whisper-medium-tg_tj / trainer_state.json
muhtasham's picture
End of training
85ba1db
raw
history blame contribute delete
No virus
25.8 kB
{
"best_metric": 23.153018764230197,
"best_model_checkpoint": "./checkpoint-5000",
"epoch": 333.001,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.0,
"learning_rate": 4.4e-07,
"loss": 2.3677,
"step": 25
},
{
"epoch": 3.0,
"learning_rate": 9.400000000000001e-07,
"loss": 1.6958,
"step": 50
},
{
"epoch": 4.0,
"learning_rate": 1.44e-06,
"loss": 1.085,
"step": 75
},
{
"epoch": 6.0,
"learning_rate": 1.94e-06,
"loss": 0.6455,
"step": 100
},
{
"epoch": 8.0,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.4216,
"step": 125
},
{
"epoch": 9.0,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.2705,
"step": 150
},
{
"epoch": 11.0,
"learning_rate": 3.44e-06,
"loss": 0.1539,
"step": 175
},
{
"epoch": 13.0,
"learning_rate": 3.94e-06,
"loss": 0.0724,
"step": 200
},
{
"epoch": 14.0,
"learning_rate": 4.440000000000001e-06,
"loss": 0.0349,
"step": 225
},
{
"epoch": 16.0,
"learning_rate": 4.94e-06,
"loss": 0.0187,
"step": 250
},
{
"epoch": 18.0,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.0129,
"step": 275
},
{
"epoch": 19.0,
"learning_rate": 5.94e-06,
"loss": 0.0118,
"step": 300
},
{
"epoch": 21.0,
"learning_rate": 6.440000000000001e-06,
"loss": 0.0067,
"step": 325
},
{
"epoch": 23.0,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.0061,
"step": 350
},
{
"epoch": 24.0,
"learning_rate": 7.440000000000001e-06,
"loss": 0.0054,
"step": 375
},
{
"epoch": 26.0,
"learning_rate": 7.94e-06,
"loss": 0.0089,
"step": 400
},
{
"epoch": 28.0,
"learning_rate": 8.44e-06,
"loss": 0.0069,
"step": 425
},
{
"epoch": 29.0,
"learning_rate": 8.94e-06,
"loss": 0.0085,
"step": 450
},
{
"epoch": 31.0,
"learning_rate": 9.440000000000001e-06,
"loss": 0.006,
"step": 475
},
{
"epoch": 33.0,
"learning_rate": 9.940000000000001e-06,
"loss": 0.005,
"step": 500
},
{
"epoch": 34.0,
"learning_rate": 9.951111111111111e-06,
"loss": 0.0048,
"step": 525
},
{
"epoch": 36.0,
"learning_rate": 9.895555555555557e-06,
"loss": 0.0056,
"step": 550
},
{
"epoch": 38.0,
"learning_rate": 9.84e-06,
"loss": 0.0053,
"step": 575
},
{
"epoch": 39.0,
"learning_rate": 9.784444444444445e-06,
"loss": 0.004,
"step": 600
},
{
"epoch": 41.0,
"learning_rate": 9.72888888888889e-06,
"loss": 0.0041,
"step": 625
},
{
"epoch": 43.0,
"learning_rate": 9.673333333333334e-06,
"loss": 0.0043,
"step": 650
},
{
"epoch": 44.0,
"learning_rate": 9.617777777777778e-06,
"loss": 0.005,
"step": 675
},
{
"epoch": 46.0,
"learning_rate": 9.562222222222223e-06,
"loss": 0.0023,
"step": 700
},
{
"epoch": 48.0,
"learning_rate": 9.506666666666667e-06,
"loss": 0.0021,
"step": 725
},
{
"epoch": 49.0,
"learning_rate": 9.451111111111112e-06,
"loss": 0.0024,
"step": 750
},
{
"epoch": 51.0,
"learning_rate": 9.395555555555556e-06,
"loss": 0.0015,
"step": 775
},
{
"epoch": 53.0,
"learning_rate": 9.340000000000002e-06,
"loss": 0.0013,
"step": 800
},
{
"epoch": 54.0,
"learning_rate": 9.284444444444444e-06,
"loss": 0.0028,
"step": 825
},
{
"epoch": 56.0,
"learning_rate": 9.22888888888889e-06,
"loss": 0.0015,
"step": 850
},
{
"epoch": 58.0,
"learning_rate": 9.173333333333334e-06,
"loss": 0.0014,
"step": 875
},
{
"epoch": 59.0,
"learning_rate": 9.117777777777778e-06,
"loss": 0.001,
"step": 900
},
{
"epoch": 61.0,
"learning_rate": 9.062222222222224e-06,
"loss": 0.0011,
"step": 925
},
{
"epoch": 63.0,
"learning_rate": 9.006666666666666e-06,
"loss": 0.0019,
"step": 950
},
{
"epoch": 64.0,
"learning_rate": 8.951111111111112e-06,
"loss": 0.0021,
"step": 975
},
{
"epoch": 66.0,
"learning_rate": 8.895555555555556e-06,
"loss": 0.0016,
"step": 1000
},
{
"epoch": 66.0,
"eval_loss": 0.6928982734680176,
"eval_runtime": 1591.0145,
"eval_samples_per_second": 0.377,
"eval_steps_per_second": 0.024,
"eval_wer": 24.29928554604695,
"step": 1000
},
{
"epoch": 68.0,
"learning_rate": 8.84e-06,
"loss": 0.0008,
"step": 1025
},
{
"epoch": 69.0,
"learning_rate": 8.784444444444446e-06,
"loss": 0.0005,
"step": 1050
},
{
"epoch": 71.0,
"learning_rate": 8.72888888888889e-06,
"loss": 0.0003,
"step": 1075
},
{
"epoch": 73.0,
"learning_rate": 8.673333333333334e-06,
"loss": 0.0003,
"step": 1100
},
{
"epoch": 74.0,
"learning_rate": 8.617777777777778e-06,
"loss": 0.0003,
"step": 1125
},
{
"epoch": 76.0,
"learning_rate": 8.562222222222224e-06,
"loss": 0.0003,
"step": 1150
},
{
"epoch": 78.0,
"learning_rate": 8.506666666666668e-06,
"loss": 0.0003,
"step": 1175
},
{
"epoch": 79.0,
"learning_rate": 8.451111111111112e-06,
"loss": 0.0002,
"step": 1200
},
{
"epoch": 81.0,
"learning_rate": 8.395555555555557e-06,
"loss": 0.0002,
"step": 1225
},
{
"epoch": 83.0,
"learning_rate": 8.34e-06,
"loss": 0.0002,
"step": 1250
},
{
"epoch": 84.0,
"learning_rate": 8.284444444444446e-06,
"loss": 0.0002,
"step": 1275
},
{
"epoch": 86.0,
"learning_rate": 8.22888888888889e-06,
"loss": 0.0002,
"step": 1300
},
{
"epoch": 88.0,
"learning_rate": 8.173333333333334e-06,
"loss": 0.0002,
"step": 1325
},
{
"epoch": 89.0,
"learning_rate": 8.11777777777778e-06,
"loss": 0.0002,
"step": 1350
},
{
"epoch": 91.0,
"learning_rate": 8.062222222222222e-06,
"loss": 0.0002,
"step": 1375
},
{
"epoch": 93.0,
"learning_rate": 8.006666666666667e-06,
"loss": 0.0002,
"step": 1400
},
{
"epoch": 94.0,
"learning_rate": 7.951111111111111e-06,
"loss": 0.0002,
"step": 1425
},
{
"epoch": 96.0,
"learning_rate": 7.895555555555557e-06,
"loss": 0.0002,
"step": 1450
},
{
"epoch": 98.0,
"learning_rate": 7.840000000000001e-06,
"loss": 0.0002,
"step": 1475
},
{
"epoch": 99.0,
"learning_rate": 7.784444444444445e-06,
"loss": 0.0002,
"step": 1500
},
{
"epoch": 101.0,
"learning_rate": 7.72888888888889e-06,
"loss": 0.0002,
"step": 1525
},
{
"epoch": 103.0,
"learning_rate": 7.673333333333333e-06,
"loss": 0.0002,
"step": 1550
},
{
"epoch": 104.0,
"learning_rate": 7.617777777777778e-06,
"loss": 0.0002,
"step": 1575
},
{
"epoch": 106.0,
"learning_rate": 7.562222222222223e-06,
"loss": 0.0002,
"step": 1600
},
{
"epoch": 108.0,
"learning_rate": 7.506666666666668e-06,
"loss": 0.0002,
"step": 1625
},
{
"epoch": 109.0,
"learning_rate": 7.451111111111111e-06,
"loss": 0.0002,
"step": 1650
},
{
"epoch": 111.0,
"learning_rate": 7.395555555555556e-06,
"loss": 0.0002,
"step": 1675
},
{
"epoch": 113.0,
"learning_rate": 7.340000000000001e-06,
"loss": 0.0002,
"step": 1700
},
{
"epoch": 114.0,
"learning_rate": 7.284444444444445e-06,
"loss": 0.0002,
"step": 1725
},
{
"epoch": 116.0,
"learning_rate": 7.22888888888889e-06,
"loss": 0.0002,
"step": 1750
},
{
"epoch": 118.0,
"learning_rate": 7.173333333333335e-06,
"loss": 0.0002,
"step": 1775
},
{
"epoch": 119.0,
"learning_rate": 7.117777777777778e-06,
"loss": 0.0002,
"step": 1800
},
{
"epoch": 121.0,
"learning_rate": 7.062222222222223e-06,
"loss": 0.0002,
"step": 1825
},
{
"epoch": 123.0,
"learning_rate": 7.006666666666667e-06,
"loss": 0.0001,
"step": 1850
},
{
"epoch": 124.0,
"learning_rate": 6.951111111111112e-06,
"loss": 0.0001,
"step": 1875
},
{
"epoch": 126.0,
"learning_rate": 6.8955555555555565e-06,
"loss": 0.0001,
"step": 1900
},
{
"epoch": 128.0,
"learning_rate": 6.8400000000000014e-06,
"loss": 0.0001,
"step": 1925
},
{
"epoch": 129.0,
"learning_rate": 6.784444444444445e-06,
"loss": 0.0001,
"step": 1950
},
{
"epoch": 131.0,
"learning_rate": 6.7288888888888895e-06,
"loss": 0.0001,
"step": 1975
},
{
"epoch": 133.0,
"learning_rate": 6.6733333333333335e-06,
"loss": 0.0001,
"step": 2000
},
{
"epoch": 133.0,
"eval_loss": 0.8053555488586426,
"eval_runtime": 1617.841,
"eval_samples_per_second": 0.371,
"eval_steps_per_second": 0.023,
"eval_wer": 23.302190468713196,
"step": 2000
},
{
"epoch": 134.0,
"learning_rate": 6.617777777777778e-06,
"loss": 0.0001,
"step": 2025
},
{
"epoch": 136.0,
"learning_rate": 6.562222222222223e-06,
"loss": 0.0001,
"step": 2050
},
{
"epoch": 138.0,
"learning_rate": 6.5066666666666665e-06,
"loss": 0.0001,
"step": 2075
},
{
"epoch": 139.0,
"learning_rate": 6.451111111111111e-06,
"loss": 0.0001,
"step": 2100
},
{
"epoch": 141.0,
"learning_rate": 6.395555555555556e-06,
"loss": 0.0001,
"step": 2125
},
{
"epoch": 143.0,
"learning_rate": 6.34e-06,
"loss": 0.0001,
"step": 2150
},
{
"epoch": 144.0,
"learning_rate": 6.284444444444445e-06,
"loss": 0.0001,
"step": 2175
},
{
"epoch": 146.0,
"learning_rate": 6.22888888888889e-06,
"loss": 0.0001,
"step": 2200
},
{
"epoch": 148.0,
"learning_rate": 6.173333333333333e-06,
"loss": 0.0001,
"step": 2225
},
{
"epoch": 149.0,
"learning_rate": 6.117777777777778e-06,
"loss": 0.0001,
"step": 2250
},
{
"epoch": 151.0,
"learning_rate": 6.062222222222223e-06,
"loss": 0.0001,
"step": 2275
},
{
"epoch": 153.0,
"learning_rate": 6.006666666666667e-06,
"loss": 0.0001,
"step": 2300
},
{
"epoch": 154.0,
"learning_rate": 5.951111111111112e-06,
"loss": 0.0001,
"step": 2325
},
{
"epoch": 156.0,
"learning_rate": 5.895555555555557e-06,
"loss": 0.0001,
"step": 2350
},
{
"epoch": 158.0,
"learning_rate": 5.84e-06,
"loss": 0.0001,
"step": 2375
},
{
"epoch": 159.0,
"learning_rate": 5.784444444444445e-06,
"loss": 0.0001,
"step": 2400
},
{
"epoch": 161.0,
"learning_rate": 5.72888888888889e-06,
"loss": 0.0001,
"step": 2425
},
{
"epoch": 163.0,
"learning_rate": 5.673333333333334e-06,
"loss": 0.0001,
"step": 2450
},
{
"epoch": 164.0,
"learning_rate": 5.617777777777779e-06,
"loss": 0.0001,
"step": 2475
},
{
"epoch": 166.0,
"learning_rate": 5.562222222222222e-06,
"loss": 0.0001,
"step": 2500
},
{
"epoch": 168.0,
"learning_rate": 5.506666666666667e-06,
"loss": 0.0001,
"step": 2525
},
{
"epoch": 169.0,
"learning_rate": 5.451111111111112e-06,
"loss": 0.0001,
"step": 2550
},
{
"epoch": 171.0,
"learning_rate": 5.3955555555555565e-06,
"loss": 0.0001,
"step": 2575
},
{
"epoch": 173.0,
"learning_rate": 5.3400000000000005e-06,
"loss": 0.0001,
"step": 2600
},
{
"epoch": 174.0,
"learning_rate": 5.2844444444444454e-06,
"loss": 0.0001,
"step": 2625
},
{
"epoch": 176.0,
"learning_rate": 5.228888888888889e-06,
"loss": 0.0001,
"step": 2650
},
{
"epoch": 178.0,
"learning_rate": 5.1733333333333335e-06,
"loss": 0.0001,
"step": 2675
},
{
"epoch": 179.0,
"learning_rate": 5.117777777777778e-06,
"loss": 0.0001,
"step": 2700
},
{
"epoch": 181.0,
"learning_rate": 5.062222222222222e-06,
"loss": 0.0001,
"step": 2725
},
{
"epoch": 183.0,
"learning_rate": 5.006666666666667e-06,
"loss": 0.0001,
"step": 2750
},
{
"epoch": 184.0,
"learning_rate": 4.951111111111111e-06,
"loss": 0.0001,
"step": 2775
},
{
"epoch": 186.0,
"learning_rate": 4.895555555555556e-06,
"loss": 0.0001,
"step": 2800
},
{
"epoch": 188.0,
"learning_rate": 4.84e-06,
"loss": 0.0001,
"step": 2825
},
{
"epoch": 189.0,
"learning_rate": 4.784444444444445e-06,
"loss": 0.0001,
"step": 2850
},
{
"epoch": 191.0,
"learning_rate": 4.728888888888889e-06,
"loss": 0.0001,
"step": 2875
},
{
"epoch": 193.0,
"learning_rate": 4.673333333333333e-06,
"loss": 0.0001,
"step": 2900
},
{
"epoch": 194.0,
"learning_rate": 4.617777777777778e-06,
"loss": 0.0001,
"step": 2925
},
{
"epoch": 196.0,
"learning_rate": 4.562222222222222e-06,
"loss": 0.0001,
"step": 2950
},
{
"epoch": 198.0,
"learning_rate": 4.506666666666667e-06,
"loss": 0.0001,
"step": 2975
},
{
"epoch": 199.0,
"learning_rate": 4.451111111111112e-06,
"loss": 0.0001,
"step": 3000
},
{
"epoch": 199.0,
"eval_loss": 0.8651753664016724,
"eval_runtime": 1594.5855,
"eval_samples_per_second": 0.376,
"eval_steps_per_second": 0.024,
"eval_wer": 23.22367904530109,
"step": 3000
},
{
"epoch": 201.0,
"learning_rate": 4.395555555555556e-06,
"loss": 0.0001,
"step": 3025
},
{
"epoch": 203.0,
"learning_rate": 4.34e-06,
"loss": 0.0001,
"step": 3050
},
{
"epoch": 204.0,
"learning_rate": 4.284444444444445e-06,
"loss": 0.0001,
"step": 3075
},
{
"epoch": 206.0,
"learning_rate": 4.228888888888889e-06,
"loss": 0.0001,
"step": 3100
},
{
"epoch": 208.0,
"learning_rate": 4.173333333333334e-06,
"loss": 0.0001,
"step": 3125
},
{
"epoch": 209.0,
"learning_rate": 4.117777777777779e-06,
"loss": 0.0001,
"step": 3150
},
{
"epoch": 211.0,
"learning_rate": 4.062222222222223e-06,
"loss": 0.0001,
"step": 3175
},
{
"epoch": 213.0,
"learning_rate": 4.006666666666667e-06,
"loss": 0.0001,
"step": 3200
},
{
"epoch": 214.0,
"learning_rate": 3.951111111111112e-06,
"loss": 0.0001,
"step": 3225
},
{
"epoch": 216.0,
"learning_rate": 3.895555555555556e-06,
"loss": 0.0001,
"step": 3250
},
{
"epoch": 218.0,
"learning_rate": 3.8400000000000005e-06,
"loss": 0.0001,
"step": 3275
},
{
"epoch": 219.0,
"learning_rate": 3.784444444444445e-06,
"loss": 0.0001,
"step": 3300
},
{
"epoch": 221.0,
"learning_rate": 3.728888888888889e-06,
"loss": 0.0001,
"step": 3325
},
{
"epoch": 223.0,
"learning_rate": 3.673333333333334e-06,
"loss": 0.0001,
"step": 3350
},
{
"epoch": 224.0,
"learning_rate": 3.617777777777778e-06,
"loss": 0.0001,
"step": 3375
},
{
"epoch": 226.0,
"learning_rate": 3.5622222222222224e-06,
"loss": 0.0001,
"step": 3400
},
{
"epoch": 228.0,
"learning_rate": 3.5066666666666673e-06,
"loss": 0.0001,
"step": 3425
},
{
"epoch": 229.0,
"learning_rate": 3.4511111111111113e-06,
"loss": 0.0001,
"step": 3450
},
{
"epoch": 231.0,
"learning_rate": 3.3955555555555558e-06,
"loss": 0.0001,
"step": 3475
},
{
"epoch": 233.0,
"learning_rate": 3.3400000000000006e-06,
"loss": 0.0001,
"step": 3500
},
{
"epoch": 234.0,
"learning_rate": 3.2844444444444447e-06,
"loss": 0.0001,
"step": 3525
},
{
"epoch": 236.0,
"learning_rate": 3.228888888888889e-06,
"loss": 0.0001,
"step": 3550
},
{
"epoch": 238.0,
"learning_rate": 3.173333333333334e-06,
"loss": 0.0001,
"step": 3575
},
{
"epoch": 239.0,
"learning_rate": 3.117777777777778e-06,
"loss": 0.0001,
"step": 3600
},
{
"epoch": 241.0,
"learning_rate": 3.0622222222222225e-06,
"loss": 0.0,
"step": 3625
},
{
"epoch": 243.0,
"learning_rate": 3.0066666666666674e-06,
"loss": 0.0001,
"step": 3650
},
{
"epoch": 244.0,
"learning_rate": 2.9511111111111114e-06,
"loss": 0.0,
"step": 3675
},
{
"epoch": 246.0,
"learning_rate": 2.895555555555556e-06,
"loss": 0.0,
"step": 3700
},
{
"epoch": 248.0,
"learning_rate": 2.84e-06,
"loss": 0.0,
"step": 3725
},
{
"epoch": 249.0,
"learning_rate": 2.784444444444445e-06,
"loss": 0.0,
"step": 3750
},
{
"epoch": 251.0,
"learning_rate": 2.7288888888888893e-06,
"loss": 0.0,
"step": 3775
},
{
"epoch": 253.0,
"learning_rate": 2.6733333333333333e-06,
"loss": 0.0,
"step": 3800
},
{
"epoch": 254.0,
"learning_rate": 2.617777777777778e-06,
"loss": 0.0,
"step": 3825
},
{
"epoch": 256.0,
"learning_rate": 2.5622222222222226e-06,
"loss": 0.0,
"step": 3850
},
{
"epoch": 258.0,
"learning_rate": 2.5066666666666667e-06,
"loss": 0.0,
"step": 3875
},
{
"epoch": 259.0,
"learning_rate": 2.451111111111111e-06,
"loss": 0.0,
"step": 3900
},
{
"epoch": 261.0,
"learning_rate": 2.3955555555555556e-06,
"loss": 0.0,
"step": 3925
},
{
"epoch": 263.0,
"learning_rate": 2.3400000000000005e-06,
"loss": 0.0,
"step": 3950
},
{
"epoch": 264.0,
"learning_rate": 2.2844444444444445e-06,
"loss": 0.0,
"step": 3975
},
{
"epoch": 266.0,
"learning_rate": 2.228888888888889e-06,
"loss": 0.0,
"step": 4000
},
{
"epoch": 266.0,
"eval_loss": 0.9019195437431335,
"eval_runtime": 1593.9856,
"eval_samples_per_second": 0.376,
"eval_steps_per_second": 0.024,
"eval_wer": 23.239381329983512,
"step": 4000
},
{
"epoch": 268.0,
"learning_rate": 2.1733333333333334e-06,
"loss": 0.0,
"step": 4025
},
{
"epoch": 269.0,
"learning_rate": 2.117777777777778e-06,
"loss": 0.0,
"step": 4050
},
{
"epoch": 271.0,
"learning_rate": 2.0622222222222223e-06,
"loss": 0.0,
"step": 4075
},
{
"epoch": 273.0,
"learning_rate": 2.006666666666667e-06,
"loss": 0.0,
"step": 4100
},
{
"epoch": 274.0,
"learning_rate": 1.9511111111111113e-06,
"loss": 0.0,
"step": 4125
},
{
"epoch": 276.0,
"learning_rate": 1.8955555555555557e-06,
"loss": 0.0,
"step": 4150
},
{
"epoch": 278.0,
"learning_rate": 1.8400000000000002e-06,
"loss": 0.0,
"step": 4175
},
{
"epoch": 279.0,
"learning_rate": 1.7844444444444444e-06,
"loss": 0.0,
"step": 4200
},
{
"epoch": 281.0,
"learning_rate": 1.728888888888889e-06,
"loss": 0.0,
"step": 4225
},
{
"epoch": 283.0,
"learning_rate": 1.6733333333333335e-06,
"loss": 0.0,
"step": 4250
},
{
"epoch": 284.0,
"learning_rate": 1.6177777777777778e-06,
"loss": 0.0,
"step": 4275
},
{
"epoch": 286.0,
"learning_rate": 1.5622222222222225e-06,
"loss": 0.0,
"step": 4300
},
{
"epoch": 288.0,
"learning_rate": 1.506666666666667e-06,
"loss": 0.0,
"step": 4325
},
{
"epoch": 289.0,
"learning_rate": 1.4511111111111112e-06,
"loss": 0.0,
"step": 4350
},
{
"epoch": 291.0,
"learning_rate": 1.3955555555555556e-06,
"loss": 0.0,
"step": 4375
},
{
"epoch": 293.0,
"learning_rate": 1.34e-06,
"loss": 0.0,
"step": 4400
},
{
"epoch": 294.0,
"learning_rate": 1.2844444444444445e-06,
"loss": 0.0,
"step": 4425
},
{
"epoch": 296.0,
"learning_rate": 1.228888888888889e-06,
"loss": 0.0,
"step": 4450
},
{
"epoch": 298.0,
"learning_rate": 1.1733333333333335e-06,
"loss": 0.0,
"step": 4475
},
{
"epoch": 299.0,
"learning_rate": 1.117777777777778e-06,
"loss": 0.0,
"step": 4500
},
{
"epoch": 301.0,
"learning_rate": 1.0622222222222222e-06,
"loss": 0.0,
"step": 4525
},
{
"epoch": 303.0,
"learning_rate": 1.0066666666666668e-06,
"loss": 0.0,
"step": 4550
},
{
"epoch": 304.0,
"learning_rate": 9.511111111111111e-07,
"loss": 0.0,
"step": 4575
},
{
"epoch": 306.0,
"learning_rate": 8.955555555555557e-07,
"loss": 0.0,
"step": 4600
},
{
"epoch": 308.0,
"learning_rate": 8.400000000000001e-07,
"loss": 0.0,
"step": 4625
},
{
"epoch": 309.0,
"learning_rate": 7.844444444444445e-07,
"loss": 0.0,
"step": 4650
},
{
"epoch": 311.0,
"learning_rate": 7.28888888888889e-07,
"loss": 0.0,
"step": 4675
},
{
"epoch": 313.0,
"learning_rate": 6.733333333333334e-07,
"loss": 0.0,
"step": 4700
},
{
"epoch": 314.0,
"learning_rate": 6.177777777777778e-07,
"loss": 0.0,
"step": 4725
},
{
"epoch": 316.0,
"learning_rate": 5.622222222222223e-07,
"loss": 0.0,
"step": 4750
},
{
"epoch": 318.0,
"learning_rate": 5.066666666666667e-07,
"loss": 0.0,
"step": 4775
},
{
"epoch": 319.0,
"learning_rate": 4.511111111111111e-07,
"loss": 0.0,
"step": 4800
},
{
"epoch": 321.0,
"learning_rate": 3.9555555555555557e-07,
"loss": 0.0,
"step": 4825
},
{
"epoch": 323.0,
"learning_rate": 3.4000000000000003e-07,
"loss": 0.0,
"step": 4850
},
{
"epoch": 324.0,
"learning_rate": 2.844444444444445e-07,
"loss": 0.0,
"step": 4875
},
{
"epoch": 326.0,
"learning_rate": 2.2888888888888892e-07,
"loss": 0.0,
"step": 4900
},
{
"epoch": 328.0,
"learning_rate": 1.7333333333333335e-07,
"loss": 0.0,
"step": 4925
},
{
"epoch": 329.0,
"learning_rate": 1.1777777777777778e-07,
"loss": 0.0,
"step": 4950
},
{
"epoch": 331.0,
"learning_rate": 6.222222222222223e-08,
"loss": 0.0,
"step": 4975
},
{
"epoch": 333.0,
"learning_rate": 6.666666666666667e-09,
"loss": 0.0,
"step": 5000
},
{
"epoch": 333.0,
"eval_loss": 0.9217017889022827,
"eval_runtime": 1592.7118,
"eval_samples_per_second": 0.377,
"eval_steps_per_second": 0.024,
"eval_wer": 23.153018764230197,
"step": 5000
},
{
"epoch": 333.0,
"step": 5000,
"total_flos": 1.6125761659797504e+20,
"train_loss": 0.034576853090384975,
"train_runtime": 59221.7156,
"train_samples_per_second": 2.702,
"train_steps_per_second": 0.084
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 1.6125761659797504e+20,
"trial_name": null,
"trial_params": null
}