Grogros's picture
Training in progress, step 2000, checkpoint
5797f8d verified
raw
history blame
30.3 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.5,
"eval_steps": 500,
"global_step": 2000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0025,
"grad_norm": 19.25,
"learning_rate": 1.1250000000000001e-07,
"loss": 1.8141,
"step": 10
},
{
"epoch": 0.005,
"grad_norm": 15.25,
"learning_rate": 2.3750000000000003e-07,
"loss": 1.7844,
"step": 20
},
{
"epoch": 0.0075,
"grad_norm": 18.25,
"learning_rate": 3.625e-07,
"loss": 1.7555,
"step": 30
},
{
"epoch": 0.01,
"grad_norm": 14.0625,
"learning_rate": 4.875000000000001e-07,
"loss": 1.7414,
"step": 40
},
{
"epoch": 0.0125,
"grad_norm": 17.75,
"learning_rate": 6.125000000000001e-07,
"loss": 1.7531,
"step": 50
},
{
"epoch": 0.015,
"grad_norm": 15.5,
"learning_rate": 7.375e-07,
"loss": 1.7852,
"step": 60
},
{
"epoch": 0.0175,
"grad_norm": 17.875,
"learning_rate": 8.625e-07,
"loss": 1.7844,
"step": 70
},
{
"epoch": 0.02,
"grad_norm": 14.9375,
"learning_rate": 9.875e-07,
"loss": 1.7883,
"step": 80
},
{
"epoch": 0.0225,
"grad_norm": 11.0,
"learning_rate": 1.1125000000000001e-06,
"loss": 1.7312,
"step": 90
},
{
"epoch": 0.025,
"grad_norm": 12.9375,
"learning_rate": 1.2375e-06,
"loss": 1.75,
"step": 100
},
{
"epoch": 0.0275,
"grad_norm": 12.0625,
"learning_rate": 1.3625000000000003e-06,
"loss": 1.7477,
"step": 110
},
{
"epoch": 0.03,
"grad_norm": 14.3125,
"learning_rate": 1.4875000000000002e-06,
"loss": 1.7312,
"step": 120
},
{
"epoch": 0.0325,
"grad_norm": 12.375,
"learning_rate": 1.6125000000000002e-06,
"loss": 1.7414,
"step": 130
},
{
"epoch": 0.035,
"grad_norm": 12.3125,
"learning_rate": 1.7375e-06,
"loss": 1.7367,
"step": 140
},
{
"epoch": 0.0375,
"grad_norm": 15.1875,
"learning_rate": 1.8625000000000002e-06,
"loss": 1.7656,
"step": 150
},
{
"epoch": 0.04,
"grad_norm": 13.25,
"learning_rate": 1.9875000000000005e-06,
"loss": 1.7523,
"step": 160
},
{
"epoch": 0.0425,
"grad_norm": 10.125,
"learning_rate": 2.1125e-06,
"loss": 1.6953,
"step": 170
},
{
"epoch": 0.045,
"grad_norm": 13.375,
"learning_rate": 2.2375e-06,
"loss": 1.7109,
"step": 180
},
{
"epoch": 0.0475,
"grad_norm": 9.5625,
"learning_rate": 2.3625000000000003e-06,
"loss": 1.7047,
"step": 190
},
{
"epoch": 0.05,
"grad_norm": 11.875,
"learning_rate": 2.4875000000000003e-06,
"loss": 1.675,
"step": 200
},
{
"epoch": 0.0525,
"grad_norm": 14.5,
"learning_rate": 2.6125e-06,
"loss": 1.6836,
"step": 210
},
{
"epoch": 0.055,
"grad_norm": 12.875,
"learning_rate": 2.7375e-06,
"loss": 1.7133,
"step": 220
},
{
"epoch": 0.0575,
"grad_norm": 11.0625,
"learning_rate": 2.8625e-06,
"loss": 1.6461,
"step": 230
},
{
"epoch": 0.06,
"grad_norm": 13.625,
"learning_rate": 2.9875e-06,
"loss": 1.6242,
"step": 240
},
{
"epoch": 0.0625,
"grad_norm": 15.375,
"learning_rate": 3.1125000000000007e-06,
"loss": 1.6523,
"step": 250
},
{
"epoch": 0.065,
"grad_norm": 10.5,
"learning_rate": 3.2375e-06,
"loss": 1.6141,
"step": 260
},
{
"epoch": 0.0675,
"grad_norm": 13.0625,
"learning_rate": 3.3625000000000004e-06,
"loss": 1.5766,
"step": 270
},
{
"epoch": 0.07,
"grad_norm": 13.6875,
"learning_rate": 3.4875000000000005e-06,
"loss": 1.5625,
"step": 280
},
{
"epoch": 0.0725,
"grad_norm": 19.0,
"learning_rate": 3.6125000000000006e-06,
"loss": 1.5367,
"step": 290
},
{
"epoch": 0.075,
"grad_norm": 18.75,
"learning_rate": 3.7375000000000006e-06,
"loss": 1.4852,
"step": 300
},
{
"epoch": 0.0775,
"grad_norm": 11.3125,
"learning_rate": 3.8625e-06,
"loss": 1.4773,
"step": 310
},
{
"epoch": 0.08,
"grad_norm": 14.9375,
"learning_rate": 3.9875e-06,
"loss": 1.4672,
"step": 320
},
{
"epoch": 0.0825,
"grad_norm": 12.3125,
"learning_rate": 4.1125e-06,
"loss": 1.4109,
"step": 330
},
{
"epoch": 0.085,
"grad_norm": 12.625,
"learning_rate": 4.2375000000000005e-06,
"loss": 1.375,
"step": 340
},
{
"epoch": 0.0875,
"grad_norm": 19.0,
"learning_rate": 4.362500000000001e-06,
"loss": 1.3438,
"step": 350
},
{
"epoch": 0.09,
"grad_norm": 13.0625,
"learning_rate": 4.4875e-06,
"loss": 1.2742,
"step": 360
},
{
"epoch": 0.0925,
"grad_norm": 12.625,
"learning_rate": 4.6125e-06,
"loss": 1.2508,
"step": 370
},
{
"epoch": 0.095,
"grad_norm": 10.0625,
"learning_rate": 4.737500000000001e-06,
"loss": 1.268,
"step": 380
},
{
"epoch": 0.0975,
"grad_norm": 10.125,
"learning_rate": 4.8625000000000005e-06,
"loss": 1.2055,
"step": 390
},
{
"epoch": 0.1,
"grad_norm": 12.25,
"learning_rate": 4.987500000000001e-06,
"loss": 1.1742,
"step": 400
},
{
"epoch": 0.1025,
"grad_norm": 12.4375,
"learning_rate": 4.999922894111975e-06,
"loss": 1.1199,
"step": 410
},
{
"epoch": 0.105,
"grad_norm": 9.8125,
"learning_rate": 4.999656361346094e-06,
"loss": 1.1195,
"step": 420
},
{
"epoch": 0.1075,
"grad_norm": 11.1875,
"learning_rate": 4.999199470070484e-06,
"loss": 1.0727,
"step": 430
},
{
"epoch": 0.11,
"grad_norm": 11.625,
"learning_rate": 4.998552255079182e-06,
"loss": 1.0711,
"step": 440
},
{
"epoch": 0.1125,
"grad_norm": 9.3125,
"learning_rate": 4.9977147656601196e-06,
"loss": 1.0414,
"step": 450
},
{
"epoch": 0.115,
"grad_norm": 10.3125,
"learning_rate": 4.996687065591355e-06,
"loss": 1.0367,
"step": 460
},
{
"epoch": 0.1175,
"grad_norm": 10.5,
"learning_rate": 4.9954692331362295e-06,
"loss": 1.0125,
"step": 470
},
{
"epoch": 0.12,
"grad_norm": 10.625,
"learning_rate": 4.9940613610373974e-06,
"loss": 0.9828,
"step": 480
},
{
"epoch": 0.1225,
"grad_norm": 10.125,
"learning_rate": 4.992463556509772e-06,
"loss": 0.9695,
"step": 490
},
{
"epoch": 0.125,
"grad_norm": 10.75,
"learning_rate": 4.990675941232353e-06,
"loss": 0.9395,
"step": 500
},
{
"epoch": 0.1275,
"grad_norm": 9.0,
"learning_rate": 4.988698651338965e-06,
"loss": 0.9137,
"step": 510
},
{
"epoch": 0.13,
"grad_norm": 11.375,
"learning_rate": 4.986531837407891e-06,
"loss": 0.9082,
"step": 520
},
{
"epoch": 0.1325,
"grad_norm": 12.5,
"learning_rate": 4.9841756644503965e-06,
"loss": 0.8809,
"step": 530
},
{
"epoch": 0.135,
"grad_norm": 18.875,
"learning_rate": 4.981630311898178e-06,
"loss": 0.8648,
"step": 540
},
{
"epoch": 0.1375,
"grad_norm": 9.8125,
"learning_rate": 4.978895973589686e-06,
"loss": 0.8633,
"step": 550
},
{
"epoch": 0.14,
"grad_norm": 9.3125,
"learning_rate": 4.975972857755369e-06,
"loss": 0.8293,
"step": 560
},
{
"epoch": 0.1425,
"grad_norm": 9.9375,
"learning_rate": 4.972861187001815e-06,
"loss": 0.8242,
"step": 570
},
{
"epoch": 0.145,
"grad_norm": 9.25,
"learning_rate": 4.9695611982947995e-06,
"loss": 0.8031,
"step": 580
},
{
"epoch": 0.1475,
"grad_norm": 9.0625,
"learning_rate": 4.966073142941239e-06,
"loss": 0.775,
"step": 590
},
{
"epoch": 0.15,
"grad_norm": 9.4375,
"learning_rate": 4.962397286570053e-06,
"loss": 0.7996,
"step": 600
},
{
"epoch": 0.1525,
"grad_norm": 10.5,
"learning_rate": 4.958533909111936e-06,
"loss": 0.7492,
"step": 610
},
{
"epoch": 0.155,
"grad_norm": 11.1875,
"learning_rate": 4.95448330477804e-06,
"loss": 0.7184,
"step": 620
},
{
"epoch": 0.1575,
"grad_norm": 9.6875,
"learning_rate": 4.950245782037566e-06,
"loss": 0.7023,
"step": 630
},
{
"epoch": 0.16,
"grad_norm": 10.0625,
"learning_rate": 4.945821663594277e-06,
"loss": 0.6949,
"step": 640
},
{
"epoch": 0.1625,
"grad_norm": 9.125,
"learning_rate": 4.941211286361922e-06,
"loss": 0.7043,
"step": 650
},
{
"epoch": 0.165,
"grad_norm": 9.5625,
"learning_rate": 4.936415001438577e-06,
"loss": 0.6633,
"step": 660
},
{
"epoch": 0.1675,
"grad_norm": 9.25,
"learning_rate": 4.9314331740799084e-06,
"loss": 0.652,
"step": 670
},
{
"epoch": 0.17,
"grad_norm": 9.0625,
"learning_rate": 4.926266183671356e-06,
"loss": 0.6637,
"step": 680
},
{
"epoch": 0.1725,
"grad_norm": 9.75,
"learning_rate": 4.920914423699247e-06,
"loss": 0.6262,
"step": 690
},
{
"epoch": 0.175,
"grad_norm": 8.9375,
"learning_rate": 4.915378301720822e-06,
"loss": 0.6398,
"step": 700
},
{
"epoch": 0.1775,
"grad_norm": 9.25,
"learning_rate": 4.909658239333203e-06,
"loss": 0.6195,
"step": 710
},
{
"epoch": 0.18,
"grad_norm": 9.4375,
"learning_rate": 4.903754672141288e-06,
"loss": 0.5977,
"step": 720
},
{
"epoch": 0.1825,
"grad_norm": 8.5,
"learning_rate": 4.897668049724574e-06,
"loss": 0.5766,
"step": 730
},
{
"epoch": 0.185,
"grad_norm": 9.1875,
"learning_rate": 4.891398835602925e-06,
"loss": 0.5656,
"step": 740
},
{
"epoch": 0.1875,
"grad_norm": 8.4375,
"learning_rate": 4.884947507201268e-06,
"loss": 0.5781,
"step": 750
},
{
"epoch": 0.19,
"grad_norm": 9.0,
"learning_rate": 4.878314555813237e-06,
"loss": 0.559,
"step": 760
},
{
"epoch": 0.1925,
"grad_norm": 8.5625,
"learning_rate": 4.8715004865637616e-06,
"loss": 0.5496,
"step": 770
},
{
"epoch": 0.195,
"grad_norm": 8.5,
"learning_rate": 4.8645058183705976e-06,
"loss": 0.5264,
"step": 780
},
{
"epoch": 0.1975,
"grad_norm": 9.0625,
"learning_rate": 4.8573310839048085e-06,
"loss": 0.5404,
"step": 790
},
{
"epoch": 0.2,
"grad_norm": 8.625,
"learning_rate": 4.8499768295502e-06,
"loss": 0.5117,
"step": 800
},
{
"epoch": 0.2025,
"grad_norm": 8.4375,
"learning_rate": 4.842443615361718e-06,
"loss": 0.5195,
"step": 810
},
{
"epoch": 0.205,
"grad_norm": 8.4375,
"learning_rate": 4.834732015022786e-06,
"loss": 0.4818,
"step": 820
},
{
"epoch": 0.2075,
"grad_norm": 8.25,
"learning_rate": 4.826842615801628e-06,
"loss": 0.4992,
"step": 830
},
{
"epoch": 0.21,
"grad_norm": 8.375,
"learning_rate": 4.818776018506538e-06,
"loss": 0.4975,
"step": 840
},
{
"epoch": 0.2125,
"grad_norm": 8.3125,
"learning_rate": 4.810532837440134e-06,
"loss": 0.4752,
"step": 850
},
{
"epoch": 0.215,
"grad_norm": 8.0,
"learning_rate": 4.802113700352567e-06,
"loss": 0.4924,
"step": 860
},
{
"epoch": 0.2175,
"grad_norm": 7.6875,
"learning_rate": 4.793519248393721e-06,
"loss": 0.4564,
"step": 870
},
{
"epoch": 0.22,
"grad_norm": 8.125,
"learning_rate": 4.78475013606439e-06,
"loss": 0.443,
"step": 880
},
{
"epoch": 0.2225,
"grad_norm": 7.65625,
"learning_rate": 4.775807031166428e-06,
"loss": 0.4428,
"step": 890
},
{
"epoch": 0.225,
"grad_norm": 7.3125,
"learning_rate": 4.766690614751897e-06,
"loss": 0.4443,
"step": 900
},
{
"epoch": 0.2275,
"grad_norm": 7.625,
"learning_rate": 4.757401581071203e-06,
"loss": 0.4482,
"step": 910
},
{
"epoch": 0.23,
"grad_norm": 7.09375,
"learning_rate": 4.747940637520226e-06,
"loss": 0.4332,
"step": 920
},
{
"epoch": 0.2325,
"grad_norm": 7.65625,
"learning_rate": 4.738308504586445e-06,
"loss": 0.4215,
"step": 930
},
{
"epoch": 0.235,
"grad_norm": 6.53125,
"learning_rate": 4.7285059157940765e-06,
"loss": 0.4332,
"step": 940
},
{
"epoch": 0.2375,
"grad_norm": 7.03125,
"learning_rate": 4.718533617648209e-06,
"loss": 0.4092,
"step": 950
},
{
"epoch": 0.24,
"grad_norm": 6.96875,
"learning_rate": 4.7083923695779546e-06,
"loss": 0.4297,
"step": 960
},
{
"epoch": 0.2425,
"grad_norm": 6.78125,
"learning_rate": 4.6980829438786176e-06,
"loss": 0.3949,
"step": 970
},
{
"epoch": 0.245,
"grad_norm": 7.46875,
"learning_rate": 4.687606125652882e-06,
"loss": 0.4023,
"step": 980
},
{
"epoch": 0.2475,
"grad_norm": 6.75,
"learning_rate": 4.676962712751015e-06,
"loss": 0.3988,
"step": 990
},
{
"epoch": 0.25,
"grad_norm": 6.5,
"learning_rate": 4.666153515710118e-06,
"loss": 0.3975,
"step": 1000
},
{
"epoch": 0.2525,
"grad_norm": 6.84375,
"learning_rate": 4.655179357692396e-06,
"loss": 0.4049,
"step": 1010
},
{
"epoch": 0.255,
"grad_norm": 6.375,
"learning_rate": 4.644041074422469e-06,
"loss": 0.4037,
"step": 1020
},
{
"epoch": 0.2575,
"grad_norm": 7.03125,
"learning_rate": 4.632739514123733e-06,
"loss": 0.3846,
"step": 1030
},
{
"epoch": 0.26,
"grad_norm": 6.09375,
"learning_rate": 4.6212755374537596e-06,
"loss": 0.376,
"step": 1040
},
{
"epoch": 0.2625,
"grad_norm": 6.59375,
"learning_rate": 4.609650017438757e-06,
"loss": 0.4123,
"step": 1050
},
{
"epoch": 0.265,
"grad_norm": 6.84375,
"learning_rate": 4.5978638394070835e-06,
"loss": 0.3816,
"step": 1060
},
{
"epoch": 0.2675,
"grad_norm": 6.1875,
"learning_rate": 4.58591790092183e-06,
"loss": 0.3854,
"step": 1070
},
{
"epoch": 0.27,
"grad_norm": 6.75,
"learning_rate": 4.5738131117124605e-06,
"loss": 0.3783,
"step": 1080
},
{
"epoch": 0.2725,
"grad_norm": 5.84375,
"learning_rate": 4.561550393605541e-06,
"loss": 0.384,
"step": 1090
},
{
"epoch": 0.275,
"grad_norm": 7.625,
"learning_rate": 4.549130680454532e-06,
"loss": 0.41,
"step": 1100
},
{
"epoch": 0.2775,
"grad_norm": 5.9375,
"learning_rate": 4.536554918068673e-06,
"loss": 0.3664,
"step": 1110
},
{
"epoch": 0.28,
"grad_norm": 5.03125,
"learning_rate": 4.523824064140961e-06,
"loss": 0.3727,
"step": 1120
},
{
"epoch": 0.2825,
"grad_norm": 6.09375,
"learning_rate": 4.510939088175211e-06,
"loss": 0.3764,
"step": 1130
},
{
"epoch": 0.285,
"grad_norm": 5.875,
"learning_rate": 4.49790097141223e-06,
"loss": 0.358,
"step": 1140
},
{
"epoch": 0.2875,
"grad_norm": 5.875,
"learning_rate": 4.484710706755087e-06,
"loss": 0.3549,
"step": 1150
},
{
"epoch": 0.29,
"grad_norm": 5.84375,
"learning_rate": 4.471369298693505e-06,
"loss": 0.3553,
"step": 1160
},
{
"epoch": 0.2925,
"grad_norm": 5.75,
"learning_rate": 4.457877763227361e-06,
"loss": 0.3623,
"step": 1170
},
{
"epoch": 0.295,
"grad_norm": 5.5625,
"learning_rate": 4.444237127789315e-06,
"loss": 0.3629,
"step": 1180
},
{
"epoch": 0.2975,
"grad_norm": 6.96875,
"learning_rate": 4.430448431166567e-06,
"loss": 0.3434,
"step": 1190
},
{
"epoch": 0.3,
"grad_norm": 5.9375,
"learning_rate": 4.416512723421752e-06,
"loss": 0.3549,
"step": 1200
},
{
"epoch": 0.3025,
"grad_norm": 5.53125,
"learning_rate": 4.402431065812968e-06,
"loss": 0.3461,
"step": 1210
},
{
"epoch": 0.305,
"grad_norm": 5.375,
"learning_rate": 4.388204530712959e-06,
"loss": 0.3547,
"step": 1220
},
{
"epoch": 0.3075,
"grad_norm": 5.90625,
"learning_rate": 4.373834201527457e-06,
"loss": 0.3383,
"step": 1230
},
{
"epoch": 0.31,
"grad_norm": 5.6875,
"learning_rate": 4.359321172612664e-06,
"loss": 0.3414,
"step": 1240
},
{
"epoch": 0.3125,
"grad_norm": 5.15625,
"learning_rate": 4.344666549191921e-06,
"loss": 0.3285,
"step": 1250
},
{
"epoch": 0.315,
"grad_norm": 5.4375,
"learning_rate": 4.329871447271541e-06,
"loss": 0.3352,
"step": 1260
},
{
"epoch": 0.3175,
"grad_norm": 5.0625,
"learning_rate": 4.314936993555816e-06,
"loss": 0.3441,
"step": 1270
},
{
"epoch": 0.32,
"grad_norm": 5.09375,
"learning_rate": 4.299864325361217e-06,
"loss": 0.3523,
"step": 1280
},
{
"epoch": 0.3225,
"grad_norm": 5.71875,
"learning_rate": 4.284654590529784e-06,
"loss": 0.3348,
"step": 1290
},
{
"epoch": 0.325,
"grad_norm": 5.28125,
"learning_rate": 4.269308947341711e-06,
"loss": 0.3252,
"step": 1300
},
{
"epoch": 0.3275,
"grad_norm": 4.8125,
"learning_rate": 4.25382856442714e-06,
"loss": 0.3439,
"step": 1310
},
{
"epoch": 0.33,
"grad_norm": 5.4375,
"learning_rate": 4.238214620677164e-06,
"loss": 0.3326,
"step": 1320
},
{
"epoch": 0.3325,
"grad_norm": 4.75,
"learning_rate": 4.222468305154052e-06,
"loss": 0.3332,
"step": 1330
},
{
"epoch": 0.335,
"grad_norm": 5.25,
"learning_rate": 4.206590817000695e-06,
"loss": 0.3277,
"step": 1340
},
{
"epoch": 0.3375,
"grad_norm": 5.03125,
"learning_rate": 4.190583365349289e-06,
"loss": 0.3363,
"step": 1350
},
{
"epoch": 0.34,
"grad_norm": 5.0,
"learning_rate": 4.174447169229252e-06,
"loss": 0.3412,
"step": 1360
},
{
"epoch": 0.3425,
"grad_norm": 5.28125,
"learning_rate": 4.158183457474392e-06,
"loss": 0.326,
"step": 1370
},
{
"epoch": 0.345,
"grad_norm": 5.3125,
"learning_rate": 4.141793468629327e-06,
"loss": 0.334,
"step": 1380
},
{
"epoch": 0.3475,
"grad_norm": 5.0,
"learning_rate": 4.125278450855165e-06,
"loss": 0.3367,
"step": 1390
},
{
"epoch": 0.35,
"grad_norm": 6.28125,
"learning_rate": 4.1086396618344474e-06,
"loss": 0.3168,
"step": 1400
},
{
"epoch": 0.3525,
"grad_norm": 5.15625,
"learning_rate": 4.09187836867538e-06,
"loss": 0.3193,
"step": 1410
},
{
"epoch": 0.355,
"grad_norm": 5.15625,
"learning_rate": 4.074995847815331e-06,
"loss": 0.3225,
"step": 1420
},
{
"epoch": 0.3575,
"grad_norm": 4.96875,
"learning_rate": 4.057993384923626e-06,
"loss": 0.332,
"step": 1430
},
{
"epoch": 0.36,
"grad_norm": 4.9375,
"learning_rate": 4.0408722748036426e-06,
"loss": 0.3221,
"step": 1440
},
{
"epoch": 0.3625,
"grad_norm": 4.8125,
"learning_rate": 4.023633821294203e-06,
"loss": 0.3211,
"step": 1450
},
{
"epoch": 0.365,
"grad_norm": 5.125,
"learning_rate": 4.006279337170283e-06,
"loss": 0.3195,
"step": 1460
},
{
"epoch": 0.3675,
"grad_norm": 4.96875,
"learning_rate": 3.988810144043041e-06,
"loss": 0.3225,
"step": 1470
},
{
"epoch": 0.37,
"grad_norm": 4.6875,
"learning_rate": 3.971227572259167e-06,
"loss": 0.3242,
"step": 1480
},
{
"epoch": 0.3725,
"grad_norm": 5.0625,
"learning_rate": 3.953532960799577e-06,
"loss": 0.3205,
"step": 1490
},
{
"epoch": 0.375,
"grad_norm": 4.71875,
"learning_rate": 3.935727657177439e-06,
"loss": 0.3152,
"step": 1500
},
{
"epoch": 0.3775,
"grad_norm": 4.90625,
"learning_rate": 3.917813017335562e-06,
"loss": 0.2998,
"step": 1510
},
{
"epoch": 0.38,
"grad_norm": 5.15625,
"learning_rate": 3.899790405543129e-06,
"loss": 0.3229,
"step": 1520
},
{
"epoch": 0.3825,
"grad_norm": 4.46875,
"learning_rate": 3.881661194291805e-06,
"loss": 0.3088,
"step": 1530
},
{
"epoch": 0.385,
"grad_norm": 4.9375,
"learning_rate": 3.863426764191216e-06,
"loss": 0.3205,
"step": 1540
},
{
"epoch": 0.3875,
"grad_norm": 5.15625,
"learning_rate": 3.845088503863813e-06,
"loss": 0.3137,
"step": 1550
},
{
"epoch": 0.39,
"grad_norm": 4.5,
"learning_rate": 3.826647809839119e-06,
"loss": 0.3055,
"step": 1560
},
{
"epoch": 0.3925,
"grad_norm": 5.15625,
"learning_rate": 3.8081060864473794e-06,
"loss": 0.316,
"step": 1570
},
{
"epoch": 0.395,
"grad_norm": 5.0625,
"learning_rate": 3.7894647457126188e-06,
"loss": 0.3215,
"step": 1580
},
{
"epoch": 0.3975,
"grad_norm": 5.03125,
"learning_rate": 3.770725207245106e-06,
"loss": 0.3125,
"step": 1590
},
{
"epoch": 0.4,
"grad_norm": 4.84375,
"learning_rate": 3.751888898133249e-06,
"loss": 0.3168,
"step": 1600
},
{
"epoch": 0.4025,
"grad_norm": 5.03125,
"learning_rate": 3.7329572528349145e-06,
"loss": 0.3074,
"step": 1610
},
{
"epoch": 0.405,
"grad_norm": 4.90625,
"learning_rate": 3.7139317130681886e-06,
"loss": 0.3248,
"step": 1620
},
{
"epoch": 0.4075,
"grad_norm": 4.59375,
"learning_rate": 3.694813727701584e-06,
"loss": 0.3041,
"step": 1630
},
{
"epoch": 0.41,
"grad_norm": 5.09375,
"learning_rate": 3.675604752643706e-06,
"loss": 0.3049,
"step": 1640
},
{
"epoch": 0.4125,
"grad_norm": 4.625,
"learning_rate": 3.6563062507323752e-06,
"loss": 0.3285,
"step": 1650
},
{
"epoch": 0.415,
"grad_norm": 4.75,
"learning_rate": 3.6369196916232297e-06,
"loss": 0.2977,
"step": 1660
},
{
"epoch": 0.4175,
"grad_norm": 4.71875,
"learning_rate": 3.6174465516778032e-06,
"loss": 0.3074,
"step": 1670
},
{
"epoch": 0.42,
"grad_norm": 4.625,
"learning_rate": 3.5978883138510963e-06,
"loss": 0.3055,
"step": 1680
},
{
"epoch": 0.4225,
"grad_norm": 4.53125,
"learning_rate": 3.578246467578642e-06,
"loss": 0.3096,
"step": 1690
},
{
"epoch": 0.425,
"grad_norm": 4.84375,
"learning_rate": 3.558522508663081e-06,
"loss": 0.3102,
"step": 1700
},
{
"epoch": 0.4275,
"grad_norm": 4.6875,
"learning_rate": 3.538717939160249e-06,
"loss": 0.2945,
"step": 1710
},
{
"epoch": 0.43,
"grad_norm": 4.625,
"learning_rate": 3.5188342672647897e-06,
"loss": 0.2992,
"step": 1720
},
{
"epoch": 0.4325,
"grad_norm": 4.5,
"learning_rate": 3.4988730071953005e-06,
"loss": 0.3107,
"step": 1730
},
{
"epoch": 0.435,
"grad_norm": 4.375,
"learning_rate": 3.478835679079019e-06,
"loss": 0.3031,
"step": 1740
},
{
"epoch": 0.4375,
"grad_norm": 4.8125,
"learning_rate": 3.4587238088360605e-06,
"loss": 0.317,
"step": 1750
},
{
"epoch": 0.44,
"grad_norm": 4.5625,
"learning_rate": 3.438538928063208e-06,
"loss": 0.2965,
"step": 1760
},
{
"epoch": 0.4425,
"grad_norm": 4.65625,
"learning_rate": 3.4182825739172826e-06,
"loss": 0.2899,
"step": 1770
},
{
"epoch": 0.445,
"grad_norm": 4.375,
"learning_rate": 3.3979562889980777e-06,
"loss": 0.3107,
"step": 1780
},
{
"epoch": 0.4475,
"grad_norm": 4.46875,
"learning_rate": 3.377561621230887e-06,
"loss": 0.2855,
"step": 1790
},
{
"epoch": 0.45,
"grad_norm": 4.375,
"learning_rate": 3.357100123748621e-06,
"loss": 0.3061,
"step": 1800
},
{
"epoch": 0.4525,
"grad_norm": 4.53125,
"learning_rate": 3.3365733547735334e-06,
"loss": 0.3068,
"step": 1810
},
{
"epoch": 0.455,
"grad_norm": 4.5,
"learning_rate": 3.315982877498555e-06,
"loss": 0.2834,
"step": 1820
},
{
"epoch": 0.4575,
"grad_norm": 4.34375,
"learning_rate": 3.2953302599682487e-06,
"loss": 0.3029,
"step": 1830
},
{
"epoch": 0.46,
"grad_norm": 4.5625,
"learning_rate": 3.2746170749593998e-06,
"loss": 0.3121,
"step": 1840
},
{
"epoch": 0.4625,
"grad_norm": 4.46875,
"learning_rate": 3.2538448998612394e-06,
"loss": 0.293,
"step": 1850
},
{
"epoch": 0.465,
"grad_norm": 4.875,
"learning_rate": 3.233015316555326e-06,
"loss": 0.2988,
"step": 1860
},
{
"epoch": 0.4675,
"grad_norm": 4.53125,
"learning_rate": 3.212129911295074e-06,
"loss": 0.3002,
"step": 1870
},
{
"epoch": 0.47,
"grad_norm": 7.6875,
"learning_rate": 3.1911902745849526e-06,
"loss": 0.3146,
"step": 1880
},
{
"epoch": 0.4725,
"grad_norm": 4.3125,
"learning_rate": 3.17019800105937e-06,
"loss": 0.2926,
"step": 1890
},
{
"epoch": 0.475,
"grad_norm": 4.46875,
"learning_rate": 3.1491546893612296e-06,
"loss": 0.2998,
"step": 1900
},
{
"epoch": 0.4775,
"grad_norm": 4.71875,
"learning_rate": 3.128061942020189e-06,
"loss": 0.3104,
"step": 1910
},
{
"epoch": 0.48,
"grad_norm": 4.75,
"learning_rate": 3.1069213653306242e-06,
"loss": 0.2955,
"step": 1920
},
{
"epoch": 0.4825,
"grad_norm": 4.4375,
"learning_rate": 3.0857345692292968e-06,
"loss": 0.2937,
"step": 1930
},
{
"epoch": 0.485,
"grad_norm": 4.25,
"learning_rate": 3.0645031671727598e-06,
"loss": 0.2878,
"step": 1940
},
{
"epoch": 0.4875,
"grad_norm": 4.53125,
"learning_rate": 3.0432287760144797e-06,
"loss": 0.2988,
"step": 1950
},
{
"epoch": 0.49,
"grad_norm": 4.21875,
"learning_rate": 3.0219130158817093e-06,
"loss": 0.3043,
"step": 1960
},
{
"epoch": 0.4925,
"grad_norm": 4.8125,
"learning_rate": 3.0005575100521115e-06,
"loss": 0.2842,
"step": 1970
},
{
"epoch": 0.495,
"grad_norm": 4.5625,
"learning_rate": 2.979163884830137e-06,
"loss": 0.2922,
"step": 1980
},
{
"epoch": 0.4975,
"grad_norm": 4.9375,
"learning_rate": 2.957733769423174e-06,
"loss": 0.2998,
"step": 1990
},
{
"epoch": 0.5,
"grad_norm": 4.5625,
"learning_rate": 2.93626879581748e-06,
"loss": 0.2908,
"step": 2000
}
],
"logging_steps": 10,
"max_steps": 4000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 2000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 2.5153945731072e+17,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}