xls-r-300m-te / trainer_state.json
chmanoj's picture
End of training
51d7925
raw
history blame
18 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 99.99760191846522,
"global_step": 10400,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.96,
"learning_rate": 4.9500000000000004e-05,
"loss": 14.732,
"step": 100
},
{
"epoch": 1.92,
"learning_rate": 9.95e-05,
"loss": 4.5706,
"step": 200
},
{
"epoch": 2.88,
"learning_rate": 0.0001495,
"loss": 3.4967,
"step": 300
},
{
"epoch": 3.84,
"learning_rate": 0.00019950000000000002,
"loss": 3.3326,
"step": 400
},
{
"epoch": 4.81,
"learning_rate": 0.0002495,
"loss": 3.0304,
"step": 500
},
{
"epoch": 4.81,
"eval_loss": 1.5675914287567139,
"eval_runtime": 78.6195,
"eval_samples_per_second": 14.144,
"eval_steps_per_second": 3.536,
"eval_wer": 1.0554260089686098,
"step": 500
},
{
"epoch": 5.77,
"learning_rate": 0.0002995,
"loss": 2.0254,
"step": 600
},
{
"epoch": 6.73,
"learning_rate": 0.0003495,
"loss": 1.7226,
"step": 700
},
{
"epoch": 7.69,
"learning_rate": 0.0003995,
"loss": 1.6212,
"step": 800
},
{
"epoch": 8.65,
"learning_rate": 0.00044950000000000003,
"loss": 1.5773,
"step": 900
},
{
"epoch": 9.61,
"learning_rate": 0.0004995,
"loss": 1.5263,
"step": 1000
},
{
"epoch": 9.61,
"eval_loss": 0.46925902366638184,
"eval_runtime": 52.7689,
"eval_samples_per_second": 21.073,
"eval_steps_per_second": 5.268,
"eval_wer": 0.8023318385650224,
"step": 1000
},
{
"epoch": 10.58,
"learning_rate": 0.0005495,
"loss": 1.5324,
"step": 1100
},
{
"epoch": 11.54,
"learning_rate": 0.0005995000000000001,
"loss": 1.5058,
"step": 1200
},
{
"epoch": 12.5,
"learning_rate": 0.0006495,
"loss": 1.512,
"step": 1300
},
{
"epoch": 13.46,
"learning_rate": 0.0006995,
"loss": 1.5222,
"step": 1400
},
{
"epoch": 14.42,
"learning_rate": 0.0007495000000000001,
"loss": 1.5299,
"step": 1500
},
{
"epoch": 14.42,
"eval_loss": 0.43676647543907166,
"eval_runtime": 52.3008,
"eval_samples_per_second": 21.262,
"eval_steps_per_second": 5.315,
"eval_wer": 0.7311210762331839,
"step": 1500
},
{
"epoch": 15.38,
"learning_rate": 0.0007995,
"loss": 1.5168,
"step": 1600
},
{
"epoch": 16.35,
"learning_rate": 0.0008495000000000001,
"loss": 1.5392,
"step": 1700
},
{
"epoch": 17.31,
"learning_rate": 0.0008995,
"loss": 1.5365,
"step": 1800
},
{
"epoch": 18.27,
"learning_rate": 0.0009495,
"loss": 1.4991,
"step": 1900
},
{
"epoch": 19.23,
"learning_rate": 0.0009995000000000002,
"loss": 1.5063,
"step": 2000
},
{
"epoch": 19.23,
"eval_loss": 0.43596330285072327,
"eval_runtime": 50.4313,
"eval_samples_per_second": 22.05,
"eval_steps_per_second": 5.512,
"eval_wer": 0.7302242152466367,
"step": 2000
},
{
"epoch": 20.19,
"learning_rate": 0.0009882142857142856,
"loss": 1.5283,
"step": 2100
},
{
"epoch": 21.15,
"learning_rate": 0.0009763095238095238,
"loss": 1.5146,
"step": 2200
},
{
"epoch": 22.12,
"learning_rate": 0.0009644047619047619,
"loss": 1.5236,
"step": 2300
},
{
"epoch": 23.08,
"learning_rate": 0.0009525,
"loss": 1.4972,
"step": 2400
},
{
"epoch": 24.04,
"learning_rate": 0.0009405952380952381,
"loss": 1.455,
"step": 2500
},
{
"epoch": 24.04,
"eval_loss": 0.4213278293609619,
"eval_runtime": 50.1354,
"eval_samples_per_second": 22.18,
"eval_steps_per_second": 5.545,
"eval_wer": 0.6692376681614349,
"step": 2500
},
{
"epoch": 25.0,
"learning_rate": 0.0009286904761904762,
"loss": 1.4424,
"step": 2600
},
{
"epoch": 25.96,
"learning_rate": 0.0009167857142857144,
"loss": 1.4434,
"step": 2700
},
{
"epoch": 26.92,
"learning_rate": 0.0009048809523809524,
"loss": 1.4513,
"step": 2800
},
{
"epoch": 27.88,
"learning_rate": 0.0008929761904761905,
"loss": 1.4328,
"step": 2900
},
{
"epoch": 28.84,
"learning_rate": 0.0008811904761904762,
"loss": 1.4755,
"step": 3000
},
{
"epoch": 28.84,
"eval_loss": 0.4329236149787903,
"eval_runtime": 50.0668,
"eval_samples_per_second": 22.21,
"eval_steps_per_second": 5.553,
"eval_wer": 0.5942600896860987,
"step": 3000
},
{
"epoch": 29.81,
"learning_rate": 0.0008692857142857144,
"loss": 1.4344,
"step": 3100
},
{
"epoch": 30.77,
"learning_rate": 0.0008573809523809523,
"loss": 1.4326,
"step": 3200
},
{
"epoch": 31.73,
"learning_rate": 0.0008454761904761905,
"loss": 1.4016,
"step": 3300
},
{
"epoch": 32.69,
"learning_rate": 0.0008335714285714285,
"loss": 1.3705,
"step": 3400
},
{
"epoch": 33.65,
"learning_rate": 0.0008216666666666667,
"loss": 1.352,
"step": 3500
},
{
"epoch": 33.65,
"eval_loss": 0.40741658210754395,
"eval_runtime": 76.8047,
"eval_samples_per_second": 14.478,
"eval_steps_per_second": 3.62,
"eval_wer": 0.5765022421524664,
"step": 3500
},
{
"epoch": 34.61,
"learning_rate": 0.0008097619047619048,
"loss": 1.3511,
"step": 3600
},
{
"epoch": 35.58,
"learning_rate": 0.0007978571428571428,
"loss": 1.3237,
"step": 3700
},
{
"epoch": 36.54,
"learning_rate": 0.000785952380952381,
"loss": 1.3165,
"step": 3800
},
{
"epoch": 37.5,
"learning_rate": 0.0007740476190476191,
"loss": 1.3342,
"step": 3900
},
{
"epoch": 38.46,
"learning_rate": 0.0007621428571428572,
"loss": 1.3122,
"step": 4000
},
{
"epoch": 38.46,
"eval_loss": 0.38659143447875977,
"eval_runtime": 80.318,
"eval_samples_per_second": 13.845,
"eval_steps_per_second": 3.461,
"eval_wer": 0.56304932735426,
"step": 4000
},
{
"epoch": 39.42,
"learning_rate": 0.0007502380952380953,
"loss": 1.3098,
"step": 4100
},
{
"epoch": 40.38,
"learning_rate": 0.0007383333333333334,
"loss": 1.3253,
"step": 4200
},
{
"epoch": 41.35,
"learning_rate": 0.0007264285714285714,
"loss": 1.2787,
"step": 4300
},
{
"epoch": 42.31,
"learning_rate": 0.0007146428571428572,
"loss": 1.274,
"step": 4400
},
{
"epoch": 43.27,
"learning_rate": 0.0007027380952380952,
"loss": 1.2799,
"step": 4500
},
{
"epoch": 43.27,
"eval_loss": 0.3859865069389343,
"eval_runtime": 77.9764,
"eval_samples_per_second": 14.261,
"eval_steps_per_second": 3.565,
"eval_wer": 0.5479820627802691,
"step": 4500
},
{
"epoch": 44.23,
"learning_rate": 0.0006908333333333333,
"loss": 1.2525,
"step": 4600
},
{
"epoch": 45.19,
"learning_rate": 0.0006789285714285714,
"loss": 1.2245,
"step": 4700
},
{
"epoch": 46.15,
"learning_rate": 0.0006670238095238096,
"loss": 1.2158,
"step": 4800
},
{
"epoch": 47.12,
"learning_rate": 0.0006551190476190476,
"loss": 1.2028,
"step": 4900
},
{
"epoch": 48.08,
"learning_rate": 0.0006432142857142857,
"loss": 1.212,
"step": 5000
},
{
"epoch": 48.08,
"eval_loss": 0.358958899974823,
"eval_runtime": 78.2642,
"eval_samples_per_second": 14.208,
"eval_steps_per_second": 3.552,
"eval_wer": 0.5316591928251121,
"step": 5000
},
{
"epoch": 49.04,
"learning_rate": 0.0006313095238095238,
"loss": 1.1808,
"step": 5100
},
{
"epoch": 50.0,
"learning_rate": 0.0006194047619047619,
"loss": 1.1783,
"step": 5200
},
{
"epoch": 50.96,
"learning_rate": 0.0006075000000000001,
"loss": 1.1675,
"step": 5300
},
{
"epoch": 51.92,
"learning_rate": 0.0005955952380952381,
"loss": 1.1867,
"step": 5400
},
{
"epoch": 52.88,
"learning_rate": 0.0005836904761904763,
"loss": 1.1645,
"step": 5500
},
{
"epoch": 52.88,
"eval_loss": 0.328298419713974,
"eval_runtime": 77.7284,
"eval_samples_per_second": 14.306,
"eval_steps_per_second": 3.577,
"eval_wer": 0.475695067264574,
"step": 5500
},
{
"epoch": 53.84,
"learning_rate": 0.0005717857142857142,
"loss": 1.1407,
"step": 5600
},
{
"epoch": 54.81,
"learning_rate": 0.0005598809523809523,
"loss": 1.1217,
"step": 5700
},
{
"epoch": 55.77,
"learning_rate": 0.0005479761904761905,
"loss": 1.118,
"step": 5800
},
{
"epoch": 56.73,
"learning_rate": 0.0005360714285714285,
"loss": 1.0727,
"step": 5900
},
{
"epoch": 57.69,
"learning_rate": 0.0005241666666666667,
"loss": 1.0854,
"step": 6000
},
{
"epoch": 57.69,
"eval_loss": 0.3161650598049164,
"eval_runtime": 82.3243,
"eval_samples_per_second": 13.508,
"eval_steps_per_second": 3.377,
"eval_wer": 0.4686995515695067,
"step": 6000
},
{
"epoch": 58.65,
"learning_rate": 0.0005122619047619048,
"loss": 1.0673,
"step": 6100
},
{
"epoch": 59.61,
"learning_rate": 0.0005003571428571429,
"loss": 1.0595,
"step": 6200
},
{
"epoch": 60.58,
"learning_rate": 0.000488452380952381,
"loss": 1.0474,
"step": 6300
},
{
"epoch": 61.54,
"learning_rate": 0.00047654761904761906,
"loss": 1.0424,
"step": 6400
},
{
"epoch": 62.5,
"learning_rate": 0.00046464285714285715,
"loss": 1.0292,
"step": 6500
},
{
"epoch": 62.5,
"eval_loss": 0.3125934600830078,
"eval_runtime": 78.0974,
"eval_samples_per_second": 14.239,
"eval_steps_per_second": 3.56,
"eval_wer": 0.44161434977578473,
"step": 6500
},
{
"epoch": 63.46,
"learning_rate": 0.00045285714285714287,
"loss": 1.0259,
"step": 6600
},
{
"epoch": 64.42,
"learning_rate": 0.00044095238095238096,
"loss": 1.0141,
"step": 6700
},
{
"epoch": 65.38,
"learning_rate": 0.00042916666666666667,
"loss": 0.9839,
"step": 6800
},
{
"epoch": 66.35,
"learning_rate": 0.00041726190476190476,
"loss": 0.9553,
"step": 6900
},
{
"epoch": 67.31,
"learning_rate": 0.00040535714285714285,
"loss": 0.9607,
"step": 7000
},
{
"epoch": 67.31,
"eval_loss": 0.2990323007106781,
"eval_runtime": 78.5216,
"eval_samples_per_second": 14.162,
"eval_steps_per_second": 3.54,
"eval_wer": 0.40663677130044845,
"step": 7000
},
{
"epoch": 68.27,
"learning_rate": 0.00039345238095238094,
"loss": 0.944,
"step": 7100
},
{
"epoch": 69.23,
"learning_rate": 0.00038154761904761903,
"loss": 0.9203,
"step": 7200
},
{
"epoch": 70.19,
"learning_rate": 0.0003696428571428572,
"loss": 0.9164,
"step": 7300
},
{
"epoch": 71.15,
"learning_rate": 0.00035773809523809527,
"loss": 0.8996,
"step": 7400
},
{
"epoch": 72.12,
"learning_rate": 0.00034583333333333335,
"loss": 0.9156,
"step": 7500
},
{
"epoch": 72.12,
"eval_loss": 0.28695425391197205,
"eval_runtime": 77.1644,
"eval_samples_per_second": 14.411,
"eval_steps_per_second": 3.603,
"eval_wer": 0.4008968609865471,
"step": 7500
},
{
"epoch": 73.08,
"learning_rate": 0.0003339285714285714,
"loss": 0.8916,
"step": 7600
},
{
"epoch": 74.04,
"learning_rate": 0.00032202380952380953,
"loss": 0.8818,
"step": 7700
},
{
"epoch": 75.0,
"learning_rate": 0.0003101190476190476,
"loss": 0.8714,
"step": 7800
},
{
"epoch": 75.96,
"learning_rate": 0.0002982142857142857,
"loss": 0.8529,
"step": 7900
},
{
"epoch": 76.92,
"learning_rate": 0.0002863095238095238,
"loss": 0.8329,
"step": 8000
},
{
"epoch": 76.92,
"eval_loss": 0.27907443046569824,
"eval_runtime": 76.6666,
"eval_samples_per_second": 14.504,
"eval_steps_per_second": 3.626,
"eval_wer": 0.39085201793721974,
"step": 8000
},
{
"epoch": 77.88,
"learning_rate": 0.00027440476190476195,
"loss": 0.822,
"step": 8100
},
{
"epoch": 78.84,
"learning_rate": 0.00026250000000000004,
"loss": 0.8164,
"step": 8200
},
{
"epoch": 79.81,
"learning_rate": 0.0002505952380952381,
"loss": 0.8086,
"step": 8300
},
{
"epoch": 80.77,
"learning_rate": 0.00023869047619047622,
"loss": 0.7969,
"step": 8400
},
{
"epoch": 81.73,
"learning_rate": 0.00022678571428571428,
"loss": 0.7979,
"step": 8500
},
{
"epoch": 81.73,
"eval_loss": 0.27704960107803345,
"eval_runtime": 76.769,
"eval_samples_per_second": 14.485,
"eval_steps_per_second": 3.621,
"eval_wer": 0.3669955156950673,
"step": 8500
},
{
"epoch": 82.69,
"learning_rate": 0.0002148809523809524,
"loss": 0.7862,
"step": 8600
},
{
"epoch": 83.65,
"learning_rate": 0.0002029761904761905,
"loss": 0.7707,
"step": 8700
},
{
"epoch": 84.61,
"learning_rate": 0.00019107142857142855,
"loss": 0.7472,
"step": 8800
},
{
"epoch": 85.58,
"learning_rate": 0.00017916666666666667,
"loss": 0.7583,
"step": 8900
},
{
"epoch": 86.54,
"learning_rate": 0.00016726190476190476,
"loss": 0.7144,
"step": 9000
},
{
"epoch": 86.54,
"eval_loss": 0.2840667963027954,
"eval_runtime": 77.8679,
"eval_samples_per_second": 14.281,
"eval_steps_per_second": 3.57,
"eval_wer": 0.36609865470852015,
"step": 9000
},
{
"epoch": 87.5,
"learning_rate": 0.00015535714285714287,
"loss": 0.7302,
"step": 9100
},
{
"epoch": 88.46,
"learning_rate": 0.0001435714285714286,
"loss": 0.7214,
"step": 9200
},
{
"epoch": 89.42,
"learning_rate": 0.00013166666666666665,
"loss": 0.7179,
"step": 9300
},
{
"epoch": 90.38,
"learning_rate": 0.00011976190476190477,
"loss": 0.7037,
"step": 9400
},
{
"epoch": 91.35,
"learning_rate": 0.00010785714285714286,
"loss": 0.6997,
"step": 9500
},
{
"epoch": 91.35,
"eval_loss": 0.2721162438392639,
"eval_runtime": 77.3441,
"eval_samples_per_second": 14.377,
"eval_steps_per_second": 3.594,
"eval_wer": 0.3485201793721973,
"step": 9500
},
{
"epoch": 92.31,
"learning_rate": 9.595238095238096e-05,
"loss": 0.676,
"step": 9600
},
{
"epoch": 93.27,
"learning_rate": 8.404761904761905e-05,
"loss": 0.6748,
"step": 9700
},
{
"epoch": 94.23,
"learning_rate": 7.214285714285715e-05,
"loss": 0.6668,
"step": 9800
},
{
"epoch": 95.19,
"learning_rate": 6.023809523809524e-05,
"loss": 0.662,
"step": 9900
},
{
"epoch": 96.15,
"learning_rate": 4.8333333333333334e-05,
"loss": 0.6568,
"step": 10000
},
{
"epoch": 96.15,
"eval_loss": 0.2680588364601135,
"eval_runtime": 76.864,
"eval_samples_per_second": 14.467,
"eval_steps_per_second": 3.617,
"eval_wer": 0.34367713004484307,
"step": 10000
},
{
"epoch": 97.12,
"learning_rate": 3.642857142857143e-05,
"loss": 0.6517,
"step": 10100
},
{
"epoch": 98.08,
"learning_rate": 2.4523809523809523e-05,
"loss": 0.6518,
"step": 10200
},
{
"epoch": 99.04,
"learning_rate": 1.2619047619047618e-05,
"loss": 0.6285,
"step": 10300
},
{
"epoch": 100.0,
"learning_rate": 7.142857142857143e-07,
"loss": 0.6282,
"step": 10400
},
{
"epoch": 100.0,
"step": 10400,
"total_flos": 4.70570622386933e+19,
"train_loss": 1.3617363874728863,
"train_runtime": 41188.5513,
"train_samples_per_second": 8.099,
"train_steps_per_second": 0.252
}
],
"max_steps": 10400,
"num_train_epochs": 100,
"total_flos": 4.70570622386933e+19,
"trial_name": null,
"trial_params": null
}