Files
tofu_Llama-3.2-1B-Instruct_…/trainer_state.json

827 lines
20 KiB
JSON
Raw Normal View History

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.977777777777778,
"eval_steps": 500,
"global_step": 560,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.044444444444444446,
"grad_norm": 81.70519960215101,
"learning_rate": 4.4642857142857147e-07,
"loss": 2.6592,
"step": 5
},
{
"epoch": 0.08888888888888889,
"grad_norm": 65.23941572305874,
"learning_rate": 8.928571428571429e-07,
"loss": 2.4229,
"step": 10
},
{
"epoch": 0.13333333333333333,
"grad_norm": 23.030445392122235,
"learning_rate": 1.3392857142857143e-06,
"loss": 2.3129,
"step": 15
},
{
"epoch": 0.17777777777777778,
"grad_norm": 15.10886078640394,
"learning_rate": 1.7857142857142859e-06,
"loss": 2.1464,
"step": 20
},
{
"epoch": 0.2222222222222222,
"grad_norm": 15.158918433245649,
"learning_rate": 2.2321428571428573e-06,
"loss": 1.9653,
"step": 25
},
{
"epoch": 0.26666666666666666,
"grad_norm": 10.452753266560764,
"learning_rate": 2.6785714285714285e-06,
"loss": 1.9509,
"step": 30
},
{
"epoch": 0.3111111111111111,
"grad_norm": 9.403469397121878,
"learning_rate": 3.125e-06,
"loss": 1.939,
"step": 35
},
{
"epoch": 0.35555555555555557,
"grad_norm": 11.056967394101143,
"learning_rate": 3.5714285714285718e-06,
"loss": 1.8875,
"step": 40
},
{
"epoch": 0.4,
"grad_norm": 9.328316923554103,
"learning_rate": 4.017857142857143e-06,
"loss": 1.8571,
"step": 45
},
{
"epoch": 0.4444444444444444,
"grad_norm": 8.513590804020774,
"learning_rate": 4.464285714285715e-06,
"loss": 1.8091,
"step": 50
},
{
"epoch": 0.4888888888888889,
"grad_norm": 8.195744354512959,
"learning_rate": 4.910714285714286e-06,
"loss": 1.7601,
"step": 55
},
{
"epoch": 0.5333333333333333,
"grad_norm": 8.438758464497496,
"learning_rate": 5.357142857142857e-06,
"loss": 1.7015,
"step": 60
},
{
"epoch": 0.5777777777777777,
"grad_norm": 7.858160572018814,
"learning_rate": 5.8035714285714295e-06,
"loss": 1.7021,
"step": 65
},
{
"epoch": 0.6222222222222222,
"grad_norm": 8.910231103765815,
"learning_rate": 6.25e-06,
"loss": 1.6429,
"step": 70
},
{
"epoch": 0.6666666666666666,
"grad_norm": 7.384754032842493,
"learning_rate": 6.696428571428571e-06,
"loss": 1.6268,
"step": 75
},
{
"epoch": 0.7111111111111111,
"grad_norm": 7.692581371932647,
"learning_rate": 7.1428571428571436e-06,
"loss": 1.6956,
"step": 80
},
{
"epoch": 0.7555555555555555,
"grad_norm": 7.8951737417986445,
"learning_rate": 7.589285714285714e-06,
"loss": 1.6613,
"step": 85
},
{
"epoch": 0.8,
"grad_norm": 7.678660178384809,
"learning_rate": 8.035714285714286e-06,
"loss": 1.6367,
"step": 90
},
{
"epoch": 0.8444444444444444,
"grad_norm": 7.9498820978188895,
"learning_rate": 8.482142857142858e-06,
"loss": 1.6568,
"step": 95
},
{
"epoch": 0.8888888888888888,
"grad_norm": 7.48336511323642,
"learning_rate": 8.92857142857143e-06,
"loss": 1.6309,
"step": 100
},
{
"epoch": 0.9333333333333333,
"grad_norm": 7.481491041447778,
"learning_rate": 9.375000000000001e-06,
"loss": 1.6108,
"step": 105
},
{
"epoch": 0.9777777777777777,
"grad_norm": 7.720239899968915,
"learning_rate": 9.821428571428573e-06,
"loss": 1.5646,
"step": 110
},
{
"epoch": 1.0222222222222221,
"grad_norm": 7.0981539773430615,
"learning_rate": 9.933035714285715e-06,
"loss": 1.4956,
"step": 115
},
{
"epoch": 1.0666666666666667,
"grad_norm": 7.510833590279441,
"learning_rate": 9.821428571428573e-06,
"loss": 1.3929,
"step": 120
},
{
"epoch": 1.1111111111111112,
"grad_norm": 7.248134123936718,
"learning_rate": 9.70982142857143e-06,
"loss": 1.4206,
"step": 125
},
{
"epoch": 1.1555555555555554,
"grad_norm": 8.174657015727915,
"learning_rate": 9.598214285714287e-06,
"loss": 1.3426,
"step": 130
},
{
"epoch": 1.2,
"grad_norm": 7.352102037038091,
"learning_rate": 9.486607142857144e-06,
"loss": 1.3171,
"step": 135
},
{
"epoch": 1.2444444444444445,
"grad_norm": 7.5635904551535225,
"learning_rate": 9.375000000000001e-06,
"loss": 1.3779,
"step": 140
},
{
"epoch": 1.2888888888888888,
"grad_norm": 9.033956503305323,
"learning_rate": 9.263392857142858e-06,
"loss": 1.345,
"step": 145
},
{
"epoch": 1.3333333333333333,
"grad_norm": 8.224641097275477,
"learning_rate": 9.151785714285715e-06,
"loss": 1.4234,
"step": 150
},
{
"epoch": 1.3777777777777778,
"grad_norm": 8.830468339631146,
"learning_rate": 9.040178571428572e-06,
"loss": 1.2946,
"step": 155
},
{
"epoch": 1.4222222222222223,
"grad_norm": 8.443856362876113,
"learning_rate": 8.92857142857143e-06,
"loss": 1.3646,
"step": 160
},
{
"epoch": 1.4666666666666668,
"grad_norm": 8.043363545508665,
"learning_rate": 8.816964285714286e-06,
"loss": 1.3936,
"step": 165
},
{
"epoch": 1.511111111111111,
"grad_norm": 8.616005497888816,
"learning_rate": 8.705357142857143e-06,
"loss": 1.3557,
"step": 170
},
{
"epoch": 1.5555555555555556,
"grad_norm": 8.255699646794918,
"learning_rate": 8.59375e-06,
"loss": 1.3142,
"step": 175
},
{
"epoch": 1.6,
"grad_norm": 8.215553491915397,
"learning_rate": 8.482142857142858e-06,
"loss": 1.3464,
"step": 180
},
{
"epoch": 1.6444444444444444,
"grad_norm": 7.885356619483398,
"learning_rate": 8.370535714285715e-06,
"loss": 1.3553,
"step": 185
},
{
"epoch": 1.6888888888888889,
"grad_norm": 8.12112906938901,
"learning_rate": 8.258928571428572e-06,
"loss": 1.3527,
"step": 190
},
{
"epoch": 1.7333333333333334,
"grad_norm": 8.350294666975472,
"learning_rate": 8.147321428571429e-06,
"loss": 1.2518,
"step": 195
},
{
"epoch": 1.7777777777777777,
"grad_norm": 7.905524590675295,
"learning_rate": 8.035714285714286e-06,
"loss": 1.3491,
"step": 200
},
{
"epoch": 1.8222222222222222,
"grad_norm": 7.871873965564558,
"learning_rate": 7.924107142857143e-06,
"loss": 1.382,
"step": 205
},
{
"epoch": 1.8666666666666667,
"grad_norm": 8.12507505022624,
"learning_rate": 7.8125e-06,
"loss": 1.3336,
"step": 210
},
{
"epoch": 1.911111111111111,
"grad_norm": 8.058616824355166,
"learning_rate": 7.700892857142857e-06,
"loss": 1.3512,
"step": 215
},
{
"epoch": 1.9555555555555557,
"grad_norm": 7.59358946949836,
"learning_rate": 7.589285714285714e-06,
"loss": 1.3093,
"step": 220
},
{
"epoch": 2.0,
"grad_norm": 8.88478802727675,
"learning_rate": 7.4776785714285714e-06,
"loss": 1.3261,
"step": 225
},
{
"epoch": 2.0444444444444443,
"grad_norm": 7.6072417424325,
"learning_rate": 7.366071428571429e-06,
"loss": 0.9253,
"step": 230
},
{
"epoch": 2.088888888888889,
"grad_norm": 8.772960686537047,
"learning_rate": 7.2544642857142865e-06,
"loss": 0.8591,
"step": 235
},
{
"epoch": 2.1333333333333333,
"grad_norm": 9.452930658317475,
"learning_rate": 7.1428571428571436e-06,
"loss": 0.8771,
"step": 240
},
{
"epoch": 2.1777777777777776,
"grad_norm": 9.3220762052475,
"learning_rate": 7.031250000000001e-06,
"loss": 0.8593,
"step": 245
},
{
"epoch": 2.2222222222222223,
"grad_norm": 8.32864095740941,
"learning_rate": 6.919642857142858e-06,
"loss": 0.9117,
"step": 250
},
{
"epoch": 2.2666666666666666,
"grad_norm": 8.792144175107566,
"learning_rate": 6.808035714285715e-06,
"loss": 0.8772,
"step": 255
},
{
"epoch": 2.311111111111111,
"grad_norm": 9.98800755135626,
"learning_rate": 6.696428571428571e-06,
"loss": 0.858,
"step": 260
},
{
"epoch": 2.3555555555555556,
"grad_norm": 9.018987249678533,
"learning_rate": 6.58482142857143e-06,
"loss": 0.8708,
"step": 265
},
{
"epoch": 2.4,
"grad_norm": 8.547777859748455,
"learning_rate": 6.473214285714287e-06,
"loss": 0.872,
"step": 270
},
{
"epoch": 2.4444444444444446,
"grad_norm": 8.771861798383114,
"learning_rate": 6.361607142857143e-06,
"loss": 0.8603,
"step": 275
},
{
"epoch": 2.488888888888889,
"grad_norm": 10.206792686412927,
"learning_rate": 6.25e-06,
"loss": 0.9321,
"step": 280
},
{
"epoch": 2.533333333333333,
"grad_norm": 8.15300521664367,
"learning_rate": 6.138392857142857e-06,
"loss": 0.7993,
"step": 285
},
{
"epoch": 2.5777777777777775,
"grad_norm": 8.916029898752333,
"learning_rate": 6.0267857142857145e-06,
"loss": 0.9229,
"step": 290
},
{
"epoch": 2.6222222222222222,
"grad_norm": 9.688980370039602,
"learning_rate": 5.9151785714285716e-06,
"loss": 0.8954,
"step": 295
},
{
"epoch": 2.6666666666666665,
"grad_norm": 8.749780229882957,
"learning_rate": 5.8035714285714295e-06,
"loss": 0.8473,
"step": 300
},
{
"epoch": 2.7111111111111112,
"grad_norm": 9.261115491971523,
"learning_rate": 5.691964285714287e-06,
"loss": 0.8772,
"step": 305
},
{
"epoch": 2.7555555555555555,
"grad_norm": 9.010412339583716,
"learning_rate": 5.580357142857144e-06,
"loss": 0.8786,
"step": 310
},
{
"epoch": 2.8,
"grad_norm": 9.17920752101898,
"learning_rate": 5.468750000000001e-06,
"loss": 0.851,
"step": 315
},
{
"epoch": 2.8444444444444446,
"grad_norm": 8.254891628439802,
"learning_rate": 5.357142857142857e-06,
"loss": 0.8981,
"step": 320
},
{
"epoch": 2.888888888888889,
"grad_norm": 8.642637449945827,
"learning_rate": 5.245535714285714e-06,
"loss": 0.8717,
"step": 325
},
{
"epoch": 2.9333333333333336,
"grad_norm": 8.815270770583437,
"learning_rate": 5.133928571428571e-06,
"loss": 0.8865,
"step": 330
},
{
"epoch": 2.977777777777778,
"grad_norm": 9.062651499166742,
"learning_rate": 5.022321428571429e-06,
"loss": 0.8776,
"step": 335
},
{
"epoch": 3.022222222222222,
"grad_norm": 7.480764915704419,
"learning_rate": 4.910714285714286e-06,
"loss": 0.7016,
"step": 340
},
{
"epoch": 3.066666666666667,
"grad_norm": 8.739944237211384,
"learning_rate": 4.799107142857143e-06,
"loss": 0.5105,
"step": 345
},
{
"epoch": 3.111111111111111,
"grad_norm": 11.051330955646318,
"learning_rate": 4.6875000000000004e-06,
"loss": 0.5077,
"step": 350
},
{
"epoch": 3.1555555555555554,
"grad_norm": 8.402494171987351,
"learning_rate": 4.5758928571428575e-06,
"loss": 0.4504,
"step": 355
},
{
"epoch": 3.2,
"grad_norm": 9.554895055203646,
"learning_rate": 4.464285714285715e-06,
"loss": 0.4525,
"step": 360
},
{
"epoch": 3.2444444444444445,
"grad_norm": 9.849412176242506,
"learning_rate": 4.352678571428572e-06,
"loss": 0.438,
"step": 365
},
{
"epoch": 3.2888888888888888,
"grad_norm": 9.879820678510647,
"learning_rate": 4.241071428571429e-06,
"loss": 0.4634,
"step": 370
},
{
"epoch": 3.3333333333333335,
"grad_norm": 9.560743040510404,
"learning_rate": 4.129464285714286e-06,
"loss": 0.4804,
"step": 375
},
{
"epoch": 3.3777777777777778,
"grad_norm": 9.122163123715797,
"learning_rate": 4.017857142857143e-06,
"loss": 0.4346,
"step": 380
},
{
"epoch": 3.422222222222222,
"grad_norm": 9.57226826749949,
"learning_rate": 3.90625e-06,
"loss": 0.4483,
"step": 385
},
{
"epoch": 3.466666666666667,
"grad_norm": 10.306235127578079,
"learning_rate": 3.794642857142857e-06,
"loss": 0.4713,
"step": 390
},
{
"epoch": 3.511111111111111,
"grad_norm": 9.64313411436284,
"learning_rate": 3.6830357142857147e-06,
"loss": 0.4359,
"step": 395
},
{
"epoch": 3.5555555555555554,
"grad_norm": 10.420285208662374,
"learning_rate": 3.5714285714285718e-06,
"loss": 0.451,
"step": 400
},
{
"epoch": 3.6,
"grad_norm": 9.87846382809359,
"learning_rate": 3.459821428571429e-06,
"loss": 0.4763,
"step": 405
},
{
"epoch": 3.6444444444444444,
"grad_norm": 9.876467528663943,
"learning_rate": 3.3482142857142855e-06,
"loss": 0.4537,
"step": 410
},
{
"epoch": 3.688888888888889,
"grad_norm": 8.675841435317817,
"learning_rate": 3.2366071428571435e-06,
"loss": 0.437,
"step": 415
},
{
"epoch": 3.7333333333333334,
"grad_norm": 9.325507180126642,
"learning_rate": 3.125e-06,
"loss": 0.4497,
"step": 420
},
{
"epoch": 3.7777777777777777,
"grad_norm": 10.153198500901361,
"learning_rate": 3.0133928571428572e-06,
"loss": 0.4441,
"step": 425
},
{
"epoch": 3.822222222222222,
"grad_norm": 10.645689697884485,
"learning_rate": 2.9017857142857148e-06,
"loss": 0.4706,
"step": 430
},
{
"epoch": 3.8666666666666667,
"grad_norm": 9.668794366955733,
"learning_rate": 2.790178571428572e-06,
"loss": 0.4576,
"step": 435
},
{
"epoch": 3.911111111111111,
"grad_norm": 8.850840067616335,
"learning_rate": 2.6785714285714285e-06,
"loss": 0.4327,
"step": 440
},
{
"epoch": 3.9555555555555557,
"grad_norm": 9.06809450411661,
"learning_rate": 2.5669642857142856e-06,
"loss": 0.489,
"step": 445
},
{
"epoch": 4.0,
"grad_norm": 9.875435470206298,
"learning_rate": 2.455357142857143e-06,
"loss": 0.4454,
"step": 450
},
{
"epoch": 4.044444444444444,
"grad_norm": 14.31106135217529,
"learning_rate": 2.3437500000000002e-06,
"loss": 0.2196,
"step": 455
},
{
"epoch": 4.088888888888889,
"grad_norm": 8.609859535415907,
"learning_rate": 2.2321428571428573e-06,
"loss": 0.2121,
"step": 460
},
{
"epoch": 4.133333333333334,
"grad_norm": 13.280395274888972,
"learning_rate": 2.1205357142857144e-06,
"loss": 0.2035,
"step": 465
},
{
"epoch": 4.177777777777778,
"grad_norm": 8.75211366566662,
"learning_rate": 2.0089285714285715e-06,
"loss": 0.1911,
"step": 470
},
{
"epoch": 4.222222222222222,
"grad_norm": 7.090600469273079,
"learning_rate": 1.8973214285714286e-06,
"loss": 0.1923,
"step": 475
},
{
"epoch": 4.266666666666667,
"grad_norm": 6.939677578044171,
"learning_rate": 1.7857142857142859e-06,
"loss": 0.1895,
"step": 480
},
{
"epoch": 4.311111111111111,
"grad_norm": 7.862142692655875,
"learning_rate": 1.6741071428571428e-06,
"loss": 0.1977,
"step": 485
},
{
"epoch": 4.355555555555555,
"grad_norm": 7.583820440823111,
"learning_rate": 1.5625e-06,
"loss": 0.1751,
"step": 490
},
{
"epoch": 4.4,
"grad_norm": 8.947913314563216,
"learning_rate": 1.4508928571428574e-06,
"loss": 0.1838,
"step": 495
},
{
"epoch": 4.444444444444445,
"grad_norm": 7.658722193186348,
"learning_rate": 1.3392857142857143e-06,
"loss": 0.2006,
"step": 500
},
{
"epoch": 4.488888888888889,
"grad_norm": 8.207622619508404,
"learning_rate": 1.2276785714285716e-06,
"loss": 0.186,
"step": 505
},
{
"epoch": 4.533333333333333,
"grad_norm": 8.120063690560118,
"learning_rate": 1.1160714285714287e-06,
"loss": 0.2101,
"step": 510
},
{
"epoch": 4.5777777777777775,
"grad_norm": 7.850134563130463,
"learning_rate": 1.0044642857142857e-06,
"loss": 0.1954,
"step": 515
},
{
"epoch": 4.622222222222222,
"grad_norm": 10.423756147314887,
"learning_rate": 8.928571428571429e-07,
"loss": 0.1877,
"step": 520
},
{
"epoch": 4.666666666666667,
"grad_norm": 7.945666215863678,
"learning_rate": 7.8125e-07,
"loss": 0.1988,
"step": 525
},
{
"epoch": 4.711111111111111,
"grad_norm": 8.106328872459784,
"learning_rate": 6.696428571428571e-07,
"loss": 0.2015,
"step": 530
},
{
"epoch": 4.7555555555555555,
"grad_norm": 8.767972813276023,
"learning_rate": 5.580357142857143e-07,
"loss": 0.188,
"step": 535
},
{
"epoch": 4.8,
"grad_norm": 9.369887507108524,
"learning_rate": 4.4642857142857147e-07,
"loss": 0.1859,
"step": 540
},
{
"epoch": 4.844444444444444,
"grad_norm": 8.892651474952375,
"learning_rate": 3.3482142857142856e-07,
"loss": 0.1826,
"step": 545
},
{
"epoch": 4.888888888888889,
"grad_norm": 7.261334614879305,
"learning_rate": 2.2321428571428574e-07,
"loss": 0.1852,
"step": 550
},
{
"epoch": 4.933333333333334,
"grad_norm": 9.36687860609368,
"learning_rate": 1.1160714285714287e-07,
"loss": 0.1889,
"step": 555
},
{
"epoch": 4.977777777777778,
"grad_norm": 10.485619564355174,
"learning_rate": 0.0,
"loss": 0.1811,
"step": 560
},
{
"epoch": 4.977777777777778,
"step": 560,
"total_flos": 770201321472.0,
"train_loss": 0.950020797018494,
"train_runtime": 694.3412,
"train_samples_per_second": 25.924,
"train_steps_per_second": 0.807
}
],
"logging_steps": 5,
"max_steps": 560,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 770201321472.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}