wav2vec2-bert-CV16-en-libri / trainer_state.json
ylacombe's picture
ylacombe HF staff
End of training
d58c6c2 verified
raw
history blame
46.8 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.996216897856242,
"eval_steps": 250,
"global_step": 1980,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.5000000000000002e-08,
"loss": 4.6482,
"step": 5
},
{
"epoch": 0.03,
"learning_rate": 3.0000000000000004e-08,
"loss": 4.7026,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 4.5e-08,
"loss": 4.6854,
"step": 15
},
{
"epoch": 0.05,
"learning_rate": 6.000000000000001e-08,
"loss": 4.7824,
"step": 20
},
{
"epoch": 0.06,
"learning_rate": 7.500000000000001e-08,
"loss": 4.6708,
"step": 25
},
{
"epoch": 0.08,
"learning_rate": 9e-08,
"loss": 4.6248,
"step": 30
},
{
"epoch": 0.09,
"learning_rate": 1.05e-07,
"loss": 4.7019,
"step": 35
},
{
"epoch": 0.1,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.6747,
"step": 40
},
{
"epoch": 0.11,
"learning_rate": 1.35e-07,
"loss": 4.6209,
"step": 45
},
{
"epoch": 0.13,
"learning_rate": 1.5000000000000002e-07,
"loss": 4.6223,
"step": 50
},
{
"epoch": 0.14,
"learning_rate": 1.6499999999999998e-07,
"loss": 4.5145,
"step": 55
},
{
"epoch": 0.15,
"learning_rate": 1.8e-07,
"loss": 4.5692,
"step": 60
},
{
"epoch": 0.16,
"learning_rate": 1.9499999999999999e-07,
"loss": 4.4948,
"step": 65
},
{
"epoch": 0.18,
"learning_rate": 2.1e-07,
"loss": 4.5019,
"step": 70
},
{
"epoch": 0.19,
"learning_rate": 2.25e-07,
"loss": 4.4425,
"step": 75
},
{
"epoch": 0.2,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.4318,
"step": 80
},
{
"epoch": 0.21,
"learning_rate": 2.5500000000000005e-07,
"loss": 4.428,
"step": 85
},
{
"epoch": 0.23,
"learning_rate": 2.7e-07,
"loss": 4.2983,
"step": 90
},
{
"epoch": 0.24,
"learning_rate": 2.85e-07,
"loss": 4.228,
"step": 95
},
{
"epoch": 0.25,
"learning_rate": 2.97e-07,
"loss": 4.2784,
"step": 100
},
{
"epoch": 0.26,
"learning_rate": 3.12e-07,
"loss": 4.1254,
"step": 105
},
{
"epoch": 0.28,
"learning_rate": 3.27e-07,
"loss": 4.0566,
"step": 110
},
{
"epoch": 0.29,
"learning_rate": 3.42e-07,
"loss": 4.0241,
"step": 115
},
{
"epoch": 0.3,
"learning_rate": 3.5700000000000003e-07,
"loss": 4.0261,
"step": 120
},
{
"epoch": 0.32,
"learning_rate": 3.72e-07,
"loss": 3.864,
"step": 125
},
{
"epoch": 0.33,
"learning_rate": 3.87e-07,
"loss": 3.8068,
"step": 130
},
{
"epoch": 0.34,
"learning_rate": 4.02e-07,
"loss": 3.7553,
"step": 135
},
{
"epoch": 0.35,
"learning_rate": 4.17e-07,
"loss": 3.6628,
"step": 140
},
{
"epoch": 0.37,
"learning_rate": 4.32e-07,
"loss": 3.5843,
"step": 145
},
{
"epoch": 0.38,
"learning_rate": 4.47e-07,
"loss": 3.5552,
"step": 150
},
{
"epoch": 0.39,
"learning_rate": 4.6200000000000003e-07,
"loss": 3.4307,
"step": 155
},
{
"epoch": 0.4,
"learning_rate": 4.77e-07,
"loss": 3.3634,
"step": 160
},
{
"epoch": 0.42,
"learning_rate": 4.92e-07,
"loss": 3.3041,
"step": 165
},
{
"epoch": 0.43,
"learning_rate": 5.07e-07,
"loss": 3.2511,
"step": 170
},
{
"epoch": 0.44,
"learning_rate": 5.219999999999999e-07,
"loss": 3.1583,
"step": 175
},
{
"epoch": 0.45,
"learning_rate": 5.37e-07,
"loss": 3.1013,
"step": 180
},
{
"epoch": 0.47,
"learning_rate": 5.52e-07,
"loss": 3.0769,
"step": 185
},
{
"epoch": 0.48,
"learning_rate": 5.67e-07,
"loss": 3.0122,
"step": 190
},
{
"epoch": 0.49,
"learning_rate": 5.82e-07,
"loss": 2.9856,
"step": 195
},
{
"epoch": 0.5,
"learning_rate": 5.970000000000001e-07,
"loss": 2.9903,
"step": 200
},
{
"epoch": 0.52,
"learning_rate": 6.12e-07,
"loss": 2.9308,
"step": 205
},
{
"epoch": 0.53,
"learning_rate": 6.27e-07,
"loss": 2.9159,
"step": 210
},
{
"epoch": 0.54,
"learning_rate": 6.42e-07,
"loss": 2.9075,
"step": 215
},
{
"epoch": 0.55,
"learning_rate": 6.57e-07,
"loss": 2.8958,
"step": 220
},
{
"epoch": 0.57,
"learning_rate": 6.72e-07,
"loss": 2.8698,
"step": 225
},
{
"epoch": 0.58,
"learning_rate": 6.87e-07,
"loss": 2.8697,
"step": 230
},
{
"epoch": 0.59,
"learning_rate": 7.02e-07,
"loss": 2.8911,
"step": 235
},
{
"epoch": 0.61,
"learning_rate": 7.170000000000001e-07,
"loss": 2.8506,
"step": 240
},
{
"epoch": 0.62,
"learning_rate": 7.32e-07,
"loss": 2.8522,
"step": 245
},
{
"epoch": 0.63,
"learning_rate": 7.47e-07,
"loss": 2.8812,
"step": 250
},
{
"epoch": 0.63,
"eval_cer": 0.9999960063100302,
"eval_loss": 2.892289161682129,
"eval_runtime": 59.7132,
"eval_samples_per_second": 42.336,
"eval_steps_per_second": 1.189,
"eval_wer": 1.0,
"step": 250
},
{
"epoch": 0.64,
"learning_rate": 7.62e-07,
"loss": 2.8429,
"step": 255
},
{
"epoch": 0.66,
"learning_rate": 7.77e-07,
"loss": 2.8424,
"step": 260
},
{
"epoch": 0.67,
"learning_rate": 7.92e-07,
"loss": 2.8437,
"step": 265
},
{
"epoch": 0.68,
"learning_rate": 8.070000000000001e-07,
"loss": 2.8541,
"step": 270
},
{
"epoch": 0.69,
"learning_rate": 8.22e-07,
"loss": 2.8411,
"step": 275
},
{
"epoch": 0.71,
"learning_rate": 8.370000000000001e-07,
"loss": 2.8387,
"step": 280
},
{
"epoch": 0.72,
"learning_rate": 8.520000000000001e-07,
"loss": 2.8561,
"step": 285
},
{
"epoch": 0.73,
"learning_rate": 8.67e-07,
"loss": 2.8254,
"step": 290
},
{
"epoch": 0.74,
"learning_rate": 8.82e-07,
"loss": 2.8244,
"step": 295
},
{
"epoch": 0.76,
"learning_rate": 8.97e-07,
"loss": 2.8409,
"step": 300
},
{
"epoch": 0.77,
"learning_rate": 9.12e-07,
"loss": 2.816,
"step": 305
},
{
"epoch": 0.78,
"learning_rate": 9.270000000000001e-07,
"loss": 2.8173,
"step": 310
},
{
"epoch": 0.79,
"learning_rate": 9.419999999999999e-07,
"loss": 2.8253,
"step": 315
},
{
"epoch": 0.81,
"learning_rate": 9.57e-07,
"loss": 2.8185,
"step": 320
},
{
"epoch": 0.82,
"learning_rate": 9.72e-07,
"loss": 2.8046,
"step": 325
},
{
"epoch": 0.83,
"learning_rate": 9.87e-07,
"loss": 2.8048,
"step": 330
},
{
"epoch": 0.84,
"learning_rate": 1.002e-06,
"loss": 2.8177,
"step": 335
},
{
"epoch": 0.86,
"learning_rate": 1.017e-06,
"loss": 2.775,
"step": 340
},
{
"epoch": 0.87,
"learning_rate": 1.032e-06,
"loss": 2.7762,
"step": 345
},
{
"epoch": 0.88,
"learning_rate": 1.047e-06,
"loss": 2.7735,
"step": 350
},
{
"epoch": 0.9,
"learning_rate": 1.062e-06,
"loss": 2.7319,
"step": 355
},
{
"epoch": 0.91,
"learning_rate": 1.077e-06,
"loss": 2.71,
"step": 360
},
{
"epoch": 0.92,
"learning_rate": 1.092e-06,
"loss": 2.6868,
"step": 365
},
{
"epoch": 0.93,
"learning_rate": 1.1070000000000002e-06,
"loss": 2.6709,
"step": 370
},
{
"epoch": 0.95,
"learning_rate": 1.122e-06,
"loss": 2.6141,
"step": 375
},
{
"epoch": 0.96,
"learning_rate": 1.137e-06,
"loss": 2.5824,
"step": 380
},
{
"epoch": 0.97,
"learning_rate": 1.152e-06,
"loss": 2.5791,
"step": 385
},
{
"epoch": 0.98,
"learning_rate": 1.167e-06,
"loss": 2.5124,
"step": 390
},
{
"epoch": 1.0,
"learning_rate": 1.179e-06,
"loss": 2.4623,
"step": 395
},
{
"epoch": 1.01,
"learning_rate": 1.1940000000000001e-06,
"loss": 2.4341,
"step": 400
},
{
"epoch": 1.02,
"learning_rate": 1.2090000000000002e-06,
"loss": 2.3441,
"step": 405
},
{
"epoch": 1.03,
"learning_rate": 1.224e-06,
"loss": 2.2861,
"step": 410
},
{
"epoch": 1.05,
"learning_rate": 1.2390000000000001e-06,
"loss": 2.2725,
"step": 415
},
{
"epoch": 1.06,
"learning_rate": 1.254e-06,
"loss": 2.1615,
"step": 420
},
{
"epoch": 1.07,
"learning_rate": 1.269e-06,
"loss": 2.0869,
"step": 425
},
{
"epoch": 1.08,
"learning_rate": 1.284e-06,
"loss": 2.0713,
"step": 430
},
{
"epoch": 1.1,
"learning_rate": 1.299e-06,
"loss": 1.9363,
"step": 435
},
{
"epoch": 1.11,
"learning_rate": 1.314e-06,
"loss": 1.8633,
"step": 440
},
{
"epoch": 1.12,
"learning_rate": 1.3290000000000001e-06,
"loss": 1.8307,
"step": 445
},
{
"epoch": 1.13,
"learning_rate": 1.344e-06,
"loss": 1.7683,
"step": 450
},
{
"epoch": 1.15,
"learning_rate": 1.359e-06,
"loss": 1.6742,
"step": 455
},
{
"epoch": 1.16,
"learning_rate": 1.374e-06,
"loss": 1.6215,
"step": 460
},
{
"epoch": 1.17,
"learning_rate": 1.389e-06,
"loss": 1.6537,
"step": 465
},
{
"epoch": 1.19,
"learning_rate": 1.404e-06,
"loss": 1.5039,
"step": 470
},
{
"epoch": 1.2,
"learning_rate": 1.419e-06,
"loss": 1.4679,
"step": 475
},
{
"epoch": 1.21,
"learning_rate": 1.4340000000000002e-06,
"loss": 1.4565,
"step": 480
},
{
"epoch": 1.22,
"learning_rate": 1.449e-06,
"loss": 1.3904,
"step": 485
},
{
"epoch": 1.24,
"learning_rate": 1.464e-06,
"loss": 1.3057,
"step": 490
},
{
"epoch": 1.25,
"learning_rate": 1.479e-06,
"loss": 1.2998,
"step": 495
},
{
"epoch": 1.26,
"learning_rate": 1.494e-06,
"loss": 1.2899,
"step": 500
},
{
"epoch": 1.26,
"eval_cer": 0.2562710916751533,
"eval_loss": 1.147094488143921,
"eval_runtime": 60.746,
"eval_samples_per_second": 41.616,
"eval_steps_per_second": 1.169,
"eval_wer": 0.7029631516386483,
"step": 500
},
{
"epoch": 1.27,
"learning_rate": 1.5089999999999999e-06,
"loss": 1.2025,
"step": 505
},
{
"epoch": 1.29,
"learning_rate": 1.524e-06,
"loss": 1.1523,
"step": 510
},
{
"epoch": 1.3,
"learning_rate": 1.539e-06,
"loss": 1.1846,
"step": 515
},
{
"epoch": 1.31,
"learning_rate": 1.554e-06,
"loss": 1.0974,
"step": 520
},
{
"epoch": 1.32,
"learning_rate": 1.569e-06,
"loss": 1.0635,
"step": 525
},
{
"epoch": 1.34,
"learning_rate": 1.584e-06,
"loss": 1.0922,
"step": 530
},
{
"epoch": 1.35,
"learning_rate": 1.599e-06,
"loss": 1.0167,
"step": 535
},
{
"epoch": 1.36,
"learning_rate": 1.6140000000000001e-06,
"loss": 0.9946,
"step": 540
},
{
"epoch": 1.37,
"learning_rate": 1.629e-06,
"loss": 1.0021,
"step": 545
},
{
"epoch": 1.39,
"learning_rate": 1.644e-06,
"loss": 1.006,
"step": 550
},
{
"epoch": 1.4,
"learning_rate": 1.6590000000000001e-06,
"loss": 0.9312,
"step": 555
},
{
"epoch": 1.41,
"learning_rate": 1.6740000000000002e-06,
"loss": 0.9275,
"step": 560
},
{
"epoch": 1.42,
"learning_rate": 1.689e-06,
"loss": 0.9743,
"step": 565
},
{
"epoch": 1.44,
"learning_rate": 1.7040000000000001e-06,
"loss": 0.8707,
"step": 570
},
{
"epoch": 1.45,
"learning_rate": 1.719e-06,
"loss": 0.8765,
"step": 575
},
{
"epoch": 1.46,
"learning_rate": 1.734e-06,
"loss": 0.9403,
"step": 580
},
{
"epoch": 1.48,
"learning_rate": 1.749e-06,
"loss": 0.8242,
"step": 585
},
{
"epoch": 1.49,
"learning_rate": 1.764e-06,
"loss": 0.7783,
"step": 590
},
{
"epoch": 1.5,
"learning_rate": 1.779e-06,
"loss": 0.8126,
"step": 595
},
{
"epoch": 1.51,
"learning_rate": 1.794e-06,
"loss": 0.8544,
"step": 600
},
{
"epoch": 1.53,
"learning_rate": 1.809e-06,
"loss": 0.7637,
"step": 605
},
{
"epoch": 1.54,
"learning_rate": 1.824e-06,
"loss": 0.7481,
"step": 610
},
{
"epoch": 1.55,
"learning_rate": 1.839e-06,
"loss": 0.8198,
"step": 615
},
{
"epoch": 1.56,
"learning_rate": 1.8540000000000002e-06,
"loss": 0.7086,
"step": 620
},
{
"epoch": 1.58,
"learning_rate": 1.869e-06,
"loss": 0.7167,
"step": 625
},
{
"epoch": 1.59,
"learning_rate": 1.8810000000000003e-06,
"loss": 0.7675,
"step": 630
},
{
"epoch": 1.6,
"learning_rate": 1.8960000000000001e-06,
"loss": 0.6906,
"step": 635
},
{
"epoch": 1.61,
"learning_rate": 1.9110000000000004e-06,
"loss": 0.6739,
"step": 640
},
{
"epoch": 1.63,
"learning_rate": 1.926e-06,
"loss": 0.7041,
"step": 645
},
{
"epoch": 1.64,
"learning_rate": 1.9409999999999997e-06,
"loss": 0.6874,
"step": 650
},
{
"epoch": 1.65,
"learning_rate": 1.956e-06,
"loss": 0.6738,
"step": 655
},
{
"epoch": 1.66,
"learning_rate": 1.971e-06,
"loss": 0.6507,
"step": 660
},
{
"epoch": 1.68,
"learning_rate": 1.9859999999999997e-06,
"loss": 0.6857,
"step": 665
},
{
"epoch": 1.69,
"learning_rate": 2.001e-06,
"loss": 0.6291,
"step": 670
},
{
"epoch": 1.7,
"learning_rate": 2.016e-06,
"loss": 0.5821,
"step": 675
},
{
"epoch": 1.72,
"learning_rate": 2.031e-06,
"loss": 0.6688,
"step": 680
},
{
"epoch": 1.73,
"learning_rate": 2.046e-06,
"loss": 0.5958,
"step": 685
},
{
"epoch": 1.74,
"learning_rate": 2.061e-06,
"loss": 0.5807,
"step": 690
},
{
"epoch": 1.75,
"learning_rate": 2.076e-06,
"loss": 0.5991,
"step": 695
},
{
"epoch": 1.77,
"learning_rate": 2.091e-06,
"loss": 0.5904,
"step": 700
},
{
"epoch": 1.78,
"learning_rate": 2.106e-06,
"loss": 0.5715,
"step": 705
},
{
"epoch": 1.79,
"learning_rate": 2.121e-06,
"loss": 0.563,
"step": 710
},
{
"epoch": 1.8,
"learning_rate": 2.136e-06,
"loss": 0.6291,
"step": 715
},
{
"epoch": 1.82,
"learning_rate": 2.1510000000000002e-06,
"loss": 0.5513,
"step": 720
},
{
"epoch": 1.83,
"learning_rate": 2.166e-06,
"loss": 0.539,
"step": 725
},
{
"epoch": 1.84,
"learning_rate": 2.181e-06,
"loss": 0.6234,
"step": 730
},
{
"epoch": 1.85,
"learning_rate": 2.1960000000000002e-06,
"loss": 0.5196,
"step": 735
},
{
"epoch": 1.87,
"learning_rate": 2.211e-06,
"loss": 0.5122,
"step": 740
},
{
"epoch": 1.88,
"learning_rate": 2.226e-06,
"loss": 0.5331,
"step": 745
},
{
"epoch": 1.89,
"learning_rate": 2.2410000000000002e-06,
"loss": 0.5276,
"step": 750
},
{
"epoch": 1.89,
"eval_cer": 0.11271790570897981,
"eval_loss": 0.46866554021835327,
"eval_runtime": 59.8626,
"eval_samples_per_second": 42.23,
"eval_steps_per_second": 1.186,
"eval_wer": 0.41144506707420614,
"step": 750
},
{
"epoch": 1.9,
"learning_rate": 2.256e-06,
"loss": 0.49,
"step": 755
},
{
"epoch": 1.92,
"learning_rate": 2.2710000000000004e-06,
"loss": 0.4961,
"step": 760
},
{
"epoch": 1.93,
"learning_rate": 2.2860000000000002e-06,
"loss": 0.5207,
"step": 765
},
{
"epoch": 1.94,
"learning_rate": 2.301e-06,
"loss": 0.4767,
"step": 770
},
{
"epoch": 1.95,
"learning_rate": 2.3160000000000004e-06,
"loss": 0.4761,
"step": 775
},
{
"epoch": 1.97,
"learning_rate": 2.3310000000000002e-06,
"loss": 0.5454,
"step": 780
},
{
"epoch": 1.98,
"learning_rate": 2.346e-06,
"loss": 0.4688,
"step": 785
},
{
"epoch": 1.99,
"learning_rate": 2.3610000000000003e-06,
"loss": 0.4609,
"step": 790
},
{
"epoch": 2.01,
"learning_rate": 2.376e-06,
"loss": 0.4967,
"step": 795
},
{
"epoch": 2.02,
"learning_rate": 2.391e-06,
"loss": 0.4613,
"step": 800
},
{
"epoch": 2.03,
"learning_rate": 2.406e-06,
"loss": 0.448,
"step": 805
},
{
"epoch": 2.04,
"learning_rate": 2.4209999999999998e-06,
"loss": 0.5151,
"step": 810
},
{
"epoch": 2.06,
"learning_rate": 2.436e-06,
"loss": 0.4411,
"step": 815
},
{
"epoch": 2.07,
"learning_rate": 2.451e-06,
"loss": 0.43,
"step": 820
},
{
"epoch": 2.08,
"learning_rate": 2.4659999999999998e-06,
"loss": 0.4318,
"step": 825
},
{
"epoch": 2.09,
"learning_rate": 2.481e-06,
"loss": 0.4442,
"step": 830
},
{
"epoch": 2.11,
"learning_rate": 2.496e-06,
"loss": 0.4014,
"step": 835
},
{
"epoch": 2.12,
"learning_rate": 2.5109999999999998e-06,
"loss": 0.4062,
"step": 840
},
{
"epoch": 2.13,
"learning_rate": 2.526e-06,
"loss": 0.4431,
"step": 845
},
{
"epoch": 2.14,
"learning_rate": 2.541e-06,
"loss": 0.4057,
"step": 850
},
{
"epoch": 2.16,
"learning_rate": 2.556e-06,
"loss": 0.3873,
"step": 855
},
{
"epoch": 2.17,
"learning_rate": 2.571e-06,
"loss": 0.4497,
"step": 860
},
{
"epoch": 2.18,
"learning_rate": 2.586e-06,
"loss": 0.3917,
"step": 865
},
{
"epoch": 2.19,
"learning_rate": 2.601e-06,
"loss": 0.363,
"step": 870
},
{
"epoch": 2.21,
"learning_rate": 2.616e-06,
"loss": 0.4173,
"step": 875
},
{
"epoch": 2.22,
"learning_rate": 2.631e-06,
"loss": 0.4128,
"step": 880
},
{
"epoch": 2.23,
"learning_rate": 2.646e-06,
"loss": 0.3761,
"step": 885
},
{
"epoch": 2.24,
"learning_rate": 2.661e-06,
"loss": 0.3817,
"step": 890
},
{
"epoch": 2.26,
"learning_rate": 2.6760000000000003e-06,
"loss": 0.4256,
"step": 895
},
{
"epoch": 2.27,
"learning_rate": 2.691e-06,
"loss": 0.3778,
"step": 900
},
{
"epoch": 2.28,
"learning_rate": 2.706e-06,
"loss": 0.3571,
"step": 905
},
{
"epoch": 2.3,
"learning_rate": 2.7210000000000003e-06,
"loss": 0.4345,
"step": 910
},
{
"epoch": 2.31,
"learning_rate": 2.736e-06,
"loss": 0.3557,
"step": 915
},
{
"epoch": 2.32,
"learning_rate": 2.751e-06,
"loss": 0.3616,
"step": 920
},
{
"epoch": 2.33,
"learning_rate": 2.7660000000000003e-06,
"loss": 0.3813,
"step": 925
},
{
"epoch": 2.35,
"learning_rate": 2.781e-06,
"loss": 0.3828,
"step": 930
},
{
"epoch": 2.36,
"learning_rate": 2.7960000000000004e-06,
"loss": 0.3234,
"step": 935
},
{
"epoch": 2.37,
"learning_rate": 2.8110000000000003e-06,
"loss": 0.3353,
"step": 940
},
{
"epoch": 2.38,
"learning_rate": 2.826e-06,
"loss": 0.3856,
"step": 945
},
{
"epoch": 2.4,
"learning_rate": 2.8410000000000004e-06,
"loss": 0.3371,
"step": 950
},
{
"epoch": 2.41,
"learning_rate": 2.8560000000000003e-06,
"loss": 0.316,
"step": 955
},
{
"epoch": 2.42,
"learning_rate": 2.8709999999999997e-06,
"loss": 0.3904,
"step": 960
},
{
"epoch": 2.43,
"learning_rate": 2.886e-06,
"loss": 0.3297,
"step": 965
},
{
"epoch": 2.45,
"learning_rate": 2.901e-06,
"loss": 0.3163,
"step": 970
},
{
"epoch": 2.46,
"learning_rate": 2.916e-06,
"loss": 0.3396,
"step": 975
},
{
"epoch": 2.47,
"learning_rate": 2.931e-06,
"loss": 0.3422,
"step": 980
},
{
"epoch": 2.48,
"learning_rate": 2.946e-06,
"loss": 0.3075,
"step": 985
},
{
"epoch": 2.5,
"learning_rate": 2.961e-06,
"loss": 0.3132,
"step": 990
},
{
"epoch": 2.51,
"learning_rate": 2.976e-06,
"loss": 0.3504,
"step": 995
},
{
"epoch": 2.52,
"learning_rate": 2.991e-06,
"loss": 0.3313,
"step": 1000
},
{
"epoch": 2.52,
"eval_cer": 0.065871922362667,
"eval_loss": 0.287024587392807,
"eval_runtime": 60.4937,
"eval_samples_per_second": 41.789,
"eval_steps_per_second": 1.174,
"eval_wer": 0.2576838172864663,
"step": 1000
},
{
"epoch": 2.54,
"learning_rate": 3.006e-06,
"loss": 0.292,
"step": 1005
},
{
"epoch": 2.55,
"learning_rate": 3.0179999999999997e-06,
"loss": 0.3656,
"step": 1010
},
{
"epoch": 2.56,
"learning_rate": 3.033e-06,
"loss": 0.3029,
"step": 1015
},
{
"epoch": 2.58,
"learning_rate": 3.048e-06,
"loss": 0.2976,
"step": 1020
},
{
"epoch": 2.59,
"learning_rate": 3.063e-06,
"loss": 0.3674,
"step": 1025
},
{
"epoch": 2.6,
"learning_rate": 3.078e-06,
"loss": 0.3063,
"step": 1030
},
{
"epoch": 2.61,
"learning_rate": 3.093e-06,
"loss": 0.2821,
"step": 1035
},
{
"epoch": 2.63,
"learning_rate": 3.108e-06,
"loss": 0.2804,
"step": 1040
},
{
"epoch": 2.64,
"learning_rate": 3.123e-06,
"loss": 0.3475,
"step": 1045
},
{
"epoch": 2.65,
"learning_rate": 3.138e-06,
"loss": 0.2895,
"step": 1050
},
{
"epoch": 2.66,
"learning_rate": 3.153e-06,
"loss": 0.278,
"step": 1055
},
{
"epoch": 2.68,
"learning_rate": 3.168e-06,
"loss": 0.3217,
"step": 1060
},
{
"epoch": 2.69,
"learning_rate": 3.1830000000000003e-06,
"loss": 0.2822,
"step": 1065
},
{
"epoch": 2.7,
"learning_rate": 3.198e-06,
"loss": 0.2736,
"step": 1070
},
{
"epoch": 2.71,
"learning_rate": 3.213e-06,
"loss": 0.3326,
"step": 1075
},
{
"epoch": 2.73,
"learning_rate": 3.2280000000000003e-06,
"loss": 0.2738,
"step": 1080
},
{
"epoch": 2.74,
"learning_rate": 3.243e-06,
"loss": 0.2712,
"step": 1085
},
{
"epoch": 2.75,
"learning_rate": 3.258e-06,
"loss": 0.2858,
"step": 1090
},
{
"epoch": 2.76,
"learning_rate": 3.2730000000000003e-06,
"loss": 0.2962,
"step": 1095
},
{
"epoch": 2.78,
"learning_rate": 3.288e-06,
"loss": 0.2485,
"step": 1100
},
{
"epoch": 2.79,
"learning_rate": 3.3030000000000004e-06,
"loss": 0.2566,
"step": 1105
},
{
"epoch": 2.8,
"learning_rate": 3.315e-06,
"loss": 0.2978,
"step": 1110
},
{
"epoch": 2.81,
"learning_rate": 3.3300000000000003e-06,
"loss": 0.2629,
"step": 1115
},
{
"epoch": 2.83,
"learning_rate": 3.345e-06,
"loss": 0.2559,
"step": 1120
},
{
"epoch": 2.84,
"learning_rate": 3.36e-06,
"loss": 0.2628,
"step": 1125
},
{
"epoch": 2.85,
"learning_rate": 3.3750000000000003e-06,
"loss": 0.2455,
"step": 1130
},
{
"epoch": 2.87,
"learning_rate": 3.39e-06,
"loss": 0.2501,
"step": 1135
},
{
"epoch": 2.88,
"learning_rate": 3.405e-06,
"loss": 0.265,
"step": 1140
},
{
"epoch": 2.89,
"learning_rate": 3.417e-06,
"loss": 0.2922,
"step": 1145
},
{
"epoch": 2.9,
"learning_rate": 3.4320000000000003e-06,
"loss": 0.2492,
"step": 1150
},
{
"epoch": 2.92,
"learning_rate": 3.447e-06,
"loss": 0.2417,
"step": 1155
},
{
"epoch": 2.93,
"learning_rate": 3.462e-06,
"loss": 0.27,
"step": 1160
},
{
"epoch": 2.94,
"learning_rate": 3.4770000000000003e-06,
"loss": 0.2335,
"step": 1165
},
{
"epoch": 2.95,
"learning_rate": 3.492e-06,
"loss": 0.2327,
"step": 1170
},
{
"epoch": 2.97,
"learning_rate": 3.507e-06,
"loss": 0.2634,
"step": 1175
},
{
"epoch": 2.98,
"learning_rate": 3.5220000000000003e-06,
"loss": 0.2389,
"step": 1180
},
{
"epoch": 2.99,
"learning_rate": 3.537e-06,
"loss": 0.223,
"step": 1185
},
{
"epoch": 3.0,
"learning_rate": 3.552e-06,
"loss": 0.2773,
"step": 1190
},
{
"epoch": 3.02,
"learning_rate": 3.5670000000000003e-06,
"loss": 0.2285,
"step": 1195
},
{
"epoch": 3.03,
"learning_rate": 3.582e-06,
"loss": 0.2183,
"step": 1200
},
{
"epoch": 3.04,
"learning_rate": 3.5970000000000005e-06,
"loss": 0.2967,
"step": 1205
},
{
"epoch": 3.05,
"learning_rate": 3.612e-06,
"loss": 0.2179,
"step": 1210
},
{
"epoch": 3.07,
"learning_rate": 3.6269999999999997e-06,
"loss": 0.231,
"step": 1215
},
{
"epoch": 3.08,
"learning_rate": 3.642e-06,
"loss": 0.2366,
"step": 1220
},
{
"epoch": 3.09,
"learning_rate": 3.657e-06,
"loss": 0.2462,
"step": 1225
},
{
"epoch": 3.1,
"learning_rate": 3.6719999999999997e-06,
"loss": 0.2151,
"step": 1230
},
{
"epoch": 3.12,
"learning_rate": 3.687e-06,
"loss": 0.2162,
"step": 1235
},
{
"epoch": 3.13,
"learning_rate": 3.702e-06,
"loss": 0.2532,
"step": 1240
},
{
"epoch": 3.14,
"learning_rate": 3.717e-06,
"loss": 0.2035,
"step": 1245
},
{
"epoch": 3.16,
"learning_rate": 3.732e-06,
"loss": 0.2089,
"step": 1250
},
{
"epoch": 3.16,
"eval_cer": 0.044489706264102716,
"eval_loss": 0.2078969031572342,
"eval_runtime": 113.5687,
"eval_samples_per_second": 22.26,
"eval_steps_per_second": 0.625,
"eval_wer": 0.17657921548650027,
"step": 1250
},
{
"epoch": 3.17,
"learning_rate": 3.747e-06,
"loss": 0.2531,
"step": 1255
},
{
"epoch": 3.18,
"learning_rate": 3.7620000000000006e-06,
"loss": 0.2165,
"step": 1260
},
{
"epoch": 3.19,
"learning_rate": 3.7770000000000004e-06,
"loss": 0.212,
"step": 1265
},
{
"epoch": 3.21,
"learning_rate": 3.7920000000000003e-06,
"loss": 0.2218,
"step": 1270
},
{
"epoch": 3.22,
"learning_rate": 3.8070000000000006e-06,
"loss": 0.2297,
"step": 1275
},
{
"epoch": 3.23,
"learning_rate": 3.822000000000001e-06,
"loss": 0.2095,
"step": 1280
},
{
"epoch": 3.24,
"learning_rate": 3.837000000000001e-06,
"loss": 0.2028,
"step": 1285
},
{
"epoch": 3.26,
"learning_rate": 3.852e-06,
"loss": 0.2619,
"step": 1290
},
{
"epoch": 3.27,
"learning_rate": 3.8669999999999996e-06,
"loss": 0.1918,
"step": 1295
},
{
"epoch": 3.28,
"learning_rate": 3.8819999999999994e-06,
"loss": 0.1945,
"step": 1300
},
{
"epoch": 3.29,
"learning_rate": 3.897e-06,
"loss": 0.2347,
"step": 1305
},
{
"epoch": 3.31,
"learning_rate": 3.912e-06,
"loss": 0.2111,
"step": 1310
},
{
"epoch": 3.32,
"learning_rate": 3.927e-06,
"loss": 0.207,
"step": 1315
},
{
"epoch": 3.33,
"learning_rate": 3.942e-06,
"loss": 0.2281,
"step": 1320
},
{
"epoch": 3.34,
"learning_rate": 3.9569999999999996e-06,
"loss": 0.2334,
"step": 1325
},
{
"epoch": 3.36,
"learning_rate": 3.971999999999999e-06,
"loss": 0.1804,
"step": 1330
},
{
"epoch": 3.37,
"learning_rate": 3.987e-06,
"loss": 0.1943,
"step": 1335
},
{
"epoch": 3.38,
"learning_rate": 4.002e-06,
"loss": 0.2298,
"step": 1340
},
{
"epoch": 3.39,
"learning_rate": 4.017e-06,
"loss": 0.1911,
"step": 1345
},
{
"epoch": 3.41,
"learning_rate": 4.032e-06,
"loss": 0.1967,
"step": 1350
},
{
"epoch": 3.42,
"learning_rate": 4.0469999999999995e-06,
"loss": 0.2311,
"step": 1355
},
{
"epoch": 3.43,
"learning_rate": 4.062e-06,
"loss": 0.201,
"step": 1360
},
{
"epoch": 3.45,
"learning_rate": 4.077e-06,
"loss": 0.1956,
"step": 1365
},
{
"epoch": 3.46,
"learning_rate": 4.092e-06,
"loss": 0.2068,
"step": 1370
},
{
"epoch": 3.47,
"learning_rate": 4.107e-06,
"loss": 0.2211,
"step": 1375
},
{
"epoch": 3.48,
"learning_rate": 4.122e-06,
"loss": 0.187,
"step": 1380
},
{
"epoch": 3.5,
"learning_rate": 4.137e-06,
"loss": 0.1901,
"step": 1385
},
{
"epoch": 3.51,
"learning_rate": 4.152e-06,
"loss": 0.2077,
"step": 1390
},
{
"epoch": 3.52,
"learning_rate": 4.167e-06,
"loss": 0.1781,
"step": 1395
},
{
"epoch": 3.53,
"learning_rate": 4.182e-06,
"loss": 0.1806,
"step": 1400
},
{
"epoch": 3.55,
"learning_rate": 4.197e-06,
"loss": 0.2346,
"step": 1405
},
{
"epoch": 3.56,
"learning_rate": 4.212e-06,
"loss": 0.1864,
"step": 1410
},
{
"epoch": 3.57,
"learning_rate": 4.227e-06,
"loss": 0.1682,
"step": 1415
},
{
"epoch": 3.58,
"learning_rate": 4.242e-06,
"loss": 0.1944,
"step": 1420
},
{
"epoch": 3.6,
"learning_rate": 4.257e-06,
"loss": 0.2108,
"step": 1425
},
{
"epoch": 3.61,
"learning_rate": 4.272e-06,
"loss": 0.1731,
"step": 1430
},
{
"epoch": 3.62,
"learning_rate": 4.287e-06,
"loss": 0.1734,
"step": 1435
},
{
"epoch": 3.63,
"learning_rate": 4.3020000000000005e-06,
"loss": 0.2366,
"step": 1440
},
{
"epoch": 3.65,
"learning_rate": 4.317e-06,
"loss": 0.1858,
"step": 1445
},
{
"epoch": 3.66,
"learning_rate": 4.332e-06,
"loss": 0.1837,
"step": 1450
},
{
"epoch": 3.67,
"learning_rate": 4.347e-06,
"loss": 0.2047,
"step": 1455
},
{
"epoch": 3.68,
"learning_rate": 4.362e-06,
"loss": 0.1838,
"step": 1460
},
{
"epoch": 3.7,
"learning_rate": 4.377e-06,
"loss": 0.1684,
"step": 1465
},
{
"epoch": 3.71,
"learning_rate": 4.3920000000000005e-06,
"loss": 0.1965,
"step": 1470
},
{
"epoch": 3.72,
"learning_rate": 4.407e-06,
"loss": 0.2069,
"step": 1475
},
{
"epoch": 3.74,
"learning_rate": 4.422e-06,
"loss": 0.1751,
"step": 1480
},
{
"epoch": 3.75,
"learning_rate": 4.437e-06,
"loss": 0.1724,
"step": 1485
},
{
"epoch": 3.76,
"learning_rate": 4.452e-06,
"loss": 0.2224,
"step": 1490
},
{
"epoch": 3.77,
"learning_rate": 4.467000000000001e-06,
"loss": 0.1684,
"step": 1495
},
{
"epoch": 3.79,
"learning_rate": 4.4820000000000005e-06,
"loss": 0.1634,
"step": 1500
},
{
"epoch": 3.79,
"eval_cer": 0.03657820643383454,
"eval_loss": 0.16873595118522644,
"eval_runtime": 60.2617,
"eval_samples_per_second": 41.95,
"eval_steps_per_second": 1.178,
"eval_wer": 0.14106809305484802,
"step": 1500
},
{
"epoch": 3.8,
"learning_rate": 4.497e-06,
"loss": 0.2137,
"step": 1505
},
{
"epoch": 3.81,
"learning_rate": 4.5089999999999995e-06,
"loss": 0.1759,
"step": 1510
},
{
"epoch": 3.82,
"learning_rate": 4.524e-06,
"loss": 0.1698,
"step": 1515
},
{
"epoch": 3.84,
"learning_rate": 4.539e-06,
"loss": 0.1703,
"step": 1520
},
{
"epoch": 3.85,
"learning_rate": 4.554e-06,
"loss": 0.1841,
"step": 1525
},
{
"epoch": 3.86,
"learning_rate": 4.569e-06,
"loss": 0.1783,
"step": 1530
},
{
"epoch": 3.87,
"learning_rate": 4.584e-06,
"loss": 0.1633,
"step": 1535
},
{
"epoch": 3.89,
"learning_rate": 4.5989999999999995e-06,
"loss": 0.1839,
"step": 1540
},
{
"epoch": 3.9,
"learning_rate": 4.614e-06,
"loss": 0.153,
"step": 1545
},
{
"epoch": 3.91,
"learning_rate": 4.629e-06,
"loss": 0.1553,
"step": 1550
},
{
"epoch": 3.92,
"learning_rate": 4.644e-06,
"loss": 0.1834,
"step": 1555
},
{
"epoch": 3.94,
"learning_rate": 4.659e-06,
"loss": 0.1582,
"step": 1560
},
{
"epoch": 3.95,
"learning_rate": 4.674e-06,
"loss": 0.1592,
"step": 1565
},
{
"epoch": 3.96,
"learning_rate": 4.689e-06,
"loss": 0.1823,
"step": 1570
},
{
"epoch": 3.97,
"learning_rate": 4.704e-06,
"loss": 0.1782,
"step": 1575
},
{
"epoch": 3.99,
"learning_rate": 4.719e-06,
"loss": 0.1543,
"step": 1580
},
{
"epoch": 4.0,
"learning_rate": 4.734e-06,
"loss": 0.1943,
"step": 1585
},
{
"epoch": 4.01,
"learning_rate": 4.749e-06,
"loss": 0.1491,
"step": 1590
},
{
"epoch": 4.03,
"learning_rate": 4.764e-06,
"loss": 0.1656,
"step": 1595
},
{
"epoch": 4.04,
"learning_rate": 4.779e-06,
"loss": 0.1507,
"step": 1600
},
{
"epoch": 4.05,
"learning_rate": 4.794e-06,
"loss": 0.1589,
"step": 1605
},
{
"epoch": 4.06,
"learning_rate": 4.809e-06,
"loss": 0.1443,
"step": 1610
},
{
"epoch": 4.08,
"learning_rate": 4.824e-06,
"loss": 0.1487,
"step": 1615
},
{
"epoch": 4.09,
"learning_rate": 4.839e-06,
"loss": 0.1766,
"step": 1620
},
{
"epoch": 4.1,
"learning_rate": 4.8540000000000005e-06,
"loss": 0.1548,
"step": 1625
},
{
"epoch": 4.11,
"learning_rate": 4.869e-06,
"loss": 0.1494,
"step": 1630
},
{
"epoch": 4.13,
"learning_rate": 4.884e-06,
"loss": 0.1793,
"step": 1635
},
{
"epoch": 4.14,
"learning_rate": 4.899e-06,
"loss": 0.1417,
"step": 1640
},
{
"epoch": 4.15,
"learning_rate": 4.914e-06,
"loss": 0.1399,
"step": 1645
},
{
"epoch": 4.16,
"learning_rate": 4.929000000000001e-06,
"loss": 0.166,
"step": 1650
},
{
"epoch": 4.18,
"learning_rate": 4.9440000000000004e-06,
"loss": 0.1748,
"step": 1655
},
{
"epoch": 4.19,
"learning_rate": 4.959e-06,
"loss": 0.1378,
"step": 1660
},
{
"epoch": 4.2,
"learning_rate": 4.974e-06,
"loss": 0.144,
"step": 1665
},
{
"epoch": 4.21,
"learning_rate": 4.989e-06,
"loss": 0.1794,
"step": 1670
},
{
"epoch": 4.23,
"learning_rate": 5.004e-06,
"loss": 0.146,
"step": 1675
},
{
"epoch": 4.24,
"learning_rate": 5.0190000000000006e-06,
"loss": 0.141,
"step": 1680
},
{
"epoch": 4.25,
"learning_rate": 5.034e-06,
"loss": 0.1757,
"step": 1685
},
{
"epoch": 4.26,
"learning_rate": 5.049e-06,
"loss": 0.1423,
"step": 1690
},
{
"epoch": 4.28,
"learning_rate": 5.064e-06,
"loss": 0.1414,
"step": 1695
},
{
"epoch": 4.29,
"learning_rate": 5.079e-06,
"loss": 0.1556,
"step": 1700
},
{
"epoch": 4.3,
"learning_rate": 5.094000000000001e-06,
"loss": 0.1599,
"step": 1705
},
{
"epoch": 4.32,
"learning_rate": 5.1090000000000006e-06,
"loss": 0.1436,
"step": 1710
},
{
"epoch": 4.33,
"learning_rate": 5.124e-06,
"loss": 0.1346,
"step": 1715
},
{
"epoch": 4.34,
"learning_rate": 5.139e-06,
"loss": 0.1702,
"step": 1720
},
{
"epoch": 4.35,
"learning_rate": 5.154e-06,
"loss": 0.1397,
"step": 1725
},
{
"epoch": 4.37,
"learning_rate": 5.169e-06,
"loss": 0.1373,
"step": 1730
},
{
"epoch": 4.38,
"learning_rate": 5.184000000000001e-06,
"loss": 0.1929,
"step": 1735
},
{
"epoch": 4.39,
"learning_rate": 5.1990000000000005e-06,
"loss": 0.1426,
"step": 1740
},
{
"epoch": 4.4,
"learning_rate": 5.214e-06,
"loss": 0.1481,
"step": 1745
},
{
"epoch": 4.42,
"learning_rate": 5.229e-06,
"loss": 0.163,
"step": 1750
},
{
"epoch": 4.42,
"eval_cer": 0.029768965035244312,
"eval_loss": 0.1489591896533966,
"eval_runtime": 60.6618,
"eval_samples_per_second": 41.674,
"eval_steps_per_second": 1.17,
"eval_wer": 0.1163397860417728,
"step": 1750
},
{
"epoch": 4.43,
"learning_rate": 5.244e-06,
"loss": 0.1816,
"step": 1755
},
{
"epoch": 4.44,
"learning_rate": 5.259000000000001e-06,
"loss": 0.1444,
"step": 1760
},
{
"epoch": 4.45,
"learning_rate": 5.274000000000001e-06,
"loss": 0.1423,
"step": 1765
},
{
"epoch": 4.47,
"learning_rate": 5.2890000000000005e-06,
"loss": 0.1605,
"step": 1770
},
{
"epoch": 4.48,
"learning_rate": 5.304e-06,
"loss": 0.1355,
"step": 1775
},
{
"epoch": 4.49,
"learning_rate": 5.319e-06,
"loss": 0.1401,
"step": 1780
},
{
"epoch": 4.5,
"learning_rate": 5.334000000000001e-06,
"loss": 0.1593,
"step": 1785
},
{
"epoch": 4.52,
"learning_rate": 5.349e-06,
"loss": 0.1361,
"step": 1790
},
{
"epoch": 4.53,
"learning_rate": 5.364e-06,
"loss": 0.1363,
"step": 1795
},
{
"epoch": 4.54,
"learning_rate": 5.379e-06,
"loss": 0.1527,
"step": 1800
},
{
"epoch": 4.55,
"learning_rate": 5.3939999999999995e-06,
"loss": 0.1645,
"step": 1805
},
{
"epoch": 4.57,
"learning_rate": 5.408999999999999e-06,
"loss": 0.1285,
"step": 1810
},
{
"epoch": 4.58,
"learning_rate": 5.424e-06,
"loss": 0.1366,
"step": 1815
},
{
"epoch": 4.59,
"learning_rate": 5.439e-06,
"loss": 0.1808,
"step": 1820
},
{
"epoch": 4.61,
"learning_rate": 5.454e-06,
"loss": 0.1428,
"step": 1825
},
{
"epoch": 4.62,
"learning_rate": 5.469e-06,
"loss": 0.1314,
"step": 1830
},
{
"epoch": 4.63,
"learning_rate": 5.4839999999999995e-06,
"loss": 0.1479,
"step": 1835
},
{
"epoch": 4.64,
"learning_rate": 5.499e-06,
"loss": 0.135,
"step": 1840
},
{
"epoch": 4.66,
"learning_rate": 5.514e-06,
"loss": 0.1253,
"step": 1845
},
{
"epoch": 4.67,
"learning_rate": 5.529e-06,
"loss": 0.13,
"step": 1850
},
{
"epoch": 4.68,
"learning_rate": 5.544e-06,
"loss": 0.1484,
"step": 1855
},
{
"epoch": 4.69,
"learning_rate": 5.559e-06,
"loss": 0.1201,
"step": 1860
},
{
"epoch": 4.71,
"learning_rate": 5.574e-06,
"loss": 0.1263,
"step": 1865
},
{
"epoch": 4.72,
"learning_rate": 5.589e-06,
"loss": 0.1493,
"step": 1870
},
{
"epoch": 4.73,
"learning_rate": 5.604e-06,
"loss": 0.1276,
"step": 1875
},
{
"epoch": 4.74,
"learning_rate": 5.619e-06,
"loss": 0.1236,
"step": 1880
},
{
"epoch": 4.76,
"learning_rate": 5.634e-06,
"loss": 0.1718,
"step": 1885
},
{
"epoch": 4.77,
"learning_rate": 5.649e-06,
"loss": 0.1292,
"step": 1890
},
{
"epoch": 4.78,
"learning_rate": 5.664e-06,
"loss": 0.1175,
"step": 1895
},
{
"epoch": 4.79,
"learning_rate": 5.679e-06,
"loss": 0.1317,
"step": 1900
},
{
"epoch": 4.81,
"learning_rate": 5.694e-06,
"loss": 0.1412,
"step": 1905
},
{
"epoch": 4.82,
"learning_rate": 5.709e-06,
"loss": 0.1222,
"step": 1910
},
{
"epoch": 4.83,
"learning_rate": 5.724e-06,
"loss": 0.1193,
"step": 1915
},
{
"epoch": 4.84,
"learning_rate": 5.7390000000000004e-06,
"loss": 0.16,
"step": 1920
},
{
"epoch": 4.86,
"learning_rate": 5.754e-06,
"loss": 0.1228,
"step": 1925
},
{
"epoch": 4.87,
"learning_rate": 5.769e-06,
"loss": 0.1195,
"step": 1930
},
{
"epoch": 4.88,
"learning_rate": 5.784e-06,
"loss": 0.1608,
"step": 1935
},
{
"epoch": 4.9,
"learning_rate": 5.799e-06,
"loss": 0.1234,
"step": 1940
},
{
"epoch": 4.91,
"learning_rate": 5.814e-06,
"loss": 0.1265,
"step": 1945
},
{
"epoch": 4.92,
"learning_rate": 5.8290000000000004e-06,
"loss": 0.1431,
"step": 1950
},
{
"epoch": 4.93,
"learning_rate": 5.844e-06,
"loss": 0.1366,
"step": 1955
},
{
"epoch": 4.95,
"learning_rate": 5.859e-06,
"loss": 0.1091,
"step": 1960
},
{
"epoch": 4.96,
"learning_rate": 5.874e-06,
"loss": 0.139,
"step": 1965
},
{
"epoch": 4.97,
"learning_rate": 5.886000000000001e-06,
"loss": 0.1726,
"step": 1970
},
{
"epoch": 4.98,
"learning_rate": 5.901000000000001e-06,
"loss": 0.1148,
"step": 1975
},
{
"epoch": 5.0,
"learning_rate": 5.916e-06,
"loss": 0.1296,
"step": 1980
},
{
"epoch": 5.0,
"step": 1980,
"total_flos": 5.2867996741439324e+19,
"train_loss": 0.09581804365822763,
"train_runtime": 4900.8909,
"train_samples_per_second": 29.115,
"train_steps_per_second": 0.404
}
],
"logging_steps": 5,
"max_steps": 1980,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 500,
"total_flos": 5.2867996741439324e+19,
"train_batch_size": 12,
"trial_name": null,
"trial_params": null
}