376464236684e-05,
"loss": 0.0111,
"step": 18940
},
{
"grad_norm": 0.2890534996986389,
"learning_rate": 3.273164171916806e-05,
"loss": 0.0113,
"step": 18950
},
{
"grad_norm": 0.4101027548313141,
"learning_rate": 3.267992795678306e-05,
"loss": 0.0182,
"step": 18960
},
{
"grad_norm": 0.4673645496368408,
"learning_rate": 3.2628235239918744e-05,
"loss": 0.0155,
"step": 18970
},
{
"grad_norm": 0.38021376729011536,
"learning_rate": 3.2576563631386695e-05,
"loss": 0.0153,
"step": 18980
},
{
"grad_norm": 0.5109327435493469,
"learning_rate": 3.252491319397275e-05,
"loss": 0.0121,
"step": 18990
},
{
"grad_norm": 0.466889351606369,
"learning_rate": 3.247328399043706e-05,
"loss": 0.0184,
"step": 19000
},
{
"grad_norm": 0.42074355483055115,
"learning_rate": 3.242167608351399e-05,
"loss": 0.0166,
"step": 19010
},
{
"grad_norm": 0.5015917420387268,
"learning_rate": 3.2370089535911986e-05,
"loss": 0.0138,
"step": 19020
},
{
"grad_norm": 0.3733307421207428,
"learning_rate": 3.23185244103136e-05,
"loss": 0.0117,
"step": 19030
},
{
"grad_norm": 0.3340395390987396,
"learning_rate": 3.22669807693753e-05,
"loss": 0.0179,
"step": 19040
},
{
"grad_norm": 0.3702498972415924,
"learning_rate": 3.2215458675727495e-05,
"loss": 0.0157,
"step": 19050
},
{
"grad_norm": 0.4231097400188446,
"learning_rate": 3.216395819197438e-05,
"loss": 0.0152,
"step": 19060
},
{
"grad_norm": 0.4234424829483032,
"learning_rate": 3.211247938069387e-05,
"loss": 0.0215,
"step": 19070
},
{
"grad_norm": 0.39571961760520935,
"learning_rate": 3.206102230443759e-05,
"loss": 0.0141,
"step": 19080
},
{
"grad_norm": 0.40672567486763,
"learning_rate": 3.2009587025730764e-05,
"loss": 0.0229,
"step": 19090
},
{
"grad_norm": 0.29872676730155945,
"learning_rate": 3.195817360707207e-05,
"loss": 0.0202,
"step": 19100
},
{
"grad_norm": 0.35335221886634827,
"learning_rate": 3.19067821109337e-05,
"loss": 0.015,
"step": 19110
},
{
"grad_norm": 0.3291389048099518,
"learning_rate": 3.185541259976114e-05,
"loss": 0.0108,
"step": 19120
},
{
"grad_norm": 0.38024139404296875,
"learning_rate": 3.180406513597316e-05,
"loss": 0.0261,
"step": 19130
},
{
"grad_norm": 0.38877490162849426,
"learning_rate": 3.1752739781961835e-05,
"loss": 0.0108,
"step": 19140
},
{
"grad_norm": 0.40875089168548584,
"learning_rate": 3.170143660009228e-05,
"loss": 0.0077,
"step": 19150
},
{
"grad_norm": 0.4513576626777649,
"learning_rate": 3.16501556527027e-05,
"loss": 0.0149,
"step": 19160
},
{
"grad_norm": 0.3993852436542511,
"learning_rate": 3.1598897002104265e-05,
"loss": 0.0128,
"step": 19170
},
{
"grad_norm": 0.44140565395355225,
"learning_rate": 3.154766071058108e-05,
"loss": 0.0128,
"step": 19180
},
{
"grad_norm": 0.5038141012191772,
"learning_rate": 3.149644684039008e-05,
"loss": 0.0275,
"step": 19190
},
{
"grad_norm": 0.41715896129608154,
"learning_rate": 3.144525545376095e-05,
"loss": 0.0161,
"step": 19200
},
{
"grad_norm": 0.4028128981590271,
"learning_rate": 3.139408661289603e-05,
"loss": 0.0113,
"step": 19210
},
{
"grad_norm": 0.4047220051288605,
"learning_rate": 3.134294037997032e-05,
"loss": 0.0108,
"step": 19220
},
{
"grad_norm": 0.4830787777900696,
"learning_rate": 3.129181681713127e-05,
"loss": 0.0163,
"step": 19230
},
{
"grad_norm": 0.31932514905929565,
"learning_rate": 3.1240715986498855e-05,
"loss": 0.0148,
"step": 19240
},
{
"grad_norm": 0.3803325593471527,
"learning_rate": 3.1189637950165394e-05,
"loss": 0.0131,
"step": 19250
},
{
"grad_norm": 0.3831503689289093,
"learning_rate": 3.1138582770195544e-05,
"loss": 0.0112,
"step": 19260
},
{
"grad_norm": 0.37478360533714294,
"learning_rate": 3.108755050862615e-05,
"loss": 0.0095,
"step": 19270
},
{
"grad_norm": 0.378351092338562,
"learning_rate": 3.1036541227466204e-05,
"loss": 0.0133,
"step": 19280
},
{
"grad_norm": 0.5320636034011841,
"learning_rate": 3.0985554988696786e-05,
"loss": 0.0205,
"step": 19290
},
{
"grad_norm": 0.3494134545326233,
"learning_rate": 3.093459185427102e-05,
"loss": 0.0144,
"step": 19300
},
{
"grad_norm": 0.42025282979011536,
"learning_rate": 3.0883651886113905e-05,
"loss": 0.0129,
"step": 19310
},
{
"grad_norm": 0.5231170058250427,
"learning_rate": 3.0832735146122294e-05,
"loss": 0.0194,
"step": 19320
},
{
"grad_norm": 0.48453816771507263,
"learning_rate": 3.078184169616485e-05,
"loss": 0.0292,
"step": 19330
},
{
"grad_norm": 0.6170175671577454,
"learning_rate": 3.073097159808187e-05,
"loss": 0.0306,
"step": 19340
},
{
"grad_norm": 0.42114177346229553,
"learning_rate": 3.068012491368537e-05,
"loss": 0.0194,
"step": 19350
},
{
"grad_norm": 0.3489767909049988,
"learning_rate": 3.062930170475885e-05,
"loss": 0.0098,
"step": 19360
},
{
"grad_norm": 0.39335355162620544,
"learning_rate": 3.057850203305729e-05,
"loss": 0.0121,
"step": 19370
},
{
"grad_norm": 0.48086848855018616,
"learning_rate": 3.052772596030708e-05,
"loss": 0.0242,
"step": 19380
},
{
"grad_norm": 0.400884211063385,
"learning_rate": 3.0476973548205943e-05,
"loss": 0.0151,
"step": 19390
},
{
"grad_norm": 0.3511633276939392,
"learning_rate": 3.0426244858422847e-05,
"loss": 0.0142,
"step": 19400
},
{
"grad_norm": 0.4084427058696747,
"learning_rate": 3.0375539952597943e-05,
"loss": 0.0129,
"step": 19410
},
{
"grad_norm": 0.46946483850479126,
"learning_rate": 3.0324858892342468e-05,
"loss": 0.0113,
"step": 19420
},
{
"grad_norm": 0.573488175868988,
"learning_rate": 3.0274201739238672e-05,
"loss": 0.0247,
"step": 19430
},
{
"grad_norm": 0.3897065222263336,
"learning_rate": 3.0223568554839786e-05,
"loss": 0.013,
"step": 19440
},
{
"grad_norm": 0.3924945294857025,
"learning_rate": 3.0172959400669886e-05,
"loss": 0.0129,
"step": 19450
},
{
"grad_norm": 0.39307934045791626,
"learning_rate": 3.0122374338223902e-05,
"loss": 0.0144,
"step": 19460
},
{
"grad_norm": 0.30573663115501404,
"learning_rate": 3.0071813428967427e-05,
"loss": 0.0197,
"step": 19470
},
{
"grad_norm": 0.38188767433166504,
"learning_rate": 3.0021276734336746e-05,
"loss": 0.0107,
"step": 19480
},
{
"grad_norm": 0.4230223000049591,
"learning_rate": 2.997076431573871e-05,
"loss": 0.0266,
"step": 19490
},
{
"grad_norm": 0.41215723752975464,
"learning_rate": 2.9920276234550636e-05,
"loss": 0.0132,
"step": 19500
},
{
"grad_norm": 0.4712877869606018,
"learning_rate": 2.986981255212035e-05,
"loss": 0.0243,
"step": 19510
},
{
"grad_norm": 0.45658454298973083,
"learning_rate": 2.9819373329765977e-05,
"loss": 0.0121,
"step": 19520
},
{
"grad_norm": 0.4474101960659027,
"learning_rate": 2.9768958628775902e-05,
"loss": 0.0119,
"step": 19530
},
{
"grad_norm": 0.3338969349861145,
"learning_rate": 2.9718568510408762e-05,
"loss": 0.0111,
"step": 19540
},
{
"grad_norm": 0.4368281066417694,
"learning_rate": 2.9668203035893272e-05,
"loss": 0.0148,
"step": 19550
},
{
"grad_norm": 0.36956822872161865,
"learning_rate": 2.9617862266428287e-05,
"loss": 0.0197,
"step": 19560
},
{
"grad_norm": 0.34172481298446655,
"learning_rate": 2.9567546263182556e-05,
"loss": 0.0143,
"step": 19570
},
{
"grad_norm": 0.41531264781951904,
"learning_rate": 2.951725508729476e-05,
"loss": 0.0165,
"step": 19580
},
{
"grad_norm": 0.4311200976371765,
"learning_rate": 2.946698879987344e-05,
"loss": 0.0141,
"step": 19590
},
{
"grad_norm": 0.5514088869094849,
"learning_rate": 2.9416747461996853e-05,
"loss": 0.0173,
"step": 19600
},
{
"grad_norm": 0.42411795258522034,
"learning_rate": 2.9366531134712973e-05,
"loss": 0.0135,
"step": 19610
},
{
"grad_norm": 0.4155210256576538,
"learning_rate": 2.9316339879039367e-05,
"loss": 0.0148,
"step": 19620
},
{
"grad_norm": 0.3770257830619812,
"learning_rate": 2.9266173755963167e-05,
"loss": 0.0157,
"step": 19630
},
{
"grad_norm": 0.3567594885826111,
"learning_rate": 2.9216032826440926e-05,
"loss": 0.0108,
"step": 19640
},
{
"grad_norm": 0.4395619034767151,
"learning_rate": 2.9165917151398592e-05,
"loss": 0.0148,
"step": 19650
},
{
"grad_norm": 0.3320671617984772,
"learning_rate": 2.9115826791731428e-05,
"loss": 0.0124,
"step": 19660
},
{
"grad_norm": 0.3949275016784668,
"learning_rate": 2.906576180830398e-05,
"loss": 0.0112,
"step": 19670
},
{
"grad_norm": 0.3372920751571655,
"learning_rate": 2.9015722261949917e-05,
"loss": 0.0113,
"step": 19680
},
{
"grad_norm": 0.320241242647171,
"learning_rate": 2.8965708213471986e-05,
"loss": 0.0181,
"step": 19690
},
{
"grad_norm": 0.5423064827919006,
"learning_rate": 2.8915719723641975e-05,
"loss": 0.0128,
"step": 19700
},
{
"grad_norm": 0.2837228775024414,
"learning_rate": 2.8865756853200604e-05,
"loss": 0.0162,
"step": 19710
},
{
"grad_norm": 0.40356123447418213,
"learning_rate": 2.8815819662857503e-05,
"loss": 0.0168,
"step": 19720
},
{
"grad_norm": 0.36055412888526917,
"learning_rate": 2.876590821329105e-05,
"loss": 0.0127,
"step": 19730
},
{
"grad_norm": 0.4096890985965729,
"learning_rate": 2.871602256514836e-05,
"loss": 0.0132,
"step": 19740
},
{
"grad_norm": 0.2852310836315155,
"learning_rate": 2.8666162779045203e-05,
"loss": 0.0099,
"step": 19750
},
{
"grad_norm": 0.47316622734069824,
"learning_rate": 2.8616328915565904e-05,
"loss": 0.0244,
"step": 19760
},
{
"grad_norm": 0.359597772359848,
"learning_rate": 2.856652103526334e-05,
"loss": 0.0122,
"step": 19770
},
{
"grad_norm": 0.430913507938385,
"learning_rate": 2.8516739198658755e-05,
"loss": 0.0161,
"step": 19780
},
{
"grad_norm": 0.38381829857826233,
"learning_rate": 2.846698346624177e-05,
"loss": 0.0213,
"step": 19790
},
{
"grad_norm": 0.372914582490921,
"learning_rate": 2.841725389847032e-05,
"loss": 0.0134,
"step": 19800
},
{
"grad_norm": 0.3621099591255188,
"learning_rate": 2.8367550555770506e-05,
"loss": 0.015,
"step": 19810
},
{
"grad_norm": 0.317513108253479,
"learning_rate": 2.831787349853655e-05,
"loss": 0.0156,
"step": 19820
},
{
"grad_norm": 0.34738773107528687,
"learning_rate": 2.8268222787130806e-05,
"loss": 0.0093,
"step": 19830
},
{
"grad_norm": 0.27453070878982544,
"learning_rate": 2.821859848188355e-05,
"loss": 0.0104,
"step": 19840
},
{
"grad_norm": 0.45254310965538025,
"learning_rate": 2.816900064309299e-05,
"loss": 0.0259,
"step": 19850
},
{
"grad_norm": 0.48578882217407227,
"learning_rate": 2.811942933102517e-05,
"loss": 0.0234,
"step": 19860
},
{
"grad_norm": 0.5218461155891418,
"learning_rate": 2.806988460591391e-05,
"loss": 0.0116,
"step": 19870
},
{
"grad_norm": 0.44902318716049194,
"learning_rate": 2.802036652796074e-05,
"loss": 0.0201,
"step": 19880
},
{
"grad_norm": 0.3212738633155823,
"learning_rate": 2.797087515733478e-05,
"loss": 0.0165,
"step": 19890
},
{
"grad_norm": 0.4487264156341553,
"learning_rate": 2.7921410554172723e-05,
"loss": 0.033,
"step": 19900
},
{
"grad_norm": 0.40605974197387695,
"learning_rate": 2.787197277857871e-05,
"loss": 0.0143,
"step": 19910
},
{
"grad_norm": 0.468168705701828,
"learning_rate": 2.782256189062429e-05,
"loss": 0.013,
"step": 19920
},
{
"grad_norm": 0.4021473526954651,
"learning_rate": 2.777317795034839e-05,
"loss": 0.0138,
"step": 19930
},
{
"grad_norm": 0.3589460253715515,
"learning_rate": 2.7723821017757112e-05,
"loss": 0.0134,
"step": 19940
},
{
"grad_norm": 0.43409234285354614,
"learning_rate": 2.7674491152823822e-05,
"loss": 0.0129,
"step": 19950
},
{
"grad_norm": 0.37429261207580566,
"learning_rate": 2.7625188415488944e-05,
"loss": 0.0123,
"step": 19960
},
{
"grad_norm": 0.436652809381485,
"learning_rate": 2.7575912865659924e-05,
"loss": 0.012,
"step": 19970
},
{
"grad_norm": 0.5572507977485657,
"learning_rate": 2.7526664563211245e-05,
"loss": 0.0143,
"step": 19980
},
{
"grad_norm": 0.36153894662857056,
"learning_rate": 2.7477443567984224e-05,
"loss": 0.0149,
"step": 19990
},
{
"grad_norm": 0.45379918813705444,
"learning_rate": 2.7428249939787e-05,
"loss": 0.0142,
"step": 20000
},
{
"grad_norm": 0.3497525751590729,
"learning_rate": 2.7379083738394484e-05,
"loss": 0.0133,
"step": 20010
},
{
"grad_norm": 0.3281993865966797,
"learning_rate": 2.732994502354823e-05,
"loss": 0.0117,
"step": 20020
},
{
"grad_norm": 0.41283950209617615,
"learning_rate": 2.72808338549564e-05,
"loss": 0.0113,
"step": 20030
},
{
"grad_norm": 0.40923282504081726,
"learning_rate": 2.723175029229374e-05,
"loss": 0.026,
"step": 20040
},
{
"grad_norm": 0.41566795110702515,
"learning_rate": 2.718269439520138e-05,
"loss": 0.0161,
"step": 20050
},
{
"grad_norm": 0.2717386484146118,
"learning_rate": 2.7133666223286858e-05,
"loss": 0.0096,
"step": 20060
},
{
"grad_norm": 0.5419766306877136,
"learning_rate": 2.7084665836124006e-05,
"loss": 0.0209,
"step": 20070
},
{
"grad_norm": 0.4489488899707794,
"learning_rate": 2.703569329325296e-05,
"loss": 0.0122,
"step": 20080
},
{
"grad_norm": 0.436644971370697,
"learning_rate": 2.698674865417994e-05,
"loss": 0.0171,
"step": 20090
},
{
"grad_norm": 0.39454108476638794,
"learning_rate": 2.6937831978377325e-05,
"loss": 0.0127,
"step": 20100
},
{
"grad_norm": 0.3795982301235199,
"learning_rate": 2.6888943325283485e-05,
"loss": 0.0117,
"step": 20110
},
{
"grad_norm": 0.40096431970596313,
"learning_rate": 2.6840082754302733e-05,
"loss": 0.0131,
"step": 20120
},
{
"grad_norm": 0.3219280242919922,
"learning_rate": 2.679125032480525e-05,
"loss": 0.0149,
"step": 20130
},
{
"grad_norm": 0.4019109606742859,
"learning_rate": 2.674244609612708e-05,
"loss": 0.0118,
"step": 20140
},
{
"grad_norm": 0.3212689757347107,
"learning_rate": 2.6693670127569958e-05,
"loss": 0.0137,
"step": 20150
},
{
"grad_norm": 0.3412405252456665,
"learning_rate": 2.6644922478401268e-05,
"loss": 0.0139,
"step": 20160
},
{
"grad_norm": 0.4117426574230194,
"learning_rate": 2.6596203207854004e-05,
"loss": 0.0198,
"step": 20170
},
{
"grad_norm": 0.4132918417453766,
"learning_rate": 2.654751237512666e-05,
"loss": 0.0204,
"step": 20180
},
{
"grad_norm": 0.3232990801334381,
"learning_rate": 2.649885003938323e-05,
"loss": 0.0114,
"step": 20190
},
{
"grad_norm": 0.3695841431617737,
"learning_rate": 2.6450216259753e-05,
"loss": 0.0118,
"step": 20200
},
{
"grad_norm": 0.3901251554489136,
"learning_rate": 2.640161109533063e-05,
"loss": 0.011,
"step": 20210
},
{
"grad_norm": 0.3948058784008026,
"learning_rate": 2.6353034605175936e-05,
"loss": 0.012,
"step": 20220
},
{
"grad_norm": 0.457886666059494,
"learning_rate": 2.630448684831398e-05,
"loss": 0.0205,
"step": 20230
},
{
"grad_norm": 0.39184537529945374,
"learning_rate": 2.625596788373482e-05,
"loss": 0.0123,
"step": 20240
},
{
"grad_norm": 0.3909154534339905,
"learning_rate": 2.620747777039363e-05,
"loss": 0.012,
"step": 20250
},
{
"grad_norm": 0.3841375708580017,
"learning_rate": 2.6159016567210424e-05,
"loss": 0.0103,
"step": 20260
},
{
"grad_norm": 0.356472909450531,
"learning_rate": 2.6110584333070153e-05,
"loss": 0.011,
"step": 20270
},
{
"grad_norm": 0.3356530964374542,
"learning_rate": 2.6062181126822537e-05,
"loss": 0.0085,
"step": 20280
},
{
"grad_norm": 0.2907436192035675,
"learning_rate": 2.601380700728203e-05,
"loss": 0.0169,
"step": 20290
},
{
"grad_norm": 0.29357242584228516,
"learning_rate": 2.596546203322777e-05,
"loss": 0.0104,
"step": 20300
},
{
"grad_norm": 0.3168618977069855,
"learning_rate": 2.5917146263403457e-05,
"loss": 0.0169,
"step": 20310
},
{
"grad_norm": 0.30221450328826904,
"learning_rate": 2.5868859756517296e-05,
"loss": 0.0116,
"step": 20320
},
{
"grad_norm": 0.4158114492893219,
"learning_rate": 2.5820602571241947e-05,
"loss": 0.0151,
"step": 20330
},
{
"grad_norm": 0.3601737916469574,
"learning_rate": 2.577237476621442e-05,
"loss": 0.0161,
"step": 20340
},
{
"grad_norm": 0.49637776613235474,
"learning_rate": 2.572417640003609e-05,
"loss": 0.0152,
"step": 20350
},
{
"grad_norm": 0.26550450921058655,
"learning_rate": 2.5676007531272473e-05,
"loss": 0.0148,
"step": 20360
},
{
"grad_norm": 0.3872086703777313,
"learning_rate": 2.5627868218453332e-05,
"loss": 0.0165,
"step": 20370
},
{
"grad_norm": 0.41861099004745483,
"learning_rate": 2.5579758520072443e-05,
"loss": 0.0246,
"step": 20380
},
{
"grad_norm": 0.4253310561180115,
"learning_rate": 2.5531678494587615e-05,
"loss": 0.0133,
"step": 20390
},
{
"grad_norm": 0.27560749650001526,
"learning_rate": 2.5483628200420644e-05,
"loss": 0.0113,
"step": 20400
},
{
"grad_norm": 0.3707086145877838,
"learning_rate": 2.5435607695957154e-05,
"loss": 0.0133,
"step": 20410
},
{
"grad_norm": 0.4503679871559143,
"learning_rate": 2.5387617039546586e-05,
"loss": 0.0115,
"step": 20420
},
{
"grad_norm": 0.351266086101532,
"learning_rate": 2.5339656289502106e-05,
"loss": 0.01,
"step": 20430
},
{
"grad_norm": 0.44697830080986023,
"learning_rate": 2.529172550410056e-05,
"loss": 0.015,
"step": 20440
},
{
"grad_norm": 0.3603799045085907,
"learning_rate": 2.524382474158234e-05,
"loss": 0.0137,
"step": 20450
},
{
"grad_norm": 0.3506656289100647,
"learning_rate": 2.5195954060151433e-05,
"loss": 0.0363,
"step": 20460
},
{
"grad_norm": 0.36947423219680786,
"learning_rate": 2.5148113517975212e-05,
"loss": 0.0202,
"step": 20470
},
{
"grad_norm": 0.40593433380126953,
"learning_rate": 2.5100303173184447e-05,
"loss": 0.0135,
"step": 20480
},
{
"grad_norm": 0.3566000461578369,
"learning_rate": 2.5052523083873196e-05,
"loss": 0.0098,
"step": 20490
},
{
"grad_norm": 0.35194283723831177,
"learning_rate": 2.500477330809881e-05,
"loss": 0.0119,
"step": 20500
},
{
"grad_norm": 0.38939598202705383,
"learning_rate": 2.4957053903881733e-05,
"loss": 0.0123,
"step": 20510
},
{
"grad_norm": 0.3778887987136841,
"learning_rate": 2.4909364929205576e-05,
"loss": 0.0126,
"step": 20520
},
{
"grad_norm": 0.33012786507606506,
"learning_rate": 2.4861706442016923e-05,
"loss": 0.019,
"step": 20530
},
{
"grad_norm": 0.4363628625869751,
"learning_rate": 2.481407850022533e-05,
"loss": 0.0099,
"step": 20540
},
{
"grad_norm": 0.4167017638683319,
"learning_rate": 2.4766481161703213e-05,
"loss": 0.0136,
"step": 20550
},
{
"grad_norm": 0.4654170572757721,
"learning_rate": 2.4718914484285876e-05,
"loss": 0.0126,
"step": 20560
},
{
"grad_norm": 0.49404433369636536,
"learning_rate": 2.467137852577129e-05,
"loss": 0.0103,
"step": 20570
},
{
"grad_norm": 0.41351452469825745,
"learning_rate": 2.4623873343920123e-05,
"loss": 0.0105,
"step": 20580
},
{
"grad_norm": 0.3306029736995697,
"learning_rate": 2.4576398996455656e-05,
"loss": 0.0131,
"step": 20590
},
{
"grad_norm": 0.3410484790802002,
"learning_rate": 2.452895554106368e-05,
"loss": 0.0161,
"step": 20600
},
{
"grad_norm": 0.41956689953804016,
"learning_rate": 2.4481543035392506e-05,
"loss": 0.0205,
"step": 20610
},
{
"grad_norm": 0.3482762277126312,
"learning_rate": 2.4434161537052774e-05,
"loss": 0.0123,
"step": 20620
},
{
"grad_norm": 0.3954264521598816,
"learning_rate": 2.4386811103617474e-05,
"loss": 0.0151,
"step": 20630
},
{
"grad_norm": 0.3699767589569092,
"learning_rate": 2.4339491792621833e-05,
"loss": 0.0136,
"step": 20640
},
{
"grad_norm": 0.3135485351085663,
"learning_rate": 2.4292203661563312e-05,
"loss": 0.009,
"step": 20650
},
{
"grad_norm": 0.31240224838256836,
"learning_rate": 2.424494676790141e-05,
"loss": 0.0092,
"step": 20660
},
{
"grad_norm": 0.414908230304718,
"learning_rate": 2.4197721169057745e-05,
"loss": 0.0081,
"step": 20670
},
{
"grad_norm": 0.32103878259658813,
"learning_rate": 2.4150526922415857e-05,
"loss": 0.0105,
"step": 20680
},
{
"grad_norm": 0.419786661863327,
"learning_rate": 2.41033640853212e-05,
"loss": 0.0118,
"step": 20690
},
{
"grad_norm": 0.337995707988739,
"learning_rate": 2.405623271508108e-05,
"loss": 0.014,
"step": 20700
},
{
"grad_norm": 0.363885760307312,
"learning_rate": 2.4009132868964522e-05,
"loss": 0.0098,
"step": 20710
},
{
"grad_norm": 0.3196462094783783,
"learning_rate": 2.3962064604202327e-05,
"loss": 0.0172,
"step": 20720
},
{
"grad_norm": 0.3419994115829468,
"learning_rate": 2.3915027977986858e-05,
"loss": 0.022,
"step": 20730
},
{
"grad_norm": 0.2821756601333618,
"learning_rate": 2.386802304747205e-05,
"loss": 0.0092,
"step": 20740
},
{
"grad_norm": 0.42504966259002686,
"learning_rate": 2.382104986977332e-05,
"loss": 0.0163,
"step": 20750
},
{
"grad_norm": 0.4496871531009674,
"learning_rate": 2.377410850196749e-05,
"loss": 0.0167,
"step": 20760
},
{
"grad_norm": 0.4693288505077362,
"learning_rate": 2.3727199001092797e-05,
"loss": 0.0222,
"step": 20770
},
{
"grad_norm": 0.33022254705429077,
"learning_rate": 2.3680321424148677e-05,
"loss": 0.0156,
"step": 20780
},
{
"grad_norm": 0.3460859954357147,
"learning_rate": 2.3633475828095793e-05,
"loss": 0.01,
"step": 20790
},
{
"grad_norm": 0.35404422879219055,
"learning_rate": 2.358666226985599e-05,
"loss": 0.0195,
"step": 20800
},
{
"grad_norm": 0.4058339595794678,
"learning_rate": 2.353988080631213e-05,
"loss": 0.0095,
"step": 20810
},
{
"grad_norm": 0.41160592436790466,
"learning_rate": 2.3493131494308142e-05,
"loss": 0.0212,
"step": 20820
},
{
"grad_norm": 0.45322394371032715,
"learning_rate": 2.344641439064881e-05,
"loss": 0.0105,
"step": 20830
},
{
"grad_norm": 0.37024086713790894,
"learning_rate": 2.339972955209984e-05,
"loss": 0.0151,
"step": 20840
},
{
"grad_norm": 0.2993146777153015,
"learning_rate": 2.335307703538771e-05,
"loss": 0.0128,
"step": 20850
},
{
"grad_norm": 0.3020086884498596,
"learning_rate": 2.330645689719962e-05,
"loss": 0.0164,
"step": 20860
},
{
"grad_norm": 0.37857043743133545,
"learning_rate": 2.3259869194183414e-05,
"loss": 0.0119,
"step": 20870
},
{
"grad_norm": 0.3025279939174652,
"learning_rate": 2.321331398294759e-05,
"loss": 0.0129,
"step": 20880
},
{
"grad_norm": 0.45002081990242004,
"learning_rate": 2.3166791320061094e-05,
"loss": 0.0155,
"step": 20890
},
{
"grad_norm": 0.4132204055786133,
"learning_rate": 2.312030126205335e-05,
"loss": 0.0128,
"step": 20900
},
{
"grad_norm": 0.3252829313278198,
"learning_rate": 2.3073843865414162e-05,
"loss": 0.0115,
"step": 20910
},
{
"grad_norm": 0.3615618348121643,
"learning_rate": 2.302741918659363e-05,
"loss": 0.011,
"step": 20920
},
{
"grad_norm": 0.27461525797843933,
"learning_rate": 2.2981027282002154e-05,
"loss": 0.0086,
"step": 20930
},
{
"grad_norm": 0.2634499669075012,
"learning_rate": 2.2934668208010236e-05,
"loss": 0.0111,
"step": 20940
},
{
"grad_norm": 0.4315398633480072,
"learning_rate": 2.2888342020948555e-05,
"loss": 0.0139,
"step": 20950
},
{
"grad_norm": 0.34742870926856995,
"learning_rate": 2.2842048777107783e-05,
"loss": 0.013,
"step": 20960
},
{
"grad_norm": 0.2923104763031006,
"learning_rate": 2.2795788532738553e-05,
"loss": 0.009,
"step": 20970
},
{
"grad_norm": 0.3700637221336365,
"learning_rate": 2.2749561344051468e-05,
"loss": 0.0086,
"step": 20980
},
{
"grad_norm": 0.333911657333374,
"learning_rate": 2.2703367267216896e-05,
"loss": 0.0097,
"step": 20990
},
{
"grad_norm": 0.35266971588134766,
"learning_rate": 2.2657206358364997e-05,
"loss": 0.0135,
"step": 21000
},
{
"grad_norm": 0.3777472972869873,
"learning_rate": 2.261107867358563e-05,
"loss": 0.0186,
"step": 21010
},
{
"grad_norm": 0.4071866273880005,
"learning_rate": 2.2564984268928263e-05,
"loss": 0.0112,
"step": 21020
},
{
"grad_norm": 0.40689295530319214,
"learning_rate": 2.251892320040198e-05,
"loss": 0.0149,
"step": 21030
},
{
"grad_norm": 0.40095871686935425,
"learning_rate": 2.2472895523975312e-05,
"loss": 0.0151,
"step": 21040
},
{
"grad_norm": 0.37624242901802063,
"learning_rate": 2.2426901295576212e-05,
"loss": 0.0146,
"step": 21050
},
{
"grad_norm": 0.40135475993156433,
"learning_rate": 2.2380940571092013e-05,
"loss": 0.0236,
"step": 21060
},
{
"grad_norm": 0.35573506355285645,
"learning_rate": 2.2335013406369302e-05,
"loss": 0.0147,
"step": 21070
},
{
"grad_norm": 0.3900549113750458,
"learning_rate": 2.228911985721397e-05,
"loss": 0.0093,
"step": 21080
},
{
"grad_norm": 0.2615865170955658,
"learning_rate": 2.224325997939095e-05,
"loss": 0.0144,
"step": 21090
},
{
"grad_norm": 0.4485124349594116,
"learning_rate": 2.219743382862437e-05,
"loss": 0.0194,
"step": 21100
},
{
"grad_norm": 0.37226536870002747,
"learning_rate": 2.2151641460597294e-05,
"loss": 0.0105,
"step": 21110
},
{
"grad_norm": 0.40864190459251404,
"learning_rate": 2.210588293095177e-05,
"loss": 0.0098,
"step": 21120
},
{
"grad_norm": 0.3558749854564667,
"learning_rate": 2.2060158295288714e-05,
"loss": 0.0184,
"step": 21130
},
{
"grad_norm": 0.4341464638710022,
"learning_rate": 2.2014467609167905e-05,
"loss": 0.0134,
"step": 21140
},
{
"grad_norm": 0.3542465567588806,
"learning_rate": 2.1968810928107806e-05,
"loss": 0.0107,
"step": 21150
},
{
"grad_norm": 0.337018758058548,
"learning_rate": 2.1923188307585606e-05,
"loss": 0.0133,
"step": 21160
},
{
"grad_norm": 0.330608993768692,
"learning_rate": 2.187759980303708e-05,
"loss": 0.0123,
"step": 21170
},
{
"grad_norm": 0.3944160044193268,
"learning_rate": 2.1832045469856545e-05,
"loss": 0.0106,
"step": 21180
},
{
"grad_norm": 0.46322861313819885,
"learning_rate": 2.1786525363396844e-05,
"loss": 0.014,
"step": 21190
},
{
"grad_norm": 0.45094573497772217,
"learning_rate": 2.1741039538969183e-05,
"loss": 0.0129,
"step": 21200
},
{
"grad_norm": 0.3929062485694885,
"learning_rate": 2.169558805184313e-05,
"loss": 0.0111,
"step": 21210
},
{
"grad_norm": 0.2959206700325012,
"learning_rate": 2.165017095724651e-05,
"loss": 0.0131,
"step": 21220
},
{
"grad_norm": 0.3811589777469635,
"learning_rate": 2.1604788310365405e-05,
"loss": 0.0149,
"step": 21230
},
{
"grad_norm": 0.37613344192504883,
"learning_rate": 2.155944016634398e-05,
"loss": 0.0103,
"step": 21240
},
{
"grad_norm": 0.3552970588207245,
"learning_rate": 2.1514126580284538e-05,
"loss": 0.0123,
"step": 21250
},
{
"grad_norm": 0.4903419613838196,
"learning_rate": 2.1468847607247344e-05,
"loss": 0.0158,
"step": 21260
},
{
"grad_norm": 0.2657133936882019,
"learning_rate": 2.1423603302250623e-05,
"loss": 0.0184,
"step": 21270
},
{
"grad_norm": 0.30880206823349,
"learning_rate": 2.1378393720270468e-05,
"loss": 0.0134,
"step": 21280
},
{
"grad_norm": 0.3779557943344116,
"learning_rate": 2.133321891624076e-05,
"loss": 0.0115,
"step": 21290
},
{
"grad_norm": 0.33579719066619873,
"learning_rate": 2.1288078945053193e-05,
"loss": 0.0109,
"step": 21300
},
{
"grad_norm": 0.40141651034355164,
"learning_rate": 2.1242973861557064e-05,
"loss": 0.0109,
"step": 21310
},
{
"grad_norm": 0.3551847040653229,
"learning_rate": 2.1197903720559304e-05,
"loss": 0.0111,
"step": 21320
},
{
"grad_norm": 0.48017919063568115,
"learning_rate": 2.115286857682438e-05,
"loss": 0.0217,
"step": 21330
},
{
"grad_norm": 0.3843500316143036,
"learning_rate": 2.1107868485074228e-05,
"loss": 0.0111,
"step": 21340
},
{
"grad_norm": 0.34404993057250977,
"learning_rate": 2.1062903499988234e-05,
"loss": 0.0134,
"step": 21350
},
{
"grad_norm": 0.41454455256462097,
"learning_rate": 2.101797367620308e-05,
"loss": 0.0182,
"step": 21360
},
{
"grad_norm": 0.36549112200737,
"learning_rate": 2.0973079068312713e-05,
"loss": 0.0122,
"step": 21370
},
{
"grad_norm": 0.31137940287590027,
"learning_rate": 2.0928219730868358e-05,
"loss": 0.0106,
"step": 21380
},
{
"grad_norm": 0.31257331371307373,
"learning_rate": 2.0883395718378302e-05,
"loss": 0.0111,
"step": 21390
},
{
"grad_norm": 0.35328856110572815,
"learning_rate": 2.083860708530798e-05,
"loss": 0.0172,
"step": 21400
},
{
"grad_norm": 0.37517714500427246,
"learning_rate": 2.0793853886079796e-05,
"loss": 0.0176,
"step": 21410
},
{
"grad_norm": 0.34892457723617554,
"learning_rate": 2.074913617507309e-05,
"loss": 0.0139,
"step": 21420
},
{
"grad_norm": 0.2689675986766815,
"learning_rate": 2.070445400662411e-05,
"loss": 0.0102,
"step": 21430
},
{
"grad_norm": 0.293985515832901,
"learning_rate": 2.0659807435025907e-05,
"loss": 0.0126,
"step": 21440
},
{
"grad_norm": 0.3371945321559906,
"learning_rate": 2.061519651452825e-05,
"loss": 0.0215,
"step": 21450
},
{
"grad_norm": 0.4146438241004944,
"learning_rate": 2.0570621299337656e-05,
"loss": 0.012,
"step": 21460
},
{
"grad_norm": 0.48507800698280334,
"learning_rate": 2.0526081843617183e-05,
"loss": 0.0192,
"step": 21470
},
{
"grad_norm": 0.32043278217315674,
"learning_rate": 2.0481578201486485e-05,
"loss": 0.0111,
"step": 21480
},
{
"grad_norm": 0.364612340927124,
"learning_rate": 2.0437110427021677e-05,
"loss": 0.0129,
"step": 21490
},
{
"grad_norm": 0.3997472822666168,
"learning_rate": 2.0392678574255282e-05,
"loss": 0.0119,
"step": 21500
},
{
"grad_norm": 0.38202276825904846,
"learning_rate": 2.0348282697176217e-05,
"loss": 0.0155,
"step": 21510
},
{
"grad_norm": 0.27920272946357727,
"learning_rate": 2.030392284972964e-05,
"loss": 0.0112,
"step": 21520
},
{
"grad_norm": 0.34634798765182495,
"learning_rate": 2.0259599085816973e-05,
"loss": 0.0106,
"step": 21530
},
{
"grad_norm": 0.47075319290161133,
"learning_rate": 2.0215311459295755e-05,
"loss": 0.0113,
"step": 21540
},
{
"grad_norm": 0.3236309885978699,
"learning_rate": 2.0171060023979606e-05,
"loss": 0.012,
"step": 21550
},
{
"grad_norm": 0.3418712317943573,
"learning_rate": 2.012684483363823e-05,
"loss": 0.0132,
"step": 21560
},
{
"grad_norm": 0.3409408628940582,
"learning_rate": 2.0082665941997237e-05,
"loss": 0.0119,
"step": 21570
},
{
"grad_norm": 0.33321118354797363,
"learning_rate": 2.0038523402738146e-05,
"loss": 0.0094,
"step": 21580
},
{
"grad_norm": 0.3876037001609802,
"learning_rate": 1.99944172694983e-05,
"loss": 0.0155,
"step": 21590
},
{
"grad_norm": 0.31007063388824463,
"learning_rate": 1.99503475958708e-05,
"loss": 0.0108,
"step": 21600
},
{
"grad_norm": 0.35042881965637207,
"learning_rate": 1.9906314435404484e-05,
"loss": 0.0181,
"step": 21610
},
{
"grad_norm": 0.2942892014980316,
"learning_rate": 1.986231784160378e-05,
"loss": 0.0104,
"step": 21620
},
{
"grad_norm": 0.3781718611717224,
"learning_rate": 1.9818357867928693e-05,
"loss": 0.0132,
"step": 21630
},
{
"grad_norm": 0.3913798928260803,
"learning_rate": 1.977443456779474e-05,
"loss": 0.0094,
"step": 21640
},
{
"grad_norm": 0.28578534722328186,
"learning_rate": 1.9730547994572858e-05,
"loss": 0.0101,
"step": 21650
},
{
"grad_norm": 0.297516793012619,
"learning_rate": 1.9686698201589393e-05,
"loss": 0.0135,
"step": 21660
},
{
"grad_norm": 0.4046097695827484,
"learning_rate": 1.9642885242125964e-05,
"loss": 0.0096,
"step": 21670
},
{
"grad_norm": 0.27653828263282776,
"learning_rate": 1.9599109169419466e-05,
"loss": 0.0086,
"step": 21680
},
{
"grad_norm": 0.3131943643093109,
"learning_rate": 1.9555370036661948e-05,
"loss": 0.0116,
"step": 21690
},
{
"grad_norm": 0.3272033929824829,
"learning_rate": 1.9511667897000574e-05,
"loss": 0.0122,
"step": 21700
},
{
"grad_norm": 0.32891756296157837,
"learning_rate": 1.946800280353755e-05,
"loss": 0.0087,
"step": 21710
},
{
"grad_norm": 0.4825040102005005,
"learning_rate": 1.9424374809330114e-05,
"loss": 0.0163,
"step": 21720
},
{
"grad_norm": 0.38544732332229614,
"learning_rate": 1.9380783967390382e-05,
"loss": 0.0092,
"step": 21730
},
{
"grad_norm": 0.2996348440647125,
"learning_rate": 1.933723033068533e-05,
"loss": 0.0114,
"step": 21740
},
{
"grad_norm": 0.3079988956451416,
"learning_rate": 1.929371395213674e-05,
"loss": 0.0087,
"step": 21750
},
{
"grad_norm": 0.35408520698547363,
"learning_rate": 1.9250234884621092e-05,
"loss": 0.0239,
"step": 21760
},
{
"grad_norm": 0.44580671191215515,
"learning_rate": 1.920679318096959e-05,
"loss": 0.0139,
"step": 21770
},
{
"grad_norm": 0.3188534080982208,
"learning_rate": 1.9163388893967982e-05,
"loss": 0.0261,
"step": 21780
},
{
"grad_norm": 0.3553329110145569,
"learning_rate": 1.9120022076356575e-05,
"loss": 0.0099,
"step": 21790
},
{
"grad_norm": 0.3797047436237335,
"learning_rate": 1.9076692780830114e-05,
"loss": 0.0152,
"step": 21800
},
{
"grad_norm": 0.3217658996582031,
"learning_rate": 1.903340106003782e-05,
"loss": 0.0075,
"step": 21810
},
{
"grad_norm": 0.24950285255908966,
"learning_rate": 1.8990146966583183e-05,
"loss": 0.0094,
"step": 21820
},
{
"grad_norm": 0.4152776300907135,
"learning_rate": 1.8946930553024035e-05,
"loss": 0.0171,
"step": 21830
},
{
"grad_norm": 0.2997531294822693,
"learning_rate": 1.8903751871872378e-05,
"loss": 0.0111,
"step": 21840
},
{
"grad_norm": 0.2773745059967041,
"learning_rate": 1.8860610975594382e-05,
"loss": 0.0115,
"step": 21850
},
{
"grad_norm": 0.34694525599479675,
"learning_rate": 1.8817507916610307e-05,
"loss": 0.0178,
"step": 21860
},
{
"grad_norm": 0.3405821621417999,
"learning_rate": 1.8774442747294407e-05,
"loss": 0.0113,
"step": 21870
},
{
"grad_norm": 0.3691660761833191,
"learning_rate": 1.8731415519974965e-05,
"loss": 0.0089,
"step": 21880
},
{
"grad_norm": 0.28235411643981934,
"learning_rate": 1.86884262869341e-05,
"loss": 0.0091,
"step": 21890
},
{
"grad_norm": 0.32195115089416504,
"learning_rate": 1.8645475100407788e-05,
"loss": 0.0103,
"step": 21900
},
{
"grad_norm": 0.41995030641555786,
"learning_rate": 1.8602562012585768e-05,
"loss": 0.0135,
"step": 21910
},
{
"grad_norm": 0.3368965685367584,
"learning_rate": 1.8559687075611466e-05,
"loss": 0.0122,
"step": 21920
},
{
"grad_norm": 0.347684770822525,
"learning_rate": 1.8516850341582017e-05,
"loss": 0.0081,
"step": 21930
},
{
"grad_norm": 0.5414378046989441,
"learning_rate": 1.847405186254807e-05,
"loss": 0.0112,
"step": 21940
},
{
"grad_norm": 0.3861631453037262,
"learning_rate": 1.843129169051379e-05,
"loss": 0.0086,
"step": 21950
},
{
"grad_norm": 0.33482974767684937,
"learning_rate": 1.8388569877436863e-05,
"loss": 0.0103,
"step": 21960
},
{
"grad_norm": 0.30579087138175964,
"learning_rate": 1.834588647522828e-05,
"loss": 0.0129,
"step": 21970
},
{
"grad_norm": 0.4538271129131317,
"learning_rate": 1.8303241535752435e-05,
"loss": 0.0106,
"step": 21980
},
{
"grad_norm": 0.34982365369796753,
"learning_rate": 1.8260635110826934e-05,
"loss": 0.01,
"step": 21990
},
{
"grad_norm": 0.35144829750061035,
"learning_rate": 1.8218067252222597e-05,
"loss": 0.0104,
"step": 22000
},
{
"grad_norm": 0.3769899904727936,
"learning_rate": 1.817553801166339e-05,
"loss": 0.01,
"step": 22010
},
{
"grad_norm": 0.3560279309749603,
"learning_rate": 1.8133047440826333e-05,
"loss": 0.0146,
"step": 22020
},
{
"grad_norm": 0.29010841250419617,
"learning_rate": 1.809059559134151e-05,
"loss": 0.0159,
"step": 22030
},
{
"grad_norm": 0.2513137757778168,
"learning_rate": 1.80481825147919e-05,
"loss": 0.0123,
"step": 22040
},
{
"grad_norm": 0.2472553700208664,
"learning_rate": 1.8005808262713396e-05,
"loss": 0.0108,
"step": 22050
},
{
"grad_norm": 0.4036372900009155,
"learning_rate": 1.7963472886594714e-05,
"loss": 0.0084,
"step": 22060
},
{
"grad_norm": 0.2563245892524719,
"learning_rate": 1.7921176437877302e-05,
"loss": 0.0104,
"step": 22070
},
{
"grad_norm": 0.3508543372154236,
"learning_rate": 1.7878918967955366e-05,
"loss": 0.0108,
"step": 22080
},
{
"grad_norm": 0.28297942876815796,
"learning_rate": 1.7836700528175693e-05,
"loss": 0.0163,
"step": 22090
},
{
"grad_norm": 0.34369349479675293,
"learning_rate": 1.779452116983769e-05,
"loss": 0.011,
"step": 22100
},
{
"grad_norm": 0.41121354699134827,
"learning_rate": 1.775238094419325e-05,
"loss": 0.0217,
"step": 22110
},
{
"grad_norm": 0.3452577590942383,
"learning_rate": 1.7710279902446708e-05,
"loss": 0.0229,
"step": 22120
},
{
"grad_norm": 0.3425699770450592,
"learning_rate": 1.7668218095754795e-05,
"loss": 0.0139,
"step": 22130
},
{
"grad_norm": 0.37657833099365234,
"learning_rate": 1.7626195575226594e-05,
"loss": 0.0106,
"step": 22140
},
{
"grad_norm": 0.2201521396636963,
"learning_rate": 1.758421239192343e-05,
"loss": 0.0102,
"step": 22150
},
{
"grad_norm": 0.28997448086738586,
"learning_rate": 1.7542268596858814e-05,
"loss": 0.0085,
"step": 22160
},
{
"grad_norm": 0.24955101311206818,
"learning_rate": 1.750036424099841e-05,
"loss": 0.0074,
"step": 22170
},
{
"grad_norm": 0.425052285194397,
"learning_rate": 1.7458499375259956e-05,
"loss": 0.0095,
"step": 22180
},
{
"grad_norm": 0.3093009889125824,
"learning_rate": 1.7416674050513245e-05,
"loss": 0.0112,
"step": 22190
},
{
"grad_norm": 0.26229673624038696,
"learning_rate": 1.737488831757997e-05,
"loss": 0.0114,
"step": 22200
},
{
"grad_norm": 0.37094616889953613,
"learning_rate": 1.733314222723373e-05,
"loss": 0.0141,
"step": 22210
},
{
"grad_norm": 0.3314245641231537,
"learning_rate": 1.7291435830199954e-05,
"loss": 0.0106,
"step": 22220
},
{
"grad_norm": 0.27559182047843933,
"learning_rate": 1.7249769177155878e-05,
"loss": 0.0078,
"step": 22230
},
{
"grad_norm": 0.3598017990589142,
"learning_rate": 1.720814231873038e-05,
"loss": 0.0153,
"step": 22240
},
{
"grad_norm": 0.36678168177604675,
"learning_rate": 1.716655530550405e-05,
"loss": 0.0075,
"step": 22250
},
{
"grad_norm": 0.45120978355407715,
"learning_rate": 1.7125008188009017e-05,
"loss": 0.0133,
"step": 22260
},
{
"grad_norm": 0.3139858841896057,
"learning_rate": 1.708350101672894e-05,
"loss": 0.0084,
"step": 22270
},
{
"grad_norm": 0.32317620515823364,
"learning_rate": 1.7042033842098957e-05,
"loss": 0.0126,
"step": 22280
},
{
"grad_norm": 0.3026094138622284,
"learning_rate": 1.7000606714505567e-05,
"loss": 0.0102,
"step": 22290
},
{
"grad_norm": 0.29565656185150146,
"learning_rate": 1.695921968428668e-05,
"loss": 0.0102,
"step": 22300
},
{
"grad_norm": 0.2802305519580841,
"learning_rate": 1.6917872801731416e-05,
"loss": 0.0116,
"step": 22310
},
{
"grad_norm": 0.32743433117866516,
"learning_rate": 1.687656611708014e-05,
"loss": 0.0146,
"step": 22320
},
{
"grad_norm": 0.3033479154109955,
"learning_rate": 1.6835299680524368e-05,
"loss": 0.0087,
"step": 22330
},
{
"grad_norm": 0.3348958194255829,
"learning_rate": 1.679407354220669e-05,
"loss": 0.0179,
"step": 22340
},
{
"grad_norm": 0.325706422328949,
"learning_rate": 1.6752887752220793e-05,
"loss": 0.0129,
"step": 22350
},
{
"grad_norm": 0.216952845454216,
"learning_rate": 1.6711742360611277e-05,
"loss": 0.0089,
"step": 22360
},
{
"grad_norm": 0.3161250352859497,
"learning_rate": 1.6670637417373652e-05,
"loss": 0.0216,
"step": 22370
},
{
"grad_norm": 0.33046531677246094,
"learning_rate": 1.6629572972454333e-05,
"loss": 0.0104,
"step": 22380
},
{
"grad_norm": 0.29106393456459045,
"learning_rate": 1.6588549075750464e-05,
"loss": 0.0072,
"step": 22390
},
{
"grad_norm": 0.45629608631134033,
"learning_rate": 1.6547565777109976e-05,
"loss": 0.0153,
"step": 22400
},
{
"grad_norm": 0.3548107147216797,
"learning_rate": 1.650662312633143e-05,
"loss": 0.0235,
"step": 22410
},
{
"grad_norm": 0.5296837091445923,
"learning_rate": 1.6465721173164002e-05,
"loss": 0.0141,
"step": 22420
},
{
"grad_norm": 0.37373632192611694,
"learning_rate": 1.6424859967307428e-05,
"loss": 0.0085,
"step": 22430
},
{
"grad_norm": 0.3687719404697418,
"learning_rate": 1.6384039558411903e-05,
"loss": 0.0092,
"step": 22440
},
{
"grad_norm": 0.3597353994846344,
"learning_rate": 1.634325999607811e-05,
"loss": 0.0142,
"step": 22450
},
{
"grad_norm": 0.4023669958114624,
"learning_rate": 1.6302521329857045e-05,
"loss": 0.0131,
"step": 22460
},
{
"grad_norm": 0.3980092704296112,
"learning_rate": 1.6261823609250027e-05,
"loss": 0.013,
"step": 22470
},
{
"grad_norm": 0.37898752093315125,
"learning_rate": 1.6221166883708627e-05,
"loss": 0.0108,
"step": 22480
},
{
"grad_norm": 0.3727536201477051,
"learning_rate": 1.6180551202634603e-05,
"loss": 0.0122,
"step": 22490
},
{
"grad_norm": 0.3873298466205597,
"learning_rate": 1.613997661537981e-05,
"loss": 0.0093,
"step": 22500
},
{
"grad_norm": 0.3590852916240692,
"learning_rate": 1.609944317124624e-05,
"loss": 0.0073,
"step": 22510
},
{
"grad_norm": 0.2902836203575134,
"learning_rate": 1.605895091948582e-05,
"loss": 0.0151,
"step": 22520
},
{
"grad_norm": 0.41301897168159485,
"learning_rate": 1.6018499909300477e-05,
"loss": 0.0188,
"step": 22530
},
{
"grad_norm": 0.2694375813007355,
"learning_rate": 1.597809018984199e-05,
"loss": 0.0075,
"step": 22540
},
{
"grad_norm": 0.23874306678771973,
"learning_rate": 1.593772181021196e-05,
"loss": 0.0178,
"step": 22550
},
{
"grad_norm": 0.3758929371833801,
"learning_rate": 1.5897394819461813e-05,
"loss": 0.0232,
"step": 22560
},
{
"grad_norm": 0.3254625201225281,
"learning_rate": 1.5857109266592625e-05,
"loss": 0.009,
"step": 22570
},
{
"grad_norm": 0.2561032176017761,
"learning_rate": 1.581686520055514e-05,
"loss": 0.0088,
"step": 22580
},
{
"grad_norm": 0.28711187839508057,
"learning_rate": 1.5776662670249704e-05,
"loss": 0.0087,
"step": 22590
},
{
"grad_norm": 0.26026955246925354,
"learning_rate": 1.573650172452615e-05,
"loss": 0.011,
"step": 22600
},
{
"grad_norm": 0.4011930525302887,
"learning_rate": 1.5696382412183852e-05,
"loss": 0.0086,
"step": 22610
},
{
"grad_norm": 0.358429491519928,
"learning_rate": 1.565630478197155e-05,
"loss": 0.0123,
"step": 22620
},
{
"grad_norm": 0.30070945620536804,
"learning_rate": 1.5616268882587332e-05,
"loss": 0.0092,
"step": 22630
},
{
"grad_norm": 0.36825692653656006,
"learning_rate": 1.5576274762678594e-05,
"loss": 0.0171,
"step": 22640
},
{
"grad_norm": 0.3957798182964325,
"learning_rate": 1.5536322470841952e-05,
"loss": 0.0123,
"step": 22650
},
{
"grad_norm": 0.39087849855422974,
"learning_rate": 1.549641205562324e-05,
"loss": 0.0126,
"step": 22660
},
{
"grad_norm": 0.30627167224884033,
"learning_rate": 1.5456543565517334e-05,
"loss": 0.0084,
"step": 22670
},
{
"grad_norm": 0.40238896012306213,
"learning_rate": 1.541671704896825e-05,
"loss": 0.0108,
"step": 22680
},
{
"grad_norm": 0.37371826171875,
"learning_rate": 1.5376932554368938e-05,
"loss": 0.0095,
"step": 22690
},
{
"grad_norm": 0.35578683018684387,
"learning_rate": 1.5337190130061318e-05,
"loss": 0.0098,
"step": 22700
},
{
"grad_norm": 0.43002694845199585,
"learning_rate": 1.5297489824336165e-05,
"loss": 0.0199,
"step": 22710
},
{
"grad_norm": 0.4216368496417999,
"learning_rate": 1.5257831685433121e-05,
"loss": 0.0084,
"step": 22720
},
{
"grad_norm": 0.3711147606372833,
"learning_rate": 1.521821576154055e-05,
"loss": 0.0102,
"step": 22730
},
{
"grad_norm": 0.24717780947685242,
"learning_rate": 1.5178642100795542e-05,
"loss": 0.0096,
"step": 22740
},
{
"grad_norm": 0.38126954436302185,
"learning_rate": 1.5139110751283819e-05,
"loss": 0.0163,
"step": 22750
},
{
"grad_norm": 0.37915271520614624,
"learning_rate": 1.5099621761039683e-05,
"loss": 0.0177,
"step": 22760
},
{
"grad_norm": 0.3047991693019867,
"learning_rate": 1.5060175178046016e-05,
"loss": 0.011,
"step": 22770
},
{
"grad_norm": 0.32326024770736694,
"learning_rate": 1.5020771050234117e-05,
"loss": 0.009,
"step": 22780
},
{
"grad_norm": 0.2917804718017578,
"learning_rate": 1.4981409425483716e-05,
"loss": 0.0079,
"step": 22790
},
{
"grad_norm": 0.27987033128738403,
"learning_rate": 1.4942090351622883e-05,
"loss": 0.0073,
"step": 22800
},
{
"grad_norm": 0.34002649784088135,
"learning_rate": 1.490281387642804e-05,
"loss": 0.0138,
"step": 22810
},
{
"grad_norm": 0.3108595609664917,
"learning_rate": 1.4863580047623765e-05,
"loss": 0.0176,
"step": 22820
},
{
"grad_norm": 0.31535854935646057,
"learning_rate": 1.4824388912882897e-05,
"loss": 0.0088,
"step": 22830
},
{
"grad_norm": 0.29399374127388,
"learning_rate": 1.4785240519826344e-05,
"loss": 0.0141,
"step": 22840
},
{
"grad_norm": 0.4593169093132019,
"learning_rate": 1.4746134916023097e-05,
"loss": 0.0247,
"step": 22850
},
{
"grad_norm": 0.3179449141025543,
"learning_rate": 1.4707072148990141e-05,
"loss": 0.01,
"step": 22860
},
{
"grad_norm": 0.41211646795272827,
"learning_rate": 1.4668052266192423e-05,
"loss": 0.0101,
"step": 22870
},
{
"grad_norm": 0.38413020968437195,
"learning_rate": 1.4629075315042795e-05,
"loss": 0.016,
"step": 22880
},
{
"grad_norm": 0.31453660130500793,
"learning_rate": 1.4590141342901925e-05,
"loss": 0.009,
"step": 22890
},
{
"grad_norm": 0.44798779487609863,
"learning_rate": 1.4551250397078253e-05,
"loss": 0.0191,
"step": 22900
},
{
"grad_norm": 0.3584701418876648,
"learning_rate": 1.4512402524827945e-05,
"loss": 0.0142,
"step": 22910
},
{
"grad_norm": 0.3955673575401306,
"learning_rate": 1.447359777335482e-05,
"loss": 0.0167,
"step": 22920
},
{
"grad_norm": 0.3681682050228119,
"learning_rate": 1.4434836189810335e-05,
"loss": 0.009,
"step": 22930
},
{
"grad_norm": 0.24958840012550354,
"learning_rate": 1.4396117821293453e-05,
"loss": 0.0074,
"step": 22940
},
{
"grad_norm": 0.4173002243041992,
"learning_rate": 1.4357442714850633e-05,
"loss": 0.0208,
"step": 22950
},
{
"grad_norm": 0.35516002774238586,
"learning_rate": 1.43188109174758e-05,
"loss": 0.0077,
"step": 22960
},
{
"grad_norm": 0.2839513421058655,
"learning_rate": 1.4280222476110206e-05,
"loss": 0.0082,
"step": 22970
},
{
"grad_norm": 0.2738654315471649,
"learning_rate": 1.4241677437642469e-05,
"loss": 0.0104,
"step": 22980
},
{
"grad_norm": 0.4107381999492645,
"learning_rate": 1.420317584890844e-05,
"loss": 0.0129,
"step": 22990
},
{
"grad_norm": 0.34337741136550903,
"learning_rate": 1.4164717756691176e-05,
"loss": 0.0084,
"step": 23000
},
{
"grad_norm": 0.22585877776145935,
"learning_rate": 1.4126303207720882e-05,
"loss": 0.0088,
"step": 23010
},
{
"grad_norm": 0.4082593321800232,
"learning_rate": 1.408793224867484e-05,
"loss": 0.0092,
"step": 23020
},
{
"grad_norm": 0.3571750521659851,
"learning_rate": 1.4049604926177423e-05,
"loss": 0.0122,
"step": 23030
},
{
"grad_norm": 0.3942554295063019,
"learning_rate": 1.4011321286799916e-05,
"loss": 0.01,
"step": 23040
},
{
"grad_norm": 0.40147796273231506,
"learning_rate": 1.3973081377060565e-05,
"loss": 0.0191,
"step": 23050
},
{
"grad_norm": 0.39444950222969055,
"learning_rate": 1.393488524342445e-05,
"loss": 0.0099,
"step": 23060
},
{
"grad_norm": 0.4191432595252991,
"learning_rate": 1.3896732932303486e-05,
"loss": 0.0076,
"step": 23070
},
{
"grad_norm": 0.33953857421875,
"learning_rate": 1.3858624490056304e-05,
"loss": 0.0089,
"step": 23080
},
{
"grad_norm": 0.28210094571113586,
"learning_rate": 1.3820559962988299e-05,
"loss": 0.0081,
"step": 23090
},
{
"grad_norm": 0.4590706527233124,
"learning_rate": 1.3782539397351418e-05,
"loss": 0.0163,
"step": 23100
},
{
"grad_norm": 0.2451314926147461,
"learning_rate": 1.3744562839344266e-05,
"loss": 0.015,
"step": 23110
},
{
"grad_norm": 0.31231650710105896,
"learning_rate": 1.370663033511193e-05,
"loss": 0.0106,
"step": 23120
},
{
"grad_norm": 0.3804459571838379,
"learning_rate": 1.3668741930745964e-05,
"loss": 0.009,
"step": 23130
},
{
"grad_norm": 0.3504636883735657,
"learning_rate": 1.3630897672284382e-05,
"loss": 0.0069,
"step": 23140
},
{
"grad_norm": 0.32438281178474426,
"learning_rate": 1.3593097605711508e-05,
"loss": 0.0256,
"step": 23150
},
{
"grad_norm": 0.259868860244751,
"learning_rate": 1.3555341776957992e-05,
"loss": 0.0075,
"step": 23160
},
{
"grad_norm": 0.37076303362846375,
"learning_rate": 1.3517630231900724e-05,
"loss": 0.0114,
"step": 23170
},
{
"grad_norm": 0.30132490396499634,
"learning_rate": 1.3479963016362767e-05,
"loss": 0.0076,
"step": 23180
},
{
"grad_norm": 0.3498169779777527,
"learning_rate": 1.3442340176113377e-05,
"loss": 0.0134,
"step": 23190
},
{
"grad_norm": 0.36172422766685486,
"learning_rate": 1.340476175686784e-05,
"loss": 0.0098,
"step": 23200
},
{
"grad_norm": 0.24315716326236725,
"learning_rate": 1.3367227804287469e-05,
"loss": 0.0097,
"step": 23210
},
{
"grad_norm": 0.3055903911590576,
"learning_rate": 1.332973836397956e-05,
"loss": 0.0087,
"step": 23220
},
{
"grad_norm": 0.33361735939979553,
"learning_rate": 1.329229348149731e-05,
"loss": 0.013,
"step": 23230
},
{
"grad_norm": 0.2884477376937866,
"learning_rate": 1.3254893202339796e-05,
"loss": 0.01,
"step": 23240
},
{
"grad_norm": 0.24087662994861603,
"learning_rate": 1.3217537571951872e-05,
"loss": 0.0119,
"step": 23250
},
{
"grad_norm": 0.19424206018447876,
"learning_rate": 1.3180226635724169e-05,
"loss": 0.0081,
"step": 23260
},
{
"grad_norm": 0.3618966341018677,
"learning_rate": 1.314296043899298e-05,
"loss": 0.0108,
"step": 23270
},
{
"grad_norm": 0.26040786504745483,
"learning_rate": 1.3105739027040248e-05,
"loss": 0.0102,
"step": 23280
},
{
"grad_norm": 0.3066410720348358,
"learning_rate": 1.3068562445093491e-05,
"loss": 0.0207,
"step": 23290
},
{
"grad_norm": 0.396871954202652,
"learning_rate": 1.3031430738325778e-05,
"loss": 0.0105,
"step": 23300
},
{
"grad_norm": 0.3348749577999115,
"learning_rate": 1.299434395185563e-05,
"loss": 0.0145,
"step": 23310
},
{
"grad_norm": 0.5619756579399109,
"learning_rate": 1.295730213074699e-05,
"loss": 0.0118,
"step": 23320
},
{
"grad_norm": 0.30852410197257996,
"learning_rate": 1.2920305320009152e-05,
"loss": 0.0097,
"step": 23330
},
{
"grad_norm": 0.3252132534980774,
"learning_rate": 1.2883353564596729e-05,
"loss": 0.0122,
"step": 23340
},
{
"grad_norm": 0.35987356305122375,
"learning_rate": 1.2846446909409604e-05,
"loss": 0.0128,
"step": 23350
},
{
"grad_norm": 0.39850160479545593,
"learning_rate": 1.280958539929284e-05,
"loss": 0.0109,
"step": 23360
},
{
"grad_norm": 0.32098791003227234,
"learning_rate": 1.277276907903664e-05,
"loss": 0.0151,
"step": 23370
},
{
"grad_norm": 0.31667283177375793,
"learning_rate": 1.2735997993376298e-05,
"loss": 0.0164,
"step": 23380
},
{
"grad_norm": 0.38768160343170166,
"learning_rate": 1.2699272186992166e-05,
"loss": 0.0208,
"step": 23390
},
{
"grad_norm": 0.2805112600326538,
"learning_rate": 1.2662591704509547e-05,
"loss": 0.018,
"step": 23400
},
{
"grad_norm": 0.31703510880470276,
"learning_rate": 1.2625956590498711e-05,
"loss": 0.0175,
"step": 23410
},
{
"grad_norm": 0.25492143630981445,
"learning_rate": 1.2589366889474757e-05,
"loss": 0.0074,
"step": 23420
},
{
"grad_norm": 0.299262136220932,
"learning_rate": 1.2552822645897621e-05,
"loss": 0.0101,
"step": 23430
},
{
"grad_norm": 0.37217268347740173,
"learning_rate": 1.2516323904172e-05,
"loss": 0.0102,
"step": 23440
},
{
"grad_norm": 0.39470577239990234,
"learning_rate": 1.2479870708647324e-05,
"loss": 0.0128,
"step": 23450
},
{
"grad_norm": 0.2928149402141571,
"learning_rate": 1.2443463103617658e-05,
"loss": 0.0097,
"step": 23460
},
{
"grad_norm": 0.321454256772995,
"learning_rate": 1.240710113332167e-05,
"loss": 0.0125,
"step": 23470
},
{
"grad_norm": 0.24818837642669678,
"learning_rate": 1.2370784841942584e-05,
"loss": 0.008,
"step": 23480
},
{
"grad_norm": 0.33395469188690186,
"learning_rate": 1.2334514273608117e-05,
"loss": 0.0144,
"step": 23490
},
{
"grad_norm": 0.28099456429481506,
"learning_rate": 1.2298289472390417e-05,
"loss": 0.0214,
"step": 23500
},
{
"grad_norm": 0.2921864986419678,
"learning_rate": 1.226211048230606e-05,
"loss": 0.0102,
"step": 23510
},
{
"grad_norm": 0.4434816241264343,
"learning_rate": 1.2225977347315921e-05,
"loss": 0.0288,
"step": 23520
},
{
"grad_norm": 0.27457571029663086,
"learning_rate": 1.2189890111325148e-05,
"loss": 0.0069,
"step": 23530
},
{
"grad_norm": 0.3128238022327423,
"learning_rate": 1.2153848818183161e-05,
"loss": 0.0149,
"step": 23540
},
{
"grad_norm": 0.3143909275531769,
"learning_rate": 1.2117853511683507e-05,
"loss": 0.0088,
"step": 23550
},
{
"grad_norm": 0.37578046321868896,
"learning_rate": 1.2081904235563906e-05,
"loss": 0.0102,
"step": 23560
},
{
"grad_norm": 0.30693739652633667,
"learning_rate": 1.20460010335061e-05,
"loss": 0.0091,
"step": 23570
},
{
"grad_norm": 0.38705846667289734,
"learning_rate": 1.2010143949135865e-05,
"loss": 0.0104,
"step": 23580
},
{
"grad_norm": 0.4208884537220001,
"learning_rate": 1.1974333026022938e-05,
"loss": 0.0209,
"step": 23590
},
{
"grad_norm": 0.3216271996498108,
"learning_rate": 1.1938568307680964e-05,
"loss": 0.013,
"step": 23600
},
{
"grad_norm": 0.2008558213710785,
"learning_rate": 1.1902849837567464e-05,
"loss": 0.0143,
"step": 23610
},
{
"grad_norm": 0.34728798270225525,
"learning_rate": 1.1867177659083739e-05,
"loss": 0.0101,
"step": 23620
},
{
"grad_norm": 0.25428062677383423,
"learning_rate": 1.1831551815574848e-05,
"loss": 0.008,
"step": 23630
},
{
"grad_norm": 0.36646169424057007,
"learning_rate": 1.1795972350329554e-05,
"loss": 0.0189,
"step": 23640
},
{
"grad_norm": 0.41484108567237854,
"learning_rate": 1.176043930658024e-05,
"loss": 0.0116,
"step": 23650
},
{
"grad_norm": 0.2762801945209503,
"learning_rate": 1.1724952727502942e-05,
"loss": 0.0093,
"step": 23660
},
{
"grad_norm": 0.22622156143188477,
"learning_rate": 1.1689512656217178e-05,
"loss": 0.0095,
"step": 23670
},
{
"grad_norm": 0.22927750647068024,
"learning_rate": 1.1654119135785962e-05,
"loss": 0.0202,
"step": 23680
},
{
"grad_norm": 0.3763860762119293,
"learning_rate": 1.1618772209215794e-05,
"loss": 0.0088,
"step": 23690
},
{
"grad_norm": 0.35425642132759094,
"learning_rate": 1.1583471919456506e-05,
"loss": 0.0114,
"step": 23700
},
{
"grad_norm": 0.25686004757881165,
"learning_rate": 1.1548218309401266e-05,
"loss": 0.0084,
"step": 23710
},
{
"grad_norm": 0.299022376537323,
"learning_rate": 1.1513011421886555e-05,
"loss": 0.0098,
"step": 23720
},
{
"grad_norm": 0.47105270624160767,
"learning_rate": 1.1477851299692056e-05,
"loss": 0.0152,
"step": 23730
},
{
"grad_norm": 0.36134469509124756,
"learning_rate": 1.144273798554063e-05,
"loss": 0.0094,
"step": 23740
},
{
"grad_norm": 0.2725926339626312,
"learning_rate": 1.1407671522098262e-05,
"loss": 0.01,
"step": 23750
},
{
"grad_norm": 0.27360621094703674,
"learning_rate": 1.1372651951974e-05,
"loss": 0.0106,
"step": 23760
},
{
"grad_norm": 0.31498315930366516,
"learning_rate": 1.1337679317719952e-05,
"loss": 0.0079,
"step": 23770
},
{
"grad_norm": 0.35107123851776123,
"learning_rate": 1.1302753661831151e-05,
"loss": 0.0092,
"step": 23780
},
{
"grad_norm": 0.3604271709918976,
"learning_rate": 1.1267875026745561e-05,
"loss": 0.0091,
"step": 23790
},
{
"grad_norm": 0.2548331022262573,
"learning_rate": 1.1233043454844016e-05,
"loss": 0.0154,
"step": 23800
},
{
"grad_norm": 0.2613779306411743,
"learning_rate": 1.1198258988450144e-05,
"loss": 0.0096,
"step": 23810
},
{
"grad_norm": 0.378664493560791,
"learning_rate": 1.1163521669830368e-05,
"loss": 0.0077,
"step": 23820
},
{
"grad_norm": 0.2799585461616516,
"learning_rate": 1.1128831541193818e-05,
"loss": 0.0079,
"step": 23830
},
{
"grad_norm": 0.2942686975002289,
"learning_rate": 1.1094188644692255e-05,
"loss": 0.0133,
"step": 23840
},
{
"grad_norm": 0.23858754336833954,
"learning_rate": 1.1059593022420067e-05,
"loss": 0.0063,
"step": 23850
},
{
"grad_norm": 0.3071526885032654,
"learning_rate": 1.1025044716414184e-05,
"loss": 0.008,
"step": 23860
},
{
"grad_norm": 0.383394330739975,
"learning_rate": 1.0990543768654082e-05,
"loss": 0.0148,
"step": 23870
},
{
"grad_norm": 0.24383921921253204,
"learning_rate": 1.0956090221061655e-05,
"loss": 0.0068,
"step": 23880
},
{
"grad_norm": 0.21446874737739563,
"learning_rate": 1.0921684115501208e-05,
"loss": 0.0129,
"step": 23890
},
{
"grad_norm": 0.2625412940979004,
"learning_rate": 1.0887325493779405e-05,
"loss": 0.0086,
"step": 23900
},
{
"grad_norm": 0.355925977230072,
"learning_rate": 1.0853014397645207e-05,
"loss": 0.0086,
"step": 23910
},
{
"grad_norm": 0.3924808204174042,
"learning_rate": 1.0818750868789829e-05,
"loss": 0.0166,
"step": 23920
},
{
"grad_norm": 0.3411644995212555,
"learning_rate": 1.0784534948846703e-05,
"loss": 0.0083,
"step": 23930
},
{
"grad_norm": 0.2530641555786133,
"learning_rate": 1.0750366679391393e-05,
"loss": 0.0067,
"step": 23940
},
{
"grad_norm": 0.18815849721431732,
"learning_rate": 1.0716246101941557e-05,
"loss": 0.006,
"step": 23950
},
{
"grad_norm": 0.3481179475784302,
"learning_rate": 1.0682173257956934e-05,
"loss": 0.022,
"step": 23960
},
{
"grad_norm": 0.25817424058914185,
"learning_rate": 1.064814818883922e-05,
"loss": 0.0129,
"step": 23970
},
{
"grad_norm": 0.34998685121536255,
"learning_rate": 1.06141709359321e-05,
"loss": 0.0091,
"step": 23980
},
{
"grad_norm": 0.21976082026958466,
"learning_rate": 1.058024154052114e-05,
"loss": 0.0125,
"step": 23990
},
{
"grad_norm": 0.3316846489906311,
"learning_rate": 1.0546360043833747e-05,
"loss": 0.0086,
"step": 24000
},
{
"grad_norm": 0.27323266863822937,
"learning_rate": 1.0512526487039138e-05,
"loss": 0.0064,
"step": 24010
},
{
"grad_norm": 0.24928197264671326,
"learning_rate": 1.0478740911248258e-05,
"loss": 0.0088,
"step": 24020
},
{
"grad_norm": 0.24796555936336517,
"learning_rate": 1.0445003357513793e-05,
"loss": 0.0105,
"step": 24030
},
{
"grad_norm": 0.3656565546989441,
"learning_rate": 1.0411313866830042e-05,
"loss": 0.0109,
"step": 24040
},
{
"grad_norm": 0.3154156804084778,
"learning_rate": 1.0377672480132915e-05,
"loss": 0.0097,
"step": 24050
},
{
"grad_norm": 0.22947734594345093,
"learning_rate": 1.0344079238299864e-05,
"loss": 0.0125,
"step": 24060
},
{
"grad_norm": 0.276332825422287,
"learning_rate": 1.0310534182149833e-05,
"loss": 0.0068,
"step": 24070
},
{
"grad_norm": 0.42141395807266235,
"learning_rate": 1.0277037352443258e-05,
"loss": 0.0147,
"step": 24080
},
{
"grad_norm": 0.22027672827243805,
"learning_rate": 1.024358878988193e-05,
"loss": 0.0071,
"step": 24090
},
{
"grad_norm": 0.3094993531703949,
"learning_rate": 1.0210188535108995e-05,
"loss": 0.0122,
"step": 24100
},
{
"grad_norm": 0.2136320322751999,
"learning_rate": 1.0176836628708935e-05,
"loss": 0.0087,
"step": 24110
},
{
"grad_norm": 0.32316121459007263,
"learning_rate": 1.0143533111207454e-05,
"loss": 0.0068,
"step": 24120
},
{
"grad_norm": 0.23130331933498383,
"learning_rate": 1.0110278023071446e-05,
"loss": 0.016,
"step": 24130
},
{
"grad_norm": 0.2642119824886322,
"learning_rate": 1.007707140470901e-05,
"loss": 0.0152,
"step": 24140
},
{
"grad_norm": 0.3908197283744812,
"learning_rate": 1.0043913296469298e-05,
"loss": 0.0097,
"step": 24150
},
{
"grad_norm": 0.2090320587158203,
"learning_rate": 1.001080373864255e-05,
"loss": 0.01,
"step": 24160
},
{
"grad_norm": 0.285714328289032,
"learning_rate": 9.977742771459991e-06,
"loss": 0.0077,
"step": 24170
},
{
"grad_norm": 0.22159405052661896,
"learning_rate": 9.944730435093802e-06,
"loss": 0.0081,
"step": 24180
},
{
"grad_norm": 0.3165198564529419,
"learning_rate": 9.911766769657115e-06,
"loss": 0.0084,
"step": 24190
},
{
"grad_norm": 0.24157467484474182,
"learning_rate": 9.878851815203882e-06,
"loss": 0.0067,
"step": 24200
},
{
"grad_norm": 0.3204723298549652,
"learning_rate": 9.845985611728864e-06,
"loss": 0.0075,
"step": 24210
},
{
"grad_norm": 0.25934937596321106,
"learning_rate": 9.813168199167606e-06,
"loss": 0.0155,
"step": 24220
},
{
"grad_norm": 0.320283979177475,
"learning_rate": 9.780399617396341e-06,
"loss": 0.0106,
"step": 24230
},
{
"grad_norm": 0.22559820115566254,
"learning_rate": 9.747679906232016e-06,
"loss": 0.0081,
"step": 24240
},
{
"grad_norm": 0.31670987606048584,
"learning_rate": 9.715009105432138e-06,
"loss": 0.0109,
"step": 24250
},
{
"grad_norm": 0.33870166540145874,
"learning_rate": 9.682387254694835e-06,
"loss": 0.009,
"step": 24260
},
{
"grad_norm": 0.34368398785591125,
"learning_rate": 9.649814393658723e-06,
"loss": 0.0101,
"step": 24270
},
{
"grad_norm": 0.3852478861808777,
"learning_rate": 9.617290561902881e-06,
"loss": 0.0106,
"step": 24280
},
{
"grad_norm": 0.4689268171787262,
"learning_rate": 9.584815798946861e-06,
"loss": 0.0141,
"step": 24290
},
{
"grad_norm": 0.21284765005111694,
"learning_rate": 9.55239014425055e-06,
"loss": 0.0164,
"step": 24300
},
{
"grad_norm": 0.38204118609428406,
"learning_rate": 9.520013637214176e-06,
"loss": 0.0133,
"step": 24310
},
{
"grad_norm": 0.3181707561016083,
"learning_rate": 9.48768631717824e-06,
"loss": 0.0092,
"step": 24320
},
{
"grad_norm": 0.1879541277885437,
"learning_rate": 9.455408223423496e-06,
"loss": 0.0068,
"step": 24330
},
{
"grad_norm": 0.21481190621852875,
"learning_rate": 9.423179395170845e-06,
"loss": 0.012,
"step": 24340
},
{
"grad_norm": 0.29301491379737854,
"learning_rate": 9.390999871581391e-06,
"loss": 0.0138,
"step": 24350
},
{
"grad_norm": 0.23717325925827026,
"learning_rate": 9.358869691756273e-06,
"loss": 0.0088,
"step": 24360
},
{
"grad_norm": 0.20246140658855438,
"learning_rate": 9.326788894736688e-06,
"loss": 0.009,
"step": 24370
},
{
"grad_norm": 0.258135586977005,
"learning_rate": 9.294757519503811e-06,
"loss": 0.0071,
"step": 24380
},
{
"grad_norm": 0.1878809779882431,
"learning_rate": 9.262775604978819e-06,
"loss": 0.0063,
"step": 24390
},
{
"grad_norm": 0.34809255599975586,
"learning_rate": 9.230843190022725e-06,
"loss": 0.0093,
"step": 24400
},
{
"grad_norm": 0.28799232840538025,
"learning_rate": 9.198960313436445e-06,
"loss": 0.0078,
"step": 24410
},
{
"grad_norm": 0.31469273567199707,
"learning_rate": 9.16712701396067e-06,
"loss": 0.0099,
"step": 24420
},
{
"grad_norm": 0.3378365635871887,
"learning_rate": 9.135343330275864e-06,
"loss": 0.0111,
"step": 24430
},
{
"grad_norm": 0.28182870149612427,
"learning_rate": 9.10360930100218e-06,
"loss": 0.0058,
"step": 24440
},
{
"grad_norm": 0.385036826133728,
"learning_rate": 9.07192496469949e-06,
"loss": 0.0143,
"step": 24450
},
{
"grad_norm": 0.2500513792037964,
"learning_rate": 9.040290359867231e-06,
"loss": 0.01,
"step": 24460
},
{
"grad_norm": 0.29241782426834106,
"learning_rate": 9.008705524944438e-06,
"loss": 0.0121,
"step": 24470
},
{
"grad_norm": 0.23506666719913483,
"learning_rate": 8.977170498309651e-06,
"loss": 0.0084,
"step": 24480
},
{
"grad_norm": 0.311430960893631,
"learning_rate": 8.945685318280916e-06,
"loss": 0.0085,
"step": 24490
},
{
"grad_norm": 0.3923851549625397,
"learning_rate": 8.914250023115672e-06,
"loss": 0.0107,
"step": 24500
},
{
"grad_norm": 0.31323307752609253,
"learning_rate": 8.882864651010798e-06,
"loss": 0.0067,
"step": 24510
},
{
"grad_norm": 0.21854744851589203,
"learning_rate": 8.851529240102462e-06,
"loss": 0.013,
"step": 24520
},
{
"grad_norm": 0.25315526127815247,
"learning_rate": 8.820243828466134e-06,
"loss": 0.0068,
"step": 24530
},
{
"grad_norm": 0.3651484549045563,
"learning_rate": 8.789008454116566e-06,
"loss": 0.0079,
"step": 24540
},
{
"grad_norm": 0.2921467125415802,
"learning_rate": 8.757823155007655e-06,
"loss": 0.0111,
"step": 24550
},
{
"grad_norm": 0.1809811145067215,
"learning_rate": 8.72668796903251e-06,
"loss": 0.0063,
"step": 24560
},
{
"grad_norm": 0.23706315457820892,
"learning_rate": 8.695602934023301e-06,
"loss": 0.007,
"step": 24570
},
{
"grad_norm": 0.322356253862381,
"learning_rate": 8.664568087751274e-06,
"loss": 0.0096,
"step": 24580
},
{
"grad_norm": 0.26752379536628723,
"learning_rate": 8.633583467926697e-06,
"loss": 0.0086,
"step": 24590
},
{
"grad_norm": 0.16770339012145996,
"learning_rate": 8.602649112198796e-06,
"loss": 0.007,
"step": 24600
},
{
"grad_norm": 0.24515877664089203,
"learning_rate": 8.571765058155745e-06,
"loss": 0.0154,
"step": 24610
},
{
"grad_norm": 0.1980920135974884,
"learning_rate": 8.540931343324583e-06,
"loss": 0.0077,
"step": 24620
},
{
"grad_norm": 0.1964527666568756,
"learning_rate": 8.51014800517117e-06,
"loss": 0.0168,
"step": 24630
},
{
"grad_norm": 0.32992860674858093,
"learning_rate": 8.479415081100167e-06,
"loss": 0.0088,
"step": 24640
},
{
"grad_norm": 0.22021183371543884,
"learning_rate": 8.448732608454967e-06,
"loss": 0.0075,
"step": 24650
},
{
"grad_norm": 0.2648811936378479,
"learning_rate": 8.418100624517688e-06,
"loss": 0.011,
"step": 24660
},
{
"grad_norm": 0.259725421667099,
"learning_rate": 8.387519166509062e-06,
"loss": 0.0136,
"step": 24670
},
{
"grad_norm": 0.3602302074432373,
"learning_rate": 8.356988271588445e-06,
"loss": 0.0094,
"step": 24680
},
{
"grad_norm": 0.3162817358970642,
"learning_rate": 8.326507976853765e-06,
"loss": 0.0096,
"step": 24690
},
{
"grad_norm": 0.3423268795013428,
"learning_rate": 8.296078319341443e-06,
"loss": 0.0115,
"step": 24700
},
{
"grad_norm": 0.36489230394363403,
"learning_rate": 8.265699336026383e-06,
"loss": 0.0115,
"step": 24710
},
{
"grad_norm": 0.35590094327926636,
"learning_rate": 8.235371063821923e-06,
"loss": 0.0103,
"step": 24720
},
{
"grad_norm": 0.2977840304374695,
"learning_rate": 8.205093539579767e-06,
"loss": 0.0081,
"step": 24730
},
{
"grad_norm": 0.28951936960220337,
"learning_rate": 8.174866800089963e-06,
"loss": 0.009,
"step": 24740
},
{
"grad_norm": 0.3270035982131958,
"learning_rate": 8.144690882080853e-06,
"loss": 0.0105,
"step": 24750
},
{
"grad_norm": 0.351190447807312,
"learning_rate": 8.114565822219006e-06,
"loss": 0.0153,
"step": 24760
},
{
"grad_norm": 0.314729779958725,
"learning_rate": 8.084491657109233e-06,
"loss": 0.0108,
"step": 24770
},
{
"grad_norm": 0.35927537083625793,
"learning_rate": 8.054468423294476e-06,
"loss": 0.0112,
"step": 24780
},
{
"grad_norm": 0.22085341811180115,
"learning_rate": 8.024496157255785e-06,
"loss": 0.0089,
"step": 24790
},
{
"grad_norm": 0.2041126936674118,
"learning_rate": 7.994574895412293e-06,
"loss": 0.0062,
"step": 24800
},
{
"grad_norm": 0.2654230296611786,
"learning_rate": 7.964704674121149e-06,
"loss": 0.0086,
"step": 24810
},
{
"grad_norm": 0.25436100363731384,
"learning_rate": 7.934885529677505e-06,
"loss": 0.0086,
"step": 24820
},
{
"grad_norm": 0.203522190451622,
"learning_rate": 7.905117498314412e-06,
"loss": 0.0109,
"step": 24830
},
{
"grad_norm": 0.32665109634399414,
"learning_rate": 7.87540061620286e-06,
"loss": 0.0081,
"step": 24840
},
{
"grad_norm": 0.2379113733768463,
"learning_rate": 7.845734919451647e-06,
"loss": 0.0093,
"step": 24850
},
{
"grad_norm": 0.3950441777706146,
"learning_rate": 7.816120444107383e-06,
"loss": 0.0085,
"step": 24860
},
{
"grad_norm": 0.3268023431301117,
"learning_rate": 7.786557226154473e-06,
"loss": 0.0155,
"step": 24870
},
{
"grad_norm": 0.19815684854984283,
"learning_rate": 7.757045301514998e-06,
"loss": 0.0059,
"step": 24880
},
{
"grad_norm": 0.27587947249412537,
"learning_rate": 7.727584706048735e-06,
"loss": 0.0102,
"step": 24890
},
{
"grad_norm": 0.35937270522117615,
"learning_rate": 7.698175475553077e-06,
"loss": 0.0118,
"step": 24900
},
{
"grad_norm": 0.25024354457855225,
"learning_rate": 7.668817645763021e-06,
"loss": 0.0073,
"step": 24910
},
{
"grad_norm": 0.22915396094322205,
"learning_rate": 7.639511252351088e-06,
"loss": 0.0071,
"step": 24920
},
{
"grad_norm": 0.1825212687253952,
"learning_rate": 7.610256330927324e-06,
"loss": 0.0067,
"step": 24930
},
{
"grad_norm": 0.18780112266540527,
"learning_rate": 7.581052917039211e-06,
"loss": 0.0103,
"step": 24940
},
{
"grad_norm": 0.37713614106178284,
"learning_rate": 7.551901046171645e-06,
"loss": 0.0118,
"step": 24950
},
{
"grad_norm": 0.3918404281139374,
"learning_rate": 7.522800753746895e-06,
"loss": 0.0099,
"step": 24960
},
{
"grad_norm": 0.24894756078720093,
"learning_rate": 7.493752075124577e-06,
"loss": 0.0097,
"step": 24970
},
{
"grad_norm": 0.28078004717826843,
"learning_rate": 7.464755045601557e-06,
"loss": 0.0087,
"step": 24980
},
{
"grad_norm": 0.2317744940519333,
"learning_rate": 7.435809700411972e-06,
"loss": 0.0224,
"step": 24990
},
{
"grad_norm": 0.3898012936115265,
"learning_rate": 7.40691607472715e-06,
"loss": 0.0089,
"step": 25000
},
{
"grad_norm": 0.25833478569984436,
"learning_rate": 7.3780742036555606e-06,
"loss": 0.0107,
"step": 25010
},
{
"grad_norm": 0.34780773520469666,
"learning_rate": 7.349284122242783e-06,
"loss": 0.0101,
"step": 25020
},
{
"grad_norm": 0.28763628005981445,
"learning_rate": 7.320545865471512e-06,
"loss": 0.02,
"step": 25030
},
{
"grad_norm": 0.20789121091365814,
"learning_rate": 7.291859468261425e-06,
"loss": 0.008,
"step": 25040
},
{
"grad_norm": 0.24685008823871613,
"learning_rate": 7.263224965469195e-06,
"loss": 0.0096,
"step": 25050
},
{
"grad_norm": 0.285519003868103,
"learning_rate": 7.234642391888446e-06,
"loss": 0.0094,
"step": 25060
},
{
"grad_norm": 0.2459411919116974,
"learning_rate": 7.206111782249697e-06,
"loss": 0.009,
"step": 25070
},
{
"grad_norm": 0.18150289356708527,
"learning_rate": 7.1776331712203385e-06,
"loss": 0.0106,
"step": 25080
},
{
"grad_norm": 0.3443770110607147,
"learning_rate": 7.149206593404561e-06,
"loss": 0.0115,
"step": 25090
},
{
"grad_norm": 0.25675782561302185,
"learning_rate": 7.120832083343337e-06,
"loss": 0.0101,
"step": 25100
},
{
"grad_norm": 0.2560029923915863,
"learning_rate": 7.092509675514369e-06,
"loss": 0.0095,
"step": 25110
},
{
"grad_norm": 0.34635189175605774,
"learning_rate": 7.064239404332063e-06,
"loss": 0.0117,
"step": 25120
},
{
"grad_norm": 0.29154613614082336,
"learning_rate": 7.03602130414745e-06,
"loss": 0.0214,
"step": 25130
},
{
"grad_norm": 0.20932748913764954,
"learning_rate": 7.007855409248198e-06,
"loss": 0.0101,
"step": 25140
},
{
"grad_norm": 0.25224676728248596,
"learning_rate": 6.979741753858521e-06,
"loss": 0.0079,
"step": 25150
},
{
"grad_norm": 0.2278524786233902,
"learning_rate": 6.951680372139158e-06,
"loss": 0.0093,
"step": 25160
},
{
"grad_norm": 0.36583197116851807,
"learning_rate": 6.923671298187334e-06,
"loss": 0.0058,
"step": 25170
},
{
"grad_norm": 0.29362985491752625,
"learning_rate": 6.895714566036704e-06,
"loss": 0.0088,
"step": 25180
},
{
"grad_norm": 0.24151550233364105,
"learning_rate": 6.86781020965736e-06,
"loss": 0.0113,
"step": 25190
},
{
"grad_norm": 0.24941690266132355,
"learning_rate": 6.839958262955709e-06,
"loss": 0.0066,
"step": 25200
},
{
"grad_norm": 0.28753578662872314,
"learning_rate": 6.8121587597744894e-06,
"loss": 0.0145,
"step": 25210
},
{
"grad_norm": 0.2187775820493698,
"learning_rate": 6.784411733892732e-06,
"loss": 0.0099,
"step": 25220
},
{
"grad_norm": 0.344710111618042,
"learning_rate": 6.756717219025666e-06,
"loss": 0.0121,
"step": 25230
},
{
"grad_norm": 0.4404401183128357,
"learning_rate": 6.729075248824762e-06,
"loss": 0.008,
"step": 25240
},
{
"grad_norm": 0.20994271337985992,
"learning_rate": 6.701485856877615e-06,
"loss": 0.0055,
"step": 25250
},
{
"grad_norm": 0.2949792444705963,
"learning_rate": 6.673949076707925e-06,
"loss": 0.0064,
"step": 25260
},
{
"grad_norm": 0.23669622838497162,
"learning_rate": 6.646464941775499e-06,
"loss": 0.0092,
"step": 25270
},
{
"grad_norm": 0.25692516565322876,
"learning_rate": 6.619033485476128e-06,
"loss": 0.0096,
"step": 25280
},
{
"grad_norm": 0.26228588819503784,
"learning_rate": 6.59165474114164e-06,
"loss": 0.0092,
"step": 25290
},
{
"grad_norm": 0.24011710286140442,
"learning_rate": 6.564328742039783e-06,
"loss": 0.0119,
"step": 25300
},
{
"grad_norm": 0.3565312922000885,
"learning_rate": 6.537055521374219e-06,
"loss": 0.0093,
"step": 25310
},
{
"grad_norm": 0.30972421169281006,
"learning_rate": 6.509835112284485e-06,
"loss": 0.0071,
"step": 25320
},
{
"grad_norm": 0.28988102078437805,
"learning_rate": 6.482667547845944e-06,
"loss": 0.0165,
"step": 25330
},
{
"grad_norm": 0.23721013963222504,
"learning_rate": 6.455552861069736e-06,
"loss": 0.0065,
"step": 25340
},
{
"grad_norm": 0.3081379532814026,
"learning_rate": 6.428491084902788e-06,
"loss": 0.0086,
"step": 25350
},
{
"grad_norm": 0.35128065943717957,
"learning_rate": 6.401482252227698e-06,
"loss": 0.0084,
"step": 25360
},
{
"grad_norm": 0.3011689782142639,
"learning_rate": 6.37452639586274e-06,
"loss": 0.0106,
"step": 25370
},
{
"grad_norm": 0.20033995807170868,
"learning_rate": 6.347623548561826e-06,
"loss": 0.0065,
"step": 25380
},
{
"grad_norm": 0.18873929977416992,
"learning_rate": 6.320773743014441e-06,
"loss": 0.0076,
"step": 25390
},
{
"grad_norm": 0.2524699568748474,
"learning_rate": 6.293977011845648e-06,
"loss": 0.0075,
"step": 25400
},
{
"grad_norm": 0.3948061764240265,
"learning_rate": 6.2672333876159835e-06,
"loss": 0.0087,
"step": 25410
},
{
"grad_norm": 0.3653682768344879,
"learning_rate": 6.2405429028215e-06,
"loss": 0.0092,
"step": 25420
},
{
"grad_norm": 0.44343745708465576,
"learning_rate": 6.213905589893631e-06,
"loss": 0.0127,
"step": 25430
},
{
"grad_norm": 0.2481200397014618,
"learning_rate": 6.187321481199221e-06,
"loss": 0.0111,
"step": 25440
},
{
"grad_norm": 0.20366159081459045,
"learning_rate": 6.1607906090404895e-06,
"loss": 0.0186,
"step": 25450
},
{
"grad_norm": 0.4059971570968628,
"learning_rate": 6.134313005654929e-06,
"loss": 0.0101,
"step": 25460
},
{
"grad_norm": 0.4176698923110962,
"learning_rate": 6.107888703215337e-06,
"loss": 0.0147,
"step": 25470
},
{
"grad_norm": 0.3053951859474182,
"learning_rate": 6.081517733829722e-06,
"loss": 0.0079,
"step": 25480
},
{
"grad_norm": 0.21016788482666016,
"learning_rate": 6.055200129541294e-06,
"loss": 0.0071,
"step": 25490
},
{
"grad_norm": 0.3120420575141907,
"learning_rate": 6.0289359223284434e-06,
"loss": 0.011,
"step": 25500
},
{
"grad_norm": 0.19099359214305878,
"learning_rate": 6.002725144104648e-06,
"loss": 0.0094,
"step": 25510
},
{
"grad_norm": 0.3218839466571808,
"learning_rate": 5.976567826718477e-06,
"loss": 0.0085,
"step": 25520
},
{
"grad_norm": 0.2995322048664093,
"learning_rate": 5.9504640019535315e-06,
"loss": 0.0139,
"step": 25530
},
{
"grad_norm": 0.4251338243484497,
"learning_rate": 5.924413701528414e-06,
"loss": 0.0109,
"step": 25540
},
{
"grad_norm": 0.35200032591819763,
"learning_rate": 5.898416957096703e-06,
"loss": 0.0089,
"step": 25550
},
{
"grad_norm": 0.32996129989624023,
"learning_rate": 5.872473800246913e-06,
"loss": 0.0106,
"step": 25560
},
{
"grad_norm": 0.42911210656166077,
"learning_rate": 5.846584262502402e-06,
"loss": 0.0101,
"step": 25570
},
{
"grad_norm": 0.17232227325439453,
"learning_rate": 5.820748375321411e-06,
"loss": 0.0087,
"step": 25580
},
{
"grad_norm": 0.17928582429885864,
"learning_rate": 5.794966170096977e-06,
"loss": 0.0102,
"step": 25590
},
{
"grad_norm": 0.2240273803472519,
"learning_rate": 5.769237678156897e-06,
"loss": 0.0065,
"step": 25600
},
{
"grad_norm": 0.20052582025527954,
"learning_rate": 5.743562930763735e-06,
"loss": 0.0122,
"step": 25610
},
{
"grad_norm": 0.2960312068462372,
"learning_rate": 5.717941959114726e-06,
"loss": 0.0099,
"step": 25620
},
{
"grad_norm": 0.34251827001571655,
"learning_rate": 5.69237479434176e-06,
"loss": 0.0093,
"step": 25630
},
{
"grad_norm": 0.33590638637542725,
"learning_rate": 5.666861467511353e-06,
"loss": 0.0064,
"step": 25640
},
{
"grad_norm": 0.3605147898197174,
"learning_rate": 5.641402009624591e-06,
"loss": 0.0122,
"step": 25650
},
{
"grad_norm": 0.16599352657794952,
"learning_rate": 5.615996451617145e-06,
"loss": 0.0051,
"step": 25660
},
{
"grad_norm": 0.23967251181602478,
"learning_rate": 5.590644824359148e-06,
"loss": 0.0153,
"step": 25670
},
{
"grad_norm": 0.2549436390399933,
"learning_rate": 5.56534715865521e-06,
"loss": 0.0068,
"step": 25680
},
{
"grad_norm": 0.20421919226646423,
"learning_rate": 5.5401034852443965e-06,
"loss": 0.0084,
"step": 25690
},
{
"grad_norm": 0.33924439549446106,
"learning_rate": 5.514913834800134e-06,
"loss": 0.0089,
"step": 25700
},
{
"grad_norm": 0.30444881319999695,
"learning_rate": 5.489778237930238e-06,
"loss": 0.0088,
"step": 25710
},
{
"grad_norm": 0.24199800193309784,
"learning_rate": 5.464696725176827e-06,
"loss": 0.0065,
"step": 25720
},
{
"grad_norm": 0.2547299265861511,
"learning_rate": 5.439669327016294e-06,
"loss": 0.0104,
"step": 25730
},
{
"grad_norm": 0.29023823142051697,
"learning_rate": 5.4146960738592985e-06,
"loss": 0.0074,
"step": 25740
},
{
"grad_norm": 0.2345302402973175,
"learning_rate": 5.389776996050694e-06,
"loss": 0.0075,
"step": 25750
},
{
"grad_norm": 0.29668354988098145,
"learning_rate": 5.364912123869492e-06,
"loss": 0.0086,
"step": 25760
},
{
"grad_norm": 0.19641797244548798,
"learning_rate": 5.3401014875288864e-06,
"loss": 0.0075,
"step": 25770
},
{
"grad_norm": 0.23984096944332123,
"learning_rate": 5.315345117176129e-06,
"loss": 0.0058,
"step": 25780
},
{
"grad_norm": 0.2512895166873932,
"learning_rate": 5.2906430428925405e-06,
"loss": 0.0085,
"step": 25790
},
{
"grad_norm": 0.21854522824287415,
"learning_rate": 5.265995294693471e-06,
"loss": 0.0082,
"step": 25800
},
{
"grad_norm": 0.2669749855995178,
"learning_rate": 5.241401902528254e-06,
"loss": 0.0071,
"step": 25810
},
{
"grad_norm": 0.15877646207809448,
"learning_rate": 5.216862896280189e-06,
"loss": 0.0057,
"step": 25820
},
{
"grad_norm": 0.22950813174247742,
"learning_rate": 5.192378305766471e-06,
"loss": 0.0078,
"step": 25830
},
{
"grad_norm": 0.2555539906024933,
"learning_rate": 5.1679481607382065e-06,
"loss": 0.0119,
"step": 25840
},
{
"grad_norm": 0.20202337205410004,
"learning_rate": 5.1435724908803105e-06,
"loss": 0.009,
"step": 25850
},
{
"grad_norm": 0.20334777235984802,
"learning_rate": 5.119251325811514e-06,
"loss": 0.0072,
"step": 25860
},
{
"grad_norm": 0.20359674096107483,
"learning_rate": 5.094984695084348e-06,
"loss": 0.0084,
"step": 25870
},
{
"grad_norm": 0.31620484590530396,
"learning_rate": 5.070772628185039e-06,
"loss": 0.0168,
"step": 25880
},
{
"grad_norm": 0.2068619579076767,
"learning_rate": 5.046615154533535e-06,
"loss": 0.0085,
"step": 25890
},
{
"grad_norm": 0.1954227089881897,
"learning_rate": 5.022512303483451e-06,
"loss": 0.006,
"step": 25900
},
{
"grad_norm": 0.20723772048950195,
"learning_rate": 4.998464104322015e-06,
"loss": 0.0152,
"step": 25910
},
{
"grad_norm": 0.35729050636291504,
"learning_rate": 4.974470586270047e-06,
"loss": 0.009,
"step": 25920
},
{
"grad_norm": 0.16327857971191406,
"learning_rate": 4.950531778481965e-06,
"loss": 0.0075,
"step": 25930
},
{
"grad_norm": 0.2857097387313843,
"learning_rate": 4.926647710045651e-06,
"loss": 0.0055,
"step": 25940
},
{
"grad_norm": 0.23355622589588165,
"learning_rate": 4.9028184099825125e-06,
"loss": 0.0099,
"step": 25950
},
{
"grad_norm": 0.16490572690963745,
"learning_rate": 4.879043907247382e-06,
"loss": 0.0116,
"step": 25960
},
{
"grad_norm": 0.28257453441619873,
"learning_rate": 4.8553242307285414e-06,
"loss": 0.009,
"step": 25970
},
{
"grad_norm": 0.3403509259223938,
"learning_rate": 4.8316594092476185e-06,
"loss": 0.0086,
"step": 25980
},
{
"grad_norm": 0.20419685542583466,
"learning_rate": 4.808049471559617e-06,
"loss": 0.0072,
"step": 25990
},
{
"grad_norm": 0.34411945939064026,
"learning_rate": 4.784494446352833e-06,
"loss": 0.0086,
"step": 26000
},
{
"grad_norm": 0.4460735321044922,
"learning_rate": 4.760994362248833e-06,
"loss": 0.0114,
"step": 26010
},
{
"grad_norm": 0.19788499176502228,
"learning_rate": 4.737549247802442e-06,
"loss": 0.0074,
"step": 26020
},
{
"grad_norm": 0.3280922770500183,
"learning_rate": 4.714159131501689e-06,
"loss": 0.0107,
"step": 26030
},
{
"grad_norm": 0.2518462836742401,
"learning_rate": 4.690824041767766e-06,
"loss": 0.0113,
"step": 26040
},
{
"grad_norm": 0.2995360493659973,
"learning_rate": 4.6675440069550105e-06,
"loss": 0.014,
"step": 26050
},
{
"grad_norm": 0.2634023129940033,
"learning_rate": 4.64431905535086e-06,
"loss": 0.0066,
"step": 26060
},
{
"grad_norm": 0.28187295794487,
"learning_rate": 4.621149215175808e-06,
"loss": 0.0131,
"step": 26070
},
{
"grad_norm": 0.33387911319732666,
"learning_rate": 4.5980345145834154e-06,
"loss": 0.02,
"step": 26080
},
{
"grad_norm": 0.2211586833000183,
"learning_rate": 4.574974981660213e-06,
"loss": 0.0129,
"step": 26090
},
{
"grad_norm": 0.26694872975349426,
"learning_rate": 4.551970644425707e-06,
"loss": 0.011,
"step": 26100
},
{
"grad_norm": 0.18771067261695862,
"learning_rate": 4.529021530832328e-06,
"loss": 0.0063,
"step": 26110
},
{
"grad_norm": 0.2316020429134369,
"learning_rate": 4.5061276687654306e-06,
"loss": 0.0078,
"step": 26120
},
{
"grad_norm": 0.23546847701072693,
"learning_rate": 4.4832890860431965e-06,
"loss": 0.0109,
"step": 26130
},
{
"grad_norm": 0.27060621976852417,
"learning_rate": 4.460505810416682e-06,
"loss": 0.0086,
"step": 26140
},
{
"grad_norm": 0.29783564805984497,
"learning_rate": 4.4377778695696984e-06,
"loss": 0.0172,
"step": 26150
},
{
"grad_norm": 0.21668773889541626,
"learning_rate": 4.415105291118843e-06,
"loss": 0.0068,
"step": 26160
},
{
"grad_norm": 0.2789166271686554,
"learning_rate": 4.392488102613435e-06,
"loss": 0.0091,
"step": 26170
},
{
"grad_norm": 0.2664152681827545,
"learning_rate": 4.369926331535473e-06,
"loss": 0.0088,
"step": 26180
},
{
"grad_norm": 0.3760206401348114,
"learning_rate": 4.347420005299668e-06,
"loss": 0.0097,
"step": 26190
},
{
"grad_norm": 0.23559440672397614,
"learning_rate": 4.324969151253317e-06,
"loss": 0.0088,
"step": 26200
},
{
"grad_norm": 0.36386236548423767,
"learning_rate": 4.302573796676312e-06,
"loss": 0.0127,
"step": 26210
},
{
"grad_norm": 0.2530874013900757,
"learning_rate": 4.2802339687811385e-06,
"loss": 0.0099,
"step": 26220
},
{
"grad_norm": 0.29895782470703125,
"learning_rate": 4.257949694712771e-06,
"loss": 0.0107,
"step": 26230
},
{
"grad_norm": 0.33973217010498047,
"learning_rate": 4.235721001548726e-06,
"loss": 0.0079,
"step": 26240
},
{
"grad_norm": 0.19557560980319977,
"learning_rate": 4.21354791629896e-06,
"loss": 0.0132,
"step": 26250
},
{
"grad_norm": 0.31418323516845703,
"learning_rate": 4.191430465905843e-06,
"loss": 0.0144,
"step": 26260
},
{
"grad_norm": 0.21686318516731262,
"learning_rate": 4.169368677244184e-06,
"loss": 0.0072,
"step": 26270
},
{
"grad_norm": 0.2362222969532013,
"learning_rate": 4.147362577121122e-06,
"loss": 0.0202,
"step": 26280
},
{
"grad_norm": 0.21339112520217896,
"learning_rate": 4.125412192276157e-06,
"loss": 0.0178,
"step": 26290
},
{
"grad_norm": 0.277231901884079,
"learning_rate": 4.10351754938107e-06,
"loss": 0.0133,
"step": 26300
},
{
"grad_norm": 0.29276028275489807,
"learning_rate": 4.081678675039913e-06,
"loss": 0.0216,
"step": 26310
},
{
"grad_norm": 0.22694578766822815,
"learning_rate": 4.0598955957889876e-06,
"loss": 0.0071,
"step": 26320
},
{
"grad_norm": 0.22576840221881866,
"learning_rate": 4.038168338096776e-06,
"loss": 0.0099,
"step": 26330
},
{
"grad_norm": 0.30577895045280457,
"learning_rate": 4.016496928363944e-06,
"loss": 0.0088,
"step": 26340
},
{
"grad_norm": 0.24181580543518066,
"learning_rate": 3.994881392923317e-06,
"loss": 0.0078,
"step": 26350
},
{
"grad_norm": 0.2284122109413147,
"learning_rate": 3.9733217580397945e-06,
"loss": 0.0122,
"step": 26360
},
{
"grad_norm": 0.16746746003627777,
"learning_rate": 3.951818049910366e-06,
"loss": 0.0066,
"step": 26370
},
{
"grad_norm": 0.2788512408733368,
"learning_rate": 3.930370294664071e-06,
"loss": 0.0087,
"step": 26380
},
{
"grad_norm": 0.22725024819374084,
"learning_rate": 3.908978518361938e-06,
"loss": 0.0114,
"step": 26390
},
{
"grad_norm": 0.2820492088794708,
"learning_rate": 3.887642746997017e-06,
"loss": 0.01,
"step": 26400
},
{
"grad_norm": 0.21025604009628296,
"learning_rate": 3.866363006494255e-06,
"loss": 0.007,
"step": 26410
},
{
"grad_norm": 0.23573805391788483,
"learning_rate": 3.8451393227105725e-06,
"loss": 0.006,
"step": 26420
},
{
"grad_norm": 0.2777254581451416,
"learning_rate": 3.823971721434727e-06,
"loss": 0.0122,
"step": 26430
},
{
"grad_norm": 0.17709200084209442,
"learning_rate": 3.8028602283873505e-06,
"loss": 0.0096,
"step": 26440
},
{
"grad_norm": 0.22949759662151337,
"learning_rate": 3.781804869220912e-06,
"loss": 0.0119,
"step": 26450
},
{
"grad_norm": 0.24804015457630157,
"learning_rate": 3.760805669519646e-06,
"loss": 0.0096,
"step": 26460
},
{
"grad_norm": 0.33381086587905884,
"learning_rate": 3.7398626547995587e-06,
"loss": 0.0082,
"step": 26470
},
{
"grad_norm": 0.2890039086341858,
"learning_rate": 3.7189758505083995e-06,
"loss": 0.0068,
"step": 26480
},
{
"grad_norm": 0.21154865622520447,
"learning_rate": 3.6981452820255837e-06,
"loss": 0.0084,
"step": 26490
},
{
"grad_norm": 0.35219600796699524,
"learning_rate": 3.6773709746622363e-06,
"loss": 0.0111,
"step": 26500
},
{
"grad_norm": 0.15033812820911407,
"learning_rate": 3.6566529536610995e-06,
"loss": 0.02,
"step": 26510
},
{
"grad_norm": 0.17892305552959442,
"learning_rate": 3.635991244196513e-06,
"loss": 0.0091,
"step": 26520
},
{
"grad_norm": 0.23761463165283203,
"learning_rate": 3.615385871374405e-06,
"loss": 0.0104,
"step": 26530
},
{
"grad_norm": 0.33376213908195496,
"learning_rate": 3.5948368602322367e-06,
"loss": 0.0081,
"step": 26540
},
{
"grad_norm": 0.4278036952018738,
"learning_rate": 3.5743442357390187e-06,
"loss": 0.0082,
"step": 26550
},
{
"grad_norm": 0.3044934868812561,
"learning_rate": 3.5539080227951937e-06,
"loss": 0.0095,
"step": 26560
},
{
"grad_norm": 0.14772175252437592,
"learning_rate": 3.533528246232709e-06,
"loss": 0.0069,
"step": 26570
},
{
"grad_norm": 0.29768723249435425,
"learning_rate": 3.5132049308149117e-06,
"loss": 0.0066,
"step": 26580
},
{
"grad_norm": 0.26002663373947144,
"learning_rate": 3.4929381012365425e-06,
"loss": 0.0078,
"step": 26590
},
{
"grad_norm": 0.26255571842193604,
"learning_rate": 3.472727782123697e-06,
"loss": 0.0111,
"step": 26600
},
{
"grad_norm": 0.20837272703647614,
"learning_rate": 3.4525739980338425e-06,
"loss": 0.0072,
"step": 26610
},
{
"grad_norm": 0.17213797569274902,
"learning_rate": 3.432476773455712e-06,
"loss": 0.012,
"step": 26620
},
{
"grad_norm": 0.2694096863269806,
"learning_rate": 3.4124361328093378e-06,
"loss": 0.0085,
"step": 26630
},
{
"grad_norm": 0.25780072808265686,
"learning_rate": 3.392452100445975e-06,
"loss": 0.0086,
"step": 26640
},
{
"grad_norm": 0.1880541443824768,
"learning_rate": 3.3725247006481163e-06,
"loss": 0.0058,
"step": 26650
},
{
"grad_norm": 0.2633858621120453,
"learning_rate": 3.3526539576294313e-06,
"loss": 0.012,
"step": 26660
},
{
"grad_norm": 0.214751198887825,
"learning_rate": 3.332839895534745e-06,
"loss": 0.0066,
"step": 26670
},
{
"grad_norm": 0.3085266351699829,
"learning_rate": 3.3130825384400155e-06,
"loss": 0.0097,
"step": 26680
},
{
"grad_norm": 0.22667960822582245,
"learning_rate": 3.293381910352278e-06,
"loss": 0.0076,
"step": 26690
},
{
"grad_norm": 0.38128751516342163,
"learning_rate": 3.273738035209678e-06,
"loss": 0.0176,
"step": 26700
},
{
"grad_norm": 0.46959632635116577,
"learning_rate": 3.254150936881356e-06,
"loss": 0.011,
"step": 26710
},
{
"grad_norm": 0.2346912920475006,
"learning_rate": 3.234620639167496e-06,
"loss": 0.0065,
"step": 26720
},
{
"grad_norm": 0.17337638139724731,
"learning_rate": 3.2151471657992482e-06,
"loss": 0.0072,
"step": 26730
},
{
"grad_norm": 0.24507002532482147,
"learning_rate": 3.195730540438718e-06,
"loss": 0.0066,
"step": 26740
},
{
"grad_norm": 0.2788914144039154,
"learning_rate": 3.1763707866789336e-06,
"loss": 0.0125,
"step": 26750
},
{
"grad_norm": 0.2104663848876953,
"learning_rate": 3.1570679280438165e-06,
"loss": 0.0089,
"step": 26760
},
{
"grad_norm": 0.3239039480686188,
"learning_rate": 3.1378219879881767e-06,
"loss": 0.0072,
"step": 26770
},
{
"grad_norm": 0.192292720079422,
"learning_rate": 3.11863298989763e-06,
"loss": 0.009,
"step": 26780
},
{
"grad_norm": 0.2783186137676239,
"learning_rate": 3.0995009570886303e-06,
"loss": 0.0104,
"step": 26790
},
{
"grad_norm": 0.29066213965415955,
"learning_rate": 3.0804259128083936e-06,
"loss": 0.008,
"step": 26800
},
{
"grad_norm": 0.143081396818161,
"learning_rate": 3.06140788023489e-06,
"loss": 0.0072,
"step": 26810
},
{
"grad_norm": 0.27613288164138794,
"learning_rate": 3.042446882476846e-06,
"loss": 0.0072,
"step": 26820
},
{
"grad_norm": 0.18693867325782776,
"learning_rate": 3.023542942573643e-06,
"loss": 0.0073,
"step": 26830
},
{
"grad_norm": 0.3633919060230255,
"learning_rate": 3.004696083495351e-06,
"loss": 0.0151,
"step": 26840
},
{
"grad_norm": 0.33856984972953796,
"learning_rate": 2.98590632814269e-06,
"loss": 0.0109,
"step": 26850
},
{
"grad_norm": 0.11361749470233917,
"learning_rate": 2.9671736993469702e-06,
"loss": 0.008,
"step": 26860
},
{
"grad_norm": 0.3445863425731659,
"learning_rate": 2.9484982198701216e-06,
"loss": 0.0079,
"step": 26870
},
{
"grad_norm": 0.3401656746864319,
"learning_rate": 2.929879912404604e-06,
"loss": 0.0094,
"step": 26880
},
{
"grad_norm": 0.1342494934797287,
"learning_rate": 2.9113187995734147e-06,
"loss": 0.0096,
"step": 26890
},
{
"grad_norm": 0.29594022035598755,
"learning_rate": 2.8928149039300524e-06,
"loss": 0.007,
"step": 26900
},
{
"grad_norm": 0.148177370429039,
"learning_rate": 2.8743682479584975e-06,
"loss": 0.0101,
"step": 26910
},
{
"grad_norm": 0.4018382728099823,
"learning_rate": 2.8559788540731824e-06,
"loss": 0.0123,
"step": 26920
},
{
"grad_norm": 0.2794182300567627,
"learning_rate": 2.8376467446189493e-06,
"loss": 0.0077,
"step": 26930
},
{
"grad_norm": 0.43325501680374146,
"learning_rate": 2.8193719418710406e-06,
"loss": 0.0181,
"step": 26940
},
{
"grad_norm": 0.23830386996269226,
"learning_rate": 2.8011544680350667e-06,
"loss": 0.0112,
"step": 26950
},
{
"grad_norm": 0.19145773351192474,
"learning_rate": 2.7829943452469753e-06,
"loss": 0.007,
"step": 26960
},
{
"grad_norm": 0.2257826030254364,
"learning_rate": 2.7648915955730213e-06,
"loss": 0.0139,
"step": 26970
},
{
"grad_norm": 0.1542842835187912,
"learning_rate": 2.746846241009765e-06,
"loss": 0.0051,
"step": 26980
},
{
"grad_norm": 0.24472612142562866,
"learning_rate": 2.7288583034839945e-06,
"loss": 0.0102,
"step": 26990
},
{
"grad_norm": 0.16731634736061096,
"learning_rate": 2.7109278048527752e-06,
"loss": 0.0066,
"step": 27000
},
{
"grad_norm": 0.18837502598762512,
"learning_rate": 2.6930547669033413e-06,
"loss": 0.0056,
"step": 27010
},
{
"grad_norm": 0.2425365298986435,
"learning_rate": 2.675239211353109e-06,
"loss": 0.0075,
"step": 27020
},
{
"grad_norm": 0.1956389844417572,
"learning_rate": 2.6574811598496786e-06,
"loss": 0.0061,
"step": 27030
},
{
"grad_norm": 0.24185949563980103,
"learning_rate": 2.6397806339707456e-06,
"loss": 0.0162,
"step": 27040
},
{
"grad_norm": 0.34461912512779236,
"learning_rate": 2.6221376552241217e-06,
"loss": 0.0064,
"step": 27050
},
{
"grad_norm": 0.17602357268333435,
"learning_rate": 2.604552245047681e-06,
"loss": 0.0062,
"step": 27060
},
{
"grad_norm": 0.16017012298107147,
"learning_rate": 2.587024424809359e-06,
"loss": 0.0072,
"step": 27070
},
{
"grad_norm": 0.2674061357975006,
"learning_rate": 2.5695542158071186e-06,
"loss": 0.0087,
"step": 27080
},
{
"grad_norm": 0.22055087983608246,
"learning_rate": 2.5521416392689068e-06,
"loss": 0.0072,
"step": 27090
},
{
"grad_norm": 0.2661854922771454,
"learning_rate": 2.5347867163526384e-06,
"loss": 0.0065,
"step": 27100
},
{
"grad_norm": 0.3940221071243286,
"learning_rate": 2.517489468146189e-06,
"loss": 0.0095,
"step": 27110
},
{
"grad_norm": 0.2543990910053253,
"learning_rate": 2.500249915667341e-06,
"loss": 0.0082,
"step": 27120
},
{
"grad_norm": 0.23193122446537018,
"learning_rate": 2.4830680798637817e-06,
"loss": 0.0063,
"step": 27130
},
{
"grad_norm": 0.23208947479724884,
"learning_rate": 2.465943981613056e-06,
"loss": 0.0145,
"step": 27140
},
{
"grad_norm": 0.22675101459026337,
"learning_rate": 2.448877641722569e-06,
"loss": 0.0072,
"step": 27150
},
{
"grad_norm": 0.19925850629806519,
"learning_rate": 2.431869080929522e-06,
"loss": 0.007,
"step": 27160
},
{
"grad_norm": 0.26392441987991333,
"learning_rate": 2.4149183199009216e-06,
"loss": 0.0123,
"step": 27170
},
{
"grad_norm": 0.4779159426689148,
"learning_rate": 2.3980253792335427e-06,
"loss": 0.014,
"step": 27180
},
{
"grad_norm": 0.22511422634124756,
"learning_rate": 2.381190279453899e-06,
"loss": 0.0091,
"step": 27190
},
{
"grad_norm": 0.2878941297531128,
"learning_rate": 2.364413041018232e-06,
"loss": 0.0166,
"step": 27200
},
{
"grad_norm": 0.2500768005847931,
"learning_rate": 2.347693684312463e-06,
"loss": 0.0069,
"step": 27210
},
{
"grad_norm": 0.2809790074825287,
"learning_rate": 2.331032229652186e-06,
"loss": 0.0068,
"step": 27220
},
{
"grad_norm": 0.22055134177207947,
"learning_rate": 2.314428697282628e-06,
"loss": 0.0068,
"step": 27230
},
{
"grad_norm": 0.12686477601528168,
"learning_rate": 2.297883107378673e-06,
"loss": 0.007,
"step": 27240
},
{
"grad_norm": 0.35691311955451965,
"learning_rate": 2.2813954800447513e-06,
"loss": 0.0148,
"step": 27250
},
{
"grad_norm": 0.25910744071006775,
"learning_rate": 2.2649658353148974e-06,
"loss": 0.0062,
"step": 27260
},
{
"grad_norm": 0.21527698636054993,
"learning_rate": 2.2485941931526645e-06,
"loss": 0.0043,
"step": 27270
},
{
"grad_norm": 0.32111385464668274,
"learning_rate": 2.232280573451151e-06,
"loss": 0.0078,
"step": 27280
},
{
"grad_norm": 0.17319099605083466,
"learning_rate": 2.216024996032945e-06,
"loss": 0.0054,
"step": 27290
},
{
"grad_norm": 0.18256959319114685,
"learning_rate": 2.1998274806501074e-06,
"loss": 0.0051,
"step": 27300
},
{
"grad_norm": 0.33494529128074646,
"learning_rate": 2.183688046984139e-06,
"loss": 0.009,
"step": 27310
},
{
"grad_norm": 0.1944282501935959,
"learning_rate": 2.167606714645981e-06,
"loss": 0.0081,
"step": 27320
},
{
"grad_norm": 0.19519659876823425,
"learning_rate": 2.151583503175958e-06,
"loss": 0.0049,
"step": 27330
},
{
"grad_norm": 0.18572774529457092,
"learning_rate": 2.1356184320437957e-06,
"loss": 0.0074,
"step": 27340
},
{
"grad_norm": 0.19207584857940674,
"learning_rate": 2.119711520648554e-06,
"loss": 0.0138,
"step": 27350
},
{
"grad_norm": 0.2535437047481537,
"learning_rate": 2.103862788318628e-06,
"loss": 0.0123,
"step": 27360
},
{
"grad_norm": 0.1590602844953537,
"learning_rate": 2.088072254311729e-06,
"loss": 0.0062,
"step": 27370
},
{
"grad_norm": 0.1926662176847458,
"learning_rate": 2.0723399378148435e-06,
"loss": 0.0077,
"step": 27380
},
{
"grad_norm": 0.22575537860393524,
"learning_rate": 2.0566658579442065e-06,
"loss": 0.0072,
"step": 27390
},
{
"grad_norm": 0.2069316953420639,
"learning_rate": 2.0410500337453176e-06,
"loss": 0.0056,
"step": 27400
},
{
"grad_norm": 0.1546366959810257,
"learning_rate": 2.0254924841928645e-06,
"loss": 0.0101,
"step": 27410
},
{
"grad_norm": 0.2288369983434677,
"learning_rate": 2.009993228190754e-06,
"loss": 0.0077,
"step": 27420
},
{
"grad_norm": 0.38559532165527344,
"learning_rate": 1.9945522845720323e-06,
"loss": 0.0094,
"step": 27430
},
{
"grad_norm": 0.35795527696609497,
"learning_rate": 1.9791696720988963e-06,
"loss": 0.0147,
"step": 27440
},
{
"grad_norm": 0.1973767727613449,
"learning_rate": 1.9638454094626833e-06,
"loss": 0.0091,
"step": 27450
},
{
"grad_norm": 0.18512430787086487,
"learning_rate": 1.9485795152838104e-06,
"loss": 0.0083,
"step": 27460
},
{
"grad_norm": 0.3037535548210144,
"learning_rate": 1.933372008111778e-06,
"loss": 0.0157,
"step": 27470
},
{
"grad_norm": 0.35140731930732727,
"learning_rate": 1.918222906425143e-06,
"loss": 0.0113,
"step": 27480
},
{
"grad_norm": 0.2775219678878784,
"learning_rate": 1.903132228631488e-06,
"loss": 0.0108,
"step": 27490
},
{
"grad_norm": 0.25528013706207275,
"learning_rate": 1.8880999930674215e-06,
"loss": 0.0155,
"step": 27500
},
{
"grad_norm": 0.3439268469810486,
"learning_rate": 1.8731262179985166e-06,
"loss": 0.0125,
"step": 27510
},
{
"grad_norm": 0.18030324578285217,
"learning_rate": 1.8582109216193244e-06,
"loss": 0.017,
"step": 27520
},
{
"grad_norm": 0.24032104015350342,
"learning_rate": 1.8433541220533368e-06,
"loss": 0.011,
"step": 27530
},
{
"grad_norm": 0.23355452716350555,
"learning_rate": 1.8285558373529577e-06,
"loss": 0.0069,
"step": 27540
},
{
"grad_norm": 0.22409330308437347,
"learning_rate": 1.8138160854995145e-06,
"loss": 0.0103,
"step": 27550
},
{
"grad_norm": 0.11287908256053925,
"learning_rate": 1.7991348844031864e-06,
"loss": 0.0059,
"step": 27560
},
{
"grad_norm": 0.3050505220890045,
"learning_rate": 1.78451225190302e-06,
"loss": 0.0086,
"step": 27570
},
{
"grad_norm": 0.20751787722110748,
"learning_rate": 1.7699482057668914e-06,
"loss": 0.0065,
"step": 27580
},
{
"grad_norm": 0.21502567827701569,
"learning_rate": 1.7554427636914895e-06,
"loss": 0.0072,
"step": 27590
},
{
"grad_norm": 0.29134464263916016,
"learning_rate": 1.7409959433022871e-06,
"loss": 0.009,
"step": 27600
},
{
"grad_norm": 0.3173362910747528,
"learning_rate": 1.7266077621535481e-06,
"loss": 0.0146,
"step": 27610
},
{
"grad_norm": 0.2597872018814087,
"learning_rate": 1.7122782377282597e-06,
"loss": 0.0076,
"step": 27620
},
{
"grad_norm": 0.14644448459148407,
"learning_rate": 1.6980073874381497e-06,
"loss": 0.0072,
"step": 27630
},
{
"grad_norm": 0.15357452630996704,
"learning_rate": 1.6837952286236413e-06,
"loss": 0.0075,
"step": 27640
},
{
"grad_norm": 0.26090824604034424,
"learning_rate": 1.6696417785538488e-06,
"loss": 0.0155,
"step": 27650
},
{
"grad_norm": 0.20437179505825043,
"learning_rate": 1.6555470544265538e-06,
"loss": 0.0176,
"step": 27660
},
{
"grad_norm": 0.21696539223194122,
"learning_rate": 1.6415110733681738e-06,
"loss": 0.0117,
"step": 27670
},
{
"grad_norm": 0.34526702761650085,
"learning_rate": 1.6275338524337436e-06,
"loss": 0.0112,
"step": 27680
},
{
"grad_norm": 0.18506178259849548,
"learning_rate": 1.6136154086069056e-06,
"loss": 0.0061,
"step": 27690
},
{
"grad_norm": 0.3490909934043884,
"learning_rate": 1.5997557587998868e-06,
"loss": 0.0109,
"step": 27700
},
{
"grad_norm": 0.24534110724925995,
"learning_rate": 1.5859549198534606e-06,
"loss": 0.0051,
"step": 27710
},
{
"grad_norm": 0.12610509991645813,
"learning_rate": 1.572212908536963e-06,
"loss": 0.0062,
"step": 27720
},
{
"grad_norm": 0.2860276997089386,
"learning_rate": 1.5585297415482202e-06,
"loss": 0.0067,
"step": 27730
},
{
"grad_norm": 0.17748025059700012,
"learning_rate": 1.5449054355135717e-06,
"loss": 0.0073,
"step": 27740
},
{
"grad_norm": 0.1624142974615097,
"learning_rate": 1.5313400069878415e-06,
"loss": 0.0067,
"step": 27750
},
{
"grad_norm": 0.17958523333072662,
"learning_rate": 1.5178334724542887e-06,
"loss": 0.0063,
"step": 27760
},
{
"grad_norm": 0.22965379059314728,
"learning_rate": 1.504385848324641e-06,
"loss": 0.0174,
"step": 27770
},
{
"grad_norm": 0.22627367079257965,
"learning_rate": 1.490997150939033e-06,
"loss": 0.008,
"step": 27780
},
{
"grad_norm": 0.32128453254699707,
"learning_rate": 1.4776673965659793e-06,
"loss": 0.0096,
"step": 27790
},
{
"grad_norm": 0.2727048397064209,
"learning_rate": 1.4643966014023957e-06,
"loss": 0.0112,
"step": 27800
},
{
"grad_norm": 0.3476164638996124,
"learning_rate": 1.4511847815735503e-06,
"loss": 0.0116,
"step": 27810
},
{
"grad_norm": 0.11902498453855515,
"learning_rate": 1.438031953133051e-06,
"loss": 0.0062,
"step": 27820
},
{
"grad_norm": 0.27174538373947144,
"learning_rate": 1.4249381320628197e-06,
"loss": 0.0074,
"step": 27830
},
{
"grad_norm": 0.15753769874572754,
"learning_rate": 1.4119033342730902e-06,
"loss": 0.0129,
"step": 27840
},
{
"grad_norm": 0.09955435246229172,
"learning_rate": 1.3989275756023656e-06,
"loss": 0.0066,
"step": 27850
},
{
"grad_norm": 0.13606804609298706,
"learning_rate": 1.386010871817417e-06,
"loss": 0.0072,
"step": 27860
},
{
"grad_norm": 0.2543242573738098,
"learning_rate": 1.3731532386132616e-06,
"loss": 0.0056,
"step": 27870
},
{
"grad_norm": 0.3300288915634155,
"learning_rate": 1.360354691613136e-06,
"loss": 0.0221,
"step": 27880
},
{
"grad_norm": 0.24282188713550568,
"learning_rate": 1.3476152463684776e-06,
"loss": 0.0075,
"step": 27890
},
{
"grad_norm": 0.18473635613918304,
"learning_rate": 1.3349349183589155e-06,
"loss": 0.0075,
"step": 27900
},
{
"grad_norm": 0.20280255377292633,
"learning_rate": 1.3223137229922356e-06,
"loss": 0.0067,
"step": 27910
},
{
"grad_norm": 0.258235365152359,
"learning_rate": 1.3097516756043981e-06,
"loss": 0.0084,
"step": 27920
},
{
"grad_norm": 0.3198598325252533,
"learning_rate": 1.29724879145946e-06,
"loss": 0.0088,
"step": 27930
},
{
"grad_norm": 0.11840188503265381,
"learning_rate": 1.284805085749613e-06,
"loss": 0.0071,
"step": 27940
},
{
"grad_norm": 0.31867799162864685,
"learning_rate": 1.2724205735951288e-06,
"loss": 0.0094,
"step": 27950
},
{
"grad_norm": 0.12260130792856216,
"learning_rate": 1.2600952700443591e-06,
"loss": 0.0163,
"step": 27960
},
{
"grad_norm": 0.1465507596731186,
"learning_rate": 1.247829190073707e-06,
"loss": 0.0062,
"step": 27970
},
{
"grad_norm": 0.14699310064315796,
"learning_rate": 1.2356223485876173e-06,
"loss": 0.0052,
"step": 27980
},
{
"grad_norm": 0.26343676447868347,
"learning_rate": 1.2234747604185526e-06,
"loss": 0.0085,
"step": 27990
},
{
"grad_norm": 0.2008315771818161,
"learning_rate": 1.2113864403269836e-06,
"loss": 0.0132,
"step": 28000
},
{
"grad_norm": 0.1349935233592987,
"learning_rate": 1.1993574030013555e-06,
"loss": 0.0052,
"step": 28010
},
{
"grad_norm": 0.4881086051464081,
"learning_rate": 1.1873876630580816e-06,
"loss": 0.0143,
"step": 28020
},
{
"grad_norm": 0.3282459080219269,
"learning_rate": 1.1754772350415278e-06,
"loss": 0.0084,
"step": 28030
},
{
"grad_norm": 0.19573692977428436,
"learning_rate": 1.163626133423984e-06,
"loss": 0.009,
"step": 28040
},
{
"grad_norm": 0.13341741263866425,
"learning_rate": 1.1518343726056591e-06,
"loss": 0.007,
"step": 28050
},
{
"grad_norm": 0.1511244773864746,
"learning_rate": 1.1401019669146474e-06,
"loss": 0.0064,
"step": 28060
},
{
"grad_norm": 0.29247981309890747,
"learning_rate": 1.128428930606934e-06,
"loss": 0.0073,
"step": 28070
},
{
"grad_norm": 0.14105579257011414,
"learning_rate": 1.1168152778663621e-06,
"loss": 0.0065,
"step": 28080
},
{
"grad_norm": 0.3257015645503998,
"learning_rate": 1.10526102280461e-06,
"loss": 0.0105,
"step": 28090
},
{
"grad_norm": 0.1435706913471222,
"learning_rate": 1.0937661794611864e-06,
"loss": 0.0077,
"step": 28100
},
{
"grad_norm": 0.23878289759159088,
"learning_rate": 1.082330761803413e-06,
"loss": 0.0084,
"step": 28110
},
{
"grad_norm": 0.31835564970970154,
"learning_rate": 1.0709547837263966e-06,
"loss": 0.011,
"step": 28120
},
{
"grad_norm": 0.12658680975437164,
"learning_rate": 1.059638259053025e-06,
"loss": 0.0065,
"step": 28130
},
{
"grad_norm": 0.258036732673645,
"learning_rate": 1.0483812015339434e-06,
"loss": 0.0064,
"step": 28140
},
{
"grad_norm": 0.3808008134365082,
"learning_rate": 1.037183624847543e-06,
"loss": 0.017,
"step": 28150
},
{
"grad_norm": 0.15945574641227722,
"learning_rate": 1.0260455425999238e-06,
"loss": 0.0121,
"step": 28160
},
{
"grad_norm": 0.1372184455394745,
"learning_rate": 1.0149669683249208e-06,
"loss": 0.0075,
"step": 28170
},
{
"grad_norm": 0.18334408104419708,
"learning_rate": 1.003947915484027e-06,
"loss": 0.0126,
"step": 28180
},
{
"grad_norm": 0.302151083946228,
"learning_rate": 9.92988397466449e-07,
"loss": 0.0256,
"step": 28190
},
{
"grad_norm": 0.40292710065841675,
"learning_rate": 9.820884275890286e-07,
"loss": 0.0114,
"step": 28200
},
{
"grad_norm": 0.14354296028614044,
"learning_rate": 9.712480190962546e-07,
"loss": 0.0084,
"step": 28210
},
{
"grad_norm": 0.2882560193538666,
"learning_rate": 9.604671851602464e-07,
"loss": 0.0078,
"step": 28220
},
{
"grad_norm": 0.34792399406433105,
"learning_rate": 9.497459388807306e-07,
"loss": 0.0076,
"step": 28230
},
{
"grad_norm": 0.2593053877353668,
"learning_rate": 9.390842932850364e-07,
"loss": 0.0072,
"step": 28240
},
{
"grad_norm": 0.4218943119049072,
"learning_rate": 9.28482261328073e-07,
"loss": 0.0166,
"step": 28250
},
{
"grad_norm": 0.2316504865884781,
"learning_rate": 9.179398558923025e-07,
"loss": 0.0099,
"step": 28260
},
{
"grad_norm": 0.21009816229343414,
"learning_rate": 9.074570897877388e-07,
"loss": 0.0053,
"step": 28270
},
{
"grad_norm": 0.2886572778224945,
"learning_rate": 8.970339757519375e-07,
"loss": 0.0116,
"step": 28280
},
{
"grad_norm": 0.1469964236021042,
"learning_rate": 8.866705264499619e-07,
"loss": 0.0058,
"step": 28290
},
{
"grad_norm": 0.1551496386528015,
"learning_rate": 8.763667544743836e-07,
"loss": 0.0067,
"step": 28300
},
{
"grad_norm": 0.23584331572055817,
"learning_rate": 8.661226723452543e-07,
"loss": 0.0067,
"step": 28310
},
{
"grad_norm": 0.1818852722644806,
"learning_rate": 8.559382925101001e-07,
"loss": 0.0069,
"step": 28320
},
{
"grad_norm": 0.15392281115055084,
"learning_rate": 8.458136273438943e-07,
"loss": 0.0086,
"step": 28330
},
{
"grad_norm": 0.19438393414020538,
"learning_rate": 8.357486891490795e-07,
"loss": 0.0085,
"step": 28340
},
{
"grad_norm": 0.2458229511976242,
"learning_rate": 8.257434901554895e-07,
"loss": 0.0066,
"step": 28350
},
{
"grad_norm": 0.12557046115398407,
"learning_rate": 8.157980425203937e-07,
"loss": 0.0061,
"step": 28360
},
{
"grad_norm": 0.21882516145706177,
"learning_rate": 8.059123583284367e-07,
"loss": 0.0122,
"step": 28370
},
{
"grad_norm": 0.24017155170440674,
"learning_rate": 7.960864495916654e-07,
"loss": 0.0089,
"step": 28380
},
{
"grad_norm": 0.17537395656108856,
"learning_rate": 7.863203282494847e-07,
"loss": 0.0113,
"step": 28390
},
{
"grad_norm": 0.2210918664932251,
"learning_rate": 7.766140061686522e-07,
"loss": 0.0062,
"step": 28400
},
{
"grad_norm": 0.4063290059566498,
"learning_rate": 7.669674951432615e-07,
"loss": 0.0088,
"step": 28410
},
{
"grad_norm": 0.24993111193180084,
"learning_rate": 7.573808068947363e-07,
"loss": 0.0137,
"step": 28420
},
{
"grad_norm": 0.36737027764320374,
"learning_rate": 7.478539530718087e-07,
"loss": 0.0097,
"step": 28430
},
{
"grad_norm": 0.14404527842998505,
"learning_rate": 7.383869452504965e-07,
"loss": 0.0058,
"step": 28440
},
{
"grad_norm": 0.14547091722488403,
"learning_rate": 7.289797949341204e-07,
"loss": 0.0076,
"step": 28450
},
{
"grad_norm": 0.23230545222759247,
"learning_rate": 7.196325135532423e-07,
"loss": 0.008,
"step": 28460
},
{
"grad_norm": 0.2384069263935089,
"learning_rate": 7.10345112465699e-07,
"loss": 0.0074,
"step": 28470
},
{
"grad_norm": 0.3890141248703003,
"learning_rate": 7.011176029565525e-07,
"loss": 0.0128,
"step": 28480
},
{
"grad_norm": 0.20122289657592773,
"learning_rate": 6.919499962381004e-07,
"loss": 0.0049,
"step": 28490
},
{
"grad_norm": 0.1418987363576889,
"learning_rate": 6.828423034498488e-07,
"loss": 0.0095,
"step": 28500
},
{
"grad_norm": 0.271654337644577,
"learning_rate": 6.737945356585007e-07,
"loss": 0.0075,
"step": 28510
},
{
"grad_norm": 0.2163088619709015,
"learning_rate": 6.648067038579508e-07,
"loss": 0.0111,
"step": 28520
},
{
"grad_norm": 0.28505653142929077,
"learning_rate": 6.558788189692578e-07,
"loss": 0.0094,
"step": 28530
},
{
"grad_norm": 0.17890380322933197,
"learning_rate": 6.470108918406492e-07,
"loss": 0.007,
"step": 28540
},
{
"grad_norm": 0.28197720646858215,
"learning_rate": 6.382029332474893e-07,
"loss": 0.0112,
"step": 28550
},
{
"grad_norm": 0.2491309642791748,
"learning_rate": 6.294549538922778e-07,
"loss": 0.0098,
"step": 28560
},
{
"grad_norm": 0.16437304019927979,
"learning_rate": 6.207669644046344e-07,
"loss": 0.0181,
"step": 28570
},
{
"grad_norm": 0.19345881044864655,
"learning_rate": 6.121389753412865e-07,
"loss": 0.0112,
"step": 28580
},
{
"grad_norm": 0.1998312771320343,
"learning_rate": 6.035709971860592e-07,
"loss": 0.0083,
"step": 28590
},
{
"grad_norm": 0.14426393806934357,
"learning_rate": 5.950630403498469e-07,
"loss": 0.0056,
"step": 28600
},
{
"grad_norm": 0.1739029884338379,
"learning_rate": 5.866151151706189e-07,
"loss": 0.0203,
"step": 28610
},
{
"grad_norm": 0.12093488872051239,
"learning_rate": 5.782272319134086e-07,
"loss": 0.0127,
"step": 28620
},
{
"grad_norm": 0.11821011453866959,
"learning_rate": 5.698994007702796e-07,
"loss": 0.0069,
"step": 28630
},
{
"grad_norm": 0.09072604775428772,
"learning_rate": 5.616316318603321e-07,
"loss": 0.0083,
"step": 28640
},
{
"grad_norm": 0.13620871305465698,
"learning_rate": 5.534239352296799e-07,
"loss": 0.0071,
"step": 28650
},
{
"grad_norm": 0.35490596294403076,
"learning_rate": 5.452763208514621e-07,
"loss": 0.0102,
"step": 28660
},
{
"grad_norm": 0.15092363953590393,
"learning_rate": 5.371887986257873e-07,
"loss": 0.0059,
"step": 28670
},
{
"grad_norm": 0.18033631145954132,
"learning_rate": 5.291613783797611e-07,
"loss": 0.0088,
"step": 28680
},
{
"grad_norm": 0.1737089902162552,
"learning_rate": 5.211940698674534e-07,
"loss": 0.0086,
"step": 28690
},
{
"grad_norm": 0.16497331857681274,
"learning_rate": 5.132868827698978e-07,
"loss": 0.0092,
"step": 28700
},
{
"grad_norm": 0.2506839632987976,
"learning_rate": 5.054398266950755e-07,
"loss": 0.0061,
"step": 28710
},
{
"grad_norm": 0.13000020384788513,
"learning_rate": 4.976529111778872e-07,
"loss": 0.0131,
"step": 28720
},
{
"grad_norm": 0.1498570591211319,
"learning_rate": 4.899261456801862e-07,
"loss": 0.024,
"step": 28730
},
{
"grad_norm": 0.10817938297986984,
"learning_rate": 4.822595395907126e-07,
"loss": 0.0051,
"step": 28740
},
{
"grad_norm": 0.19182389974594116,
"learning_rate": 4.7465310222510886e-07,
"loss": 0.0097,
"step": 28750
},
{
"grad_norm": 0.16202116012573242,
"learning_rate": 4.6710684282593175e-07,
"loss": 0.0197,
"step": 28760
},
{
"grad_norm": 0.27877798676490784,
"learning_rate": 4.5962077056257993e-07,
"loss": 0.0053,
"step": 28770
},
{
"grad_norm": 0.20378684997558594,
"learning_rate": 4.521948945313492e-07,
"loss": 0.0083,
"step": 28780
},
{
"grad_norm": 0.1535274237394333,
"learning_rate": 4.4482922375537195e-07,
"loss": 0.0102,
"step": 28790
},
{
"grad_norm": 0.11323801428079605,
"learning_rate": 4.375237671846333e-07,
"loss": 0.0069,
"step": 28800
},
{
"grad_norm": 0.11511887609958649,
"learning_rate": 4.302785336959547e-07,
"loss": 0.0049,
"step": 28810
},
{
"grad_norm": 0.11463732272386551,
"learning_rate": 4.2309353209297744e-07,
"loss": 0.0083,
"step": 28820
},
{
"grad_norm": 0.25968724489212036,
"learning_rate": 4.159687711061566e-07,
"loss": 0.0062,
"step": 28830
},
{
"grad_norm": 0.2854844331741333,
"learning_rate": 4.0890425939275055e-07,
"loss": 0.0054,
"step": 28840
},
{
"grad_norm": 0.39782264828681946,
"learning_rate": 4.0190000553679827e-07,
"loss": 0.0128,
"step": 28850
},
{
"grad_norm": 0.17746277153491974,
"learning_rate": 3.9495601804913627e-07,
"loss": 0.0081,
"step": 28860
},
{
"grad_norm": 0.14295367896556854,
"learning_rate": 3.880723053673652e-07,
"loss": 0.0097,
"step": 28870
},
{
"grad_norm": 0.11281397193670273,
"learning_rate": 3.812488758558386e-07,
"loss": 0.017,
"step": 28880
},
{
"grad_norm": 0.26936525106430054,
"learning_rate": 3.744857378056743e-07,
"loss": 0.0055,
"step": 28890
},
{
"grad_norm": 0.24043774604797363,
"learning_rate": 3.677828994347154e-07,
"loss": 0.007,
"step": 28900
},
{
"grad_norm": 0.25435757637023926,
"learning_rate": 3.61140368887547e-07,
"loss": 0.0083,
"step": 28910
},
{
"grad_norm": 0.13929742574691772,
"learning_rate": 3.545581542354681e-07,
"loss": 0.0054,
"step": 28920
},
{
"grad_norm": 0.1307041347026825,
"learning_rate": 3.480362634764922e-07,
"loss": 0.0059,
"step": 28930
},
{
"grad_norm": 0.30814027786254883,
"learning_rate": 3.4157470453533014e-07,
"loss": 0.0079,
"step": 28940
},
{
"grad_norm": 0.15271127223968506,
"learning_rate": 3.3517348526339034e-07,
"loss": 0.0097,
"step": 28950
},
{
"grad_norm": 0.1628933995962143,
"learning_rate": 3.288326134387454e-07,
"loss": 0.007,
"step": 28960
},
{
"grad_norm": 0.22971530258655548,
"learning_rate": 3.225520967661655e-07,
"loss": 0.0107,
"step": 28970
},
{
"grad_norm": 0.16180892288684845,
"learning_rate": 3.163319428770628e-07,
"loss": 0.006,
"step": 28980
},
{
"grad_norm": 0.27544841170310974,
"learning_rate": 3.1017215932951374e-07,
"loss": 0.0071,
"step": 28990
},
{
"grad_norm": 0.24542902410030365,
"learning_rate": 3.040727536082366e-07,
"loss": 0.0071,
"step": 29000
},
{
"grad_norm": 0.1707499772310257,
"learning_rate": 2.980337331245864e-07,
"loss": 0.008,
"step": 29010
},
{
"grad_norm": 0.1601923108100891,
"learning_rate": 2.9205510521653213e-07,
"loss": 0.0084,
"step": 29020
},
{
"grad_norm": 0.10638295859098434,
"learning_rate": 2.86136877148685e-07,
"loss": 0.0075,
"step": 29030
},
{
"grad_norm": 0.3442731499671936,
"learning_rate": 2.8027905611223704e-07,
"loss": 0.0101,
"step": 29040
},
{
"grad_norm": 0.09500379115343094,
"learning_rate": 2.7448164922500573e-07,
"loss": 0.0096,
"step": 29050
},
{
"grad_norm": 0.3162676990032196,
"learning_rate": 2.687446635313784e-07,
"loss": 0.0087,
"step": 29060
},
{
"grad_norm": 0.20223665237426758,
"learning_rate": 2.630681060023343e-07,
"loss": 0.0084,
"step": 29070
},
{
"grad_norm": 0.21525061130523682,
"learning_rate": 2.5745198353542833e-07,
"loss": 0.0067,
"step": 29080
},
{
"grad_norm": 0.14935867488384247,
"learning_rate": 2.518963029547794e-07,
"loss": 0.0068,
"step": 29090
},
{
"grad_norm": 0.24216340482234955,
"learning_rate": 2.464010710110598e-07,
"loss": 0.0115,
"step": 29100
},
{
"grad_norm": 0.12357182055711746,
"learning_rate": 2.4096629438150054e-07,
"loss": 0.0057,
"step": 29110
},
{
"grad_norm": 0.14813165366649628,
"learning_rate": 2.3559197966985802e-07,
"loss": 0.0068,
"step": 29120
},
{
"grad_norm": 0.17783525586128235,
"learning_rate": 2.3027813340644188e-07,
"loss": 0.0061,
"step": 29130
},
{
"grad_norm": 0.18345582485198975,
"learning_rate": 2.2502476204807055e-07,
"loss": 0.0071,
"step": 29140
},
{
"grad_norm": 0.12039521336555481,
"learning_rate": 2.1983187197808786e-07,
"loss": 0.0075,
"step": 29150
},
{
"grad_norm": 0.17816612124443054,
"learning_rate": 2.1469946950634644e-07,
"loss": 0.0121,
"step": 29160
},
{
"grad_norm": 0.32045531272888184,
"learning_rate": 2.096275608691911e-07,
"loss": 0.0144,
"step": 29170
},
{
"grad_norm": 0.23378242552280426,
"learning_rate": 2.046161522294754e-07,
"loss": 0.0095,
"step": 29180
},
{
"grad_norm": 0.24218012392520905,
"learning_rate": 1.9966524967653944e-07,
"loss": 0.0075,
"step": 29190
},
{
"grad_norm": 0.21988293528556824,
"learning_rate": 1.9477485922618222e-07,
"loss": 0.0079,
"step": 29200
},
{
"grad_norm": 0.30169081687927246,
"learning_rate": 1.899449868207004e-07,
"loss": 0.0074,
"step": 29210
},
{
"grad_norm": 0.17310160398483276,
"learning_rate": 1.851756383288439e-07,
"loss": 0.0235,
"step": 29220
},
{
"grad_norm": 0.1534615010023117,
"learning_rate": 1.8046681954581035e-07,
"loss": 0.0082,
"step": 29230
},
{
"grad_norm": 0.15709884464740753,
"learning_rate": 1.7581853619327294e-07,
"loss": 0.0166,
"step": 29240
},
{
"grad_norm": 0.30939313769340515,
"learning_rate": 1.7123079391932472e-07,
"loss": 0.0067,
"step": 29250
},
{
"grad_norm": 0.11957605928182602,
"learning_rate": 1.6670359829850657e-07,
"loss": 0.0073,
"step": 29260
},
{
"grad_norm": 0.15490154922008514,
"learning_rate": 1.6223695483179037e-07,
"loss": 0.0068,
"step": 29270
},
{
"grad_norm": 0.26335710287094116,
"learning_rate": 1.5783086894656795e-07,
"loss": 0.0081,
"step": 29280
},
{
"grad_norm": 0.26684433221817017,
"learning_rate": 1.5348534599665121e-07,
"loss": 0.0082,
"step": 29290
},
{
"grad_norm": 0.2584269940853119,
"learning_rate": 1.4920039126225527e-07,
"loss": 0.0078,
"step": 29300
},
{
"grad_norm": 0.20256619155406952,
"learning_rate": 1.449760099500097e-07,
"loss": 0.0148,
"step": 29310
},
{
"grad_norm": 0.18130625784397125,
"learning_rate": 1.4081220719293075e-07,
"loss": 0.0089,
"step": 29320
},
{
"grad_norm": 0.3089062571525574,
"learning_rate": 1.3670898805043797e-07,
"loss": 0.0087,
"step": 29330
},
{
"grad_norm": 0.1532166302204132,
"learning_rate": 1.326663575083209e-07,
"loss": 0.0056,
"step": 29340
},
{
"grad_norm": 0.3202992081642151,
"learning_rate": 1.2868432047876688e-07,
"loss": 0.0116,
"step": 29350
},
{
"grad_norm": 0.28339993953704834,
"learning_rate": 1.2476288180032213e-07,
"loss": 0.0078,
"step": 29360
},
{
"grad_norm": 0.1671651303768158,
"learning_rate": 1.209020462379029e-07,
"loss": 0.0098,
"step": 29370
},
{
"grad_norm": 0.2666459381580353,
"learning_rate": 1.1710181848278435e-07,
"loss": 0.0064,
"step": 29380
},
{
"grad_norm": 0.1355592906475067,
"learning_rate": 1.133622031526116e-07,
"loss": 0.0054,
"step": 29390
},
{
"grad_norm": 0.16878631711006165,
"learning_rate": 1.0968320479136095e-07,
"loss": 0.0104,
"step": 29400
},
{
"grad_norm": 0.3057379722595215,
"learning_rate": 1.0606482786936767e-07,
"loss": 0.009,
"step": 29410
},
{
"grad_norm": 0.13650526106357574,
"learning_rate": 1.0250707678329808e-07,
"loss": 0.0053,
"step": 29420
},
{
"grad_norm": 0.2038799524307251,
"learning_rate": 9.900995585615525e-08,
"loss": 0.0082,
"step": 29430
},
{
"grad_norm": 0.17177928984165192,
"learning_rate": 9.55734693372734e-08,
"loss": 0.0125,
"step": 29440
},
{
"grad_norm": 0.13057716190814972,
"learning_rate": 9.219762140231236e-08,
"loss": 0.0087,
"step": 29450
},
{
"grad_norm": 0.14972764253616333,
"learning_rate": 8.888241615322978e-08,
"loss": 0.0054,
"step": 29460
},
{
"grad_norm": 0.31600919365882874,
"learning_rate": 8.562785761833114e-08,
"loss": 0.0076,
"step": 29470
},
{
"grad_norm": 0.10899215936660767,
"learning_rate": 8.243394975219753e-08,
"loss": 0.0128,
"step": 29480
},
{
"grad_norm": 0.31470048427581787,
"learning_rate": 7.930069643573568e-08,
"loss": 0.0078,
"step": 29490
},
{
"grad_norm": 0.1694929301738739,
"learning_rate": 7.622810147614456e-08,
"loss": 0.0132,
"step": 29500
},
{
"grad_norm": 0.20618291199207306,
"learning_rate": 7.321616860690995e-08,
"loss": 0.0078,
"step": 29510
},
{
"grad_norm": 0.23289665579795837,
"learning_rate": 7.026490148782095e-08,
"loss": 0.0064,
"step": 29520
},
{
"grad_norm": 0.11668981611728668,
"learning_rate": 6.737430370494236e-08,
"loss": 0.0106,
"step": 29530
},
{
"grad_norm": 0.09879987686872482,
"learning_rate": 6.454437877062569e-08,
"loss": 0.0197,
"step": 29540
},
{
"grad_norm": 0.14076517522335052,
"learning_rate": 6.177513012349256e-08,
"loss": 0.0076,
"step": 29550
},
{
"grad_norm": 0.1864781230688095,
"learning_rate": 5.9066561128445775e-08,
"loss": 0.0066,
"step": 29560
},
{
"grad_norm": 0.2049897164106369,
"learning_rate": 5.6418675076641556e-08,
"loss": 0.0085,
"step": 29570
},
{
"grad_norm": 0.1332913488149643,
"learning_rate": 5.383147518552845e-08,
"loss": 0.0128,
"step": 29580
},
{
"grad_norm": 0.13722537457942963,
"learning_rate": 5.1304964598786196e-08,
"loss": 0.0082,
"step": 29590
},
{
"grad_norm": 0.2020295262336731,
"learning_rate": 4.883914638636467e-08,
"loss": 0.0062,
"step": 29600
},
{
"grad_norm": 0.10029200464487076,
"learning_rate": 4.643402354446713e-08,
"loss": 0.0046,
"step": 29610
},
{
"grad_norm": 0.26212993264198303,
"learning_rate": 4.4089598995544766e-08,
"loss": 0.007,
"step": 29620
},
{
"grad_norm": 0.21249616146087646,
"learning_rate": 4.180587558829663e-08,
"loss": 0.0133,
"step": 29630
},
{
"grad_norm": 0.3150579631328583,
"learning_rate": 3.958285609765855e-08,
"loss": 0.0079,
"step": 29640
},
{
"grad_norm": 0.12820951640605927,
"learning_rate": 3.74205432248087e-08,
"loss": 0.0096,
"step": 29650
},
{
"grad_norm": 0.12171725183725357,
"learning_rate": 3.531893959716204e-08,
"loss": 0.0126,
"step": 29660
},
{
"grad_norm": 0.2698152959346771,
"learning_rate": 3.327804776837029e-08,
"loss": 0.0097,
"step": 29670
},
{
"grad_norm": 0.16224315762519836,
"learning_rate": 3.129787021829977e-08,
"loss": 0.009,
"step": 29680
},
{
"grad_norm": 0.2528517544269562,
"learning_rate": 2.9378409353059133e-08,
"loss": 0.009,
"step": 29690
},
{
"grad_norm": 0.15599949657917023,
"learning_rate": 2.7519667504971593e-08,
"loss": 0.0058,
"step": 29700
},
{
"grad_norm": 0.40189129114151,
"learning_rate": 2.572164693258605e-08,
"loss": 0.0107,
"step": 29710
},
{
"grad_norm": 0.14458875358104706,
"learning_rate": 2.3984349820665996e-08,
"loss": 0.0091,
"step": 29720
},
{
"grad_norm": 0.2475101202726364,
"learning_rate": 2.2307778280189485e-08,
"loss": 0.0178,
"step": 29730
},
{
"grad_norm": 0.14989276230335236,
"learning_rate": 2.0691934348354704e-08,
"loss": 0.0064,
"step": 29740
},
{
"grad_norm": 0.19460299611091614,
"learning_rate": 1.9136819988557763e-08,
"loss": 0.013,
"step": 29750
},
{
"grad_norm": 0.13716372847557068,
"learning_rate": 1.7642437090414908e-08,
"loss": 0.0079,
"step": 29760
},
{
"grad_norm": 0.10082526504993439,
"learning_rate": 1.6208787469734755e-08,
"loss": 0.0116,
"step": 29770
},
{
"grad_norm": 0.1348201334476471,
"learning_rate": 1.483587286854604e-08,
"loss": 0.0087,
"step": 29780
},
{
"grad_norm": 0.13469603657722473,
"learning_rate": 1.3523694955064336e-08,
"loss": 0.007,
"step": 29790
},
{
"grad_norm": 0.2026950567960739,
"learning_rate": 1.2272255323708681e-08,
"loss": 0.0067,
"step": 29800
},
{
"grad_norm": 0.17148591578006744,
"learning_rate": 1.1081555495096042e-08,
"loss": 0.0061,
"step": 29810
},
{
"grad_norm": 0.13165178894996643,
"learning_rate": 9.951596916041306e-09,
"loss": 0.0061,
"step": 29820
},
{
"grad_norm": 0.41580843925476074,
"learning_rate": 8.882380959551739e-09,
"loss": 0.0159,
"step": 29830
},
{
"grad_norm": 0.15691199898719788,
"learning_rate": 7.873908924821428e-09,
"loss": 0.016,
"step": 29840
},
{
"grad_norm": 0.2497183084487915,
"learning_rate": 6.926182037242379e-09,
"loss": 0.0104,
"step": 29850
},
{
"grad_norm": 0.1798539161682129,
"learning_rate": 6.039201448387876e-09,
"loss": 0.0076,
"step": 29860
},
{
"grad_norm": 0.22067518532276154,
"learning_rate": 5.212968236029125e-09,
"loss": 0.006,
"step": 29870
},
{
"grad_norm": 0.2177223414182663,
"learning_rate": 4.447483404118602e-09,
"loss": 0.011,
"step": 29880
},
{
"grad_norm": 0.12921608984470367,
"learning_rate": 3.742747882784503e-09,
"loss": 0.0064,
"step": 29890
},
{
"grad_norm": 0.2243431955575943,
"learning_rate": 3.0987625283473987e-09,
"loss": 0.0166,
"step": 29900
},
{
"grad_norm": 0.2971177101135254,
"learning_rate": 2.515528123320232e-09,
"loss": 0.0074,
"step": 29910
},
{
"grad_norm": 0.3001595437526703,
"learning_rate": 1.9930453763750133e-09,
"loss": 0.0087,
"step": 29920
},
{
"grad_norm": 0.2129870355129242,
"learning_rate": 1.5313149223872281e-09,
"loss": 0.0156,
"step": 29930
},
{
"grad_norm": 0.246839702129364,
"learning_rate": 1.1303373224025305e-09,
"loss": 0.009,
"step": 29940
},
{
"grad_norm": 0.29312920570373535,
"learning_rate": 7.901130636367438e-10,
"loss": 0.0079,
"step": 29950
},
{
"grad_norm": 0.22019779682159424,
"learning_rate": 5.106425595036158e-10,
"loss": 0.0134,
"step": 29960
},
{
"grad_norm": 0.2666579782962799,
"learning_rate": 2.9192614958706286e-10,
"loss": 0.0092,
"step": 29970
},
{
"grad_norm": 0.2811462879180908,
"learning_rate": 1.3396409964117064e-10,
"loss": 0.0111,
"step": 29980
},
{
"grad_norm": 0.15274490416049957,
"learning_rate": 3.6756601606846574e-11,
"loss": 0.0087,
"step": 29990
},
{
"grad_norm": 0.21032936871051788,
"learning_rate": 3.0377360626943076e-13,
"loss": 0.0065,
"step": 30000
}
],
"logging_steps": 10,
"max_steps": 30000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 10000,
"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": 8,
"trial_name": null,
"trial_params": null
}