whisper-small-uz / trainer_state.json
BlueRaccoon's picture
End of training
67443e6
raw
history blame
25.8 kB
{
"best_metric": 23.650914047642605,
"best_model_checkpoint": "./checkpoint-4000",
"epoch": 1.405,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 5.5e-07,
"loss": 3.261,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 1.175e-06,
"loss": 2.7217,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.8000000000000001e-06,
"loss": 2.1001,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 2.425e-06,
"loss": 1.646,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 3.05e-06,
"loss": 1.2557,
"step": 125
},
{
"epoch": 0.03,
"learning_rate": 3.6750000000000004e-06,
"loss": 0.9606,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 4.3e-06,
"loss": 0.8033,
"step": 175
},
{
"epoch": 0.04,
"learning_rate": 4.925e-06,
"loss": 0.666,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 5.550000000000001e-06,
"loss": 0.58,
"step": 225
},
{
"epoch": 0.05,
"learning_rate": 6.175000000000001e-06,
"loss": 0.523,
"step": 250
},
{
"epoch": 0.06,
"learning_rate": 6.800000000000001e-06,
"loss": 0.4632,
"step": 275
},
{
"epoch": 0.06,
"learning_rate": 7.425000000000001e-06,
"loss": 0.4329,
"step": 300
},
{
"epoch": 0.07,
"learning_rate": 8.050000000000001e-06,
"loss": 0.4719,
"step": 325
},
{
"epoch": 0.07,
"learning_rate": 8.675e-06,
"loss": 0.3749,
"step": 350
},
{
"epoch": 0.07,
"learning_rate": 9.3e-06,
"loss": 0.3279,
"step": 375
},
{
"epoch": 0.08,
"learning_rate": 9.925e-06,
"loss": 0.3093,
"step": 400
},
{
"epoch": 0.09,
"learning_rate": 9.952173913043479e-06,
"loss": 0.3205,
"step": 425
},
{
"epoch": 0.09,
"learning_rate": 9.897826086956522e-06,
"loss": 0.2951,
"step": 450
},
{
"epoch": 0.1,
"learning_rate": 9.843478260869566e-06,
"loss": 0.3032,
"step": 475
},
{
"epoch": 0.1,
"learning_rate": 9.78913043478261e-06,
"loss": 0.2788,
"step": 500
},
{
"epoch": 0.1,
"learning_rate": 9.734782608695653e-06,
"loss": 0.2607,
"step": 525
},
{
"epoch": 0.11,
"learning_rate": 9.680434782608696e-06,
"loss": 0.2428,
"step": 550
},
{
"epoch": 0.12,
"learning_rate": 9.62608695652174e-06,
"loss": 0.2295,
"step": 575
},
{
"epoch": 0.12,
"learning_rate": 9.571739130434783e-06,
"loss": 0.259,
"step": 600
},
{
"epoch": 0.12,
"learning_rate": 9.517391304347827e-06,
"loss": 0.2514,
"step": 625
},
{
"epoch": 0.13,
"learning_rate": 9.46304347826087e-06,
"loss": 0.2278,
"step": 650
},
{
"epoch": 0.14,
"learning_rate": 9.408695652173914e-06,
"loss": 0.2237,
"step": 675
},
{
"epoch": 0.14,
"learning_rate": 9.354347826086957e-06,
"loss": 0.2119,
"step": 700
},
{
"epoch": 0.14,
"learning_rate": 9.3e-06,
"loss": 0.2392,
"step": 725
},
{
"epoch": 0.15,
"learning_rate": 9.245652173913044e-06,
"loss": 0.209,
"step": 750
},
{
"epoch": 0.15,
"learning_rate": 9.191304347826088e-06,
"loss": 0.1765,
"step": 775
},
{
"epoch": 0.16,
"learning_rate": 9.136956521739131e-06,
"loss": 0.1685,
"step": 800
},
{
"epoch": 0.17,
"learning_rate": 9.082608695652175e-06,
"loss": 0.179,
"step": 825
},
{
"epoch": 0.17,
"learning_rate": 9.028260869565218e-06,
"loss": 0.1626,
"step": 850
},
{
"epoch": 0.17,
"learning_rate": 8.973913043478262e-06,
"loss": 0.1576,
"step": 875
},
{
"epoch": 0.18,
"learning_rate": 8.919565217391305e-06,
"loss": 0.1596,
"step": 900
},
{
"epoch": 0.18,
"learning_rate": 8.865217391304349e-06,
"loss": 0.1498,
"step": 925
},
{
"epoch": 0.19,
"learning_rate": 8.810869565217392e-06,
"loss": 0.1488,
"step": 950
},
{
"epoch": 0.2,
"learning_rate": 8.756521739130436e-06,
"loss": 0.1523,
"step": 975
},
{
"epoch": 0.2,
"learning_rate": 8.702173913043479e-06,
"loss": 0.1542,
"step": 1000
},
{
"epoch": 0.2,
"eval_loss": 0.47108975052833557,
"eval_runtime": 3285.6824,
"eval_samples_per_second": 3.74,
"eval_steps_per_second": 0.117,
"eval_wer": 30.84127559938001,
"step": 1000
},
{
"epoch": 0.2,
"learning_rate": 8.647826086956523e-06,
"loss": 0.1321,
"step": 1025
},
{
"epoch": 0.21,
"learning_rate": 8.593478260869566e-06,
"loss": 0.1699,
"step": 1050
},
{
"epoch": 0.21,
"learning_rate": 8.53913043478261e-06,
"loss": 0.1581,
"step": 1075
},
{
"epoch": 0.22,
"learning_rate": 8.484782608695653e-06,
"loss": 0.1378,
"step": 1100
},
{
"epoch": 0.23,
"learning_rate": 8.430434782608696e-06,
"loss": 0.1242,
"step": 1125
},
{
"epoch": 0.23,
"learning_rate": 8.37608695652174e-06,
"loss": 0.1402,
"step": 1150
},
{
"epoch": 0.23,
"learning_rate": 8.321739130434783e-06,
"loss": 0.1506,
"step": 1175
},
{
"epoch": 0.24,
"learning_rate": 8.267391304347827e-06,
"loss": 0.1558,
"step": 1200
},
{
"epoch": 0.24,
"learning_rate": 8.21304347826087e-06,
"loss": 0.1549,
"step": 1225
},
{
"epoch": 0.25,
"learning_rate": 8.158695652173914e-06,
"loss": 0.1686,
"step": 1250
},
{
"epoch": 0.26,
"learning_rate": 8.104347826086957e-06,
"loss": 0.1807,
"step": 1275
},
{
"epoch": 0.26,
"learning_rate": 8.050000000000001e-06,
"loss": 0.1452,
"step": 1300
},
{
"epoch": 0.27,
"learning_rate": 7.995652173913044e-06,
"loss": 0.1507,
"step": 1325
},
{
"epoch": 0.27,
"learning_rate": 7.941304347826088e-06,
"loss": 0.162,
"step": 1350
},
{
"epoch": 0.28,
"learning_rate": 7.886956521739131e-06,
"loss": 0.1495,
"step": 1375
},
{
"epoch": 0.28,
"learning_rate": 7.832608695652175e-06,
"loss": 0.1588,
"step": 1400
},
{
"epoch": 0.28,
"learning_rate": 7.778260869565218e-06,
"loss": 0.1478,
"step": 1425
},
{
"epoch": 0.29,
"learning_rate": 7.723913043478262e-06,
"loss": 0.134,
"step": 1450
},
{
"epoch": 0.29,
"learning_rate": 7.669565217391305e-06,
"loss": 0.1453,
"step": 1475
},
{
"epoch": 0.3,
"learning_rate": 7.615217391304348e-06,
"loss": 0.1544,
"step": 1500
},
{
"epoch": 0.3,
"learning_rate": 7.560869565217392e-06,
"loss": 0.1301,
"step": 1525
},
{
"epoch": 0.31,
"learning_rate": 7.506521739130436e-06,
"loss": 0.1407,
"step": 1550
},
{
"epoch": 0.32,
"learning_rate": 7.452173913043479e-06,
"loss": 0.1505,
"step": 1575
},
{
"epoch": 0.32,
"learning_rate": 7.397826086956522e-06,
"loss": 0.1309,
"step": 1600
},
{
"epoch": 0.33,
"learning_rate": 7.343478260869565e-06,
"loss": 0.155,
"step": 1625
},
{
"epoch": 0.33,
"learning_rate": 7.28913043478261e-06,
"loss": 0.1291,
"step": 1650
},
{
"epoch": 0.34,
"learning_rate": 7.234782608695653e-06,
"loss": 0.116,
"step": 1675
},
{
"epoch": 0.34,
"learning_rate": 7.180434782608697e-06,
"loss": 0.1275,
"step": 1700
},
{
"epoch": 0.34,
"learning_rate": 7.126086956521739e-06,
"loss": 0.1059,
"step": 1725
},
{
"epoch": 0.35,
"learning_rate": 7.071739130434783e-06,
"loss": 0.094,
"step": 1750
},
{
"epoch": 0.35,
"learning_rate": 7.017391304347827e-06,
"loss": 0.1075,
"step": 1775
},
{
"epoch": 0.36,
"learning_rate": 6.9630434782608705e-06,
"loss": 0.1178,
"step": 1800
},
{
"epoch": 0.36,
"learning_rate": 6.908695652173913e-06,
"loss": 0.0955,
"step": 1825
},
{
"epoch": 0.37,
"learning_rate": 6.854347826086957e-06,
"loss": 0.0933,
"step": 1850
},
{
"epoch": 0.38,
"learning_rate": 6.800000000000001e-06,
"loss": 0.0919,
"step": 1875
},
{
"epoch": 0.38,
"learning_rate": 6.7456521739130444e-06,
"loss": 0.105,
"step": 1900
},
{
"epoch": 0.39,
"learning_rate": 6.691304347826088e-06,
"loss": 0.1006,
"step": 1925
},
{
"epoch": 0.39,
"learning_rate": 6.6369565217391306e-06,
"loss": 0.1194,
"step": 1950
},
{
"epoch": 0.4,
"learning_rate": 6.582608695652174e-06,
"loss": 0.1114,
"step": 1975
},
{
"epoch": 0.4,
"learning_rate": 6.528260869565218e-06,
"loss": 0.0976,
"step": 2000
},
{
"epoch": 0.4,
"eval_loss": 0.40396809577941895,
"eval_runtime": 3350.4547,
"eval_samples_per_second": 3.668,
"eval_steps_per_second": 0.115,
"eval_wer": 26.64643999604261,
"step": 2000
},
{
"epoch": 0.41,
"learning_rate": 6.473913043478262e-06,
"loss": 0.0989,
"step": 2025
},
{
"epoch": 0.41,
"learning_rate": 6.4195652173913045e-06,
"loss": 0.0948,
"step": 2050
},
{
"epoch": 0.41,
"learning_rate": 6.365217391304348e-06,
"loss": 0.0896,
"step": 2075
},
{
"epoch": 0.42,
"learning_rate": 6.310869565217392e-06,
"loss": 0.0883,
"step": 2100
},
{
"epoch": 0.42,
"learning_rate": 6.256521739130436e-06,
"loss": 0.0922,
"step": 2125
},
{
"epoch": 0.43,
"learning_rate": 6.2021739130434784e-06,
"loss": 0.0927,
"step": 2150
},
{
"epoch": 0.43,
"learning_rate": 6.147826086956522e-06,
"loss": 0.0889,
"step": 2175
},
{
"epoch": 0.44,
"learning_rate": 6.093478260869565e-06,
"loss": 0.0882,
"step": 2200
},
{
"epoch": 0.45,
"learning_rate": 6.03913043478261e-06,
"loss": 0.0907,
"step": 2225
},
{
"epoch": 0.45,
"learning_rate": 5.984782608695653e-06,
"loss": 0.0851,
"step": 2250
},
{
"epoch": 0.46,
"learning_rate": 5.930434782608696e-06,
"loss": 0.0914,
"step": 2275
},
{
"epoch": 0.46,
"learning_rate": 5.876086956521739e-06,
"loss": 0.0986,
"step": 2300
},
{
"epoch": 0.47,
"learning_rate": 5.821739130434783e-06,
"loss": 0.0868,
"step": 2325
},
{
"epoch": 0.47,
"learning_rate": 5.767391304347827e-06,
"loss": 0.0888,
"step": 2350
},
{
"epoch": 0.47,
"learning_rate": 5.71304347826087e-06,
"loss": 0.0845,
"step": 2375
},
{
"epoch": 0.48,
"learning_rate": 5.658695652173913e-06,
"loss": 0.0816,
"step": 2400
},
{
"epoch": 0.48,
"learning_rate": 5.604347826086957e-06,
"loss": 0.0739,
"step": 2425
},
{
"epoch": 0.49,
"learning_rate": 5.550000000000001e-06,
"loss": 0.0784,
"step": 2450
},
{
"epoch": 0.49,
"learning_rate": 5.4956521739130446e-06,
"loss": 0.0723,
"step": 2475
},
{
"epoch": 0.5,
"learning_rate": 5.441304347826087e-06,
"loss": 0.0762,
"step": 2500
},
{
"epoch": 0.51,
"learning_rate": 5.386956521739131e-06,
"loss": 0.0846,
"step": 2525
},
{
"epoch": 0.51,
"learning_rate": 5.332608695652174e-06,
"loss": 0.0774,
"step": 2550
},
{
"epoch": 0.52,
"learning_rate": 5.2782608695652185e-06,
"loss": 0.0701,
"step": 2575
},
{
"epoch": 0.52,
"learning_rate": 5.223913043478261e-06,
"loss": 0.0734,
"step": 2600
},
{
"epoch": 0.53,
"learning_rate": 5.169565217391305e-06,
"loss": 0.073,
"step": 2625
},
{
"epoch": 0.53,
"learning_rate": 5.115217391304348e-06,
"loss": 0.0715,
"step": 2650
},
{
"epoch": 0.54,
"learning_rate": 5.060869565217392e-06,
"loss": 0.0836,
"step": 2675
},
{
"epoch": 0.54,
"learning_rate": 5.006521739130436e-06,
"loss": 0.0844,
"step": 2700
},
{
"epoch": 0.55,
"learning_rate": 4.9521739130434785e-06,
"loss": 0.0784,
"step": 2725
},
{
"epoch": 0.55,
"learning_rate": 4.897826086956522e-06,
"loss": 0.0722,
"step": 2750
},
{
"epoch": 0.56,
"learning_rate": 4.8434782608695655e-06,
"loss": 0.0659,
"step": 2775
},
{
"epoch": 0.56,
"learning_rate": 4.789130434782609e-06,
"loss": 0.0651,
"step": 2800
},
{
"epoch": 0.56,
"learning_rate": 4.7347826086956525e-06,
"loss": 0.0699,
"step": 2825
},
{
"epoch": 0.57,
"learning_rate": 4.680434782608696e-06,
"loss": 0.0634,
"step": 2850
},
{
"epoch": 0.57,
"learning_rate": 4.6260869565217394e-06,
"loss": 0.0683,
"step": 2875
},
{
"epoch": 0.58,
"learning_rate": 4.571739130434783e-06,
"loss": 0.0671,
"step": 2900
},
{
"epoch": 0.58,
"learning_rate": 4.517391304347826e-06,
"loss": 0.0654,
"step": 2925
},
{
"epoch": 0.59,
"learning_rate": 4.46304347826087e-06,
"loss": 0.0674,
"step": 2950
},
{
"epoch": 0.59,
"learning_rate": 4.408695652173913e-06,
"loss": 0.0733,
"step": 2975
},
{
"epoch": 1.0,
"learning_rate": 4.354347826086957e-06,
"loss": 0.1088,
"step": 3000
},
{
"epoch": 1.0,
"eval_loss": 0.3765240013599396,
"eval_runtime": 3402.7159,
"eval_samples_per_second": 3.612,
"eval_steps_per_second": 0.113,
"eval_wer": 24.495157691081577,
"step": 3000
},
{
"epoch": 1.01,
"learning_rate": 4.3e-06,
"loss": 0.1342,
"step": 3025
},
{
"epoch": 1.01,
"learning_rate": 4.245652173913044e-06,
"loss": 0.1371,
"step": 3050
},
{
"epoch": 1.02,
"learning_rate": 4.191304347826087e-06,
"loss": 0.1497,
"step": 3075
},
{
"epoch": 1.02,
"learning_rate": 4.136956521739131e-06,
"loss": 0.108,
"step": 3100
},
{
"epoch": 1.03,
"learning_rate": 4.082608695652174e-06,
"loss": 0.0717,
"step": 3125
},
{
"epoch": 1.03,
"learning_rate": 4.028260869565218e-06,
"loss": 0.0707,
"step": 3150
},
{
"epoch": 1.04,
"learning_rate": 3.973913043478261e-06,
"loss": 0.0633,
"step": 3175
},
{
"epoch": 1.04,
"learning_rate": 3.919565217391305e-06,
"loss": 0.0575,
"step": 3200
},
{
"epoch": 1.05,
"learning_rate": 3.865217391304348e-06,
"loss": 0.0708,
"step": 3225
},
{
"epoch": 1.05,
"learning_rate": 3.8108695652173917e-06,
"loss": 0.0629,
"step": 3250
},
{
"epoch": 1.06,
"learning_rate": 3.7565217391304347e-06,
"loss": 0.0686,
"step": 3275
},
{
"epoch": 1.06,
"learning_rate": 3.7021739130434786e-06,
"loss": 0.0869,
"step": 3300
},
{
"epoch": 1.07,
"learning_rate": 3.6478260869565217e-06,
"loss": 0.0747,
"step": 3325
},
{
"epoch": 1.07,
"learning_rate": 3.5934782608695656e-06,
"loss": 0.0679,
"step": 3350
},
{
"epoch": 1.08,
"learning_rate": 3.5391304347826087e-06,
"loss": 0.0665,
"step": 3375
},
{
"epoch": 1.08,
"learning_rate": 3.4847826086956526e-06,
"loss": 0.0754,
"step": 3400
},
{
"epoch": 1.09,
"learning_rate": 3.430434782608696e-06,
"loss": 0.0687,
"step": 3425
},
{
"epoch": 1.09,
"learning_rate": 3.3760869565217395e-06,
"loss": 0.0727,
"step": 3450
},
{
"epoch": 1.1,
"learning_rate": 3.321739130434783e-06,
"loss": 0.0672,
"step": 3475
},
{
"epoch": 1.1,
"learning_rate": 3.267391304347826e-06,
"loss": 0.0695,
"step": 3500
},
{
"epoch": 1.11,
"learning_rate": 3.21304347826087e-06,
"loss": 0.0662,
"step": 3525
},
{
"epoch": 1.11,
"learning_rate": 3.158695652173913e-06,
"loss": 0.0811,
"step": 3550
},
{
"epoch": 1.12,
"learning_rate": 3.104347826086957e-06,
"loss": 0.0959,
"step": 3575
},
{
"epoch": 1.12,
"learning_rate": 3.05e-06,
"loss": 0.088,
"step": 3600
},
{
"epoch": 1.13,
"learning_rate": 2.995652173913044e-06,
"loss": 0.0824,
"step": 3625
},
{
"epoch": 1.14,
"learning_rate": 2.9413043478260874e-06,
"loss": 0.0781,
"step": 3650
},
{
"epoch": 1.14,
"learning_rate": 2.8869565217391305e-06,
"loss": 0.0816,
"step": 3675
},
{
"epoch": 1.15,
"learning_rate": 2.8326086956521744e-06,
"loss": 0.0946,
"step": 3700
},
{
"epoch": 1.15,
"learning_rate": 2.7782608695652174e-06,
"loss": 0.0799,
"step": 3725
},
{
"epoch": 1.16,
"learning_rate": 2.7239130434782613e-06,
"loss": 0.0611,
"step": 3750
},
{
"epoch": 1.16,
"learning_rate": 2.6695652173913044e-06,
"loss": 0.0561,
"step": 3775
},
{
"epoch": 1.17,
"learning_rate": 2.6152173913043483e-06,
"loss": 0.0679,
"step": 3800
},
{
"epoch": 1.17,
"learning_rate": 2.5608695652173914e-06,
"loss": 0.0603,
"step": 3825
},
{
"epoch": 1.18,
"learning_rate": 2.506521739130435e-06,
"loss": 0.0561,
"step": 3850
},
{
"epoch": 1.18,
"learning_rate": 2.4521739130434783e-06,
"loss": 0.0633,
"step": 3875
},
{
"epoch": 1.19,
"learning_rate": 2.397826086956522e-06,
"loss": 0.0545,
"step": 3900
},
{
"epoch": 1.19,
"learning_rate": 2.3434782608695657e-06,
"loss": 0.0545,
"step": 3925
},
{
"epoch": 1.2,
"learning_rate": 2.2891304347826088e-06,
"loss": 0.0603,
"step": 3950
},
{
"epoch": 1.2,
"learning_rate": 2.2347826086956523e-06,
"loss": 0.0664,
"step": 3975
},
{
"epoch": 1.21,
"learning_rate": 2.1804347826086957e-06,
"loss": 0.0527,
"step": 4000
},
{
"epoch": 1.21,
"eval_loss": 0.38717222213745117,
"eval_runtime": 3407.8282,
"eval_samples_per_second": 3.606,
"eval_steps_per_second": 0.113,
"eval_wer": 23.650914047642605,
"step": 4000
},
{
"epoch": 1.21,
"learning_rate": 2.1260869565217392e-06,
"loss": 0.0634,
"step": 4025
},
{
"epoch": 1.22,
"learning_rate": 2.0717391304347827e-06,
"loss": 0.0657,
"step": 4050
},
{
"epoch": 1.22,
"learning_rate": 2.017391304347826e-06,
"loss": 0.0571,
"step": 4075
},
{
"epoch": 1.23,
"learning_rate": 1.9630434782608697e-06,
"loss": 0.0526,
"step": 4100
},
{
"epoch": 1.23,
"learning_rate": 1.908695652173913e-06,
"loss": 0.063,
"step": 4125
},
{
"epoch": 1.23,
"learning_rate": 1.8543478260869566e-06,
"loss": 0.065,
"step": 4150
},
{
"epoch": 1.24,
"learning_rate": 1.8000000000000001e-06,
"loss": 0.0682,
"step": 4175
},
{
"epoch": 1.25,
"learning_rate": 1.7456521739130436e-06,
"loss": 0.0691,
"step": 4200
},
{
"epoch": 1.25,
"learning_rate": 1.691304347826087e-06,
"loss": 0.0707,
"step": 4225
},
{
"epoch": 1.25,
"learning_rate": 1.6369565217391306e-06,
"loss": 0.0798,
"step": 4250
},
{
"epoch": 1.26,
"learning_rate": 1.582608695652174e-06,
"loss": 0.0668,
"step": 4275
},
{
"epoch": 1.27,
"learning_rate": 1.5282608695652175e-06,
"loss": 0.0696,
"step": 4300
},
{
"epoch": 1.27,
"learning_rate": 1.4739130434782608e-06,
"loss": 0.0749,
"step": 4325
},
{
"epoch": 1.27,
"learning_rate": 1.4195652173913043e-06,
"loss": 0.0744,
"step": 4350
},
{
"epoch": 1.28,
"learning_rate": 1.3652173913043478e-06,
"loss": 0.0808,
"step": 4375
},
{
"epoch": 1.28,
"learning_rate": 1.3108695652173915e-06,
"loss": 0.0738,
"step": 4400
},
{
"epoch": 1.29,
"learning_rate": 1.256521739130435e-06,
"loss": 0.0627,
"step": 4425
},
{
"epoch": 1.29,
"learning_rate": 1.2021739130434784e-06,
"loss": 0.0763,
"step": 4450
},
{
"epoch": 1.3,
"learning_rate": 1.1478260869565217e-06,
"loss": 0.0773,
"step": 4475
},
{
"epoch": 1.3,
"learning_rate": 1.0934782608695652e-06,
"loss": 0.0614,
"step": 4500
},
{
"epoch": 1.31,
"learning_rate": 1.0391304347826089e-06,
"loss": 0.0721,
"step": 4525
},
{
"epoch": 1.31,
"learning_rate": 9.847826086956524e-07,
"loss": 0.0816,
"step": 4550
},
{
"epoch": 1.32,
"learning_rate": 9.304347826086956e-07,
"loss": 0.0741,
"step": 4575
},
{
"epoch": 1.32,
"learning_rate": 8.760869565217392e-07,
"loss": 0.0777,
"step": 4600
},
{
"epoch": 1.33,
"learning_rate": 8.217391304347827e-07,
"loss": 0.0677,
"step": 4625
},
{
"epoch": 1.33,
"learning_rate": 7.673913043478262e-07,
"loss": 0.0656,
"step": 4650
},
{
"epoch": 1.34,
"learning_rate": 7.130434782608696e-07,
"loss": 0.0642,
"step": 4675
},
{
"epoch": 1.34,
"learning_rate": 6.586956521739131e-07,
"loss": 0.057,
"step": 4700
},
{
"epoch": 1.35,
"learning_rate": 6.043478260869565e-07,
"loss": 0.0461,
"step": 4725
},
{
"epoch": 1.35,
"learning_rate": 5.5e-07,
"loss": 0.051,
"step": 4750
},
{
"epoch": 1.36,
"learning_rate": 4.956521739130435e-07,
"loss": 0.0532,
"step": 4775
},
{
"epoch": 1.36,
"learning_rate": 4.41304347826087e-07,
"loss": 0.0494,
"step": 4800
},
{
"epoch": 1.37,
"learning_rate": 3.869565217391304e-07,
"loss": 0.0483,
"step": 4825
},
{
"epoch": 1.38,
"learning_rate": 3.3260869565217396e-07,
"loss": 0.052,
"step": 4850
},
{
"epoch": 1.38,
"learning_rate": 2.7826086956521744e-07,
"loss": 0.0523,
"step": 4875
},
{
"epoch": 1.39,
"learning_rate": 2.239130434782609e-07,
"loss": 0.0516,
"step": 4900
},
{
"epoch": 1.39,
"learning_rate": 1.6956521739130438e-07,
"loss": 0.0622,
"step": 4925
},
{
"epoch": 1.4,
"learning_rate": 1.1521739130434784e-07,
"loss": 0.0548,
"step": 4950
},
{
"epoch": 1.4,
"learning_rate": 6.086956521739131e-08,
"loss": 0.0544,
"step": 4975
},
{
"epoch": 1.41,
"learning_rate": 6.521739130434784e-09,
"loss": 0.0534,
"step": 5000
},
{
"epoch": 1.41,
"eval_loss": 0.38427138328552246,
"eval_runtime": 3393.0078,
"eval_samples_per_second": 3.622,
"eval_steps_per_second": 0.113,
"eval_wer": 23.68169376380965,
"step": 5000
},
{
"epoch": 1.41,
"step": 5000,
"total_flos": 9.234415366299648e+19,
"train_loss": 0.18115086715221404,
"train_runtime": 79111.1631,
"train_samples_per_second": 4.045,
"train_steps_per_second": 0.063
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 9.234415366299648e+19,
"trial_name": null,
"trial_params": null
}