Nekodimos/Whisper-SOR / trainer_state.json
Nekodimos's picture
download
raw
35.7 kB
{
"best_global_step": 5000,
"best_metric": 25.21137586471945,
"best_model_checkpoint": "./ORO500/checkpoint-5000",
"epoch": 1.599488163787588,
"eval_steps": 1000,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.00799744081893794,
"grad_norm": 1440699.375,
"learning_rate": 2.4000000000000003e-06,
"loss": 7.2656,
"step": 25
},
{
"epoch": 0.01599488163787588,
"grad_norm": 647302.125,
"learning_rate": 4.9000000000000005e-06,
"loss": 5.3519,
"step": 50
},
{
"epoch": 0.02399232245681382,
"grad_norm": 645944.125,
"learning_rate": 7.4e-06,
"loss": 4.8477,
"step": 75
},
{
"epoch": 0.03198976327575176,
"grad_norm": 709124.625,
"learning_rate": 9.900000000000002e-06,
"loss": 4.6818,
"step": 100
},
{
"epoch": 0.0399872040946897,
"grad_norm": 740838.9375,
"learning_rate": 1.24e-05,
"loss": 4.4692,
"step": 125
},
{
"epoch": 0.04798464491362764,
"grad_norm": 777757.125,
"learning_rate": 1.49e-05,
"loss": 4.4166,
"step": 150
},
{
"epoch": 0.05598208573256558,
"grad_norm": 842929.375,
"learning_rate": 1.74e-05,
"loss": 4.2822,
"step": 175
},
{
"epoch": 0.06397952655150352,
"grad_norm": 1032184.875,
"learning_rate": 1.9900000000000003e-05,
"loss": 4.0646,
"step": 200
},
{
"epoch": 0.07197696737044146,
"grad_norm": 787339.375,
"learning_rate": 2.2400000000000002e-05,
"loss": 3.7711,
"step": 225
},
{
"epoch": 0.0799744081893794,
"grad_norm": 775550.1875,
"learning_rate": 2.4900000000000002e-05,
"loss": 3.5757,
"step": 250
},
{
"epoch": 0.08797184900831734,
"grad_norm": 903174.1875,
"learning_rate": 2.7400000000000002e-05,
"loss": 3.3609,
"step": 275
},
{
"epoch": 0.09596928982725528,
"grad_norm": 1287621.125,
"learning_rate": 2.9900000000000002e-05,
"loss": 3.1076,
"step": 300
},
{
"epoch": 0.10396673064619322,
"grad_norm": 1256387.0,
"learning_rate": 3.24e-05,
"loss": 2.6632,
"step": 325
},
{
"epoch": 0.11196417146513116,
"grad_norm": 1010871.0625,
"learning_rate": 3.49e-05,
"loss": 2.1169,
"step": 350
},
{
"epoch": 0.1199616122840691,
"grad_norm": 749345.125,
"learning_rate": 3.74e-05,
"loss": 1.5657,
"step": 375
},
{
"epoch": 0.12795905310300704,
"grad_norm": 622161.25,
"learning_rate": 3.99e-05,
"loss": 1.2129,
"step": 400
},
{
"epoch": 0.13595649392194498,
"grad_norm": 630759.6875,
"learning_rate": 4.24e-05,
"loss": 1.0569,
"step": 425
},
{
"epoch": 0.14395393474088292,
"grad_norm": 645867.75,
"learning_rate": 4.49e-05,
"loss": 0.9735,
"step": 450
},
{
"epoch": 0.15195137555982086,
"grad_norm": 570249.3125,
"learning_rate": 4.74e-05,
"loss": 0.9465,
"step": 475
},
{
"epoch": 0.1599488163787588,
"grad_norm": 613526.0,
"learning_rate": 4.99e-05,
"loss": 0.8566,
"step": 500
},
{
"epoch": 0.16794625719769674,
"grad_norm": 540352.1875,
"learning_rate": 4.973333333333334e-05,
"loss": 0.8686,
"step": 525
},
{
"epoch": 0.17594369801663468,
"grad_norm": 522272.8125,
"learning_rate": 4.945555555555556e-05,
"loss": 0.815,
"step": 550
},
{
"epoch": 0.18394113883557262,
"grad_norm": 491657.15625,
"learning_rate": 4.917777777777778e-05,
"loss": 0.7209,
"step": 575
},
{
"epoch": 0.19193857965451055,
"grad_norm": 447048.9375,
"learning_rate": 4.89e-05,
"loss": 0.7307,
"step": 600
},
{
"epoch": 0.1999360204734485,
"grad_norm": 499030.6875,
"learning_rate": 4.862222222222222e-05,
"loss": 0.7304,
"step": 625
},
{
"epoch": 0.20793346129238643,
"grad_norm": 469617.28125,
"learning_rate": 4.8344444444444447e-05,
"loss": 0.6952,
"step": 650
},
{
"epoch": 0.21593090211132437,
"grad_norm": 423677.71875,
"learning_rate": 4.806666666666667e-05,
"loss": 0.638,
"step": 675
},
{
"epoch": 0.22392834293026231,
"grad_norm": 443322.09375,
"learning_rate": 4.778888888888889e-05,
"loss": 0.668,
"step": 700
},
{
"epoch": 0.23192578374920025,
"grad_norm": 425010.1875,
"learning_rate": 4.751111111111111e-05,
"loss": 0.6433,
"step": 725
},
{
"epoch": 0.2399232245681382,
"grad_norm": 440727.375,
"learning_rate": 4.7233333333333334e-05,
"loss": 0.6465,
"step": 750
},
{
"epoch": 0.24792066538707613,
"grad_norm": 445811.65625,
"learning_rate": 4.695555555555556e-05,
"loss": 0.5966,
"step": 775
},
{
"epoch": 0.2559181062060141,
"grad_norm": 392700.03125,
"learning_rate": 4.6677777777777785e-05,
"loss": 0.6132,
"step": 800
},
{
"epoch": 0.263915547024952,
"grad_norm": 346365.21875,
"learning_rate": 4.64e-05,
"loss": 0.6177,
"step": 825
},
{
"epoch": 0.27191298784388995,
"grad_norm": 371717.78125,
"learning_rate": 4.612222222222222e-05,
"loss": 0.6013,
"step": 850
},
{
"epoch": 0.2799104286628279,
"grad_norm": 320079.03125,
"learning_rate": 4.584444444444445e-05,
"loss": 0.5826,
"step": 875
},
{
"epoch": 0.28790786948176583,
"grad_norm": 355400.8125,
"learning_rate": 4.556666666666667e-05,
"loss": 0.5779,
"step": 900
},
{
"epoch": 0.29590531030070377,
"grad_norm": 312739.21875,
"learning_rate": 4.528888888888889e-05,
"loss": 0.5789,
"step": 925
},
{
"epoch": 0.3039027511196417,
"grad_norm": 361398.8125,
"learning_rate": 4.5011111111111116e-05,
"loss": 0.5627,
"step": 950
},
{
"epoch": 0.31190019193857965,
"grad_norm": 367834.0,
"learning_rate": 4.473333333333334e-05,
"loss": 0.5691,
"step": 975
},
{
"epoch": 0.3198976327575176,
"grad_norm": 290713.9375,
"learning_rate": 4.445555555555555e-05,
"loss": 0.5194,
"step": 1000
},
{
"epoch": 0.3198976327575176,
"eval_loss": 0.4918915629386902,
"eval_runtime": 1929.0111,
"eval_samples_per_second": 0.611,
"eval_steps_per_second": 0.077,
"eval_wer": 45.857033051498846,
"step": 1000
},
{
"epoch": 0.32789507357645553,
"grad_norm": 365725.90625,
"learning_rate": 4.417777777777778e-05,
"loss": 0.5583,
"step": 1025
},
{
"epoch": 0.33589251439539347,
"grad_norm": 291537.3125,
"learning_rate": 4.39e-05,
"loss": 0.54,
"step": 1050
},
{
"epoch": 0.3438899552143314,
"grad_norm": 411854.46875,
"learning_rate": 4.362222222222223e-05,
"loss": 0.5628,
"step": 1075
},
{
"epoch": 0.35188739603326935,
"grad_norm": 337915.6875,
"learning_rate": 4.334444444444445e-05,
"loss": 0.5307,
"step": 1100
},
{
"epoch": 0.3598848368522073,
"grad_norm": 381128.53125,
"learning_rate": 4.3066666666666665e-05,
"loss": 0.4793,
"step": 1125
},
{
"epoch": 0.36788227767114523,
"grad_norm": 420617.34375,
"learning_rate": 4.278888888888889e-05,
"loss": 0.5077,
"step": 1150
},
{
"epoch": 0.37587971849008317,
"grad_norm": 308631.09375,
"learning_rate": 4.2511111111111116e-05,
"loss": 0.5096,
"step": 1175
},
{
"epoch": 0.3838771593090211,
"grad_norm": 276476.53125,
"learning_rate": 4.2233333333333334e-05,
"loss": 0.5206,
"step": 1200
},
{
"epoch": 0.39187460012795905,
"grad_norm": 295005.25,
"learning_rate": 4.195555555555556e-05,
"loss": 0.5253,
"step": 1225
},
{
"epoch": 0.399872040946897,
"grad_norm": 320991.375,
"learning_rate": 4.1677777777777785e-05,
"loss": 0.524,
"step": 1250
},
{
"epoch": 0.40786948176583493,
"grad_norm": 317325.125,
"learning_rate": 4.14e-05,
"loss": 0.5131,
"step": 1275
},
{
"epoch": 0.41586692258477287,
"grad_norm": 297112.46875,
"learning_rate": 4.112222222222222e-05,
"loss": 0.4622,
"step": 1300
},
{
"epoch": 0.4238643634037108,
"grad_norm": 321116.5625,
"learning_rate": 4.084444444444445e-05,
"loss": 0.4756,
"step": 1325
},
{
"epoch": 0.43186180422264875,
"grad_norm": 277808.53125,
"learning_rate": 4.056666666666667e-05,
"loss": 0.4557,
"step": 1350
},
{
"epoch": 0.4398592450415867,
"grad_norm": 253608.859375,
"learning_rate": 4.028888888888889e-05,
"loss": 0.4633,
"step": 1375
},
{
"epoch": 0.44785668586052463,
"grad_norm": 246855.109375,
"learning_rate": 4.001111111111111e-05,
"loss": 0.4675,
"step": 1400
},
{
"epoch": 0.45585412667946257,
"grad_norm": 376046.3125,
"learning_rate": 3.9733333333333335e-05,
"loss": 0.4902,
"step": 1425
},
{
"epoch": 0.4638515674984005,
"grad_norm": 271958.9375,
"learning_rate": 3.945555555555556e-05,
"loss": 0.4532,
"step": 1450
},
{
"epoch": 0.47184900831733845,
"grad_norm": 329253.78125,
"learning_rate": 3.917777777777778e-05,
"loss": 0.4806,
"step": 1475
},
{
"epoch": 0.4798464491362764,
"grad_norm": 276066.90625,
"learning_rate": 3.8900000000000004e-05,
"loss": 0.4511,
"step": 1500
},
{
"epoch": 0.4878438899552143,
"grad_norm": 290485.03125,
"learning_rate": 3.862222222222223e-05,
"loss": 0.4411,
"step": 1525
},
{
"epoch": 0.49584133077415227,
"grad_norm": 251736.171875,
"learning_rate": 3.834444444444444e-05,
"loss": 0.4519,
"step": 1550
},
{
"epoch": 0.5038387715930902,
"grad_norm": 257916.59375,
"learning_rate": 3.8066666666666666e-05,
"loss": 0.4505,
"step": 1575
},
{
"epoch": 0.5118362124120281,
"grad_norm": 311473.1875,
"learning_rate": 3.778888888888889e-05,
"loss": 0.4361,
"step": 1600
},
{
"epoch": 0.5198336532309661,
"grad_norm": 255816.328125,
"learning_rate": 3.7511111111111116e-05,
"loss": 0.4637,
"step": 1625
},
{
"epoch": 0.527831094049904,
"grad_norm": 313451.0625,
"learning_rate": 3.7233333333333335e-05,
"loss": 0.4198,
"step": 1650
},
{
"epoch": 0.535828534868842,
"grad_norm": 262136.203125,
"learning_rate": 3.695555555555556e-05,
"loss": 0.4275,
"step": 1675
},
{
"epoch": 0.5438259756877799,
"grad_norm": 292295.25,
"learning_rate": 3.667777777777778e-05,
"loss": 0.4627,
"step": 1700
},
{
"epoch": 0.5518234165067178,
"grad_norm": 215166.71875,
"learning_rate": 3.6400000000000004e-05,
"loss": 0.4417,
"step": 1725
},
{
"epoch": 0.5598208573256558,
"grad_norm": 252266.90625,
"learning_rate": 3.612222222222222e-05,
"loss": 0.4494,
"step": 1750
},
{
"epoch": 0.5678182981445937,
"grad_norm": 313254.59375,
"learning_rate": 3.584444444444445e-05,
"loss": 0.4071,
"step": 1775
},
{
"epoch": 0.5758157389635317,
"grad_norm": 232868.125,
"learning_rate": 3.556666666666667e-05,
"loss": 0.4159,
"step": 1800
},
{
"epoch": 0.5838131797824696,
"grad_norm": 200838.34375,
"learning_rate": 3.528888888888889e-05,
"loss": 0.4064,
"step": 1825
},
{
"epoch": 0.5918106206014075,
"grad_norm": 284412.53125,
"learning_rate": 3.501111111111111e-05,
"loss": 0.4168,
"step": 1850
},
{
"epoch": 0.5998080614203455,
"grad_norm": 244401.140625,
"learning_rate": 3.4733333333333335e-05,
"loss": 0.4427,
"step": 1875
},
{
"epoch": 0.6078055022392834,
"grad_norm": 274653.25,
"learning_rate": 3.445555555555556e-05,
"loss": 0.405,
"step": 1900
},
{
"epoch": 0.6158029430582214,
"grad_norm": 251051.59375,
"learning_rate": 3.417777777777778e-05,
"loss": 0.4001,
"step": 1925
},
{
"epoch": 0.6238003838771593,
"grad_norm": 236417.109375,
"learning_rate": 3.3900000000000004e-05,
"loss": 0.413,
"step": 1950
},
{
"epoch": 0.6317978246960972,
"grad_norm": 252315.546875,
"learning_rate": 3.362222222222222e-05,
"loss": 0.4166,
"step": 1975
},
{
"epoch": 0.6397952655150352,
"grad_norm": 230474.765625,
"learning_rate": 3.334444444444445e-05,
"loss": 0.4237,
"step": 2000
},
{
"epoch": 0.6397952655150352,
"eval_loss": 0.3680548369884491,
"eval_runtime": 2035.5053,
"eval_samples_per_second": 0.579,
"eval_steps_per_second": 0.073,
"eval_wer": 37.42505764796311,
"step": 2000
},
{
"epoch": 0.6477927063339731,
"grad_norm": 258175.65625,
"learning_rate": 3.3066666666666666e-05,
"loss": 0.4262,
"step": 2025
},
{
"epoch": 0.6557901471529111,
"grad_norm": 342532.25,
"learning_rate": 3.278888888888889e-05,
"loss": 0.4226,
"step": 2050
},
{
"epoch": 0.663787587971849,
"grad_norm": 246628.21875,
"learning_rate": 3.251111111111112e-05,
"loss": 0.4182,
"step": 2075
},
{
"epoch": 0.6717850287907869,
"grad_norm": 247210.1875,
"learning_rate": 3.2233333333333335e-05,
"loss": 0.3736,
"step": 2100
},
{
"epoch": 0.6797824696097249,
"grad_norm": 344157.65625,
"learning_rate": 3.1955555555555554e-05,
"loss": 0.3912,
"step": 2125
},
{
"epoch": 0.6877799104286628,
"grad_norm": 297428.875,
"learning_rate": 3.167777777777778e-05,
"loss": 0.3992,
"step": 2150
},
{
"epoch": 0.6957773512476008,
"grad_norm": 234133.96875,
"learning_rate": 3.1400000000000004e-05,
"loss": 0.3737,
"step": 2175
},
{
"epoch": 0.7037747920665387,
"grad_norm": 228695.609375,
"learning_rate": 3.112222222222222e-05,
"loss": 0.3883,
"step": 2200
},
{
"epoch": 0.7117722328854766,
"grad_norm": 293855.6875,
"learning_rate": 3.084444444444445e-05,
"loss": 0.3915,
"step": 2225
},
{
"epoch": 0.7197696737044146,
"grad_norm": 270283.40625,
"learning_rate": 3.0566666666666667e-05,
"loss": 0.3772,
"step": 2250
},
{
"epoch": 0.7277671145233525,
"grad_norm": 273114.5,
"learning_rate": 3.028888888888889e-05,
"loss": 0.3683,
"step": 2275
},
{
"epoch": 0.7357645553422905,
"grad_norm": 245567.890625,
"learning_rate": 3.0011111111111114e-05,
"loss": 0.356,
"step": 2300
},
{
"epoch": 0.7437619961612284,
"grad_norm": 268476.65625,
"learning_rate": 2.9733333333333336e-05,
"loss": 0.3724,
"step": 2325
},
{
"epoch": 0.7517594369801663,
"grad_norm": 247504.890625,
"learning_rate": 2.9455555555555557e-05,
"loss": 0.3522,
"step": 2350
},
{
"epoch": 0.7597568777991043,
"grad_norm": 268137.65625,
"learning_rate": 2.9177777777777783e-05,
"loss": 0.4007,
"step": 2375
},
{
"epoch": 0.7677543186180422,
"grad_norm": 278463.5,
"learning_rate": 2.8899999999999998e-05,
"loss": 0.3811,
"step": 2400
},
{
"epoch": 0.7757517594369802,
"grad_norm": 223856.390625,
"learning_rate": 2.8622222222222223e-05,
"loss": 0.359,
"step": 2425
},
{
"epoch": 0.7837492002559181,
"grad_norm": 264782.59375,
"learning_rate": 2.8344444444444445e-05,
"loss": 0.3905,
"step": 2450
},
{
"epoch": 0.791746641074856,
"grad_norm": 246214.53125,
"learning_rate": 2.806666666666667e-05,
"loss": 0.359,
"step": 2475
},
{
"epoch": 0.799744081893794,
"grad_norm": 224425.8125,
"learning_rate": 2.7788888888888892e-05,
"loss": 0.3606,
"step": 2500
},
{
"epoch": 0.8077415227127319,
"grad_norm": 334262.0,
"learning_rate": 2.751111111111111e-05,
"loss": 0.3883,
"step": 2525
},
{
"epoch": 0.8157389635316699,
"grad_norm": 301311.0625,
"learning_rate": 2.7233333333333332e-05,
"loss": 0.3566,
"step": 2550
},
{
"epoch": 0.8237364043506078,
"grad_norm": 304223.03125,
"learning_rate": 2.6955555555555558e-05,
"loss": 0.3593,
"step": 2575
},
{
"epoch": 0.8317338451695457,
"grad_norm": 237099.953125,
"learning_rate": 2.667777777777778e-05,
"loss": 0.3634,
"step": 2600
},
{
"epoch": 0.8397312859884837,
"grad_norm": 228552.375,
"learning_rate": 2.64e-05,
"loss": 0.3388,
"step": 2625
},
{
"epoch": 0.8477287268074216,
"grad_norm": 262023.40625,
"learning_rate": 2.6122222222222227e-05,
"loss": 0.3533,
"step": 2650
},
{
"epoch": 0.8557261676263596,
"grad_norm": 292732.125,
"learning_rate": 2.5844444444444442e-05,
"loss": 0.372,
"step": 2675
},
{
"epoch": 0.8637236084452975,
"grad_norm": 257013.453125,
"learning_rate": 2.5566666666666667e-05,
"loss": 0.3434,
"step": 2700
},
{
"epoch": 0.8717210492642354,
"grad_norm": 302624.0625,
"learning_rate": 2.528888888888889e-05,
"loss": 0.3446,
"step": 2725
},
{
"epoch": 0.8797184900831734,
"grad_norm": 262445.84375,
"learning_rate": 2.5011111111111114e-05,
"loss": 0.3369,
"step": 2750
},
{
"epoch": 0.8877159309021113,
"grad_norm": 221258.78125,
"learning_rate": 2.4733333333333333e-05,
"loss": 0.3555,
"step": 2775
},
{
"epoch": 0.8957133717210493,
"grad_norm": 272136.9375,
"learning_rate": 2.4455555555555558e-05,
"loss": 0.3852,
"step": 2800
},
{
"epoch": 0.9037108125399872,
"grad_norm": 233267.109375,
"learning_rate": 2.417777777777778e-05,
"loss": 0.3235,
"step": 2825
},
{
"epoch": 0.9117082533589251,
"grad_norm": 252875.796875,
"learning_rate": 2.39e-05,
"loss": 0.3285,
"step": 2850
},
{
"epoch": 0.9197056941778631,
"grad_norm": 241668.515625,
"learning_rate": 2.3622222222222223e-05,
"loss": 0.3531,
"step": 2875
},
{
"epoch": 0.927703134996801,
"grad_norm": 257490.875,
"learning_rate": 2.3344444444444445e-05,
"loss": 0.3177,
"step": 2900
},
{
"epoch": 0.935700575815739,
"grad_norm": 234095.890625,
"learning_rate": 2.3066666666666667e-05,
"loss": 0.333,
"step": 2925
},
{
"epoch": 0.9436980166346769,
"grad_norm": 290286.84375,
"learning_rate": 2.278888888888889e-05,
"loss": 0.3588,
"step": 2950
},
{
"epoch": 0.9516954574536148,
"grad_norm": 266112.71875,
"learning_rate": 2.2511111111111114e-05,
"loss": 0.3272,
"step": 2975
},
{
"epoch": 0.9596928982725528,
"grad_norm": 321205.1875,
"learning_rate": 2.2233333333333333e-05,
"loss": 0.349,
"step": 3000
},
{
"epoch": 0.9596928982725528,
"eval_loss": 0.30465400218963623,
"eval_runtime": 1844.2545,
"eval_samples_per_second": 0.639,
"eval_steps_per_second": 0.08,
"eval_wer": 30.545734050730207,
"step": 3000
},
{
"epoch": 0.9676903390914907,
"grad_norm": 237167.96875,
"learning_rate": 2.1955555555555558e-05,
"loss": 0.3497,
"step": 3025
},
{
"epoch": 0.9756877799104287,
"grad_norm": 172899.734375,
"learning_rate": 2.167777777777778e-05,
"loss": 0.3311,
"step": 3050
},
{
"epoch": 0.9836852207293666,
"grad_norm": 234054.0625,
"learning_rate": 2.1400000000000002e-05,
"loss": 0.3552,
"step": 3075
},
{
"epoch": 0.9916826615483045,
"grad_norm": 252137.0,
"learning_rate": 2.1122222222222224e-05,
"loss": 0.3146,
"step": 3100
},
{
"epoch": 0.9996801023672425,
"grad_norm": 195909.671875,
"learning_rate": 2.0844444444444446e-05,
"loss": 0.346,
"step": 3125
},
{
"epoch": 1.0076775431861804,
"grad_norm": 201506.375,
"learning_rate": 2.0566666666666667e-05,
"loss": 0.2571,
"step": 3150
},
{
"epoch": 1.0156749840051185,
"grad_norm": 223208.234375,
"learning_rate": 2.028888888888889e-05,
"loss": 0.2352,
"step": 3175
},
{
"epoch": 1.0236724248240563,
"grad_norm": 201917.8125,
"learning_rate": 2.001111111111111e-05,
"loss": 0.224,
"step": 3200
},
{
"epoch": 1.0316698656429943,
"grad_norm": 262689.75,
"learning_rate": 1.9733333333333333e-05,
"loss": 0.2359,
"step": 3225
},
{
"epoch": 1.0396673064619322,
"grad_norm": 195509.765625,
"learning_rate": 1.9455555555555558e-05,
"loss": 0.2279,
"step": 3250
},
{
"epoch": 1.0476647472808702,
"grad_norm": 247717.46875,
"learning_rate": 1.9177777777777777e-05,
"loss": 0.2359,
"step": 3275
},
{
"epoch": 1.055662188099808,
"grad_norm": 193547.796875,
"learning_rate": 1.8900000000000002e-05,
"loss": 0.234,
"step": 3300
},
{
"epoch": 1.063659628918746,
"grad_norm": 183752.125,
"learning_rate": 1.8622222222222224e-05,
"loss": 0.2241,
"step": 3325
},
{
"epoch": 1.071657069737684,
"grad_norm": 228915.0625,
"learning_rate": 1.8344444444444446e-05,
"loss": 0.2148,
"step": 3350
},
{
"epoch": 1.079654510556622,
"grad_norm": 234940.484375,
"learning_rate": 1.8066666666666668e-05,
"loss": 0.2517,
"step": 3375
},
{
"epoch": 1.0876519513755598,
"grad_norm": 201199.984375,
"learning_rate": 1.778888888888889e-05,
"loss": 0.2553,
"step": 3400
},
{
"epoch": 1.0956493921944979,
"grad_norm": 245486.96875,
"learning_rate": 1.751111111111111e-05,
"loss": 0.2468,
"step": 3425
},
{
"epoch": 1.1036468330134357,
"grad_norm": 205952.375,
"learning_rate": 1.7233333333333333e-05,
"loss": 0.2456,
"step": 3450
},
{
"epoch": 1.1116442738323737,
"grad_norm": 191502.421875,
"learning_rate": 1.6955555555555555e-05,
"loss": 0.2255,
"step": 3475
},
{
"epoch": 1.1196417146513116,
"grad_norm": 256403.328125,
"learning_rate": 1.6677777777777777e-05,
"loss": 0.2159,
"step": 3500
},
{
"epoch": 1.1276391554702494,
"grad_norm": 206577.75,
"learning_rate": 1.6400000000000002e-05,
"loss": 0.2347,
"step": 3525
},
{
"epoch": 1.1356365962891874,
"grad_norm": 242599.140625,
"learning_rate": 1.612222222222222e-05,
"loss": 0.2403,
"step": 3550
},
{
"epoch": 1.1436340371081255,
"grad_norm": 235553.34375,
"learning_rate": 1.5844444444444446e-05,
"loss": 0.2339,
"step": 3575
},
{
"epoch": 1.1516314779270633,
"grad_norm": 226743.9375,
"learning_rate": 1.5566666666666668e-05,
"loss": 0.2226,
"step": 3600
},
{
"epoch": 1.1596289187460012,
"grad_norm": 260374.8125,
"learning_rate": 1.528888888888889e-05,
"loss": 0.2453,
"step": 3625
},
{
"epoch": 1.1676263595649392,
"grad_norm": 249406.296875,
"learning_rate": 1.5011111111111112e-05,
"loss": 0.2393,
"step": 3650
},
{
"epoch": 1.1756238003838773,
"grad_norm": 226856.53125,
"learning_rate": 1.4733333333333335e-05,
"loss": 0.2039,
"step": 3675
},
{
"epoch": 1.183621241202815,
"grad_norm": 274827.625,
"learning_rate": 1.4455555555555555e-05,
"loss": 0.2337,
"step": 3700
},
{
"epoch": 1.191618682021753,
"grad_norm": 226148.796875,
"learning_rate": 1.4177777777777779e-05,
"loss": 0.2297,
"step": 3725
},
{
"epoch": 1.199616122840691,
"grad_norm": 232043.125,
"learning_rate": 1.3900000000000002e-05,
"loss": 0.2074,
"step": 3750
},
{
"epoch": 1.207613563659629,
"grad_norm": 186210.421875,
"learning_rate": 1.3622222222222223e-05,
"loss": 0.2209,
"step": 3775
},
{
"epoch": 1.2156110044785668,
"grad_norm": 225763.90625,
"learning_rate": 1.3344444444444446e-05,
"loss": 0.2201,
"step": 3800
},
{
"epoch": 1.2236084452975047,
"grad_norm": 175443.015625,
"learning_rate": 1.3066666666666666e-05,
"loss": 0.2256,
"step": 3825
},
{
"epoch": 1.2316058861164427,
"grad_norm": 211604.6875,
"learning_rate": 1.278888888888889e-05,
"loss": 0.2284,
"step": 3850
},
{
"epoch": 1.2396033269353808,
"grad_norm": 169400.21875,
"learning_rate": 1.2511111111111112e-05,
"loss": 0.2124,
"step": 3875
},
{
"epoch": 1.2476007677543186,
"grad_norm": 207476.0625,
"learning_rate": 1.2233333333333334e-05,
"loss": 0.2204,
"step": 3900
},
{
"epoch": 1.2555982085732564,
"grad_norm": 242282.1875,
"learning_rate": 1.1955555555555556e-05,
"loss": 0.2286,
"step": 3925
},
{
"epoch": 1.2635956493921945,
"grad_norm": 189784.71875,
"learning_rate": 1.1677777777777777e-05,
"loss": 0.2287,
"step": 3950
},
{
"epoch": 1.2715930902111325,
"grad_norm": 292144.40625,
"learning_rate": 1.1400000000000001e-05,
"loss": 0.2295,
"step": 3975
},
{
"epoch": 1.2795905310300704,
"grad_norm": 211320.203125,
"learning_rate": 1.1122222222222223e-05,
"loss": 0.2235,
"step": 4000
},
{
"epoch": 1.2795905310300704,
"eval_loss": 0.27581608295440674,
"eval_runtime": 1778.0071,
"eval_samples_per_second": 0.663,
"eval_steps_per_second": 0.083,
"eval_wer": 27.540353574173714,
"step": 4000
},
{
"epoch": 1.2875879718490082,
"grad_norm": 179223.859375,
"learning_rate": 1.0844444444444445e-05,
"loss": 0.2126,
"step": 4025
},
{
"epoch": 1.2955854126679462,
"grad_norm": 186728.921875,
"learning_rate": 1.0566666666666668e-05,
"loss": 0.2171,
"step": 4050
},
{
"epoch": 1.3035828534868843,
"grad_norm": 165415.40625,
"learning_rate": 1.028888888888889e-05,
"loss": 0.2234,
"step": 4075
},
{
"epoch": 1.3115802943058221,
"grad_norm": 265208.5625,
"learning_rate": 1.0011111111111112e-05,
"loss": 0.222,
"step": 4100
},
{
"epoch": 1.31957773512476,
"grad_norm": 205896.796875,
"learning_rate": 9.733333333333334e-06,
"loss": 0.2177,
"step": 4125
},
{
"epoch": 1.327575175943698,
"grad_norm": 143080.796875,
"learning_rate": 9.455555555555556e-06,
"loss": 0.2,
"step": 4150
},
{
"epoch": 1.335572616762636,
"grad_norm": 273380.78125,
"learning_rate": 9.177777777777778e-06,
"loss": 0.2128,
"step": 4175
},
{
"epoch": 1.3435700575815739,
"grad_norm": 251681.765625,
"learning_rate": 8.9e-06,
"loss": 0.2295,
"step": 4200
},
{
"epoch": 1.3515674984005117,
"grad_norm": 202703.6875,
"learning_rate": 8.622222222222223e-06,
"loss": 0.2241,
"step": 4225
},
{
"epoch": 1.3595649392194498,
"grad_norm": 207133.265625,
"learning_rate": 8.344444444444445e-06,
"loss": 0.1944,
"step": 4250
},
{
"epoch": 1.3675623800383878,
"grad_norm": 205800.71875,
"learning_rate": 8.066666666666667e-06,
"loss": 0.2047,
"step": 4275
},
{
"epoch": 1.3755598208573256,
"grad_norm": 195775.4375,
"learning_rate": 7.788888888888889e-06,
"loss": 0.2097,
"step": 4300
},
{
"epoch": 1.3835572616762635,
"grad_norm": 209468.828125,
"learning_rate": 7.511111111111112e-06,
"loss": 0.1902,
"step": 4325
},
{
"epoch": 1.3915547024952015,
"grad_norm": 206627.6875,
"learning_rate": 7.233333333333333e-06,
"loss": 0.1916,
"step": 4350
},
{
"epoch": 1.3995521433141396,
"grad_norm": 226189.765625,
"learning_rate": 6.955555555555555e-06,
"loss": 0.2081,
"step": 4375
},
{
"epoch": 1.4075495841330774,
"grad_norm": 181588.90625,
"learning_rate": 6.677777777777779e-06,
"loss": 0.1836,
"step": 4400
},
{
"epoch": 1.4155470249520152,
"grad_norm": 196607.515625,
"learning_rate": 6.4000000000000006e-06,
"loss": 0.198,
"step": 4425
},
{
"epoch": 1.4235444657709533,
"grad_norm": 221883.765625,
"learning_rate": 6.1222222222222224e-06,
"loss": 0.1859,
"step": 4450
},
{
"epoch": 1.4315419065898913,
"grad_norm": 190660.109375,
"learning_rate": 5.844444444444445e-06,
"loss": 0.1932,
"step": 4475
},
{
"epoch": 1.4395393474088292,
"grad_norm": 174721.46875,
"learning_rate": 5.566666666666667e-06,
"loss": 0.1853,
"step": 4500
},
{
"epoch": 1.447536788227767,
"grad_norm": 162856.75,
"learning_rate": 5.288888888888889e-06,
"loss": 0.1878,
"step": 4525
},
{
"epoch": 1.455534229046705,
"grad_norm": 198921.203125,
"learning_rate": 5.011111111111112e-06,
"loss": 0.197,
"step": 4550
},
{
"epoch": 1.463531669865643,
"grad_norm": 180039.671875,
"learning_rate": 4.7333333333333335e-06,
"loss": 0.19,
"step": 4575
},
{
"epoch": 1.471529110684581,
"grad_norm": 272887.4375,
"learning_rate": 4.455555555555556e-06,
"loss": 0.2046,
"step": 4600
},
{
"epoch": 1.4795265515035187,
"grad_norm": 194019.28125,
"learning_rate": 4.177777777777778e-06,
"loss": 0.1918,
"step": 4625
},
{
"epoch": 1.4875239923224568,
"grad_norm": 183209.109375,
"learning_rate": 3.9e-06,
"loss": 0.2005,
"step": 4650
},
{
"epoch": 1.4955214331413949,
"grad_norm": 221473.609375,
"learning_rate": 3.6222222222222226e-06,
"loss": 0.1799,
"step": 4675
},
{
"epoch": 1.5035188739603327,
"grad_norm": 192981.65625,
"learning_rate": 3.3444444444444445e-06,
"loss": 0.1846,
"step": 4700
},
{
"epoch": 1.5115163147792705,
"grad_norm": 200965.0,
"learning_rate": 3.066666666666667e-06,
"loss": 0.1869,
"step": 4725
},
{
"epoch": 1.5195137555982086,
"grad_norm": 217471.046875,
"learning_rate": 2.788888888888889e-06,
"loss": 0.2022,
"step": 4750
},
{
"epoch": 1.5275111964171466,
"grad_norm": 323393.96875,
"learning_rate": 2.5111111111111114e-06,
"loss": 0.2009,
"step": 4775
},
{
"epoch": 1.5355086372360844,
"grad_norm": 238671.125,
"learning_rate": 2.2333333333333333e-06,
"loss": 0.218,
"step": 4800
},
{
"epoch": 1.5435060780550223,
"grad_norm": 199357.171875,
"learning_rate": 1.9555555555555556e-06,
"loss": 0.1797,
"step": 4825
},
{
"epoch": 1.5515035188739603,
"grad_norm": 231900.140625,
"learning_rate": 1.6777777777777779e-06,
"loss": 0.194,
"step": 4850
},
{
"epoch": 1.5595009596928984,
"grad_norm": 161423.953125,
"learning_rate": 1.4000000000000001e-06,
"loss": 0.2073,
"step": 4875
},
{
"epoch": 1.5674984005118362,
"grad_norm": 205417.140625,
"learning_rate": 1.1222222222222222e-06,
"loss": 0.2166,
"step": 4900
},
{
"epoch": 1.575495841330774,
"grad_norm": 178114.84375,
"learning_rate": 8.444444444444444e-07,
"loss": 0.2061,
"step": 4925
},
{
"epoch": 1.583493282149712,
"grad_norm": 199890.453125,
"learning_rate": 5.666666666666667e-07,
"loss": 0.1665,
"step": 4950
},
{
"epoch": 1.5914907229686501,
"grad_norm": 210452.109375,
"learning_rate": 2.888888888888889e-07,
"loss": 0.1807,
"step": 4975
},
{
"epoch": 1.599488163787588,
"grad_norm": 205998.609375,
"learning_rate": 1.1111111111111112e-08,
"loss": 0.188,
"step": 5000
},
{
"epoch": 1.599488163787588,
"eval_loss": 0.2507082223892212,
"eval_runtime": 1782.8882,
"eval_samples_per_second": 0.661,
"eval_steps_per_second": 0.083,
"eval_wer": 25.21137586471945,
"step": 5000
},
{
"epoch": 1.599488163787588,
"step": 5000,
"total_flos": 2.308452334239744e+19,
"train_loss": 0.6425882095336914,
"train_runtime": 39178.1866,
"train_samples_per_second": 2.042,
"train_steps_per_second": 0.128
}
],
"logging_steps": 25,
"max_steps": 5000,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 2.308452334239744e+19,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}

Xet Storage Details

Size:
35.7 kB
·
Xet hash:
cba8e3b538c710f5e978efe918d65a32a3b4cac51ed0030e931a2fe265a0dc1f

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.