whisper-md-sr / trainer_state.json
emilios's picture
End of training
e4c7b61
{
"best_metric": 12.140833670578713,
"best_model_checkpoint": "./checkpoint-7000",
"epoch": 27.17391304347826,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.07,
"learning_rate": 1.4921499816584212e-06,
"loss": 0.9762,
"step": 25
},
{
"epoch": 0.14,
"learning_rate": 1.8585955256072678e-06,
"loss": 0.5122,
"step": 50
},
{
"epoch": 0.2,
"learning_rate": 2.0644903353287132e-06,
"loss": 0.3223,
"step": 75
},
{
"epoch": 0.27,
"learning_rate": 2.208366596599153e-06,
"loss": 0.2527,
"step": 100
},
{
"epoch": 0.34,
"learning_rate": 2.319062265222936e-06,
"loss": 0.2249,
"step": 125
},
{
"epoch": 0.41,
"learning_rate": 2.409049375798488e-06,
"loss": 0.2012,
"step": 150
},
{
"epoch": 0.48,
"learning_rate": 2.4848684238610323e-06,
"loss": 0.1792,
"step": 175
},
{
"epoch": 0.54,
"learning_rate": 2.5503798365399707e-06,
"loss": 0.1666,
"step": 200
},
{
"epoch": 0.61,
"learning_rate": 2.608053780532331e-06,
"loss": 0.1587,
"step": 225
},
{
"epoch": 0.68,
"learning_rate": 2.65956674147807e-06,
"loss": 0.145,
"step": 250
},
{
"epoch": 0.75,
"learning_rate": 2.7061088861190967e-06,
"loss": 0.1292,
"step": 275
},
{
"epoch": 0.82,
"learning_rate": 2.748555684588509e-06,
"loss": 0.1351,
"step": 300
},
{
"epoch": 0.88,
"learning_rate": 2.787569925934908e-06,
"loss": 0.1442,
"step": 325
},
{
"epoch": 0.95,
"learning_rate": 2.8236654768779034e-06,
"loss": 0.1261,
"step": 350
},
{
"epoch": 1.02,
"learning_rate": 2.8572488050095434e-06,
"loss": 0.1141,
"step": 375
},
{
"epoch": 1.09,
"learning_rate": 2.8886469682005057e-06,
"loss": 0.0723,
"step": 400
},
{
"epoch": 1.15,
"learning_rate": 2.91812704114235e-06,
"loss": 0.0762,
"step": 425
},
{
"epoch": 1.22,
"learning_rate": 2.94590994217138e-06,
"loss": 0.0748,
"step": 450
},
{
"epoch": 1.29,
"learning_rate": 2.972180492193256e-06,
"loss": 0.0565,
"step": 475
},
{
"epoch": 1.36,
"learning_rate": 2.997094874416736e-06,
"loss": 0.0704,
"step": 500
},
{
"epoch": 1.43,
"learning_rate": 3e-06,
"loss": 0.0586,
"step": 525
},
{
"epoch": 1.49,
"learning_rate": 3e-06,
"loss": 0.0634,
"step": 550
},
{
"epoch": 1.56,
"learning_rate": 3e-06,
"loss": 0.0675,
"step": 575
},
{
"epoch": 1.63,
"learning_rate": 3e-06,
"loss": 0.0605,
"step": 600
},
{
"epoch": 1.7,
"learning_rate": 3e-06,
"loss": 0.0586,
"step": 625
},
{
"epoch": 1.77,
"learning_rate": 3e-06,
"loss": 0.0562,
"step": 650
},
{
"epoch": 1.83,
"learning_rate": 3e-06,
"loss": 0.0575,
"step": 675
},
{
"epoch": 1.9,
"learning_rate": 3e-06,
"loss": 0.0602,
"step": 700
},
{
"epoch": 1.97,
"learning_rate": 3e-06,
"loss": 0.0545,
"step": 725
},
{
"epoch": 2.04,
"learning_rate": 3e-06,
"loss": 0.0388,
"step": 750
},
{
"epoch": 2.11,
"learning_rate": 3e-06,
"loss": 0.0248,
"step": 775
},
{
"epoch": 2.17,
"learning_rate": 3e-06,
"loss": 0.0254,
"step": 800
},
{
"epoch": 2.24,
"learning_rate": 3e-06,
"loss": 0.0239,
"step": 825
},
{
"epoch": 2.31,
"learning_rate": 3e-06,
"loss": 0.0224,
"step": 850
},
{
"epoch": 2.38,
"learning_rate": 3e-06,
"loss": 0.0262,
"step": 875
},
{
"epoch": 2.45,
"learning_rate": 3e-06,
"loss": 0.0263,
"step": 900
},
{
"epoch": 2.51,
"learning_rate": 3e-06,
"loss": 0.022,
"step": 925
},
{
"epoch": 2.58,
"learning_rate": 3e-06,
"loss": 0.0224,
"step": 950
},
{
"epoch": 2.65,
"learning_rate": 3e-06,
"loss": 0.0203,
"step": 975
},
{
"epoch": 2.72,
"learning_rate": 3e-06,
"loss": 0.0222,
"step": 1000
},
{
"epoch": 2.72,
"eval_loss": 0.34423828125,
"eval_runtime": 89.2882,
"eval_samples_per_second": 7.582,
"eval_steps_per_second": 0.482,
"eval_wer": 14.083367057871307,
"step": 1000
},
{
"epoch": 2.79,
"learning_rate": 3e-06,
"loss": 0.0178,
"step": 1025
},
{
"epoch": 2.85,
"learning_rate": 3e-06,
"loss": 0.0254,
"step": 1050
},
{
"epoch": 2.92,
"learning_rate": 3e-06,
"loss": 0.0254,
"step": 1075
},
{
"epoch": 2.99,
"learning_rate": 3e-06,
"loss": 0.0196,
"step": 1100
},
{
"epoch": 3.06,
"learning_rate": 3e-06,
"loss": 0.0111,
"step": 1125
},
{
"epoch": 3.12,
"learning_rate": 3e-06,
"loss": 0.0095,
"step": 1150
},
{
"epoch": 3.19,
"learning_rate": 3e-06,
"loss": 0.0086,
"step": 1175
},
{
"epoch": 3.26,
"learning_rate": 3e-06,
"loss": 0.0089,
"step": 1200
},
{
"epoch": 3.33,
"learning_rate": 3e-06,
"loss": 0.0101,
"step": 1225
},
{
"epoch": 3.4,
"learning_rate": 3e-06,
"loss": 0.0091,
"step": 1250
},
{
"epoch": 3.46,
"learning_rate": 3e-06,
"loss": 0.0104,
"step": 1275
},
{
"epoch": 3.53,
"learning_rate": 3e-06,
"loss": 0.0089,
"step": 1300
},
{
"epoch": 3.6,
"learning_rate": 3e-06,
"loss": 0.0106,
"step": 1325
},
{
"epoch": 3.67,
"learning_rate": 3e-06,
"loss": 0.0079,
"step": 1350
},
{
"epoch": 3.74,
"learning_rate": 3e-06,
"loss": 0.011,
"step": 1375
},
{
"epoch": 3.8,
"learning_rate": 3e-06,
"loss": 0.0087,
"step": 1400
},
{
"epoch": 3.87,
"learning_rate": 3e-06,
"loss": 0.0097,
"step": 1425
},
{
"epoch": 3.94,
"learning_rate": 3e-06,
"loss": 0.0081,
"step": 1450
},
{
"epoch": 4.01,
"learning_rate": 3e-06,
"loss": 0.01,
"step": 1475
},
{
"epoch": 4.08,
"learning_rate": 3e-06,
"loss": 0.0069,
"step": 1500
},
{
"epoch": 4.14,
"learning_rate": 3e-06,
"loss": 0.0049,
"step": 1525
},
{
"epoch": 4.21,
"learning_rate": 3e-06,
"loss": 0.0046,
"step": 1550
},
{
"epoch": 4.28,
"learning_rate": 3e-06,
"loss": 0.004,
"step": 1575
},
{
"epoch": 4.35,
"learning_rate": 3e-06,
"loss": 0.0043,
"step": 1600
},
{
"epoch": 4.42,
"learning_rate": 3e-06,
"loss": 0.0042,
"step": 1625
},
{
"epoch": 4.48,
"learning_rate": 3e-06,
"loss": 0.0042,
"step": 1650
},
{
"epoch": 4.55,
"learning_rate": 3e-06,
"loss": 0.0056,
"step": 1675
},
{
"epoch": 4.62,
"learning_rate": 3e-06,
"loss": 0.0041,
"step": 1700
},
{
"epoch": 4.69,
"learning_rate": 3e-06,
"loss": 0.0039,
"step": 1725
},
{
"epoch": 4.76,
"learning_rate": 3e-06,
"loss": 0.0039,
"step": 1750
},
{
"epoch": 4.82,
"learning_rate": 3e-06,
"loss": 0.0036,
"step": 1775
},
{
"epoch": 4.89,
"learning_rate": 3e-06,
"loss": 0.0046,
"step": 1800
},
{
"epoch": 4.96,
"learning_rate": 3e-06,
"loss": 0.0049,
"step": 1825
},
{
"epoch": 5.03,
"learning_rate": 3e-06,
"loss": 0.0046,
"step": 1850
},
{
"epoch": 5.1,
"learning_rate": 3e-06,
"loss": 0.0031,
"step": 1875
},
{
"epoch": 5.16,
"learning_rate": 3e-06,
"loss": 0.0032,
"step": 1900
},
{
"epoch": 5.23,
"learning_rate": 3e-06,
"loss": 0.0031,
"step": 1925
},
{
"epoch": 5.3,
"learning_rate": 3e-06,
"loss": 0.0027,
"step": 1950
},
{
"epoch": 5.37,
"learning_rate": 3e-06,
"loss": 0.0028,
"step": 1975
},
{
"epoch": 5.43,
"learning_rate": 3e-06,
"loss": 0.0032,
"step": 2000
},
{
"epoch": 5.43,
"eval_loss": 0.41064453125,
"eval_runtime": 88.6244,
"eval_samples_per_second": 7.639,
"eval_steps_per_second": 0.485,
"eval_wer": 14.52853095912586,
"step": 2000
},
{
"epoch": 5.5,
"learning_rate": 3e-06,
"loss": 0.0033,
"step": 2025
},
{
"epoch": 5.57,
"learning_rate": 3e-06,
"loss": 0.0032,
"step": 2050
},
{
"epoch": 5.64,
"learning_rate": 3e-06,
"loss": 0.0026,
"step": 2075
},
{
"epoch": 5.71,
"learning_rate": 3e-06,
"loss": 0.0033,
"step": 2100
},
{
"epoch": 5.77,
"learning_rate": 3e-06,
"loss": 0.0025,
"step": 2125
},
{
"epoch": 5.84,
"learning_rate": 3e-06,
"loss": 0.0021,
"step": 2150
},
{
"epoch": 5.91,
"learning_rate": 3e-06,
"loss": 0.0029,
"step": 2175
},
{
"epoch": 5.98,
"learning_rate": 3e-06,
"loss": 0.0024,
"step": 2200
},
{
"epoch": 6.05,
"learning_rate": 3e-06,
"loss": 0.0023,
"step": 2225
},
{
"epoch": 6.11,
"learning_rate": 3e-06,
"loss": 0.0019,
"step": 2250
},
{
"epoch": 6.18,
"learning_rate": 3e-06,
"loss": 0.0026,
"step": 2275
},
{
"epoch": 6.25,
"learning_rate": 3e-06,
"loss": 0.0025,
"step": 2300
},
{
"epoch": 6.32,
"learning_rate": 3e-06,
"loss": 0.0016,
"step": 2325
},
{
"epoch": 6.39,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 2350
},
{
"epoch": 6.45,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 2375
},
{
"epoch": 6.52,
"learning_rate": 3e-06,
"loss": 0.0016,
"step": 2400
},
{
"epoch": 6.59,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 2425
},
{
"epoch": 6.66,
"learning_rate": 3e-06,
"loss": 0.0016,
"step": 2450
},
{
"epoch": 6.73,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 2475
},
{
"epoch": 6.79,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 2500
},
{
"epoch": 6.86,
"learning_rate": 3e-06,
"loss": 0.0016,
"step": 2525
},
{
"epoch": 6.93,
"learning_rate": 3e-06,
"loss": 0.0017,
"step": 2550
},
{
"epoch": 7.0,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 2575
},
{
"epoch": 7.07,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 2600
},
{
"epoch": 7.13,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 2625
},
{
"epoch": 7.2,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 2650
},
{
"epoch": 7.27,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 2675
},
{
"epoch": 7.34,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 2700
},
{
"epoch": 7.4,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 2725
},
{
"epoch": 7.47,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 2750
},
{
"epoch": 7.54,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 2775
},
{
"epoch": 7.61,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 2800
},
{
"epoch": 7.68,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 2825
},
{
"epoch": 7.74,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 2850
},
{
"epoch": 7.81,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 2875
},
{
"epoch": 7.88,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 2900
},
{
"epoch": 7.95,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 2925
},
{
"epoch": 8.02,
"learning_rate": 3e-06,
"loss": 0.0019,
"step": 2950
},
{
"epoch": 8.08,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 2975
},
{
"epoch": 8.15,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 3000
},
{
"epoch": 8.15,
"eval_loss": 0.43310546875,
"eval_runtime": 89.2849,
"eval_samples_per_second": 7.582,
"eval_steps_per_second": 0.482,
"eval_wer": 12.869283690813434,
"step": 3000
},
{
"epoch": 8.22,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 3025
},
{
"epoch": 8.29,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 3050
},
{
"epoch": 8.36,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 3075
},
{
"epoch": 8.42,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 3100
},
{
"epoch": 8.49,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 3125
},
{
"epoch": 8.56,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 3150
},
{
"epoch": 8.63,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 3175
},
{
"epoch": 8.7,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 3200
},
{
"epoch": 8.76,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 3225
},
{
"epoch": 8.83,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 3250
},
{
"epoch": 8.9,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 3275
},
{
"epoch": 8.97,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 3300
},
{
"epoch": 9.04,
"learning_rate": 3e-06,
"loss": 0.0017,
"step": 3325
},
{
"epoch": 9.1,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 3350
},
{
"epoch": 9.17,
"learning_rate": 3e-06,
"loss": 0.0022,
"step": 3375
},
{
"epoch": 9.24,
"learning_rate": 3e-06,
"loss": 0.0019,
"step": 3400
},
{
"epoch": 9.31,
"learning_rate": 3e-06,
"loss": 0.0023,
"step": 3425
},
{
"epoch": 9.38,
"learning_rate": 3e-06,
"loss": 0.003,
"step": 3450
},
{
"epoch": 9.44,
"learning_rate": 3e-06,
"loss": 0.0036,
"step": 3475
},
{
"epoch": 9.51,
"learning_rate": 3e-06,
"loss": 0.0025,
"step": 3500
},
{
"epoch": 9.58,
"learning_rate": 3e-06,
"loss": 0.0027,
"step": 3525
},
{
"epoch": 9.65,
"learning_rate": 3e-06,
"loss": 0.0036,
"step": 3550
},
{
"epoch": 9.71,
"learning_rate": 3e-06,
"loss": 0.0038,
"step": 3575
},
{
"epoch": 9.78,
"learning_rate": 3e-06,
"loss": 0.0052,
"step": 3600
},
{
"epoch": 9.85,
"learning_rate": 3e-06,
"loss": 0.0026,
"step": 3625
},
{
"epoch": 9.92,
"learning_rate": 3e-06,
"loss": 0.0033,
"step": 3650
},
{
"epoch": 9.99,
"learning_rate": 3e-06,
"loss": 0.0031,
"step": 3675
},
{
"epoch": 10.05,
"learning_rate": 3e-06,
"loss": 0.0022,
"step": 3700
},
{
"epoch": 10.12,
"learning_rate": 3e-06,
"loss": 0.0018,
"step": 3725
},
{
"epoch": 10.19,
"learning_rate": 3e-06,
"loss": 0.0027,
"step": 3750
},
{
"epoch": 10.26,
"learning_rate": 3e-06,
"loss": 0.0037,
"step": 3775
},
{
"epoch": 10.33,
"learning_rate": 3e-06,
"loss": 0.0048,
"step": 3800
},
{
"epoch": 10.39,
"learning_rate": 3e-06,
"loss": 0.0044,
"step": 3825
},
{
"epoch": 10.46,
"learning_rate": 3e-06,
"loss": 0.0033,
"step": 3850
},
{
"epoch": 10.53,
"learning_rate": 3e-06,
"loss": 0.0048,
"step": 3875
},
{
"epoch": 10.6,
"learning_rate": 3e-06,
"loss": 0.0031,
"step": 3900
},
{
"epoch": 10.67,
"learning_rate": 3e-06,
"loss": 0.003,
"step": 3925
},
{
"epoch": 10.73,
"learning_rate": 3e-06,
"loss": 0.0033,
"step": 3950
},
{
"epoch": 10.8,
"learning_rate": 3e-06,
"loss": 0.0023,
"step": 3975
},
{
"epoch": 10.87,
"learning_rate": 3e-06,
"loss": 0.0029,
"step": 4000
},
{
"epoch": 10.87,
"eval_loss": 0.394775390625,
"eval_runtime": 89.4335,
"eval_samples_per_second": 7.57,
"eval_steps_per_second": 0.481,
"eval_wer": 12.626467017401863,
"step": 4000
},
{
"epoch": 10.94,
"learning_rate": 3e-06,
"loss": 0.0019,
"step": 4025
},
{
"epoch": 11.01,
"learning_rate": 3e-06,
"loss": 0.0017,
"step": 4050
},
{
"epoch": 11.07,
"learning_rate": 3e-06,
"loss": 0.0016,
"step": 4075
},
{
"epoch": 11.14,
"learning_rate": 3e-06,
"loss": 0.0021,
"step": 4100
},
{
"epoch": 11.21,
"learning_rate": 3e-06,
"loss": 0.0016,
"step": 4125
},
{
"epoch": 11.28,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 4150
},
{
"epoch": 11.35,
"learning_rate": 3e-06,
"loss": 0.002,
"step": 4175
},
{
"epoch": 11.41,
"learning_rate": 3e-06,
"loss": 0.0018,
"step": 4200
},
{
"epoch": 11.48,
"learning_rate": 3e-06,
"loss": 0.002,
"step": 4225
},
{
"epoch": 11.55,
"learning_rate": 3e-06,
"loss": 0.0022,
"step": 4250
},
{
"epoch": 11.62,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 4275
},
{
"epoch": 11.68,
"learning_rate": 3e-06,
"loss": 0.0017,
"step": 4300
},
{
"epoch": 11.75,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 4325
},
{
"epoch": 11.82,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 4350
},
{
"epoch": 11.89,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 4375
},
{
"epoch": 11.96,
"learning_rate": 3e-06,
"loss": 0.0034,
"step": 4400
},
{
"epoch": 12.02,
"learning_rate": 3e-06,
"loss": 0.0031,
"step": 4425
},
{
"epoch": 12.09,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 4450
},
{
"epoch": 12.16,
"learning_rate": 3e-06,
"loss": 0.0021,
"step": 4475
},
{
"epoch": 12.23,
"learning_rate": 3e-06,
"loss": 0.002,
"step": 4500
},
{
"epoch": 12.3,
"learning_rate": 3e-06,
"loss": 0.0018,
"step": 4525
},
{
"epoch": 12.36,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 4550
},
{
"epoch": 12.43,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 4575
},
{
"epoch": 12.5,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 4600
},
{
"epoch": 12.57,
"learning_rate": 3e-06,
"loss": 0.0017,
"step": 4625
},
{
"epoch": 12.64,
"learning_rate": 3e-06,
"loss": 0.0018,
"step": 4650
},
{
"epoch": 12.7,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 4675
},
{
"epoch": 12.77,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 4700
},
{
"epoch": 12.84,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 4725
},
{
"epoch": 12.91,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 4750
},
{
"epoch": 12.98,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 4775
},
{
"epoch": 13.04,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 4800
},
{
"epoch": 13.11,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 4825
},
{
"epoch": 13.18,
"learning_rate": 3e-06,
"loss": 0.0018,
"step": 4850
},
{
"epoch": 13.25,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 4875
},
{
"epoch": 13.32,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 4900
},
{
"epoch": 13.38,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 4925
},
{
"epoch": 13.45,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 4950
},
{
"epoch": 13.52,
"learning_rate": 3e-06,
"loss": 0.0018,
"step": 4975
},
{
"epoch": 13.59,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 5000
},
{
"epoch": 13.59,
"eval_loss": 0.451171875,
"eval_runtime": 89.3288,
"eval_samples_per_second": 7.579,
"eval_steps_per_second": 0.481,
"eval_wer": 12.666936462970458,
"step": 5000
},
{
"epoch": 13.65,
"learning_rate": 3e-06,
"loss": 0.0017,
"step": 5025
},
{
"epoch": 13.72,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 5050
},
{
"epoch": 13.79,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 5075
},
{
"epoch": 13.86,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 5100
},
{
"epoch": 13.93,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 5125
},
{
"epoch": 13.99,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 5150
},
{
"epoch": 14.06,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 5175
},
{
"epoch": 14.13,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 5200
},
{
"epoch": 14.2,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 5225
},
{
"epoch": 14.27,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 5250
},
{
"epoch": 14.33,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 5275
},
{
"epoch": 14.4,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 5300
},
{
"epoch": 14.47,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 5325
},
{
"epoch": 14.54,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 5350
},
{
"epoch": 14.61,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 5375
},
{
"epoch": 14.67,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 5400
},
{
"epoch": 14.74,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 5425
},
{
"epoch": 14.81,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 5450
},
{
"epoch": 14.88,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 5475
},
{
"epoch": 14.95,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 5500
},
{
"epoch": 15.01,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 5525
},
{
"epoch": 15.08,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 5550
},
{
"epoch": 15.15,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 5575
},
{
"epoch": 15.22,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 5600
},
{
"epoch": 15.29,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 5625
},
{
"epoch": 15.35,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 5650
},
{
"epoch": 15.42,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 5675
},
{
"epoch": 15.49,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 5700
},
{
"epoch": 15.56,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 5725
},
{
"epoch": 15.62,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 5750
},
{
"epoch": 15.69,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 5775
},
{
"epoch": 15.76,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 5800
},
{
"epoch": 15.83,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 5825
},
{
"epoch": 15.9,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 5850
},
{
"epoch": 15.96,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 5875
},
{
"epoch": 16.03,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 5900
},
{
"epoch": 16.1,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 5925
},
{
"epoch": 16.17,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 5950
},
{
"epoch": 16.24,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 5975
},
{
"epoch": 16.3,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 6000
},
{
"epoch": 16.3,
"eval_loss": 0.489013671875,
"eval_runtime": 89.4348,
"eval_samples_per_second": 7.57,
"eval_steps_per_second": 0.481,
"eval_wer": 12.747875354107649,
"step": 6000
},
{
"epoch": 16.37,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 6025
},
{
"epoch": 16.44,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 6050
},
{
"epoch": 16.51,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 6075
},
{
"epoch": 16.58,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 6100
},
{
"epoch": 16.64,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 6125
},
{
"epoch": 16.71,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 6150
},
{
"epoch": 16.78,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 6175
},
{
"epoch": 16.85,
"learning_rate": 3e-06,
"loss": 0.0017,
"step": 6200
},
{
"epoch": 16.92,
"learning_rate": 3e-06,
"loss": 0.0026,
"step": 6225
},
{
"epoch": 16.98,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 6250
},
{
"epoch": 17.05,
"learning_rate": 3e-06,
"loss": 0.0025,
"step": 6275
},
{
"epoch": 17.12,
"learning_rate": 3e-06,
"loss": 0.0021,
"step": 6300
},
{
"epoch": 17.19,
"learning_rate": 3e-06,
"loss": 0.0017,
"step": 6325
},
{
"epoch": 17.26,
"learning_rate": 3e-06,
"loss": 0.0017,
"step": 6350
},
{
"epoch": 17.32,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 6375
},
{
"epoch": 17.39,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 6400
},
{
"epoch": 17.46,
"learning_rate": 3e-06,
"loss": 0.0019,
"step": 6425
},
{
"epoch": 17.53,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 6450
},
{
"epoch": 17.6,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 6475
},
{
"epoch": 17.66,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 6500
},
{
"epoch": 17.73,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 6525
},
{
"epoch": 17.8,
"learning_rate": 3e-06,
"loss": 0.0018,
"step": 6550
},
{
"epoch": 17.87,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 6575
},
{
"epoch": 17.93,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 6600
},
{
"epoch": 18.0,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 6625
},
{
"epoch": 18.07,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 6650
},
{
"epoch": 18.14,
"learning_rate": 3e-06,
"loss": 0.0019,
"step": 6675
},
{
"epoch": 18.21,
"learning_rate": 3e-06,
"loss": 0.002,
"step": 6700
},
{
"epoch": 18.27,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 6725
},
{
"epoch": 18.34,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 6750
},
{
"epoch": 18.41,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 6775
},
{
"epoch": 18.48,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 6800
},
{
"epoch": 18.55,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 6825
},
{
"epoch": 18.61,
"learning_rate": 3e-06,
"loss": 0.0019,
"step": 6850
},
{
"epoch": 18.68,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 6875
},
{
"epoch": 18.75,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 6900
},
{
"epoch": 18.82,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 6925
},
{
"epoch": 18.89,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 6950
},
{
"epoch": 18.95,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 6975
},
{
"epoch": 19.02,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 7000
},
{
"epoch": 19.02,
"eval_loss": 0.48681640625,
"eval_runtime": 89.5803,
"eval_samples_per_second": 7.557,
"eval_steps_per_second": 0.48,
"eval_wer": 12.140833670578713,
"step": 7000
},
{
"epoch": 19.09,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 7025
},
{
"epoch": 19.16,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 7050
},
{
"epoch": 19.23,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 7075
},
{
"epoch": 19.29,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 7100
},
{
"epoch": 19.36,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 7125
},
{
"epoch": 19.43,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 7150
},
{
"epoch": 19.5,
"learning_rate": 3e-06,
"loss": 0.0022,
"step": 7175
},
{
"epoch": 19.57,
"learning_rate": 3e-06,
"loss": 0.0019,
"step": 7200
},
{
"epoch": 19.63,
"learning_rate": 3e-06,
"loss": 0.0025,
"step": 7225
},
{
"epoch": 19.7,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 7250
},
{
"epoch": 19.77,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 7275
},
{
"epoch": 19.84,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 7300
},
{
"epoch": 19.9,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 7325
},
{
"epoch": 19.97,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 7350
},
{
"epoch": 20.04,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 7375
},
{
"epoch": 20.11,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 7400
},
{
"epoch": 20.18,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 7425
},
{
"epoch": 20.24,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 7450
},
{
"epoch": 20.31,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 7475
},
{
"epoch": 20.38,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 7500
},
{
"epoch": 20.45,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 7525
},
{
"epoch": 20.52,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 7550
},
{
"epoch": 20.58,
"learning_rate": 3e-06,
"loss": 0.0016,
"step": 7575
},
{
"epoch": 20.65,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 7600
},
{
"epoch": 20.72,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 7625
},
{
"epoch": 20.79,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 7650
},
{
"epoch": 20.86,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 7675
},
{
"epoch": 20.92,
"learning_rate": 3e-06,
"loss": 0.001,
"step": 7700
},
{
"epoch": 20.99,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 7725
},
{
"epoch": 21.06,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 7750
},
{
"epoch": 21.13,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 7775
},
{
"epoch": 21.2,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 7800
},
{
"epoch": 21.26,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 7825
},
{
"epoch": 21.33,
"learning_rate": 3e-06,
"loss": 0.0003,
"step": 7850
},
{
"epoch": 21.4,
"learning_rate": 3e-06,
"loss": 0.0003,
"step": 7875
},
{
"epoch": 21.47,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 7900
},
{
"epoch": 21.54,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 7925
},
{
"epoch": 21.6,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 7950
},
{
"epoch": 21.67,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 7975
},
{
"epoch": 21.74,
"learning_rate": 3e-06,
"loss": 0.0016,
"step": 8000
},
{
"epoch": 21.74,
"eval_loss": 0.47802734375,
"eval_runtime": 89.1293,
"eval_samples_per_second": 7.596,
"eval_steps_per_second": 0.482,
"eval_wer": 12.707405908539052,
"step": 8000
},
{
"epoch": 21.81,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 8025
},
{
"epoch": 21.88,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8050
},
{
"epoch": 21.94,
"learning_rate": 3e-06,
"loss": 0.0018,
"step": 8075
},
{
"epoch": 22.01,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 8100
},
{
"epoch": 22.08,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 8125
},
{
"epoch": 22.15,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8150
},
{
"epoch": 22.21,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 8175
},
{
"epoch": 22.28,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 8200
},
{
"epoch": 22.35,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 8225
},
{
"epoch": 22.42,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 8250
},
{
"epoch": 22.49,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 8275
},
{
"epoch": 22.55,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 8300
},
{
"epoch": 22.62,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 8325
},
{
"epoch": 22.69,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 8350
},
{
"epoch": 22.76,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 8375
},
{
"epoch": 22.83,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 8400
},
{
"epoch": 22.89,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 8425
},
{
"epoch": 22.96,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 8450
},
{
"epoch": 23.03,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8475
},
{
"epoch": 23.1,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 8500
},
{
"epoch": 23.17,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 8525
},
{
"epoch": 23.23,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8550
},
{
"epoch": 23.3,
"learning_rate": 3e-06,
"loss": 0.0003,
"step": 8575
},
{
"epoch": 23.37,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8600
},
{
"epoch": 23.44,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 8625
},
{
"epoch": 23.51,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 8650
},
{
"epoch": 23.57,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8675
},
{
"epoch": 23.64,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8700
},
{
"epoch": 23.71,
"learning_rate": 3e-06,
"loss": 0.0029,
"step": 8725
},
{
"epoch": 23.78,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 8750
},
{
"epoch": 23.85,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8775
},
{
"epoch": 23.91,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 8800
},
{
"epoch": 23.98,
"learning_rate": 3e-06,
"loss": 0.0003,
"step": 8825
},
{
"epoch": 24.05,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8850
},
{
"epoch": 24.12,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 8875
},
{
"epoch": 24.18,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 8900
},
{
"epoch": 24.25,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 8925
},
{
"epoch": 24.32,
"learning_rate": 3e-06,
"loss": 0.0003,
"step": 8950
},
{
"epoch": 24.39,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 8975
},
{
"epoch": 24.46,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 9000
},
{
"epoch": 24.46,
"eval_loss": 0.490234375,
"eval_runtime": 89.5861,
"eval_samples_per_second": 7.557,
"eval_steps_per_second": 0.48,
"eval_wer": 12.221772561715905,
"step": 9000
},
{
"epoch": 24.52,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 9025
},
{
"epoch": 24.59,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 9050
},
{
"epoch": 24.66,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 9075
},
{
"epoch": 24.73,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 9100
},
{
"epoch": 24.8,
"learning_rate": 3e-06,
"loss": 0.0001,
"step": 9125
},
{
"epoch": 24.86,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 9150
},
{
"epoch": 24.93,
"learning_rate": 3e-06,
"loss": 0.0003,
"step": 9175
},
{
"epoch": 25.0,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 9200
},
{
"epoch": 25.07,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 9225
},
{
"epoch": 25.14,
"learning_rate": 3e-06,
"loss": 0.0002,
"step": 9250
},
{
"epoch": 25.2,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 9275
},
{
"epoch": 25.27,
"learning_rate": 3e-06,
"loss": 0.0009,
"step": 9300
},
{
"epoch": 25.34,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 9325
},
{
"epoch": 25.41,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 9350
},
{
"epoch": 25.48,
"learning_rate": 3e-06,
"loss": 0.0003,
"step": 9375
},
{
"epoch": 25.54,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 9400
},
{
"epoch": 25.61,
"learning_rate": 3e-06,
"loss": 0.0013,
"step": 9425
},
{
"epoch": 25.68,
"learning_rate": 3e-06,
"loss": 0.0019,
"step": 9450
},
{
"epoch": 25.75,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 9475
},
{
"epoch": 25.82,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 9500
},
{
"epoch": 25.88,
"learning_rate": 3e-06,
"loss": 0.0005,
"step": 9525
},
{
"epoch": 25.95,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 9550
},
{
"epoch": 26.02,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 9575
},
{
"epoch": 26.09,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 9600
},
{
"epoch": 26.15,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 9625
},
{
"epoch": 26.22,
"learning_rate": 3e-06,
"loss": 0.0004,
"step": 9650
},
{
"epoch": 26.29,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 9675
},
{
"epoch": 26.36,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 9700
},
{
"epoch": 26.43,
"learning_rate": 3e-06,
"loss": 0.0008,
"step": 9725
},
{
"epoch": 26.49,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 9750
},
{
"epoch": 26.56,
"learning_rate": 3e-06,
"loss": 0.0016,
"step": 9775
},
{
"epoch": 26.63,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 9800
},
{
"epoch": 26.7,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 9825
},
{
"epoch": 26.77,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 9850
},
{
"epoch": 26.83,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 9875
},
{
"epoch": 26.9,
"learning_rate": 3e-06,
"loss": 0.0006,
"step": 9900
},
{
"epoch": 26.97,
"learning_rate": 3e-06,
"loss": 0.0014,
"step": 9925
},
{
"epoch": 27.04,
"learning_rate": 3e-06,
"loss": 0.0011,
"step": 9950
},
{
"epoch": 27.11,
"learning_rate": 3e-06,
"loss": 0.0007,
"step": 9975
},
{
"epoch": 27.17,
"learning_rate": 3e-06,
"loss": 0.0012,
"step": 10000
},
{
"epoch": 27.17,
"eval_loss": 0.505859375,
"eval_runtime": 88.9504,
"eval_samples_per_second": 7.611,
"eval_steps_per_second": 0.483,
"eval_wer": 12.666936462970458,
"step": 10000
},
{
"epoch": 27.17,
"step": 10000,
"total_flos": 3.25959777016399e+20,
"train_loss": 0.0141860207259655,
"train_runtime": 38757.4349,
"train_samples_per_second": 8.256,
"train_steps_per_second": 0.258
}
],
"max_steps": 10000,
"num_train_epochs": 28,
"total_flos": 3.25959777016399e+20,
"trial_name": null,
"trial_params": null
}