math-phi-1-5-FULL-MWP-2k / trainer_state.json
MeedoSam's picture
Uploaded checkpoint-2000
cdc70ac verified
raw
history blame
29.9 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.7172314864622557,
"eval_steps": 2000,
"global_step": 2000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"grad_norm": 5.03125,
"learning_rate": 3.0000000000000004e-07,
"loss": 1.4133,
"step": 10
},
{
"epoch": 0.01,
"grad_norm": 9.0,
"learning_rate": 6.000000000000001e-07,
"loss": 1.5815,
"step": 20
},
{
"epoch": 0.01,
"grad_norm": 6.5625,
"learning_rate": 9e-07,
"loss": 1.3232,
"step": 30
},
{
"epoch": 0.01,
"grad_norm": 4.875,
"learning_rate": 1.2000000000000002e-06,
"loss": 1.3635,
"step": 40
},
{
"epoch": 0.02,
"grad_norm": 6.21875,
"learning_rate": 1.5e-06,
"loss": 1.3467,
"step": 50
},
{
"epoch": 0.02,
"grad_norm": 7.4375,
"learning_rate": 1.8e-06,
"loss": 1.3232,
"step": 60
},
{
"epoch": 0.03,
"grad_norm": 5.9375,
"learning_rate": 2.1000000000000002e-06,
"loss": 1.3657,
"step": 70
},
{
"epoch": 0.03,
"grad_norm": 5.0625,
"learning_rate": 2.4000000000000003e-06,
"loss": 1.3112,
"step": 80
},
{
"epoch": 0.03,
"grad_norm": 5.53125,
"learning_rate": 2.7e-06,
"loss": 1.3052,
"step": 90
},
{
"epoch": 0.04,
"grad_norm": 5.34375,
"learning_rate": 3e-06,
"loss": 1.3857,
"step": 100
},
{
"epoch": 0.04,
"grad_norm": 5.9375,
"learning_rate": 3.3e-06,
"loss": 1.4426,
"step": 110
},
{
"epoch": 0.04,
"grad_norm": 5.875,
"learning_rate": 3.6e-06,
"loss": 1.2005,
"step": 120
},
{
"epoch": 0.05,
"grad_norm": 6.65625,
"learning_rate": 3.9e-06,
"loss": 1.2581,
"step": 130
},
{
"epoch": 0.05,
"grad_norm": 6.09375,
"learning_rate": 4.2000000000000004e-06,
"loss": 1.3267,
"step": 140
},
{
"epoch": 0.05,
"grad_norm": 6.46875,
"learning_rate": 4.5e-06,
"loss": 1.3335,
"step": 150
},
{
"epoch": 0.06,
"grad_norm": 4.59375,
"learning_rate": 4.800000000000001e-06,
"loss": 1.2575,
"step": 160
},
{
"epoch": 0.06,
"grad_norm": 4.8125,
"learning_rate": 5.1e-06,
"loss": 1.2955,
"step": 170
},
{
"epoch": 0.06,
"grad_norm": 5.1875,
"learning_rate": 5.4e-06,
"loss": 1.1892,
"step": 180
},
{
"epoch": 0.07,
"grad_norm": 5.09375,
"learning_rate": 5.7000000000000005e-06,
"loss": 1.1546,
"step": 190
},
{
"epoch": 0.07,
"grad_norm": 5.0,
"learning_rate": 6e-06,
"loss": 1.0558,
"step": 200
},
{
"epoch": 0.08,
"grad_norm": 4.65625,
"learning_rate": 6.3e-06,
"loss": 1.0267,
"step": 210
},
{
"epoch": 0.08,
"grad_norm": 4.3125,
"learning_rate": 6.6e-06,
"loss": 1.1234,
"step": 220
},
{
"epoch": 0.08,
"grad_norm": 5.0,
"learning_rate": 6.900000000000001e-06,
"loss": 1.0501,
"step": 230
},
{
"epoch": 0.09,
"grad_norm": 4.15625,
"learning_rate": 7.2e-06,
"loss": 1.1591,
"step": 240
},
{
"epoch": 0.09,
"grad_norm": 3.53125,
"learning_rate": 7.5e-06,
"loss": 0.9678,
"step": 250
},
{
"epoch": 0.09,
"grad_norm": 6.0625,
"learning_rate": 7.8e-06,
"loss": 1.1697,
"step": 260
},
{
"epoch": 0.1,
"grad_norm": 3.53125,
"learning_rate": 8.1e-06,
"loss": 0.8972,
"step": 270
},
{
"epoch": 0.1,
"grad_norm": 4.78125,
"learning_rate": 8.400000000000001e-06,
"loss": 1.0512,
"step": 280
},
{
"epoch": 0.1,
"grad_norm": 5.9375,
"learning_rate": 8.7e-06,
"loss": 0.9511,
"step": 290
},
{
"epoch": 0.11,
"grad_norm": 4.09375,
"learning_rate": 9e-06,
"loss": 1.0023,
"step": 300
},
{
"epoch": 0.11,
"grad_norm": 4.1875,
"learning_rate": 9.3e-06,
"loss": 1.0908,
"step": 310
},
{
"epoch": 0.11,
"grad_norm": 2.875,
"learning_rate": 9.600000000000001e-06,
"loss": 1.0276,
"step": 320
},
{
"epoch": 0.12,
"grad_norm": 4.15625,
"learning_rate": 9.9e-06,
"loss": 0.9734,
"step": 330
},
{
"epoch": 0.12,
"grad_norm": 4.625,
"learning_rate": 1.02e-05,
"loss": 0.8525,
"step": 340
},
{
"epoch": 0.13,
"grad_norm": 3.546875,
"learning_rate": 1.05e-05,
"loss": 1.0789,
"step": 350
},
{
"epoch": 0.13,
"grad_norm": 3.21875,
"learning_rate": 1.08e-05,
"loss": 0.981,
"step": 360
},
{
"epoch": 0.13,
"grad_norm": 5.5625,
"learning_rate": 1.11e-05,
"loss": 1.0341,
"step": 370
},
{
"epoch": 0.14,
"grad_norm": 2.8125,
"learning_rate": 1.1400000000000001e-05,
"loss": 0.8214,
"step": 380
},
{
"epoch": 0.14,
"grad_norm": 5.96875,
"learning_rate": 1.1700000000000001e-05,
"loss": 0.9675,
"step": 390
},
{
"epoch": 0.14,
"grad_norm": 4.125,
"learning_rate": 1.2e-05,
"loss": 1.2041,
"step": 400
},
{
"epoch": 0.15,
"grad_norm": 5.15625,
"learning_rate": 1.2299999999999999e-05,
"loss": 1.0061,
"step": 410
},
{
"epoch": 0.15,
"grad_norm": 3.328125,
"learning_rate": 1.26e-05,
"loss": 0.8642,
"step": 420
},
{
"epoch": 0.15,
"grad_norm": 5.4375,
"learning_rate": 1.29e-05,
"loss": 1.0072,
"step": 430
},
{
"epoch": 0.16,
"grad_norm": 3.34375,
"learning_rate": 1.32e-05,
"loss": 0.9096,
"step": 440
},
{
"epoch": 0.16,
"grad_norm": 4.125,
"learning_rate": 1.3500000000000001e-05,
"loss": 1.002,
"step": 450
},
{
"epoch": 0.16,
"grad_norm": 4.78125,
"learning_rate": 1.3800000000000002e-05,
"loss": 0.8901,
"step": 460
},
{
"epoch": 0.17,
"grad_norm": 3.3125,
"learning_rate": 1.4099999999999999e-05,
"loss": 0.8891,
"step": 470
},
{
"epoch": 0.17,
"grad_norm": 5.09375,
"learning_rate": 1.44e-05,
"loss": 0.9389,
"step": 480
},
{
"epoch": 0.18,
"grad_norm": 2.28125,
"learning_rate": 1.47e-05,
"loss": 0.8838,
"step": 490
},
{
"epoch": 0.18,
"grad_norm": 3.765625,
"learning_rate": 1.5e-05,
"loss": 1.0191,
"step": 500
},
{
"epoch": 0.18,
"grad_norm": 2.328125,
"learning_rate": 1.4984210526315789e-05,
"loss": 0.9197,
"step": 510
},
{
"epoch": 0.19,
"grad_norm": 4.90625,
"learning_rate": 1.496842105263158e-05,
"loss": 0.9614,
"step": 520
},
{
"epoch": 0.19,
"grad_norm": 4.84375,
"learning_rate": 1.4952631578947368e-05,
"loss": 0.8612,
"step": 530
},
{
"epoch": 0.19,
"grad_norm": 3.765625,
"learning_rate": 1.4936842105263158e-05,
"loss": 0.91,
"step": 540
},
{
"epoch": 0.2,
"grad_norm": 3.15625,
"learning_rate": 1.4921052631578947e-05,
"loss": 0.8851,
"step": 550
},
{
"epoch": 0.2,
"grad_norm": 4.0625,
"learning_rate": 1.4905263157894737e-05,
"loss": 0.9448,
"step": 560
},
{
"epoch": 0.2,
"grad_norm": 2.859375,
"learning_rate": 1.4889473684210526e-05,
"loss": 0.8091,
"step": 570
},
{
"epoch": 0.21,
"grad_norm": 4.09375,
"learning_rate": 1.4873684210526315e-05,
"loss": 0.8932,
"step": 580
},
{
"epoch": 0.21,
"grad_norm": 5.375,
"learning_rate": 1.4857894736842107e-05,
"loss": 0.966,
"step": 590
},
{
"epoch": 0.22,
"grad_norm": 3.875,
"learning_rate": 1.4842105263157895e-05,
"loss": 0.8036,
"step": 600
},
{
"epoch": 0.22,
"grad_norm": 3.9375,
"learning_rate": 1.4826315789473686e-05,
"loss": 0.9743,
"step": 610
},
{
"epoch": 0.22,
"grad_norm": 3.09375,
"learning_rate": 1.4810526315789474e-05,
"loss": 0.8383,
"step": 620
},
{
"epoch": 0.23,
"grad_norm": 4.3125,
"learning_rate": 1.4794736842105265e-05,
"loss": 0.8982,
"step": 630
},
{
"epoch": 0.23,
"grad_norm": 3.90625,
"learning_rate": 1.4778947368421053e-05,
"loss": 0.8112,
"step": 640
},
{
"epoch": 0.23,
"grad_norm": 3.96875,
"learning_rate": 1.4763157894736842e-05,
"loss": 0.8426,
"step": 650
},
{
"epoch": 0.24,
"grad_norm": 4.03125,
"learning_rate": 1.4747368421052632e-05,
"loss": 0.8399,
"step": 660
},
{
"epoch": 0.24,
"grad_norm": 3.125,
"learning_rate": 1.4731578947368421e-05,
"loss": 0.9043,
"step": 670
},
{
"epoch": 0.24,
"grad_norm": 3.5,
"learning_rate": 1.4715789473684211e-05,
"loss": 0.8521,
"step": 680
},
{
"epoch": 0.25,
"grad_norm": 3.46875,
"learning_rate": 1.47e-05,
"loss": 0.9079,
"step": 690
},
{
"epoch": 0.25,
"grad_norm": 4.4375,
"learning_rate": 1.468421052631579e-05,
"loss": 0.7998,
"step": 700
},
{
"epoch": 0.25,
"grad_norm": 3.46875,
"learning_rate": 1.4668421052631579e-05,
"loss": 0.8204,
"step": 710
},
{
"epoch": 0.26,
"grad_norm": 3.3125,
"learning_rate": 1.4652631578947367e-05,
"loss": 0.8764,
"step": 720
},
{
"epoch": 0.26,
"grad_norm": 6.28125,
"learning_rate": 1.4636842105263158e-05,
"loss": 0.9226,
"step": 730
},
{
"epoch": 0.27,
"grad_norm": 3.921875,
"learning_rate": 1.4621052631578946e-05,
"loss": 0.8921,
"step": 740
},
{
"epoch": 0.27,
"grad_norm": 3.671875,
"learning_rate": 1.4605263157894737e-05,
"loss": 0.8764,
"step": 750
},
{
"epoch": 0.27,
"grad_norm": 3.671875,
"learning_rate": 1.4589473684210527e-05,
"loss": 0.8475,
"step": 760
},
{
"epoch": 0.28,
"grad_norm": 3.921875,
"learning_rate": 1.4573684210526317e-05,
"loss": 0.9297,
"step": 770
},
{
"epoch": 0.28,
"grad_norm": 3.109375,
"learning_rate": 1.4557894736842106e-05,
"loss": 0.9052,
"step": 780
},
{
"epoch": 0.28,
"grad_norm": 4.40625,
"learning_rate": 1.4542105263157895e-05,
"loss": 0.8695,
"step": 790
},
{
"epoch": 0.29,
"grad_norm": 4.25,
"learning_rate": 1.4526315789473685e-05,
"loss": 0.9214,
"step": 800
},
{
"epoch": 0.29,
"grad_norm": 3.25,
"learning_rate": 1.4510526315789474e-05,
"loss": 0.8403,
"step": 810
},
{
"epoch": 0.29,
"grad_norm": 3.046875,
"learning_rate": 1.4494736842105264e-05,
"loss": 0.918,
"step": 820
},
{
"epoch": 0.3,
"grad_norm": 5.53125,
"learning_rate": 1.4478947368421053e-05,
"loss": 0.8599,
"step": 830
},
{
"epoch": 0.3,
"grad_norm": 2.875,
"learning_rate": 1.4463157894736843e-05,
"loss": 0.954,
"step": 840
},
{
"epoch": 0.3,
"grad_norm": 4.21875,
"learning_rate": 1.4447368421052632e-05,
"loss": 0.8568,
"step": 850
},
{
"epoch": 0.31,
"grad_norm": 4.03125,
"learning_rate": 1.443157894736842e-05,
"loss": 0.9043,
"step": 860
},
{
"epoch": 0.31,
"grad_norm": 4.0,
"learning_rate": 1.441578947368421e-05,
"loss": 0.9108,
"step": 870
},
{
"epoch": 0.32,
"grad_norm": 4.28125,
"learning_rate": 1.44e-05,
"loss": 0.8723,
"step": 880
},
{
"epoch": 0.32,
"grad_norm": 3.859375,
"learning_rate": 1.438421052631579e-05,
"loss": 0.8514,
"step": 890
},
{
"epoch": 0.32,
"grad_norm": 2.59375,
"learning_rate": 1.4368421052631578e-05,
"loss": 0.8584,
"step": 900
},
{
"epoch": 0.33,
"grad_norm": 3.9375,
"learning_rate": 1.4352631578947369e-05,
"loss": 0.8763,
"step": 910
},
{
"epoch": 0.33,
"grad_norm": 4.4375,
"learning_rate": 1.4336842105263159e-05,
"loss": 0.7777,
"step": 920
},
{
"epoch": 0.33,
"grad_norm": 3.671875,
"learning_rate": 1.4321052631578948e-05,
"loss": 0.9353,
"step": 930
},
{
"epoch": 0.34,
"grad_norm": 3.4375,
"learning_rate": 1.4305263157894738e-05,
"loss": 0.8708,
"step": 940
},
{
"epoch": 0.34,
"grad_norm": 3.828125,
"learning_rate": 1.4289473684210527e-05,
"loss": 0.8801,
"step": 950
},
{
"epoch": 0.34,
"grad_norm": 3.5,
"learning_rate": 1.4273684210526317e-05,
"loss": 0.8625,
"step": 960
},
{
"epoch": 0.35,
"grad_norm": 5.03125,
"learning_rate": 1.4257894736842106e-05,
"loss": 0.875,
"step": 970
},
{
"epoch": 0.35,
"grad_norm": 5.21875,
"learning_rate": 1.4242105263157896e-05,
"loss": 0.9445,
"step": 980
},
{
"epoch": 0.36,
"grad_norm": 3.421875,
"learning_rate": 1.4226315789473685e-05,
"loss": 0.7317,
"step": 990
},
{
"epoch": 0.36,
"grad_norm": 3.6875,
"learning_rate": 1.4210526315789473e-05,
"loss": 0.8303,
"step": 1000
},
{
"epoch": 0.36,
"grad_norm": 4.40625,
"learning_rate": 1.4194736842105264e-05,
"loss": 0.8433,
"step": 1010
},
{
"epoch": 0.37,
"grad_norm": 3.484375,
"learning_rate": 1.4178947368421052e-05,
"loss": 0.8713,
"step": 1020
},
{
"epoch": 0.37,
"grad_norm": 3.890625,
"learning_rate": 1.4163157894736843e-05,
"loss": 0.8411,
"step": 1030
},
{
"epoch": 0.37,
"grad_norm": 4.3125,
"learning_rate": 1.4147368421052631e-05,
"loss": 0.8757,
"step": 1040
},
{
"epoch": 0.38,
"grad_norm": 3.3125,
"learning_rate": 1.4131578947368422e-05,
"loss": 0.7574,
"step": 1050
},
{
"epoch": 0.38,
"grad_norm": 5.21875,
"learning_rate": 1.411578947368421e-05,
"loss": 0.8089,
"step": 1060
},
{
"epoch": 0.38,
"grad_norm": 2.890625,
"learning_rate": 1.4099999999999999e-05,
"loss": 0.7259,
"step": 1070
},
{
"epoch": 0.39,
"grad_norm": 3.125,
"learning_rate": 1.408421052631579e-05,
"loss": 0.7544,
"step": 1080
},
{
"epoch": 0.39,
"grad_norm": 3.375,
"learning_rate": 1.406842105263158e-05,
"loss": 0.9522,
"step": 1090
},
{
"epoch": 0.39,
"grad_norm": 4.78125,
"learning_rate": 1.405263157894737e-05,
"loss": 0.8574,
"step": 1100
},
{
"epoch": 0.4,
"grad_norm": 2.78125,
"learning_rate": 1.4036842105263158e-05,
"loss": 0.9539,
"step": 1110
},
{
"epoch": 0.4,
"grad_norm": 3.34375,
"learning_rate": 1.4021052631578949e-05,
"loss": 0.8365,
"step": 1120
},
{
"epoch": 0.41,
"grad_norm": 5.34375,
"learning_rate": 1.4005263157894737e-05,
"loss": 0.7868,
"step": 1130
},
{
"epoch": 0.41,
"grad_norm": 3.515625,
"learning_rate": 1.3989473684210526e-05,
"loss": 0.9054,
"step": 1140
},
{
"epoch": 0.41,
"grad_norm": 3.015625,
"learning_rate": 1.3973684210526316e-05,
"loss": 0.841,
"step": 1150
},
{
"epoch": 0.42,
"grad_norm": 4.21875,
"learning_rate": 1.3957894736842105e-05,
"loss": 0.8132,
"step": 1160
},
{
"epoch": 0.42,
"grad_norm": 3.515625,
"learning_rate": 1.3942105263157895e-05,
"loss": 0.8678,
"step": 1170
},
{
"epoch": 0.42,
"grad_norm": 3.4375,
"learning_rate": 1.3926315789473684e-05,
"loss": 0.8434,
"step": 1180
},
{
"epoch": 0.43,
"grad_norm": 4.3125,
"learning_rate": 1.3910526315789474e-05,
"loss": 0.7466,
"step": 1190
},
{
"epoch": 0.43,
"grad_norm": 3.5625,
"learning_rate": 1.3894736842105263e-05,
"loss": 0.9028,
"step": 1200
},
{
"epoch": 0.43,
"grad_norm": 3.203125,
"learning_rate": 1.3878947368421052e-05,
"loss": 0.9181,
"step": 1210
},
{
"epoch": 0.44,
"grad_norm": 3.703125,
"learning_rate": 1.3863157894736842e-05,
"loss": 0.9427,
"step": 1220
},
{
"epoch": 0.44,
"grad_norm": 2.609375,
"learning_rate": 1.384736842105263e-05,
"loss": 0.7861,
"step": 1230
},
{
"epoch": 0.44,
"grad_norm": 5.71875,
"learning_rate": 1.3831578947368421e-05,
"loss": 0.8857,
"step": 1240
},
{
"epoch": 0.45,
"grad_norm": 3.140625,
"learning_rate": 1.3815789473684211e-05,
"loss": 0.8447,
"step": 1250
},
{
"epoch": 0.45,
"grad_norm": 3.140625,
"learning_rate": 1.3800000000000002e-05,
"loss": 0.7963,
"step": 1260
},
{
"epoch": 0.46,
"grad_norm": 4.03125,
"learning_rate": 1.378421052631579e-05,
"loss": 0.7099,
"step": 1270
},
{
"epoch": 0.46,
"grad_norm": 3.765625,
"learning_rate": 1.3768421052631579e-05,
"loss": 0.8737,
"step": 1280
},
{
"epoch": 0.46,
"grad_norm": 3.921875,
"learning_rate": 1.375263157894737e-05,
"loss": 0.8896,
"step": 1290
},
{
"epoch": 0.47,
"grad_norm": 3.640625,
"learning_rate": 1.3736842105263158e-05,
"loss": 0.8224,
"step": 1300
},
{
"epoch": 0.47,
"grad_norm": 3.796875,
"learning_rate": 1.3721052631578948e-05,
"loss": 0.7736,
"step": 1310
},
{
"epoch": 0.47,
"grad_norm": 4.65625,
"learning_rate": 1.3705263157894737e-05,
"loss": 0.8416,
"step": 1320
},
{
"epoch": 0.48,
"grad_norm": 2.828125,
"learning_rate": 1.3689473684210527e-05,
"loss": 0.7371,
"step": 1330
},
{
"epoch": 0.48,
"grad_norm": 3.3125,
"learning_rate": 1.3673684210526316e-05,
"loss": 0.9159,
"step": 1340
},
{
"epoch": 0.48,
"grad_norm": 4.625,
"learning_rate": 1.3657894736842106e-05,
"loss": 0.8392,
"step": 1350
},
{
"epoch": 0.49,
"grad_norm": 3.609375,
"learning_rate": 1.3642105263157895e-05,
"loss": 0.7774,
"step": 1360
},
{
"epoch": 0.49,
"grad_norm": 4.1875,
"learning_rate": 1.3626315789473684e-05,
"loss": 0.9497,
"step": 1370
},
{
"epoch": 0.49,
"grad_norm": 2.71875,
"learning_rate": 1.3610526315789474e-05,
"loss": 0.8797,
"step": 1380
},
{
"epoch": 0.5,
"grad_norm": 2.265625,
"learning_rate": 1.3594736842105263e-05,
"loss": 0.8666,
"step": 1390
},
{
"epoch": 0.5,
"grad_norm": 3.96875,
"learning_rate": 1.3578947368421053e-05,
"loss": 0.9691,
"step": 1400
},
{
"epoch": 0.51,
"grad_norm": 5.21875,
"learning_rate": 1.3563157894736842e-05,
"loss": 0.796,
"step": 1410
},
{
"epoch": 0.51,
"grad_norm": 2.296875,
"learning_rate": 1.3547368421052634e-05,
"loss": 0.8152,
"step": 1420
},
{
"epoch": 0.51,
"grad_norm": 4.0625,
"learning_rate": 1.3531578947368422e-05,
"loss": 0.8702,
"step": 1430
},
{
"epoch": 0.52,
"grad_norm": 3.765625,
"learning_rate": 1.3515789473684211e-05,
"loss": 0.8582,
"step": 1440
},
{
"epoch": 0.52,
"grad_norm": 4.03125,
"learning_rate": 1.3500000000000001e-05,
"loss": 0.8282,
"step": 1450
},
{
"epoch": 0.52,
"grad_norm": 3.859375,
"learning_rate": 1.348421052631579e-05,
"loss": 0.9256,
"step": 1460
},
{
"epoch": 0.53,
"grad_norm": 4.6875,
"learning_rate": 1.346842105263158e-05,
"loss": 0.9328,
"step": 1470
},
{
"epoch": 0.53,
"grad_norm": 3.5,
"learning_rate": 1.3452631578947369e-05,
"loss": 0.8571,
"step": 1480
},
{
"epoch": 0.53,
"grad_norm": 3.609375,
"learning_rate": 1.343684210526316e-05,
"loss": 0.8293,
"step": 1490
},
{
"epoch": 0.54,
"grad_norm": 2.96875,
"learning_rate": 1.3421052631578948e-05,
"loss": 0.8282,
"step": 1500
},
{
"epoch": 0.54,
"grad_norm": 3.8125,
"learning_rate": 1.3405263157894736e-05,
"loss": 0.8202,
"step": 1510
},
{
"epoch": 0.55,
"grad_norm": 3.609375,
"learning_rate": 1.3389473684210527e-05,
"loss": 0.9545,
"step": 1520
},
{
"epoch": 0.55,
"grad_norm": 4.625,
"learning_rate": 1.3373684210526315e-05,
"loss": 0.8083,
"step": 1530
},
{
"epoch": 0.55,
"grad_norm": 3.84375,
"learning_rate": 1.3357894736842106e-05,
"loss": 0.8865,
"step": 1540
},
{
"epoch": 0.56,
"grad_norm": 3.875,
"learning_rate": 1.3342105263157894e-05,
"loss": 0.7867,
"step": 1550
},
{
"epoch": 0.56,
"grad_norm": 3.203125,
"learning_rate": 1.3326315789473685e-05,
"loss": 0.7882,
"step": 1560
},
{
"epoch": 0.56,
"grad_norm": 3.90625,
"learning_rate": 1.3310526315789473e-05,
"loss": 0.7837,
"step": 1570
},
{
"epoch": 0.57,
"grad_norm": 5.125,
"learning_rate": 1.3294736842105262e-05,
"loss": 0.9198,
"step": 1580
},
{
"epoch": 0.57,
"grad_norm": 3.4375,
"learning_rate": 1.3278947368421054e-05,
"loss": 0.849,
"step": 1590
},
{
"epoch": 0.57,
"grad_norm": 2.796875,
"learning_rate": 1.3263157894736843e-05,
"loss": 0.894,
"step": 1600
},
{
"epoch": 0.58,
"grad_norm": 2.953125,
"learning_rate": 1.3247368421052633e-05,
"loss": 0.8318,
"step": 1610
},
{
"epoch": 0.58,
"grad_norm": 3.703125,
"learning_rate": 1.3231578947368422e-05,
"loss": 0.7787,
"step": 1620
},
{
"epoch": 0.58,
"grad_norm": 5.0,
"learning_rate": 1.3215789473684212e-05,
"loss": 0.8853,
"step": 1630
},
{
"epoch": 0.59,
"grad_norm": 4.1875,
"learning_rate": 1.32e-05,
"loss": 0.9163,
"step": 1640
},
{
"epoch": 0.59,
"grad_norm": 3.6875,
"learning_rate": 1.318421052631579e-05,
"loss": 0.8755,
"step": 1650
},
{
"epoch": 0.6,
"grad_norm": 3.71875,
"learning_rate": 1.316842105263158e-05,
"loss": 0.8891,
"step": 1660
},
{
"epoch": 0.6,
"grad_norm": 3.984375,
"learning_rate": 1.3152631578947368e-05,
"loss": 0.8621,
"step": 1670
},
{
"epoch": 0.6,
"grad_norm": 3.28125,
"learning_rate": 1.3136842105263159e-05,
"loss": 0.8613,
"step": 1680
},
{
"epoch": 0.61,
"grad_norm": 4.1875,
"learning_rate": 1.3121052631578947e-05,
"loss": 0.8886,
"step": 1690
},
{
"epoch": 0.61,
"grad_norm": 3.53125,
"learning_rate": 1.3105263157894738e-05,
"loss": 0.843,
"step": 1700
},
{
"epoch": 0.61,
"grad_norm": 3.75,
"learning_rate": 1.3089473684210526e-05,
"loss": 0.8877,
"step": 1710
},
{
"epoch": 0.62,
"grad_norm": 4.1875,
"learning_rate": 1.3073684210526315e-05,
"loss": 0.8785,
"step": 1720
},
{
"epoch": 0.62,
"grad_norm": 3.78125,
"learning_rate": 1.3057894736842105e-05,
"loss": 0.8878,
"step": 1730
},
{
"epoch": 0.62,
"grad_norm": 3.828125,
"learning_rate": 1.3042105263157894e-05,
"loss": 0.8794,
"step": 1740
},
{
"epoch": 0.63,
"grad_norm": 4.75,
"learning_rate": 1.3026315789473684e-05,
"loss": 0.8821,
"step": 1750
},
{
"epoch": 0.63,
"grad_norm": 4.34375,
"learning_rate": 1.3010526315789475e-05,
"loss": 0.8146,
"step": 1760
},
{
"epoch": 0.63,
"grad_norm": 3.953125,
"learning_rate": 1.2994736842105265e-05,
"loss": 0.8322,
"step": 1770
},
{
"epoch": 0.64,
"grad_norm": 2.8125,
"learning_rate": 1.2978947368421054e-05,
"loss": 0.8197,
"step": 1780
},
{
"epoch": 0.64,
"grad_norm": 4.59375,
"learning_rate": 1.2963157894736842e-05,
"loss": 0.8883,
"step": 1790
},
{
"epoch": 0.65,
"grad_norm": 3.609375,
"learning_rate": 1.2947368421052633e-05,
"loss": 0.9259,
"step": 1800
},
{
"epoch": 0.65,
"grad_norm": 4.125,
"learning_rate": 1.2931578947368421e-05,
"loss": 0.7667,
"step": 1810
},
{
"epoch": 0.65,
"grad_norm": 5.3125,
"learning_rate": 1.2915789473684212e-05,
"loss": 0.9434,
"step": 1820
},
{
"epoch": 0.66,
"grad_norm": 3.3125,
"learning_rate": 1.29e-05,
"loss": 0.8081,
"step": 1830
},
{
"epoch": 0.66,
"grad_norm": 3.875,
"learning_rate": 1.288421052631579e-05,
"loss": 0.767,
"step": 1840
},
{
"epoch": 0.66,
"grad_norm": 4.53125,
"learning_rate": 1.2868421052631579e-05,
"loss": 0.8339,
"step": 1850
},
{
"epoch": 0.67,
"grad_norm": 4.0,
"learning_rate": 1.2852631578947368e-05,
"loss": 0.8083,
"step": 1860
},
{
"epoch": 0.67,
"grad_norm": 2.828125,
"learning_rate": 1.2836842105263158e-05,
"loss": 0.8255,
"step": 1870
},
{
"epoch": 0.67,
"grad_norm": 3.65625,
"learning_rate": 1.2821052631578947e-05,
"loss": 0.8618,
"step": 1880
},
{
"epoch": 0.68,
"grad_norm": 4.15625,
"learning_rate": 1.2805263157894737e-05,
"loss": 0.913,
"step": 1890
},
{
"epoch": 0.68,
"grad_norm": 4.125,
"learning_rate": 1.2789473684210526e-05,
"loss": 0.7233,
"step": 1900
},
{
"epoch": 0.68,
"grad_norm": 4.5625,
"learning_rate": 1.2773684210526316e-05,
"loss": 0.839,
"step": 1910
},
{
"epoch": 0.69,
"grad_norm": 2.765625,
"learning_rate": 1.2757894736842106e-05,
"loss": 0.9282,
"step": 1920
},
{
"epoch": 0.69,
"grad_norm": 3.3125,
"learning_rate": 1.2742105263157895e-05,
"loss": 0.8012,
"step": 1930
},
{
"epoch": 0.7,
"grad_norm": 6.3125,
"learning_rate": 1.2726315789473685e-05,
"loss": 0.7934,
"step": 1940
},
{
"epoch": 0.7,
"grad_norm": 2.921875,
"learning_rate": 1.2710526315789474e-05,
"loss": 0.8342,
"step": 1950
},
{
"epoch": 0.7,
"grad_norm": 2.96875,
"learning_rate": 1.2694736842105264e-05,
"loss": 0.8319,
"step": 1960
},
{
"epoch": 0.71,
"grad_norm": 3.4375,
"learning_rate": 1.2678947368421053e-05,
"loss": 0.855,
"step": 1970
},
{
"epoch": 0.71,
"grad_norm": 4.21875,
"learning_rate": 1.2663157894736843e-05,
"loss": 0.8794,
"step": 1980
},
{
"epoch": 0.71,
"grad_norm": 4.03125,
"learning_rate": 1.2647368421052632e-05,
"loss": 0.8106,
"step": 1990
},
{
"epoch": 0.72,
"grad_norm": 4.84375,
"learning_rate": 1.263157894736842e-05,
"loss": 0.7944,
"step": 2000
},
{
"epoch": 0.72,
"eval_loss": 0.964277446269989,
"eval_runtime": 64.3098,
"eval_samples_per_second": 15.55,
"eval_steps_per_second": 15.55,
"step": 2000
}
],
"logging_steps": 10,
"max_steps": 10000,
"num_input_tokens_seen": 0,
"num_train_epochs": 4,
"save_steps": 2000,
"total_flos": 3.227844083712e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}