UltimoUno's picture
Uploaded checkpoint-4000
5acae67 verified
raw
history blame
59.8 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.4344629729245113,
"eval_steps": 2000,
"global_step": 4000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"grad_norm": 5.90625,
"learning_rate": 3.0000000000000004e-07,
"loss": 1.4893,
"step": 10
},
{
"epoch": 0.01,
"grad_norm": 10.0625,
"learning_rate": 6.000000000000001e-07,
"loss": 1.6671,
"step": 20
},
{
"epoch": 0.01,
"grad_norm": 7.03125,
"learning_rate": 9e-07,
"loss": 1.3866,
"step": 30
},
{
"epoch": 0.01,
"grad_norm": 5.9375,
"learning_rate": 1.2000000000000002e-06,
"loss": 1.433,
"step": 40
},
{
"epoch": 0.02,
"grad_norm": 8.1875,
"learning_rate": 1.5e-06,
"loss": 1.4135,
"step": 50
},
{
"epoch": 0.02,
"grad_norm": 8.375,
"learning_rate": 1.8e-06,
"loss": 1.3779,
"step": 60
},
{
"epoch": 0.03,
"grad_norm": 7.15625,
"learning_rate": 2.1000000000000002e-06,
"loss": 1.4444,
"step": 70
},
{
"epoch": 0.03,
"grad_norm": 7.71875,
"learning_rate": 2.4000000000000003e-06,
"loss": 1.3717,
"step": 80
},
{
"epoch": 0.03,
"grad_norm": 6.65625,
"learning_rate": 2.7e-06,
"loss": 1.3646,
"step": 90
},
{
"epoch": 0.04,
"grad_norm": 6.03125,
"learning_rate": 3e-06,
"loss": 1.4607,
"step": 100
},
{
"epoch": 0.04,
"grad_norm": 7.125,
"learning_rate": 3.3e-06,
"loss": 1.509,
"step": 110
},
{
"epoch": 0.04,
"grad_norm": 6.90625,
"learning_rate": 3.6e-06,
"loss": 1.2552,
"step": 120
},
{
"epoch": 0.05,
"grad_norm": 7.78125,
"learning_rate": 3.9e-06,
"loss": 1.3142,
"step": 130
},
{
"epoch": 0.05,
"grad_norm": 6.8125,
"learning_rate": 4.2000000000000004e-06,
"loss": 1.3774,
"step": 140
},
{
"epoch": 0.05,
"grad_norm": 8.375,
"learning_rate": 4.5e-06,
"loss": 1.3869,
"step": 150
},
{
"epoch": 0.06,
"grad_norm": 5.1875,
"learning_rate": 4.800000000000001e-06,
"loss": 1.2843,
"step": 160
},
{
"epoch": 0.06,
"grad_norm": 5.71875,
"learning_rate": 5.1e-06,
"loss": 1.3458,
"step": 170
},
{
"epoch": 0.06,
"grad_norm": 6.21875,
"learning_rate": 5.4e-06,
"loss": 1.2259,
"step": 180
},
{
"epoch": 0.07,
"grad_norm": 5.71875,
"learning_rate": 5.7000000000000005e-06,
"loss": 1.1647,
"step": 190
},
{
"epoch": 0.07,
"grad_norm": 5.875,
"learning_rate": 6e-06,
"loss": 1.0721,
"step": 200
},
{
"epoch": 0.08,
"grad_norm": 5.09375,
"learning_rate": 6.3e-06,
"loss": 1.0382,
"step": 210
},
{
"epoch": 0.08,
"grad_norm": 5.1875,
"learning_rate": 6.6e-06,
"loss": 1.1419,
"step": 220
},
{
"epoch": 0.08,
"grad_norm": 5.1875,
"learning_rate": 6.900000000000001e-06,
"loss": 1.0656,
"step": 230
},
{
"epoch": 0.09,
"grad_norm": 4.8125,
"learning_rate": 7.2e-06,
"loss": 1.1806,
"step": 240
},
{
"epoch": 0.09,
"grad_norm": 3.84375,
"learning_rate": 7.5e-06,
"loss": 0.9791,
"step": 250
},
{
"epoch": 0.09,
"grad_norm": 6.5625,
"learning_rate": 7.8e-06,
"loss": 1.1894,
"step": 260
},
{
"epoch": 0.1,
"grad_norm": 3.65625,
"learning_rate": 8.1e-06,
"loss": 0.91,
"step": 270
},
{
"epoch": 0.1,
"grad_norm": 4.90625,
"learning_rate": 8.400000000000001e-06,
"loss": 1.0708,
"step": 280
},
{
"epoch": 0.1,
"grad_norm": 6.53125,
"learning_rate": 8.7e-06,
"loss": 0.9747,
"step": 290
},
{
"epoch": 0.11,
"grad_norm": 4.28125,
"learning_rate": 9e-06,
"loss": 1.0158,
"step": 300
},
{
"epoch": 0.11,
"grad_norm": 4.34375,
"learning_rate": 9.3e-06,
"loss": 1.1078,
"step": 310
},
{
"epoch": 0.11,
"grad_norm": 3.171875,
"learning_rate": 9.600000000000001e-06,
"loss": 1.0437,
"step": 320
},
{
"epoch": 0.12,
"grad_norm": 4.25,
"learning_rate": 9.9e-06,
"loss": 0.9891,
"step": 330
},
{
"epoch": 0.12,
"grad_norm": 4.875,
"learning_rate": 1.02e-05,
"loss": 0.8667,
"step": 340
},
{
"epoch": 0.13,
"grad_norm": 3.703125,
"learning_rate": 1.05e-05,
"loss": 1.0903,
"step": 350
},
{
"epoch": 0.13,
"grad_norm": 3.390625,
"learning_rate": 1.08e-05,
"loss": 0.9947,
"step": 360
},
{
"epoch": 0.13,
"grad_norm": 5.6875,
"learning_rate": 1.11e-05,
"loss": 1.043,
"step": 370
},
{
"epoch": 0.14,
"grad_norm": 3.03125,
"learning_rate": 1.1400000000000001e-05,
"loss": 0.8305,
"step": 380
},
{
"epoch": 0.14,
"grad_norm": 6.0625,
"learning_rate": 1.1700000000000001e-05,
"loss": 0.9741,
"step": 390
},
{
"epoch": 0.14,
"grad_norm": 4.34375,
"learning_rate": 1.2e-05,
"loss": 1.219,
"step": 400
},
{
"epoch": 0.15,
"grad_norm": 5.53125,
"learning_rate": 1.2299999999999999e-05,
"loss": 1.0167,
"step": 410
},
{
"epoch": 0.15,
"grad_norm": 3.578125,
"learning_rate": 1.26e-05,
"loss": 0.87,
"step": 420
},
{
"epoch": 0.15,
"grad_norm": 5.59375,
"learning_rate": 1.29e-05,
"loss": 1.0169,
"step": 430
},
{
"epoch": 0.16,
"grad_norm": 3.453125,
"learning_rate": 1.32e-05,
"loss": 0.9178,
"step": 440
},
{
"epoch": 0.16,
"grad_norm": 4.21875,
"learning_rate": 1.3500000000000001e-05,
"loss": 1.0044,
"step": 450
},
{
"epoch": 0.16,
"grad_norm": 4.875,
"learning_rate": 1.3800000000000002e-05,
"loss": 0.897,
"step": 460
},
{
"epoch": 0.17,
"grad_norm": 3.5,
"learning_rate": 1.4099999999999999e-05,
"loss": 0.8968,
"step": 470
},
{
"epoch": 0.17,
"grad_norm": 5.15625,
"learning_rate": 1.44e-05,
"loss": 0.9446,
"step": 480
},
{
"epoch": 0.18,
"grad_norm": 2.375,
"learning_rate": 1.47e-05,
"loss": 0.8887,
"step": 490
},
{
"epoch": 0.18,
"grad_norm": 3.84375,
"learning_rate": 1.5e-05,
"loss": 1.0239,
"step": 500
},
{
"epoch": 0.18,
"grad_norm": 2.390625,
"learning_rate": 1.4984210526315789e-05,
"loss": 0.9251,
"step": 510
},
{
"epoch": 0.19,
"grad_norm": 4.9375,
"learning_rate": 1.496842105263158e-05,
"loss": 0.9681,
"step": 520
},
{
"epoch": 0.19,
"grad_norm": 4.6875,
"learning_rate": 1.4952631578947368e-05,
"loss": 0.8625,
"step": 530
},
{
"epoch": 0.19,
"grad_norm": 3.734375,
"learning_rate": 1.4936842105263158e-05,
"loss": 0.9131,
"step": 540
},
{
"epoch": 0.2,
"grad_norm": 3.1875,
"learning_rate": 1.4921052631578947e-05,
"loss": 0.8903,
"step": 550
},
{
"epoch": 0.2,
"grad_norm": 4.09375,
"learning_rate": 1.4905263157894737e-05,
"loss": 0.9482,
"step": 560
},
{
"epoch": 0.2,
"grad_norm": 2.921875,
"learning_rate": 1.4889473684210526e-05,
"loss": 0.8134,
"step": 570
},
{
"epoch": 0.21,
"grad_norm": 4.15625,
"learning_rate": 1.4873684210526315e-05,
"loss": 0.896,
"step": 580
},
{
"epoch": 0.21,
"grad_norm": 5.46875,
"learning_rate": 1.4857894736842107e-05,
"loss": 0.9682,
"step": 590
},
{
"epoch": 0.22,
"grad_norm": 3.90625,
"learning_rate": 1.4842105263157895e-05,
"loss": 0.8046,
"step": 600
},
{
"epoch": 0.22,
"grad_norm": 3.96875,
"learning_rate": 1.4826315789473686e-05,
"loss": 0.9797,
"step": 610
},
{
"epoch": 0.22,
"grad_norm": 3.171875,
"learning_rate": 1.4810526315789474e-05,
"loss": 0.8386,
"step": 620
},
{
"epoch": 0.23,
"grad_norm": 4.46875,
"learning_rate": 1.4794736842105265e-05,
"loss": 0.9006,
"step": 630
},
{
"epoch": 0.23,
"grad_norm": 4.0,
"learning_rate": 1.4778947368421053e-05,
"loss": 0.8116,
"step": 640
},
{
"epoch": 0.23,
"grad_norm": 4.03125,
"learning_rate": 1.4763157894736842e-05,
"loss": 0.8465,
"step": 650
},
{
"epoch": 0.24,
"grad_norm": 4.09375,
"learning_rate": 1.4747368421052632e-05,
"loss": 0.843,
"step": 660
},
{
"epoch": 0.24,
"grad_norm": 3.21875,
"learning_rate": 1.4731578947368421e-05,
"loss": 0.9078,
"step": 670
},
{
"epoch": 0.24,
"grad_norm": 3.53125,
"learning_rate": 1.4715789473684211e-05,
"loss": 0.855,
"step": 680
},
{
"epoch": 0.25,
"grad_norm": 3.546875,
"learning_rate": 1.47e-05,
"loss": 0.9115,
"step": 690
},
{
"epoch": 0.25,
"grad_norm": 4.5625,
"learning_rate": 1.468421052631579e-05,
"loss": 0.8013,
"step": 700
},
{
"epoch": 0.25,
"grad_norm": 3.515625,
"learning_rate": 1.4668421052631579e-05,
"loss": 0.8253,
"step": 710
},
{
"epoch": 0.26,
"grad_norm": 3.4375,
"learning_rate": 1.4652631578947367e-05,
"loss": 0.8778,
"step": 720
},
{
"epoch": 0.26,
"grad_norm": 6.4375,
"learning_rate": 1.4636842105263158e-05,
"loss": 0.9265,
"step": 730
},
{
"epoch": 0.27,
"grad_norm": 4.0625,
"learning_rate": 1.4621052631578946e-05,
"loss": 0.8955,
"step": 740
},
{
"epoch": 0.27,
"grad_norm": 3.6875,
"learning_rate": 1.4605263157894737e-05,
"loss": 0.8791,
"step": 750
},
{
"epoch": 0.27,
"grad_norm": 3.703125,
"learning_rate": 1.4589473684210527e-05,
"loss": 0.8493,
"step": 760
},
{
"epoch": 0.28,
"grad_norm": 3.984375,
"learning_rate": 1.4573684210526317e-05,
"loss": 0.9307,
"step": 770
},
{
"epoch": 0.28,
"grad_norm": 3.125,
"learning_rate": 1.4557894736842106e-05,
"loss": 0.9093,
"step": 780
},
{
"epoch": 0.28,
"grad_norm": 4.40625,
"learning_rate": 1.4542105263157895e-05,
"loss": 0.873,
"step": 790
},
{
"epoch": 0.29,
"grad_norm": 4.21875,
"learning_rate": 1.4526315789473685e-05,
"loss": 0.9238,
"step": 800
},
{
"epoch": 0.29,
"grad_norm": 3.25,
"learning_rate": 1.4510526315789474e-05,
"loss": 0.8425,
"step": 810
},
{
"epoch": 0.29,
"grad_norm": 3.125,
"learning_rate": 1.4494736842105264e-05,
"loss": 0.9208,
"step": 820
},
{
"epoch": 0.3,
"grad_norm": 5.625,
"learning_rate": 1.4478947368421053e-05,
"loss": 0.8656,
"step": 830
},
{
"epoch": 0.3,
"grad_norm": 2.890625,
"learning_rate": 1.4463157894736843e-05,
"loss": 0.9557,
"step": 840
},
{
"epoch": 0.3,
"grad_norm": 4.15625,
"learning_rate": 1.4447368421052632e-05,
"loss": 0.859,
"step": 850
},
{
"epoch": 0.31,
"grad_norm": 3.984375,
"learning_rate": 1.443157894736842e-05,
"loss": 0.9057,
"step": 860
},
{
"epoch": 0.31,
"grad_norm": 3.890625,
"learning_rate": 1.441578947368421e-05,
"loss": 0.9119,
"step": 870
},
{
"epoch": 0.32,
"grad_norm": 4.34375,
"learning_rate": 1.44e-05,
"loss": 0.8758,
"step": 880
},
{
"epoch": 0.32,
"grad_norm": 3.828125,
"learning_rate": 1.438421052631579e-05,
"loss": 0.8534,
"step": 890
},
{
"epoch": 0.32,
"grad_norm": 2.59375,
"learning_rate": 1.4368421052631578e-05,
"loss": 0.8624,
"step": 900
},
{
"epoch": 0.33,
"grad_norm": 3.828125,
"learning_rate": 1.4352631578947369e-05,
"loss": 0.8776,
"step": 910
},
{
"epoch": 0.33,
"grad_norm": 4.5625,
"learning_rate": 1.4336842105263159e-05,
"loss": 0.7811,
"step": 920
},
{
"epoch": 0.33,
"grad_norm": 3.71875,
"learning_rate": 1.4321052631578948e-05,
"loss": 0.9363,
"step": 930
},
{
"epoch": 0.34,
"grad_norm": 3.359375,
"learning_rate": 1.4305263157894738e-05,
"loss": 0.87,
"step": 940
},
{
"epoch": 0.34,
"grad_norm": 3.875,
"learning_rate": 1.4289473684210527e-05,
"loss": 0.8802,
"step": 950
},
{
"epoch": 0.34,
"grad_norm": 3.5625,
"learning_rate": 1.4273684210526317e-05,
"loss": 0.8649,
"step": 960
},
{
"epoch": 0.35,
"grad_norm": 5.1875,
"learning_rate": 1.4257894736842106e-05,
"loss": 0.8778,
"step": 970
},
{
"epoch": 0.35,
"grad_norm": 5.1875,
"learning_rate": 1.4242105263157896e-05,
"loss": 0.9465,
"step": 980
},
{
"epoch": 0.36,
"grad_norm": 3.40625,
"learning_rate": 1.4226315789473685e-05,
"loss": 0.7351,
"step": 990
},
{
"epoch": 0.36,
"grad_norm": 3.65625,
"learning_rate": 1.4210526315789473e-05,
"loss": 0.8345,
"step": 1000
},
{
"epoch": 0.36,
"grad_norm": 4.34375,
"learning_rate": 1.4194736842105264e-05,
"loss": 0.844,
"step": 1010
},
{
"epoch": 0.37,
"grad_norm": 3.546875,
"learning_rate": 1.4178947368421052e-05,
"loss": 0.8772,
"step": 1020
},
{
"epoch": 0.37,
"grad_norm": 3.859375,
"learning_rate": 1.4163157894736843e-05,
"loss": 0.8421,
"step": 1030
},
{
"epoch": 0.37,
"grad_norm": 4.1875,
"learning_rate": 1.4147368421052631e-05,
"loss": 0.8795,
"step": 1040
},
{
"epoch": 0.38,
"grad_norm": 3.28125,
"learning_rate": 1.4131578947368422e-05,
"loss": 0.7592,
"step": 1050
},
{
"epoch": 0.38,
"grad_norm": 5.375,
"learning_rate": 1.411578947368421e-05,
"loss": 0.8124,
"step": 1060
},
{
"epoch": 0.38,
"grad_norm": 2.84375,
"learning_rate": 1.4099999999999999e-05,
"loss": 0.7272,
"step": 1070
},
{
"epoch": 0.39,
"grad_norm": 3.125,
"learning_rate": 1.408421052631579e-05,
"loss": 0.7565,
"step": 1080
},
{
"epoch": 0.39,
"grad_norm": 3.34375,
"learning_rate": 1.406842105263158e-05,
"loss": 0.957,
"step": 1090
},
{
"epoch": 0.39,
"grad_norm": 4.78125,
"learning_rate": 1.405263157894737e-05,
"loss": 0.8601,
"step": 1100
},
{
"epoch": 0.4,
"grad_norm": 2.828125,
"learning_rate": 1.4036842105263158e-05,
"loss": 0.9521,
"step": 1110
},
{
"epoch": 0.4,
"grad_norm": 3.3125,
"learning_rate": 1.4021052631578949e-05,
"loss": 0.8389,
"step": 1120
},
{
"epoch": 0.41,
"grad_norm": 5.40625,
"learning_rate": 1.4005263157894737e-05,
"loss": 0.7895,
"step": 1130
},
{
"epoch": 0.41,
"grad_norm": 3.546875,
"learning_rate": 1.3989473684210526e-05,
"loss": 0.9063,
"step": 1140
},
{
"epoch": 0.41,
"grad_norm": 3.015625,
"learning_rate": 1.3973684210526316e-05,
"loss": 0.8427,
"step": 1150
},
{
"epoch": 0.42,
"grad_norm": 4.25,
"learning_rate": 1.3957894736842105e-05,
"loss": 0.8178,
"step": 1160
},
{
"epoch": 0.42,
"grad_norm": 3.546875,
"learning_rate": 1.3942105263157895e-05,
"loss": 0.8688,
"step": 1170
},
{
"epoch": 0.42,
"grad_norm": 3.4375,
"learning_rate": 1.3926315789473684e-05,
"loss": 0.8471,
"step": 1180
},
{
"epoch": 0.43,
"grad_norm": 4.28125,
"learning_rate": 1.3910526315789474e-05,
"loss": 0.7473,
"step": 1190
},
{
"epoch": 0.43,
"grad_norm": 3.609375,
"learning_rate": 1.3894736842105263e-05,
"loss": 0.903,
"step": 1200
},
{
"epoch": 0.43,
"grad_norm": 3.46875,
"learning_rate": 1.3878947368421052e-05,
"loss": 0.9214,
"step": 1210
},
{
"epoch": 0.44,
"grad_norm": 3.75,
"learning_rate": 1.3863157894736842e-05,
"loss": 0.9448,
"step": 1220
},
{
"epoch": 0.44,
"grad_norm": 2.609375,
"learning_rate": 1.384736842105263e-05,
"loss": 0.7912,
"step": 1230
},
{
"epoch": 0.44,
"grad_norm": 5.71875,
"learning_rate": 1.3831578947368421e-05,
"loss": 0.887,
"step": 1240
},
{
"epoch": 0.45,
"grad_norm": 3.15625,
"learning_rate": 1.3815789473684211e-05,
"loss": 0.8479,
"step": 1250
},
{
"epoch": 0.45,
"grad_norm": 3.125,
"learning_rate": 1.3800000000000002e-05,
"loss": 0.7985,
"step": 1260
},
{
"epoch": 0.46,
"grad_norm": 4.03125,
"learning_rate": 1.378421052631579e-05,
"loss": 0.7114,
"step": 1270
},
{
"epoch": 0.46,
"grad_norm": 3.765625,
"learning_rate": 1.3768421052631579e-05,
"loss": 0.8742,
"step": 1280
},
{
"epoch": 0.46,
"grad_norm": 3.984375,
"learning_rate": 1.375263157894737e-05,
"loss": 0.8908,
"step": 1290
},
{
"epoch": 0.47,
"grad_norm": 3.640625,
"learning_rate": 1.3736842105263158e-05,
"loss": 0.8237,
"step": 1300
},
{
"epoch": 0.47,
"grad_norm": 3.78125,
"learning_rate": 1.3721052631578948e-05,
"loss": 0.7752,
"step": 1310
},
{
"epoch": 0.47,
"grad_norm": 4.71875,
"learning_rate": 1.3705263157894737e-05,
"loss": 0.8437,
"step": 1320
},
{
"epoch": 0.48,
"grad_norm": 2.875,
"learning_rate": 1.3689473684210527e-05,
"loss": 0.7395,
"step": 1330
},
{
"epoch": 0.48,
"grad_norm": 3.375,
"learning_rate": 1.3673684210526316e-05,
"loss": 0.9191,
"step": 1340
},
{
"epoch": 0.48,
"grad_norm": 4.65625,
"learning_rate": 1.3657894736842106e-05,
"loss": 0.8401,
"step": 1350
},
{
"epoch": 0.49,
"grad_norm": 3.625,
"learning_rate": 1.3642105263157895e-05,
"loss": 0.7785,
"step": 1360
},
{
"epoch": 0.49,
"grad_norm": 4.25,
"learning_rate": 1.3626315789473684e-05,
"loss": 0.9515,
"step": 1370
},
{
"epoch": 0.49,
"grad_norm": 2.78125,
"learning_rate": 1.3610526315789474e-05,
"loss": 0.8812,
"step": 1380
},
{
"epoch": 0.5,
"grad_norm": 2.296875,
"learning_rate": 1.3594736842105263e-05,
"loss": 0.8685,
"step": 1390
},
{
"epoch": 0.5,
"grad_norm": 4.03125,
"learning_rate": 1.3578947368421053e-05,
"loss": 0.9704,
"step": 1400
},
{
"epoch": 0.51,
"grad_norm": 5.375,
"learning_rate": 1.3563157894736842e-05,
"loss": 0.7972,
"step": 1410
},
{
"epoch": 0.51,
"grad_norm": 2.3125,
"learning_rate": 1.3547368421052634e-05,
"loss": 0.8166,
"step": 1420
},
{
"epoch": 0.51,
"grad_norm": 4.0625,
"learning_rate": 1.3531578947368422e-05,
"loss": 0.8718,
"step": 1430
},
{
"epoch": 0.52,
"grad_norm": 3.75,
"learning_rate": 1.3515789473684211e-05,
"loss": 0.8607,
"step": 1440
},
{
"epoch": 0.52,
"grad_norm": 4.0625,
"learning_rate": 1.3500000000000001e-05,
"loss": 0.8328,
"step": 1450
},
{
"epoch": 0.52,
"grad_norm": 3.96875,
"learning_rate": 1.348421052631579e-05,
"loss": 0.925,
"step": 1460
},
{
"epoch": 0.53,
"grad_norm": 4.71875,
"learning_rate": 1.346842105263158e-05,
"loss": 0.9331,
"step": 1470
},
{
"epoch": 0.53,
"grad_norm": 3.421875,
"learning_rate": 1.3452631578947369e-05,
"loss": 0.8584,
"step": 1480
},
{
"epoch": 0.53,
"grad_norm": 3.578125,
"learning_rate": 1.343684210526316e-05,
"loss": 0.8312,
"step": 1490
},
{
"epoch": 0.54,
"grad_norm": 3.0,
"learning_rate": 1.3421052631578948e-05,
"loss": 0.8311,
"step": 1500
},
{
"epoch": 0.54,
"grad_norm": 3.859375,
"learning_rate": 1.3405263157894736e-05,
"loss": 0.8207,
"step": 1510
},
{
"epoch": 0.55,
"grad_norm": 3.640625,
"learning_rate": 1.3389473684210527e-05,
"loss": 0.9551,
"step": 1520
},
{
"epoch": 0.55,
"grad_norm": 4.6875,
"learning_rate": 1.3373684210526315e-05,
"loss": 0.8096,
"step": 1530
},
{
"epoch": 0.55,
"grad_norm": 3.90625,
"learning_rate": 1.3357894736842106e-05,
"loss": 0.8908,
"step": 1540
},
{
"epoch": 0.56,
"grad_norm": 3.796875,
"learning_rate": 1.3342105263157894e-05,
"loss": 0.7875,
"step": 1550
},
{
"epoch": 0.56,
"grad_norm": 3.234375,
"learning_rate": 1.3326315789473685e-05,
"loss": 0.7913,
"step": 1560
},
{
"epoch": 0.56,
"grad_norm": 3.890625,
"learning_rate": 1.3310526315789473e-05,
"loss": 0.7862,
"step": 1570
},
{
"epoch": 0.57,
"grad_norm": 5.1875,
"learning_rate": 1.3294736842105262e-05,
"loss": 0.9223,
"step": 1580
},
{
"epoch": 0.57,
"grad_norm": 3.453125,
"learning_rate": 1.3278947368421054e-05,
"loss": 0.8516,
"step": 1590
},
{
"epoch": 0.57,
"grad_norm": 2.78125,
"learning_rate": 1.3263157894736843e-05,
"loss": 0.8969,
"step": 1600
},
{
"epoch": 0.58,
"grad_norm": 2.984375,
"learning_rate": 1.3247368421052633e-05,
"loss": 0.8309,
"step": 1610
},
{
"epoch": 0.58,
"grad_norm": 3.734375,
"learning_rate": 1.3231578947368422e-05,
"loss": 0.7816,
"step": 1620
},
{
"epoch": 0.58,
"grad_norm": 5.0625,
"learning_rate": 1.3215789473684212e-05,
"loss": 0.8899,
"step": 1630
},
{
"epoch": 0.59,
"grad_norm": 4.21875,
"learning_rate": 1.32e-05,
"loss": 0.9195,
"step": 1640
},
{
"epoch": 0.59,
"grad_norm": 3.71875,
"learning_rate": 1.318421052631579e-05,
"loss": 0.8773,
"step": 1650
},
{
"epoch": 0.6,
"grad_norm": 3.75,
"learning_rate": 1.316842105263158e-05,
"loss": 0.8922,
"step": 1660
},
{
"epoch": 0.6,
"grad_norm": 4.0,
"learning_rate": 1.3152631578947368e-05,
"loss": 0.8647,
"step": 1670
},
{
"epoch": 0.6,
"grad_norm": 3.296875,
"learning_rate": 1.3136842105263159e-05,
"loss": 0.8625,
"step": 1680
},
{
"epoch": 0.61,
"grad_norm": 4.21875,
"learning_rate": 1.3121052631578947e-05,
"loss": 0.89,
"step": 1690
},
{
"epoch": 0.61,
"grad_norm": 3.546875,
"learning_rate": 1.3105263157894738e-05,
"loss": 0.8426,
"step": 1700
},
{
"epoch": 0.61,
"grad_norm": 3.734375,
"learning_rate": 1.3089473684210526e-05,
"loss": 0.8909,
"step": 1710
},
{
"epoch": 0.62,
"grad_norm": 4.25,
"learning_rate": 1.3073684210526315e-05,
"loss": 0.8788,
"step": 1720
},
{
"epoch": 0.62,
"grad_norm": 3.875,
"learning_rate": 1.3057894736842105e-05,
"loss": 0.8895,
"step": 1730
},
{
"epoch": 0.62,
"grad_norm": 4.0,
"learning_rate": 1.3042105263157894e-05,
"loss": 0.883,
"step": 1740
},
{
"epoch": 0.63,
"grad_norm": 4.78125,
"learning_rate": 1.3026315789473684e-05,
"loss": 0.8837,
"step": 1750
},
{
"epoch": 0.63,
"grad_norm": 4.34375,
"learning_rate": 1.3010526315789475e-05,
"loss": 0.8173,
"step": 1760
},
{
"epoch": 0.63,
"grad_norm": 3.96875,
"learning_rate": 1.2994736842105265e-05,
"loss": 0.8328,
"step": 1770
},
{
"epoch": 0.64,
"grad_norm": 2.84375,
"learning_rate": 1.2978947368421054e-05,
"loss": 0.8218,
"step": 1780
},
{
"epoch": 0.64,
"grad_norm": 4.625,
"learning_rate": 1.2963157894736842e-05,
"loss": 0.89,
"step": 1790
},
{
"epoch": 0.65,
"grad_norm": 3.640625,
"learning_rate": 1.2947368421052633e-05,
"loss": 0.9269,
"step": 1800
},
{
"epoch": 0.65,
"grad_norm": 4.1875,
"learning_rate": 1.2931578947368421e-05,
"loss": 0.7664,
"step": 1810
},
{
"epoch": 0.65,
"grad_norm": 5.34375,
"learning_rate": 1.2915789473684212e-05,
"loss": 0.9444,
"step": 1820
},
{
"epoch": 0.66,
"grad_norm": 3.375,
"learning_rate": 1.29e-05,
"loss": 0.8094,
"step": 1830
},
{
"epoch": 0.66,
"grad_norm": 3.90625,
"learning_rate": 1.288421052631579e-05,
"loss": 0.7694,
"step": 1840
},
{
"epoch": 0.66,
"grad_norm": 4.59375,
"learning_rate": 1.2868421052631579e-05,
"loss": 0.8341,
"step": 1850
},
{
"epoch": 0.67,
"grad_norm": 3.90625,
"learning_rate": 1.2852631578947368e-05,
"loss": 0.811,
"step": 1860
},
{
"epoch": 0.67,
"grad_norm": 2.8125,
"learning_rate": 1.2836842105263158e-05,
"loss": 0.8261,
"step": 1870
},
{
"epoch": 0.67,
"grad_norm": 3.65625,
"learning_rate": 1.2821052631578947e-05,
"loss": 0.8654,
"step": 1880
},
{
"epoch": 0.68,
"grad_norm": 4.1875,
"learning_rate": 1.2805263157894737e-05,
"loss": 0.9158,
"step": 1890
},
{
"epoch": 0.68,
"grad_norm": 4.125,
"learning_rate": 1.2789473684210526e-05,
"loss": 0.7241,
"step": 1900
},
{
"epoch": 0.68,
"grad_norm": 4.59375,
"learning_rate": 1.2773684210526316e-05,
"loss": 0.84,
"step": 1910
},
{
"epoch": 0.69,
"grad_norm": 2.765625,
"learning_rate": 1.2757894736842106e-05,
"loss": 0.9277,
"step": 1920
},
{
"epoch": 0.69,
"grad_norm": 3.390625,
"learning_rate": 1.2742105263157895e-05,
"loss": 0.8024,
"step": 1930
},
{
"epoch": 0.7,
"grad_norm": 6.34375,
"learning_rate": 1.2726315789473685e-05,
"loss": 0.7964,
"step": 1940
},
{
"epoch": 0.7,
"grad_norm": 2.875,
"learning_rate": 1.2710526315789474e-05,
"loss": 0.8335,
"step": 1950
},
{
"epoch": 0.7,
"grad_norm": 3.0,
"learning_rate": 1.2694736842105264e-05,
"loss": 0.8329,
"step": 1960
},
{
"epoch": 0.71,
"grad_norm": 3.4375,
"learning_rate": 1.2678947368421053e-05,
"loss": 0.8553,
"step": 1970
},
{
"epoch": 0.71,
"grad_norm": 4.25,
"learning_rate": 1.2663157894736843e-05,
"loss": 0.8801,
"step": 1980
},
{
"epoch": 0.71,
"grad_norm": 4.03125,
"learning_rate": 1.2647368421052632e-05,
"loss": 0.813,
"step": 1990
},
{
"epoch": 0.72,
"grad_norm": 4.8125,
"learning_rate": 1.263157894736842e-05,
"loss": 0.7953,
"step": 2000
},
{
"epoch": 0.72,
"eval_loss": 0.965668797492981,
"eval_runtime": 63.3755,
"eval_samples_per_second": 15.779,
"eval_steps_per_second": 15.779,
"step": 2000
},
{
"epoch": 0.72,
"grad_norm": 3.9375,
"learning_rate": 1.2615789473684211e-05,
"loss": 0.7854,
"step": 2010
},
{
"epoch": 0.72,
"grad_norm": 4.28125,
"learning_rate": 1.26e-05,
"loss": 0.9323,
"step": 2020
},
{
"epoch": 0.73,
"grad_norm": 3.1875,
"learning_rate": 1.258421052631579e-05,
"loss": 0.8514,
"step": 2030
},
{
"epoch": 0.73,
"grad_norm": 3.09375,
"learning_rate": 1.2568421052631579e-05,
"loss": 0.7566,
"step": 2040
},
{
"epoch": 0.74,
"grad_norm": 6.90625,
"learning_rate": 1.2552631578947369e-05,
"loss": 0.8933,
"step": 2050
},
{
"epoch": 0.74,
"grad_norm": 4.59375,
"learning_rate": 1.2536842105263158e-05,
"loss": 0.7971,
"step": 2060
},
{
"epoch": 0.74,
"grad_norm": 2.859375,
"learning_rate": 1.2521052631578946e-05,
"loss": 0.7583,
"step": 2070
},
{
"epoch": 0.75,
"grad_norm": 3.1875,
"learning_rate": 1.2505263157894737e-05,
"loss": 0.844,
"step": 2080
},
{
"epoch": 0.75,
"grad_norm": 3.921875,
"learning_rate": 1.2489473684210527e-05,
"loss": 0.8764,
"step": 2090
},
{
"epoch": 0.75,
"grad_norm": 5.09375,
"learning_rate": 1.2473684210526317e-05,
"loss": 0.8263,
"step": 2100
},
{
"epoch": 0.76,
"grad_norm": 3.796875,
"learning_rate": 1.2457894736842106e-05,
"loss": 0.876,
"step": 2110
},
{
"epoch": 0.76,
"grad_norm": 2.625,
"learning_rate": 1.2442105263157896e-05,
"loss": 0.8852,
"step": 2120
},
{
"epoch": 0.76,
"grad_norm": 3.375,
"learning_rate": 1.2426315789473685e-05,
"loss": 0.8555,
"step": 2130
},
{
"epoch": 0.77,
"grad_norm": 3.328125,
"learning_rate": 1.2410526315789474e-05,
"loss": 0.8724,
"step": 2140
},
{
"epoch": 0.77,
"grad_norm": 3.265625,
"learning_rate": 1.2394736842105264e-05,
"loss": 0.8052,
"step": 2150
},
{
"epoch": 0.77,
"grad_norm": 4.375,
"learning_rate": 1.2378947368421053e-05,
"loss": 0.6949,
"step": 2160
},
{
"epoch": 0.78,
"grad_norm": 4.4375,
"learning_rate": 1.2363157894736843e-05,
"loss": 0.8075,
"step": 2170
},
{
"epoch": 0.78,
"grad_norm": 3.09375,
"learning_rate": 1.2347368421052631e-05,
"loss": 0.8539,
"step": 2180
},
{
"epoch": 0.79,
"grad_norm": 3.640625,
"learning_rate": 1.2331578947368422e-05,
"loss": 0.7066,
"step": 2190
},
{
"epoch": 0.79,
"grad_norm": 4.125,
"learning_rate": 1.231578947368421e-05,
"loss": 0.8651,
"step": 2200
},
{
"epoch": 0.79,
"grad_norm": 4.75,
"learning_rate": 1.2299999999999999e-05,
"loss": 0.7921,
"step": 2210
},
{
"epoch": 0.8,
"grad_norm": 3.875,
"learning_rate": 1.228421052631579e-05,
"loss": 0.7711,
"step": 2220
},
{
"epoch": 0.8,
"grad_norm": 4.21875,
"learning_rate": 1.2268421052631578e-05,
"loss": 0.8328,
"step": 2230
},
{
"epoch": 0.8,
"grad_norm": 2.96875,
"learning_rate": 1.2252631578947368e-05,
"loss": 0.872,
"step": 2240
},
{
"epoch": 0.81,
"grad_norm": 3.203125,
"learning_rate": 1.2236842105263159e-05,
"loss": 0.7755,
"step": 2250
},
{
"epoch": 0.81,
"grad_norm": 3.15625,
"learning_rate": 1.2221052631578949e-05,
"loss": 0.7647,
"step": 2260
},
{
"epoch": 0.81,
"grad_norm": 3.0,
"learning_rate": 1.2205263157894738e-05,
"loss": 0.8639,
"step": 2270
},
{
"epoch": 0.82,
"grad_norm": 4.625,
"learning_rate": 1.2189473684210526e-05,
"loss": 0.8007,
"step": 2280
},
{
"epoch": 0.82,
"grad_norm": 4.78125,
"learning_rate": 1.2173684210526317e-05,
"loss": 0.8344,
"step": 2290
},
{
"epoch": 0.82,
"grad_norm": 3.140625,
"learning_rate": 1.2157894736842105e-05,
"loss": 0.9327,
"step": 2300
},
{
"epoch": 0.83,
"grad_norm": 4.09375,
"learning_rate": 1.2142105263157896e-05,
"loss": 0.8617,
"step": 2310
},
{
"epoch": 0.83,
"grad_norm": 4.1875,
"learning_rate": 1.2126315789473684e-05,
"loss": 0.8129,
"step": 2320
},
{
"epoch": 0.84,
"grad_norm": 3.171875,
"learning_rate": 1.2110526315789475e-05,
"loss": 0.9715,
"step": 2330
},
{
"epoch": 0.84,
"grad_norm": 4.8125,
"learning_rate": 1.2094736842105263e-05,
"loss": 0.8432,
"step": 2340
},
{
"epoch": 0.84,
"grad_norm": 3.71875,
"learning_rate": 1.2078947368421052e-05,
"loss": 0.7863,
"step": 2350
},
{
"epoch": 0.85,
"grad_norm": 3.671875,
"learning_rate": 1.2063157894736842e-05,
"loss": 0.8806,
"step": 2360
},
{
"epoch": 0.85,
"grad_norm": 3.390625,
"learning_rate": 1.2047368421052631e-05,
"loss": 0.8683,
"step": 2370
},
{
"epoch": 0.85,
"grad_norm": 3.5625,
"learning_rate": 1.2031578947368421e-05,
"loss": 0.8558,
"step": 2380
},
{
"epoch": 0.86,
"grad_norm": 3.328125,
"learning_rate": 1.201578947368421e-05,
"loss": 0.7749,
"step": 2390
},
{
"epoch": 0.86,
"grad_norm": 5.0,
"learning_rate": 1.2e-05,
"loss": 0.8283,
"step": 2400
},
{
"epoch": 0.86,
"grad_norm": 4.40625,
"learning_rate": 1.1984210526315789e-05,
"loss": 0.8553,
"step": 2410
},
{
"epoch": 0.87,
"grad_norm": 3.546875,
"learning_rate": 1.196842105263158e-05,
"loss": 0.8862,
"step": 2420
},
{
"epoch": 0.87,
"grad_norm": 3.59375,
"learning_rate": 1.195263157894737e-05,
"loss": 0.8069,
"step": 2430
},
{
"epoch": 0.88,
"grad_norm": 4.28125,
"learning_rate": 1.1936842105263158e-05,
"loss": 0.9325,
"step": 2440
},
{
"epoch": 0.88,
"grad_norm": 3.546875,
"learning_rate": 1.1921052631578949e-05,
"loss": 0.8751,
"step": 2450
},
{
"epoch": 0.88,
"grad_norm": 5.71875,
"learning_rate": 1.1905263157894737e-05,
"loss": 0.9068,
"step": 2460
},
{
"epoch": 0.89,
"grad_norm": 3.5,
"learning_rate": 1.1889473684210528e-05,
"loss": 0.8121,
"step": 2470
},
{
"epoch": 0.89,
"grad_norm": 3.125,
"learning_rate": 1.1873684210526316e-05,
"loss": 0.8152,
"step": 2480
},
{
"epoch": 0.89,
"grad_norm": 3.5625,
"learning_rate": 1.1857894736842105e-05,
"loss": 0.7132,
"step": 2490
},
{
"epoch": 0.9,
"grad_norm": 4.8125,
"learning_rate": 1.1842105263157895e-05,
"loss": 0.8085,
"step": 2500
},
{
"epoch": 0.9,
"grad_norm": 3.171875,
"learning_rate": 1.1826315789473684e-05,
"loss": 0.8593,
"step": 2510
},
{
"epoch": 0.9,
"grad_norm": 3.984375,
"learning_rate": 1.1810526315789474e-05,
"loss": 0.7501,
"step": 2520
},
{
"epoch": 0.91,
"grad_norm": 2.609375,
"learning_rate": 1.1794736842105263e-05,
"loss": 0.8399,
"step": 2530
},
{
"epoch": 0.91,
"grad_norm": 5.09375,
"learning_rate": 1.1778947368421053e-05,
"loss": 0.6944,
"step": 2540
},
{
"epoch": 0.91,
"grad_norm": 3.796875,
"learning_rate": 1.1763157894736842e-05,
"loss": 0.8347,
"step": 2550
},
{
"epoch": 0.92,
"grad_norm": 3.734375,
"learning_rate": 1.174736842105263e-05,
"loss": 0.8156,
"step": 2560
},
{
"epoch": 0.92,
"grad_norm": 4.25,
"learning_rate": 1.173157894736842e-05,
"loss": 0.712,
"step": 2570
},
{
"epoch": 0.93,
"grad_norm": 4.125,
"learning_rate": 1.171578947368421e-05,
"loss": 0.8281,
"step": 2580
},
{
"epoch": 0.93,
"grad_norm": 3.109375,
"learning_rate": 1.1700000000000001e-05,
"loss": 0.964,
"step": 2590
},
{
"epoch": 0.93,
"grad_norm": 3.890625,
"learning_rate": 1.168421052631579e-05,
"loss": 0.783,
"step": 2600
},
{
"epoch": 0.94,
"grad_norm": 2.9375,
"learning_rate": 1.166842105263158e-05,
"loss": 0.7572,
"step": 2610
},
{
"epoch": 0.94,
"grad_norm": 3.09375,
"learning_rate": 1.1652631578947369e-05,
"loss": 0.8726,
"step": 2620
},
{
"epoch": 0.94,
"grad_norm": 3.6875,
"learning_rate": 1.1636842105263158e-05,
"loss": 0.8499,
"step": 2630
},
{
"epoch": 0.95,
"grad_norm": 4.09375,
"learning_rate": 1.1621052631578948e-05,
"loss": 0.7741,
"step": 2640
},
{
"epoch": 0.95,
"grad_norm": 2.703125,
"learning_rate": 1.1605263157894737e-05,
"loss": 0.7522,
"step": 2650
},
{
"epoch": 0.95,
"grad_norm": 3.5,
"learning_rate": 1.1589473684210527e-05,
"loss": 0.8632,
"step": 2660
},
{
"epoch": 0.96,
"grad_norm": 3.875,
"learning_rate": 1.1573684210526316e-05,
"loss": 0.7483,
"step": 2670
},
{
"epoch": 0.96,
"grad_norm": 3.078125,
"learning_rate": 1.1557894736842106e-05,
"loss": 0.7606,
"step": 2680
},
{
"epoch": 0.96,
"grad_norm": 4.875,
"learning_rate": 1.1542105263157895e-05,
"loss": 0.7057,
"step": 2690
},
{
"epoch": 0.97,
"grad_norm": 3.015625,
"learning_rate": 1.1526315789473683e-05,
"loss": 0.7932,
"step": 2700
},
{
"epoch": 0.97,
"grad_norm": 3.234375,
"learning_rate": 1.1510526315789474e-05,
"loss": 0.7906,
"step": 2710
},
{
"epoch": 0.98,
"grad_norm": 4.03125,
"learning_rate": 1.1494736842105262e-05,
"loss": 0.8328,
"step": 2720
},
{
"epoch": 0.98,
"grad_norm": 3.234375,
"learning_rate": 1.1478947368421053e-05,
"loss": 0.7279,
"step": 2730
},
{
"epoch": 0.98,
"grad_norm": 4.09375,
"learning_rate": 1.1463157894736841e-05,
"loss": 0.8951,
"step": 2740
},
{
"epoch": 0.99,
"grad_norm": 2.875,
"learning_rate": 1.1447368421052632e-05,
"loss": 0.8575,
"step": 2750
},
{
"epoch": 0.99,
"grad_norm": 3.90625,
"learning_rate": 1.1431578947368422e-05,
"loss": 0.8459,
"step": 2760
},
{
"epoch": 0.99,
"grad_norm": 3.0625,
"learning_rate": 1.141578947368421e-05,
"loss": 0.7296,
"step": 2770
},
{
"epoch": 1.0,
"grad_norm": 4.5625,
"learning_rate": 1.1400000000000001e-05,
"loss": 0.8719,
"step": 2780
},
{
"epoch": 1.0,
"grad_norm": 3.71875,
"learning_rate": 1.138421052631579e-05,
"loss": 0.7355,
"step": 2790
},
{
"epoch": 1.0,
"grad_norm": 2.875,
"learning_rate": 1.136842105263158e-05,
"loss": 0.9747,
"step": 2800
},
{
"epoch": 1.01,
"grad_norm": 3.875,
"learning_rate": 1.1352631578947369e-05,
"loss": 0.7992,
"step": 2810
},
{
"epoch": 1.01,
"grad_norm": 3.390625,
"learning_rate": 1.1336842105263159e-05,
"loss": 0.7168,
"step": 2820
},
{
"epoch": 1.01,
"grad_norm": 3.75,
"learning_rate": 1.1321052631578948e-05,
"loss": 0.6836,
"step": 2830
},
{
"epoch": 1.02,
"grad_norm": 4.0,
"learning_rate": 1.1305263157894736e-05,
"loss": 0.8253,
"step": 2840
},
{
"epoch": 1.02,
"grad_norm": 3.0625,
"learning_rate": 1.1289473684210527e-05,
"loss": 0.8678,
"step": 2850
},
{
"epoch": 1.03,
"grad_norm": 3.234375,
"learning_rate": 1.1273684210526315e-05,
"loss": 0.7433,
"step": 2860
},
{
"epoch": 1.03,
"grad_norm": 4.28125,
"learning_rate": 1.1257894736842106e-05,
"loss": 0.8963,
"step": 2870
},
{
"epoch": 1.03,
"grad_norm": 3.796875,
"learning_rate": 1.1242105263157894e-05,
"loss": 0.9059,
"step": 2880
},
{
"epoch": 1.04,
"grad_norm": 4.21875,
"learning_rate": 1.1226315789473685e-05,
"loss": 0.7148,
"step": 2890
},
{
"epoch": 1.04,
"grad_norm": 4.03125,
"learning_rate": 1.1210526315789473e-05,
"loss": 0.8227,
"step": 2900
},
{
"epoch": 1.04,
"grad_norm": 3.421875,
"learning_rate": 1.1194736842105264e-05,
"loss": 0.7934,
"step": 2910
},
{
"epoch": 1.05,
"grad_norm": 4.03125,
"learning_rate": 1.1178947368421054e-05,
"loss": 0.8192,
"step": 2920
},
{
"epoch": 1.05,
"grad_norm": 5.1875,
"learning_rate": 1.1163157894736842e-05,
"loss": 0.7516,
"step": 2930
},
{
"epoch": 1.05,
"grad_norm": 4.46875,
"learning_rate": 1.1147368421052633e-05,
"loss": 0.8524,
"step": 2940
},
{
"epoch": 1.06,
"grad_norm": 3.265625,
"learning_rate": 1.1131578947368421e-05,
"loss": 0.7609,
"step": 2950
},
{
"epoch": 1.06,
"grad_norm": 2.625,
"learning_rate": 1.1115789473684212e-05,
"loss": 0.7896,
"step": 2960
},
{
"epoch": 1.07,
"grad_norm": 3.4375,
"learning_rate": 1.11e-05,
"loss": 0.7638,
"step": 2970
},
{
"epoch": 1.07,
"grad_norm": 4.125,
"learning_rate": 1.108421052631579e-05,
"loss": 0.7804,
"step": 2980
},
{
"epoch": 1.07,
"grad_norm": 3.75,
"learning_rate": 1.106842105263158e-05,
"loss": 0.7906,
"step": 2990
},
{
"epoch": 1.08,
"grad_norm": 4.0,
"learning_rate": 1.1052631578947368e-05,
"loss": 0.8006,
"step": 3000
},
{
"epoch": 1.08,
"grad_norm": 3.609375,
"learning_rate": 1.1036842105263158e-05,
"loss": 0.8677,
"step": 3010
},
{
"epoch": 1.08,
"grad_norm": 3.953125,
"learning_rate": 1.1021052631578947e-05,
"loss": 0.9164,
"step": 3020
},
{
"epoch": 1.09,
"grad_norm": 4.34375,
"learning_rate": 1.1005263157894737e-05,
"loss": 0.8206,
"step": 3030
},
{
"epoch": 1.09,
"grad_norm": 3.953125,
"learning_rate": 1.0989473684210526e-05,
"loss": 0.8203,
"step": 3040
},
{
"epoch": 1.09,
"grad_norm": 3.8125,
"learning_rate": 1.0973684210526316e-05,
"loss": 0.7283,
"step": 3050
},
{
"epoch": 1.1,
"grad_norm": 4.46875,
"learning_rate": 1.0957894736842105e-05,
"loss": 0.7454,
"step": 3060
},
{
"epoch": 1.1,
"grad_norm": 4.09375,
"learning_rate": 1.0942105263157894e-05,
"loss": 0.7764,
"step": 3070
},
{
"epoch": 1.1,
"grad_norm": 3.15625,
"learning_rate": 1.0926315789473684e-05,
"loss": 0.7876,
"step": 3080
},
{
"epoch": 1.11,
"grad_norm": 2.953125,
"learning_rate": 1.0910526315789474e-05,
"loss": 0.7466,
"step": 3090
},
{
"epoch": 1.11,
"grad_norm": 3.859375,
"learning_rate": 1.0894736842105265e-05,
"loss": 0.7939,
"step": 3100
},
{
"epoch": 1.12,
"grad_norm": 3.734375,
"learning_rate": 1.0878947368421053e-05,
"loss": 0.8303,
"step": 3110
},
{
"epoch": 1.12,
"grad_norm": 4.34375,
"learning_rate": 1.0863157894736844e-05,
"loss": 0.7779,
"step": 3120
},
{
"epoch": 1.12,
"grad_norm": 4.9375,
"learning_rate": 1.0847368421052632e-05,
"loss": 0.9279,
"step": 3130
},
{
"epoch": 1.13,
"grad_norm": 3.5625,
"learning_rate": 1.0831578947368421e-05,
"loss": 0.8196,
"step": 3140
},
{
"epoch": 1.13,
"grad_norm": 3.375,
"learning_rate": 1.0815789473684211e-05,
"loss": 0.7357,
"step": 3150
},
{
"epoch": 1.13,
"grad_norm": 4.5625,
"learning_rate": 1.08e-05,
"loss": 0.8071,
"step": 3160
},
{
"epoch": 1.14,
"grad_norm": 4.125,
"learning_rate": 1.078421052631579e-05,
"loss": 0.8873,
"step": 3170
},
{
"epoch": 1.14,
"grad_norm": 3.796875,
"learning_rate": 1.0768421052631579e-05,
"loss": 0.8192,
"step": 3180
},
{
"epoch": 1.14,
"grad_norm": 3.453125,
"learning_rate": 1.075263157894737e-05,
"loss": 0.7588,
"step": 3190
},
{
"epoch": 1.15,
"grad_norm": 4.03125,
"learning_rate": 1.0736842105263158e-05,
"loss": 0.7912,
"step": 3200
},
{
"epoch": 1.15,
"grad_norm": 4.25,
"learning_rate": 1.0721052631578947e-05,
"loss": 0.8396,
"step": 3210
},
{
"epoch": 1.15,
"grad_norm": 3.15625,
"learning_rate": 1.0705263157894737e-05,
"loss": 0.8006,
"step": 3220
},
{
"epoch": 1.16,
"grad_norm": 4.96875,
"learning_rate": 1.0689473684210526e-05,
"loss": 0.8191,
"step": 3230
},
{
"epoch": 1.16,
"grad_norm": 3.859375,
"learning_rate": 1.0673684210526316e-05,
"loss": 0.7822,
"step": 3240
},
{
"epoch": 1.17,
"grad_norm": 3.09375,
"learning_rate": 1.0657894736842106e-05,
"loss": 0.6842,
"step": 3250
},
{
"epoch": 1.17,
"grad_norm": 3.953125,
"learning_rate": 1.0642105263157897e-05,
"loss": 0.6702,
"step": 3260
},
{
"epoch": 1.17,
"grad_norm": 4.125,
"learning_rate": 1.0626315789473685e-05,
"loss": 0.943,
"step": 3270
},
{
"epoch": 1.18,
"grad_norm": 3.9375,
"learning_rate": 1.0610526315789474e-05,
"loss": 0.7917,
"step": 3280
},
{
"epoch": 1.18,
"grad_norm": 4.375,
"learning_rate": 1.0594736842105264e-05,
"loss": 0.7856,
"step": 3290
},
{
"epoch": 1.18,
"grad_norm": 5.1875,
"learning_rate": 1.0578947368421053e-05,
"loss": 0.7467,
"step": 3300
},
{
"epoch": 1.19,
"grad_norm": 4.375,
"learning_rate": 1.0563157894736843e-05,
"loss": 0.8295,
"step": 3310
},
{
"epoch": 1.19,
"grad_norm": 4.34375,
"learning_rate": 1.0547368421052632e-05,
"loss": 0.8781,
"step": 3320
},
{
"epoch": 1.19,
"grad_norm": 2.390625,
"learning_rate": 1.0531578947368422e-05,
"loss": 0.7586,
"step": 3330
},
{
"epoch": 1.2,
"grad_norm": 3.984375,
"learning_rate": 1.051578947368421e-05,
"loss": 0.8071,
"step": 3340
},
{
"epoch": 1.2,
"grad_norm": 3.5625,
"learning_rate": 1.05e-05,
"loss": 0.7813,
"step": 3350
},
{
"epoch": 1.2,
"grad_norm": 4.09375,
"learning_rate": 1.048421052631579e-05,
"loss": 0.7212,
"step": 3360
},
{
"epoch": 1.21,
"grad_norm": 3.921875,
"learning_rate": 1.0468421052631578e-05,
"loss": 0.8672,
"step": 3370
},
{
"epoch": 1.21,
"grad_norm": 4.15625,
"learning_rate": 1.0452631578947369e-05,
"loss": 0.7315,
"step": 3380
},
{
"epoch": 1.22,
"grad_norm": 4.03125,
"learning_rate": 1.0436842105263157e-05,
"loss": 0.7626,
"step": 3390
},
{
"epoch": 1.22,
"grad_norm": 3.890625,
"learning_rate": 1.0421052631578948e-05,
"loss": 0.7502,
"step": 3400
},
{
"epoch": 1.22,
"grad_norm": 3.203125,
"learning_rate": 1.0405263157894736e-05,
"loss": 0.813,
"step": 3410
},
{
"epoch": 1.23,
"grad_norm": 3.796875,
"learning_rate": 1.0389473684210527e-05,
"loss": 0.703,
"step": 3420
},
{
"epoch": 1.23,
"grad_norm": 3.421875,
"learning_rate": 1.0373684210526317e-05,
"loss": 0.7764,
"step": 3430
},
{
"epoch": 1.23,
"grad_norm": 3.765625,
"learning_rate": 1.0357894736842106e-05,
"loss": 0.7738,
"step": 3440
},
{
"epoch": 1.24,
"grad_norm": 3.40625,
"learning_rate": 1.0342105263157896e-05,
"loss": 0.8006,
"step": 3450
},
{
"epoch": 1.24,
"grad_norm": 3.046875,
"learning_rate": 1.0326315789473685e-05,
"loss": 0.6495,
"step": 3460
},
{
"epoch": 1.24,
"grad_norm": 3.359375,
"learning_rate": 1.0310526315789475e-05,
"loss": 0.8163,
"step": 3470
},
{
"epoch": 1.25,
"grad_norm": 2.921875,
"learning_rate": 1.0294736842105264e-05,
"loss": 0.7458,
"step": 3480
},
{
"epoch": 1.25,
"grad_norm": 4.21875,
"learning_rate": 1.0278947368421052e-05,
"loss": 0.8422,
"step": 3490
},
{
"epoch": 1.26,
"grad_norm": 4.03125,
"learning_rate": 1.0263157894736843e-05,
"loss": 0.8339,
"step": 3500
},
{
"epoch": 1.26,
"grad_norm": 2.96875,
"learning_rate": 1.0247368421052631e-05,
"loss": 0.7214,
"step": 3510
},
{
"epoch": 1.26,
"grad_norm": 4.0625,
"learning_rate": 1.0231578947368422e-05,
"loss": 0.7303,
"step": 3520
},
{
"epoch": 1.27,
"grad_norm": 4.0,
"learning_rate": 1.021578947368421e-05,
"loss": 0.9404,
"step": 3530
},
{
"epoch": 1.27,
"grad_norm": 4.0,
"learning_rate": 1.02e-05,
"loss": 0.657,
"step": 3540
},
{
"epoch": 1.27,
"grad_norm": 4.65625,
"learning_rate": 1.018421052631579e-05,
"loss": 0.7938,
"step": 3550
},
{
"epoch": 1.28,
"grad_norm": 3.671875,
"learning_rate": 1.0168421052631578e-05,
"loss": 0.769,
"step": 3560
},
{
"epoch": 1.28,
"grad_norm": 3.765625,
"learning_rate": 1.0152631578947368e-05,
"loss": 0.8255,
"step": 3570
},
{
"epoch": 1.28,
"grad_norm": 3.78125,
"learning_rate": 1.0136842105263157e-05,
"loss": 0.7263,
"step": 3580
},
{
"epoch": 1.29,
"grad_norm": 4.59375,
"learning_rate": 1.0121052631578949e-05,
"loss": 0.669,
"step": 3590
},
{
"epoch": 1.29,
"grad_norm": 4.3125,
"learning_rate": 1.0105263157894738e-05,
"loss": 0.85,
"step": 3600
},
{
"epoch": 1.29,
"grad_norm": 4.875,
"learning_rate": 1.0089473684210528e-05,
"loss": 0.7775,
"step": 3610
},
{
"epoch": 1.3,
"grad_norm": 4.0625,
"learning_rate": 1.0073684210526317e-05,
"loss": 0.7895,
"step": 3620
},
{
"epoch": 1.3,
"grad_norm": 4.25,
"learning_rate": 1.0057894736842105e-05,
"loss": 0.7614,
"step": 3630
},
{
"epoch": 1.31,
"grad_norm": 3.515625,
"learning_rate": 1.0042105263157896e-05,
"loss": 0.7206,
"step": 3640
},
{
"epoch": 1.31,
"grad_norm": 4.75,
"learning_rate": 1.0026315789473684e-05,
"loss": 0.8636,
"step": 3650
},
{
"epoch": 1.31,
"grad_norm": 3.390625,
"learning_rate": 1.0010526315789474e-05,
"loss": 0.8654,
"step": 3660
},
{
"epoch": 1.32,
"grad_norm": 4.1875,
"learning_rate": 9.994736842105263e-06,
"loss": 0.8104,
"step": 3670
},
{
"epoch": 1.32,
"grad_norm": 3.890625,
"learning_rate": 9.978947368421053e-06,
"loss": 0.7288,
"step": 3680
},
{
"epoch": 1.32,
"grad_norm": 4.25,
"learning_rate": 9.963157894736842e-06,
"loss": 0.7018,
"step": 3690
},
{
"epoch": 1.33,
"grad_norm": 2.984375,
"learning_rate": 9.94736842105263e-06,
"loss": 0.8822,
"step": 3700
},
{
"epoch": 1.33,
"grad_norm": 3.234375,
"learning_rate": 9.931578947368421e-06,
"loss": 0.8066,
"step": 3710
},
{
"epoch": 1.33,
"grad_norm": 4.0625,
"learning_rate": 9.91578947368421e-06,
"loss": 0.8029,
"step": 3720
},
{
"epoch": 1.34,
"grad_norm": 4.625,
"learning_rate": 9.9e-06,
"loss": 0.8314,
"step": 3730
},
{
"epoch": 1.34,
"grad_norm": 3.1875,
"learning_rate": 9.884210526315789e-06,
"loss": 0.7684,
"step": 3740
},
{
"epoch": 1.34,
"grad_norm": 3.734375,
"learning_rate": 9.868421052631579e-06,
"loss": 0.8704,
"step": 3750
},
{
"epoch": 1.35,
"grad_norm": 3.0625,
"learning_rate": 9.85263157894737e-06,
"loss": 0.6407,
"step": 3760
},
{
"epoch": 1.35,
"grad_norm": 3.09375,
"learning_rate": 9.836842105263158e-06,
"loss": 0.8706,
"step": 3770
},
{
"epoch": 1.36,
"grad_norm": 3.484375,
"learning_rate": 9.821052631578948e-06,
"loss": 0.6614,
"step": 3780
},
{
"epoch": 1.36,
"grad_norm": 6.25,
"learning_rate": 9.805263157894737e-06,
"loss": 0.9533,
"step": 3790
},
{
"epoch": 1.36,
"grad_norm": 3.546875,
"learning_rate": 9.789473684210527e-06,
"loss": 0.7741,
"step": 3800
},
{
"epoch": 1.37,
"grad_norm": 4.46875,
"learning_rate": 9.773684210526316e-06,
"loss": 0.8563,
"step": 3810
},
{
"epoch": 1.37,
"grad_norm": 3.71875,
"learning_rate": 9.757894736842106e-06,
"loss": 0.7783,
"step": 3820
},
{
"epoch": 1.37,
"grad_norm": 3.65625,
"learning_rate": 9.742105263157895e-06,
"loss": 0.8157,
"step": 3830
},
{
"epoch": 1.38,
"grad_norm": 3.484375,
"learning_rate": 9.726315789473684e-06,
"loss": 0.8049,
"step": 3840
},
{
"epoch": 1.38,
"grad_norm": 4.03125,
"learning_rate": 9.710526315789474e-06,
"loss": 0.7587,
"step": 3850
},
{
"epoch": 1.38,
"grad_norm": 3.59375,
"learning_rate": 9.694736842105263e-06,
"loss": 0.7946,
"step": 3860
},
{
"epoch": 1.39,
"grad_norm": 4.46875,
"learning_rate": 9.678947368421053e-06,
"loss": 0.8342,
"step": 3870
},
{
"epoch": 1.39,
"grad_norm": 3.28125,
"learning_rate": 9.663157894736842e-06,
"loss": 0.7517,
"step": 3880
},
{
"epoch": 1.4,
"grad_norm": 3.34375,
"learning_rate": 9.647368421052632e-06,
"loss": 0.8783,
"step": 3890
},
{
"epoch": 1.4,
"grad_norm": 3.40625,
"learning_rate": 9.63157894736842e-06,
"loss": 0.8379,
"step": 3900
},
{
"epoch": 1.4,
"grad_norm": 3.09375,
"learning_rate": 9.61578947368421e-06,
"loss": 0.7126,
"step": 3910
},
{
"epoch": 1.41,
"grad_norm": 3.8125,
"learning_rate": 9.600000000000001e-06,
"loss": 0.8612,
"step": 3920
},
{
"epoch": 1.41,
"grad_norm": 4.90625,
"learning_rate": 9.58421052631579e-06,
"loss": 0.729,
"step": 3930
},
{
"epoch": 1.41,
"grad_norm": 4.21875,
"learning_rate": 9.56842105263158e-06,
"loss": 0.7619,
"step": 3940
},
{
"epoch": 1.42,
"grad_norm": 4.625,
"learning_rate": 9.552631578947369e-06,
"loss": 0.8175,
"step": 3950
},
{
"epoch": 1.42,
"grad_norm": 3.1875,
"learning_rate": 9.53684210526316e-06,
"loss": 0.6401,
"step": 3960
},
{
"epoch": 1.42,
"grad_norm": 3.953125,
"learning_rate": 9.521052631578948e-06,
"loss": 0.8253,
"step": 3970
},
{
"epoch": 1.43,
"grad_norm": 4.25,
"learning_rate": 9.505263157894737e-06,
"loss": 0.7396,
"step": 3980
},
{
"epoch": 1.43,
"grad_norm": 2.96875,
"learning_rate": 9.489473684210527e-06,
"loss": 0.7403,
"step": 3990
},
{
"epoch": 1.43,
"grad_norm": 3.46875,
"learning_rate": 9.473684210526315e-06,
"loss": 0.7155,
"step": 4000
},
{
"epoch": 1.43,
"eval_loss": 0.9205163717269897,
"eval_runtime": 63.4494,
"eval_samples_per_second": 15.761,
"eval_steps_per_second": 15.761,
"step": 4000
}
],
"logging_steps": 10,
"max_steps": 10000,
"num_input_tokens_seen": 0,
"num_train_epochs": 4,
"save_steps": 2000,
"total_flos": 6.455688167424e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}