whisper-base-ko-1 / trainer_state.json
arun100's picture
End of training
5345b79 verified
{
"best_metric": 45.5026455026455,
"best_model_checkpoint": "./checkpoint-1000",
"epoch": 1333.00025,
"eval_steps": 1000,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 3.0,
"learning_rate": 4.6e-07,
"loss": 0.9339,
"step": 25
},
{
"epoch": 6.0,
"learning_rate": 9.6e-07,
"loss": 0.8592,
"step": 50
},
{
"epoch": 9.0,
"learning_rate": 9.976884422110552e-07,
"loss": 0.6668,
"step": 75
},
{
"epoch": 13.0,
"learning_rate": 9.95175879396985e-07,
"loss": 0.4854,
"step": 100
},
{
"epoch": 16.0,
"learning_rate": 9.926633165829145e-07,
"loss": 0.3883,
"step": 125
},
{
"epoch": 19.0,
"learning_rate": 9.901507537688442e-07,
"loss": 0.34,
"step": 150
},
{
"epoch": 23.0,
"learning_rate": 9.876381909547738e-07,
"loss": 0.3051,
"step": 175
},
{
"epoch": 26.0,
"learning_rate": 9.851256281407035e-07,
"loss": 0.2705,
"step": 200
},
{
"epoch": 29.0,
"learning_rate": 9.82613065326633e-07,
"loss": 0.2434,
"step": 225
},
{
"epoch": 33.0,
"learning_rate": 9.801005025125628e-07,
"loss": 0.214,
"step": 250
},
{
"epoch": 36.0,
"learning_rate": 9.775879396984923e-07,
"loss": 0.1906,
"step": 275
},
{
"epoch": 39.0,
"learning_rate": 9.75075376884422e-07,
"loss": 0.1736,
"step": 300
},
{
"epoch": 43.0,
"learning_rate": 9.725628140703516e-07,
"loss": 0.1529,
"step": 325
},
{
"epoch": 46.0,
"learning_rate": 9.700502512562814e-07,
"loss": 0.1377,
"step": 350
},
{
"epoch": 49.0,
"learning_rate": 9.67537688442211e-07,
"loss": 0.1236,
"step": 375
},
{
"epoch": 53.0,
"learning_rate": 9.650251256281406e-07,
"loss": 0.1081,
"step": 400
},
{
"epoch": 56.0,
"learning_rate": 9.625125628140704e-07,
"loss": 0.0949,
"step": 425
},
{
"epoch": 59.0,
"learning_rate": 9.6e-07,
"loss": 0.0834,
"step": 450
},
{
"epoch": 63.0,
"learning_rate": 9.574874371859297e-07,
"loss": 0.0726,
"step": 475
},
{
"epoch": 66.0,
"learning_rate": 9.549748743718594e-07,
"loss": 0.0635,
"step": 500
},
{
"epoch": 69.0,
"learning_rate": 9.524623115577888e-07,
"loss": 0.0572,
"step": 525
},
{
"epoch": 73.0,
"learning_rate": 9.499497487437186e-07,
"loss": 0.0501,
"step": 550
},
{
"epoch": 76.0,
"learning_rate": 9.474371859296481e-07,
"loss": 0.0449,
"step": 575
},
{
"epoch": 79.0,
"learning_rate": 9.449246231155779e-07,
"loss": 0.0399,
"step": 600
},
{
"epoch": 83.0,
"learning_rate": 9.424120603015075e-07,
"loss": 0.0374,
"step": 625
},
{
"epoch": 86.0,
"learning_rate": 9.398994974874371e-07,
"loss": 0.0336,
"step": 650
},
{
"epoch": 89.0,
"learning_rate": 9.373869346733668e-07,
"loss": 0.031,
"step": 675
},
{
"epoch": 93.0,
"learning_rate": 9.348743718592964e-07,
"loss": 0.0285,
"step": 700
},
{
"epoch": 96.0,
"learning_rate": 9.323618090452261e-07,
"loss": 0.0267,
"step": 725
},
{
"epoch": 99.0,
"learning_rate": 9.298492462311557e-07,
"loss": 0.025,
"step": 750
},
{
"epoch": 103.0,
"learning_rate": 9.273366834170854e-07,
"loss": 0.0235,
"step": 775
},
{
"epoch": 106.0,
"learning_rate": 9.248241206030151e-07,
"loss": 0.0221,
"step": 800
},
{
"epoch": 109.0,
"learning_rate": 9.223115577889446e-07,
"loss": 0.0209,
"step": 825
},
{
"epoch": 113.0,
"learning_rate": 9.197989949748744e-07,
"loss": 0.0196,
"step": 850
},
{
"epoch": 116.0,
"learning_rate": 9.172864321608039e-07,
"loss": 0.0188,
"step": 875
},
{
"epoch": 119.0,
"learning_rate": 9.147738693467337e-07,
"loss": 0.0176,
"step": 900
},
{
"epoch": 123.0,
"learning_rate": 9.122613065326633e-07,
"loss": 0.017,
"step": 925
},
{
"epoch": 126.0,
"learning_rate": 9.097487437185929e-07,
"loss": 0.0162,
"step": 950
},
{
"epoch": 129.0,
"learning_rate": 9.072361809045226e-07,
"loss": 0.0155,
"step": 975
},
{
"epoch": 133.0,
"learning_rate": 9.047236180904521e-07,
"loss": 0.0149,
"step": 1000
},
{
"epoch": 133.0,
"eval_loss": 0.6687414050102234,
"eval_runtime": 24.3077,
"eval_samples_per_second": 11.149,
"eval_steps_per_second": 0.37,
"eval_wer": 45.5026455026455,
"step": 1000
},
{
"epoch": 136.0,
"learning_rate": 9.022110552763819e-07,
"loss": 0.0142,
"step": 1025
},
{
"epoch": 139.0,
"learning_rate": 8.996984924623115e-07,
"loss": 0.0136,
"step": 1050
},
{
"epoch": 143.0,
"learning_rate": 8.971859296482411e-07,
"loss": 0.0131,
"step": 1075
},
{
"epoch": 146.0,
"learning_rate": 8.946733668341708e-07,
"loss": 0.0127,
"step": 1100
},
{
"epoch": 149.0,
"learning_rate": 8.921608040201004e-07,
"loss": 0.012,
"step": 1125
},
{
"epoch": 153.0,
"learning_rate": 8.896482412060301e-07,
"loss": 0.0117,
"step": 1150
},
{
"epoch": 156.0,
"learning_rate": 8.871356783919598e-07,
"loss": 0.0113,
"step": 1175
},
{
"epoch": 159.0,
"learning_rate": 8.846231155778894e-07,
"loss": 0.0109,
"step": 1200
},
{
"epoch": 163.0,
"learning_rate": 8.821105527638191e-07,
"loss": 0.0106,
"step": 1225
},
{
"epoch": 166.0,
"learning_rate": 8.795979899497486e-07,
"loss": 0.0103,
"step": 1250
},
{
"epoch": 169.0,
"learning_rate": 8.770854271356784e-07,
"loss": 0.0099,
"step": 1275
},
{
"epoch": 173.0,
"learning_rate": 8.74572864321608e-07,
"loss": 0.0096,
"step": 1300
},
{
"epoch": 176.0,
"learning_rate": 8.720603015075377e-07,
"loss": 0.0094,
"step": 1325
},
{
"epoch": 179.0,
"learning_rate": 8.695477386934673e-07,
"loss": 0.0091,
"step": 1350
},
{
"epoch": 183.0,
"learning_rate": 8.670351758793969e-07,
"loss": 0.0088,
"step": 1375
},
{
"epoch": 186.0,
"learning_rate": 8.645226130653266e-07,
"loss": 0.0085,
"step": 1400
},
{
"epoch": 189.0,
"learning_rate": 8.620100502512563e-07,
"loss": 0.0084,
"step": 1425
},
{
"epoch": 193.0,
"learning_rate": 8.594974874371859e-07,
"loss": 0.008,
"step": 1450
},
{
"epoch": 196.0,
"learning_rate": 8.569849246231155e-07,
"loss": 0.0078,
"step": 1475
},
{
"epoch": 199.0,
"learning_rate": 8.544723618090451e-07,
"loss": 0.0076,
"step": 1500
},
{
"epoch": 203.0,
"learning_rate": 8.519597989949748e-07,
"loss": 0.0074,
"step": 1525
},
{
"epoch": 206.0,
"learning_rate": 8.494472361809045e-07,
"loss": 0.0072,
"step": 1550
},
{
"epoch": 209.0,
"learning_rate": 8.469346733668341e-07,
"loss": 0.007,
"step": 1575
},
{
"epoch": 213.0,
"learning_rate": 8.444221105527638e-07,
"loss": 0.0069,
"step": 1600
},
{
"epoch": 216.0,
"learning_rate": 8.419095477386934e-07,
"loss": 0.0067,
"step": 1625
},
{
"epoch": 219.0,
"learning_rate": 8.393969849246231e-07,
"loss": 0.0065,
"step": 1650
},
{
"epoch": 223.0,
"learning_rate": 8.368844221105527e-07,
"loss": 0.0064,
"step": 1675
},
{
"epoch": 226.0,
"learning_rate": 8.343718592964824e-07,
"loss": 0.0062,
"step": 1700
},
{
"epoch": 229.0,
"learning_rate": 8.31859296482412e-07,
"loss": 0.006,
"step": 1725
},
{
"epoch": 233.0,
"learning_rate": 8.293467336683417e-07,
"loss": 0.006,
"step": 1750
},
{
"epoch": 236.0,
"learning_rate": 8.268341708542713e-07,
"loss": 0.0058,
"step": 1775
},
{
"epoch": 239.0,
"learning_rate": 8.24321608040201e-07,
"loss": 0.0057,
"step": 1800
},
{
"epoch": 243.0,
"learning_rate": 8.218090452261306e-07,
"loss": 0.0055,
"step": 1825
},
{
"epoch": 246.0,
"learning_rate": 8.192964824120603e-07,
"loss": 0.0054,
"step": 1850
},
{
"epoch": 249.0,
"learning_rate": 8.167839195979899e-07,
"loss": 0.0053,
"step": 1875
},
{
"epoch": 253.0,
"learning_rate": 8.142713567839195e-07,
"loss": 0.0052,
"step": 1900
},
{
"epoch": 256.0,
"learning_rate": 8.117587939698493e-07,
"loss": 0.0051,
"step": 1925
},
{
"epoch": 259.0,
"learning_rate": 8.092462311557788e-07,
"loss": 0.005,
"step": 1950
},
{
"epoch": 263.0,
"learning_rate": 8.067336683417085e-07,
"loss": 0.0049,
"step": 1975
},
{
"epoch": 266.0,
"learning_rate": 8.042211055276381e-07,
"loss": 0.0048,
"step": 2000
},
{
"epoch": 266.0,
"eval_loss": 0.7147858738899231,
"eval_runtime": 22.5699,
"eval_samples_per_second": 12.007,
"eval_steps_per_second": 0.399,
"eval_wer": 47.76334776334777,
"step": 2000
},
{
"epoch": 269.0,
"learning_rate": 8.017085427135678e-07,
"loss": 0.0047,
"step": 2025
},
{
"epoch": 273.0,
"learning_rate": 7.991959798994975e-07,
"loss": 0.0046,
"step": 2050
},
{
"epoch": 276.0,
"learning_rate": 7.966834170854271e-07,
"loss": 0.0045,
"step": 2075
},
{
"epoch": 279.0,
"learning_rate": 7.941708542713567e-07,
"loss": 0.0044,
"step": 2100
},
{
"epoch": 283.0,
"learning_rate": 7.916582914572864e-07,
"loss": 0.0043,
"step": 2125
},
{
"epoch": 286.0,
"learning_rate": 7.89145728643216e-07,
"loss": 0.0042,
"step": 2150
},
{
"epoch": 289.0,
"learning_rate": 7.866331658291458e-07,
"loss": 0.0041,
"step": 2175
},
{
"epoch": 293.0,
"learning_rate": 7.841206030150753e-07,
"loss": 0.0041,
"step": 2200
},
{
"epoch": 296.0,
"learning_rate": 7.816080402010051e-07,
"loss": 0.004,
"step": 2225
},
{
"epoch": 299.0,
"learning_rate": 7.790954773869346e-07,
"loss": 0.0039,
"step": 2250
},
{
"epoch": 303.0,
"learning_rate": 7.765829145728643e-07,
"loss": 0.0039,
"step": 2275
},
{
"epoch": 306.0,
"learning_rate": 7.74070351758794e-07,
"loss": 0.0038,
"step": 2300
},
{
"epoch": 309.0,
"learning_rate": 7.715577889447235e-07,
"loss": 0.0037,
"step": 2325
},
{
"epoch": 313.0,
"learning_rate": 7.690452261306533e-07,
"loss": 0.0037,
"step": 2350
},
{
"epoch": 316.0,
"learning_rate": 7.665326633165828e-07,
"loss": 0.0036,
"step": 2375
},
{
"epoch": 319.0,
"learning_rate": 7.640201005025125e-07,
"loss": 0.0035,
"step": 2400
},
{
"epoch": 323.0,
"learning_rate": 7.615075376884422e-07,
"loss": 0.0035,
"step": 2425
},
{
"epoch": 326.0,
"learning_rate": 7.589949748743718e-07,
"loss": 0.0034,
"step": 2450
},
{
"epoch": 329.0,
"learning_rate": 7.564824120603015e-07,
"loss": 0.0033,
"step": 2475
},
{
"epoch": 333.0,
"learning_rate": 7.539698492462311e-07,
"loss": 0.0033,
"step": 2500
},
{
"epoch": 336.0,
"learning_rate": 7.514572864321607e-07,
"loss": 0.0032,
"step": 2525
},
{
"epoch": 339.0,
"learning_rate": 7.489447236180905e-07,
"loss": 0.0032,
"step": 2550
},
{
"epoch": 343.0,
"learning_rate": 7.4643216080402e-07,
"loss": 0.0031,
"step": 2575
},
{
"epoch": 346.0,
"learning_rate": 7.439195979899498e-07,
"loss": 0.0031,
"step": 2600
},
{
"epoch": 349.0,
"learning_rate": 7.414070351758793e-07,
"loss": 0.003,
"step": 2625
},
{
"epoch": 353.0,
"learning_rate": 7.388944723618091e-07,
"loss": 0.003,
"step": 2650
},
{
"epoch": 356.0,
"learning_rate": 7.363819095477387e-07,
"loss": 0.0029,
"step": 2675
},
{
"epoch": 359.0,
"learning_rate": 7.338693467336683e-07,
"loss": 0.0029,
"step": 2700
},
{
"epoch": 363.0,
"learning_rate": 7.31356783919598e-07,
"loss": 0.0028,
"step": 2725
},
{
"epoch": 366.0,
"learning_rate": 7.288442211055276e-07,
"loss": 0.0028,
"step": 2750
},
{
"epoch": 369.0,
"learning_rate": 7.263316582914573e-07,
"loss": 0.0027,
"step": 2775
},
{
"epoch": 373.0,
"learning_rate": 7.23819095477387e-07,
"loss": 0.0027,
"step": 2800
},
{
"epoch": 376.0,
"learning_rate": 7.213065326633165e-07,
"loss": 0.0026,
"step": 2825
},
{
"epoch": 379.0,
"learning_rate": 7.187939698492462e-07,
"loss": 0.0026,
"step": 2850
},
{
"epoch": 383.0,
"learning_rate": 7.162814070351758e-07,
"loss": 0.0026,
"step": 2875
},
{
"epoch": 386.0,
"learning_rate": 7.137688442211055e-07,
"loss": 0.0025,
"step": 2900
},
{
"epoch": 389.0,
"learning_rate": 7.112562814070352e-07,
"loss": 0.0025,
"step": 2925
},
{
"epoch": 393.0,
"learning_rate": 7.087437185929647e-07,
"loss": 0.0025,
"step": 2950
},
{
"epoch": 396.0,
"learning_rate": 7.062311557788945e-07,
"loss": 0.0024,
"step": 2975
},
{
"epoch": 399.0,
"learning_rate": 7.03718592964824e-07,
"loss": 0.0024,
"step": 3000
},
{
"epoch": 399.0,
"eval_loss": 0.7483593821525574,
"eval_runtime": 21.6386,
"eval_samples_per_second": 12.524,
"eval_steps_per_second": 0.416,
"eval_wer": 48.484848484848484,
"step": 3000
},
{
"epoch": 403.0,
"learning_rate": 7.012060301507538e-07,
"loss": 0.0024,
"step": 3025
},
{
"epoch": 406.0,
"learning_rate": 6.986934673366834e-07,
"loss": 0.0023,
"step": 3050
},
{
"epoch": 409.0,
"learning_rate": 6.961809045226131e-07,
"loss": 0.0023,
"step": 3075
},
{
"epoch": 413.0,
"learning_rate": 6.936683417085427e-07,
"loss": 0.0023,
"step": 3100
},
{
"epoch": 416.0,
"learning_rate": 6.911557788944723e-07,
"loss": 0.0022,
"step": 3125
},
{
"epoch": 419.0,
"learning_rate": 6.88643216080402e-07,
"loss": 0.0022,
"step": 3150
},
{
"epoch": 423.0,
"learning_rate": 6.861306532663317e-07,
"loss": 0.0022,
"step": 3175
},
{
"epoch": 426.0,
"learning_rate": 6.836180904522613e-07,
"loss": 0.0021,
"step": 3200
},
{
"epoch": 429.0,
"learning_rate": 6.81105527638191e-07,
"loss": 0.0021,
"step": 3225
},
{
"epoch": 433.0,
"learning_rate": 6.785929648241205e-07,
"loss": 0.0021,
"step": 3250
},
{
"epoch": 436.0,
"learning_rate": 6.760804020100502e-07,
"loss": 0.0021,
"step": 3275
},
{
"epoch": 439.0,
"learning_rate": 6.735678391959798e-07,
"loss": 0.002,
"step": 3300
},
{
"epoch": 443.0,
"learning_rate": 6.710552763819095e-07,
"loss": 0.002,
"step": 3325
},
{
"epoch": 446.0,
"learning_rate": 6.685427135678392e-07,
"loss": 0.002,
"step": 3350
},
{
"epoch": 449.0,
"learning_rate": 6.660301507537687e-07,
"loss": 0.0019,
"step": 3375
},
{
"epoch": 453.0,
"learning_rate": 6.635175879396985e-07,
"loss": 0.0019,
"step": 3400
},
{
"epoch": 456.0,
"learning_rate": 6.61005025125628e-07,
"loss": 0.0019,
"step": 3425
},
{
"epoch": 459.0,
"learning_rate": 6.584924623115578e-07,
"loss": 0.0019,
"step": 3450
},
{
"epoch": 463.0,
"learning_rate": 6.559798994974874e-07,
"loss": 0.0018,
"step": 3475
},
{
"epoch": 466.0,
"learning_rate": 6.534673366834171e-07,
"loss": 0.0018,
"step": 3500
},
{
"epoch": 469.0,
"learning_rate": 6.509547738693467e-07,
"loss": 0.0018,
"step": 3525
},
{
"epoch": 473.0,
"learning_rate": 6.484422110552763e-07,
"loss": 0.0018,
"step": 3550
},
{
"epoch": 476.0,
"learning_rate": 6.45929648241206e-07,
"loss": 0.0017,
"step": 3575
},
{
"epoch": 479.0,
"learning_rate": 6.434170854271357e-07,
"loss": 0.0017,
"step": 3600
},
{
"epoch": 483.0,
"learning_rate": 6.409045226130653e-07,
"loss": 0.0017,
"step": 3625
},
{
"epoch": 486.0,
"learning_rate": 6.38391959798995e-07,
"loss": 0.0017,
"step": 3650
},
{
"epoch": 489.0,
"learning_rate": 6.358793969849245e-07,
"loss": 0.0017,
"step": 3675
},
{
"epoch": 493.0,
"learning_rate": 6.333668341708542e-07,
"loss": 0.0016,
"step": 3700
},
{
"epoch": 496.0,
"learning_rate": 6.308542713567839e-07,
"loss": 0.0016,
"step": 3725
},
{
"epoch": 499.0,
"learning_rate": 6.283417085427135e-07,
"loss": 0.0016,
"step": 3750
},
{
"epoch": 503.0,
"learning_rate": 6.258291457286432e-07,
"loss": 0.0016,
"step": 3775
},
{
"epoch": 506.0,
"learning_rate": 6.233165829145727e-07,
"loss": 0.0016,
"step": 3800
},
{
"epoch": 509.0,
"learning_rate": 6.208040201005025e-07,
"loss": 0.0015,
"step": 3825
},
{
"epoch": 513.0,
"learning_rate": 6.182914572864321e-07,
"loss": 0.0015,
"step": 3850
},
{
"epoch": 516.0,
"learning_rate": 6.157788944723618e-07,
"loss": 0.0015,
"step": 3875
},
{
"epoch": 519.0,
"learning_rate": 6.132663316582914e-07,
"loss": 0.0015,
"step": 3900
},
{
"epoch": 523.0,
"learning_rate": 6.107537688442211e-07,
"loss": 0.0015,
"step": 3925
},
{
"epoch": 526.0,
"learning_rate": 6.082412060301507e-07,
"loss": 0.0014,
"step": 3950
},
{
"epoch": 529.0,
"learning_rate": 6.057286432160804e-07,
"loss": 0.0014,
"step": 3975
},
{
"epoch": 533.0,
"learning_rate": 6.0321608040201e-07,
"loss": 0.0014,
"step": 4000
},
{
"epoch": 533.0,
"eval_loss": 0.7773712873458862,
"eval_runtime": 21.4192,
"eval_samples_per_second": 12.652,
"eval_steps_per_second": 0.42,
"eval_wer": 49.013949013949016,
"step": 4000
},
{
"epoch": 536.0,
"learning_rate": 6.007035175879397e-07,
"loss": 0.0014,
"step": 4025
},
{
"epoch": 539.0,
"learning_rate": 5.981909547738693e-07,
"loss": 0.0014,
"step": 4050
},
{
"epoch": 543.0,
"learning_rate": 5.95678391959799e-07,
"loss": 0.0014,
"step": 4075
},
{
"epoch": 546.0,
"learning_rate": 5.931658291457287e-07,
"loss": 0.0013,
"step": 4100
},
{
"epoch": 549.0,
"learning_rate": 5.906532663316583e-07,
"loss": 0.0013,
"step": 4125
},
{
"epoch": 553.0,
"learning_rate": 5.881407035175879e-07,
"loss": 0.0013,
"step": 4150
},
{
"epoch": 556.0,
"learning_rate": 5.856281407035175e-07,
"loss": 0.0013,
"step": 4175
},
{
"epoch": 559.0,
"learning_rate": 5.831155778894472e-07,
"loss": 0.0013,
"step": 4200
},
{
"epoch": 563.0,
"learning_rate": 5.806030150753769e-07,
"loss": 0.0013,
"step": 4225
},
{
"epoch": 566.0,
"learning_rate": 5.780904522613065e-07,
"loss": 0.0012,
"step": 4250
},
{
"epoch": 569.0,
"learning_rate": 5.755778894472361e-07,
"loss": 0.0012,
"step": 4275
},
{
"epoch": 573.0,
"learning_rate": 5.730653266331658e-07,
"loss": 0.0012,
"step": 4300
},
{
"epoch": 576.0,
"learning_rate": 5.705527638190954e-07,
"loss": 0.0012,
"step": 4325
},
{
"epoch": 579.0,
"learning_rate": 5.680402010050252e-07,
"loss": 0.0012,
"step": 4350
},
{
"epoch": 583.0,
"learning_rate": 5.655276381909547e-07,
"loss": 0.0012,
"step": 4375
},
{
"epoch": 586.0,
"learning_rate": 5.630150753768844e-07,
"loss": 0.0012,
"step": 4400
},
{
"epoch": 589.0,
"learning_rate": 5.60502512562814e-07,
"loss": 0.0012,
"step": 4425
},
{
"epoch": 593.0,
"learning_rate": 5.579899497487437e-07,
"loss": 0.0011,
"step": 4450
},
{
"epoch": 596.0,
"learning_rate": 5.554773869346734e-07,
"loss": 0.0011,
"step": 4475
},
{
"epoch": 599.0,
"learning_rate": 5.52964824120603e-07,
"loss": 0.0011,
"step": 4500
},
{
"epoch": 603.0,
"learning_rate": 5.504522613065327e-07,
"loss": 0.0011,
"step": 4525
},
{
"epoch": 606.0,
"learning_rate": 5.479396984924623e-07,
"loss": 0.0011,
"step": 4550
},
{
"epoch": 609.0,
"learning_rate": 5.454271356783919e-07,
"loss": 0.0011,
"step": 4575
},
{
"epoch": 613.0,
"learning_rate": 5.429145728643217e-07,
"loss": 0.0011,
"step": 4600
},
{
"epoch": 616.0,
"learning_rate": 5.404020100502512e-07,
"loss": 0.0011,
"step": 4625
},
{
"epoch": 619.0,
"learning_rate": 5.378894472361809e-07,
"loss": 0.0011,
"step": 4650
},
{
"epoch": 623.0,
"learning_rate": 5.353768844221105e-07,
"loss": 0.001,
"step": 4675
},
{
"epoch": 626.0,
"learning_rate": 5.328643216080401e-07,
"loss": 0.001,
"step": 4700
},
{
"epoch": 629.0,
"learning_rate": 5.303517587939699e-07,
"loss": 0.001,
"step": 4725
},
{
"epoch": 633.0,
"learning_rate": 5.278391959798994e-07,
"loss": 0.001,
"step": 4750
},
{
"epoch": 636.0,
"learning_rate": 5.253266331658292e-07,
"loss": 0.001,
"step": 4775
},
{
"epoch": 639.0,
"learning_rate": 5.228140703517587e-07,
"loss": 0.001,
"step": 4800
},
{
"epoch": 643.0,
"learning_rate": 5.203015075376884e-07,
"loss": 0.001,
"step": 4825
},
{
"epoch": 646.0,
"learning_rate": 5.177889447236181e-07,
"loss": 0.001,
"step": 4850
},
{
"epoch": 649.0,
"learning_rate": 5.152763819095477e-07,
"loss": 0.001,
"step": 4875
},
{
"epoch": 653.0,
"learning_rate": 5.127638190954774e-07,
"loss": 0.0009,
"step": 4900
},
{
"epoch": 656.0,
"learning_rate": 5.10251256281407e-07,
"loss": 0.0009,
"step": 4925
},
{
"epoch": 659.0,
"learning_rate": 5.077386934673367e-07,
"loss": 0.0009,
"step": 4950
},
{
"epoch": 663.0,
"learning_rate": 5.052261306532664e-07,
"loss": 0.0009,
"step": 4975
},
{
"epoch": 666.0,
"learning_rate": 5.027135678391959e-07,
"loss": 0.0009,
"step": 5000
},
{
"epoch": 666.0,
"eval_loss": 0.8036791682243347,
"eval_runtime": 21.4474,
"eval_samples_per_second": 12.636,
"eval_steps_per_second": 0.42,
"eval_wer": 48.821548821548824,
"step": 5000
},
{
"epoch": 669.0,
"learning_rate": 5.002010050251257e-07,
"loss": 0.0009,
"step": 5025
},
{
"epoch": 673.0,
"learning_rate": 4.976884422110553e-07,
"loss": 0.0009,
"step": 5050
},
{
"epoch": 676.0,
"learning_rate": 4.951758793969849e-07,
"loss": 0.0009,
"step": 5075
},
{
"epoch": 679.0,
"learning_rate": 4.926633165829145e-07,
"loss": 0.0009,
"step": 5100
},
{
"epoch": 683.0,
"learning_rate": 4.901507537688441e-07,
"loss": 0.0009,
"step": 5125
},
{
"epoch": 686.0,
"learning_rate": 4.876381909547738e-07,
"loss": 0.0008,
"step": 5150
},
{
"epoch": 689.0,
"learning_rate": 4.851256281407035e-07,
"loss": 0.0008,
"step": 5175
},
{
"epoch": 693.0,
"learning_rate": 4.826130653266332e-07,
"loss": 0.0008,
"step": 5200
},
{
"epoch": 696.0,
"learning_rate": 4.801005025125628e-07,
"loss": 0.0008,
"step": 5225
},
{
"epoch": 699.0,
"learning_rate": 4.775879396984925e-07,
"loss": 0.0008,
"step": 5250
},
{
"epoch": 703.0,
"learning_rate": 4.7507537688442204e-07,
"loss": 0.0008,
"step": 5275
},
{
"epoch": 706.0,
"learning_rate": 4.7256281407035173e-07,
"loss": 0.0008,
"step": 5300
},
{
"epoch": 709.0,
"learning_rate": 4.7005025125628137e-07,
"loss": 0.0008,
"step": 5325
},
{
"epoch": 713.0,
"learning_rate": 4.67537688442211e-07,
"loss": 0.0008,
"step": 5350
},
{
"epoch": 716.0,
"learning_rate": 4.6502512562814066e-07,
"loss": 0.0008,
"step": 5375
},
{
"epoch": 719.0,
"learning_rate": 4.625125628140703e-07,
"loss": 0.0008,
"step": 5400
},
{
"epoch": 723.0,
"learning_rate": 4.6e-07,
"loss": 0.0008,
"step": 5425
},
{
"epoch": 726.0,
"learning_rate": 4.5748743718592963e-07,
"loss": 0.0008,
"step": 5450
},
{
"epoch": 729.0,
"learning_rate": 4.5497487437185927e-07,
"loss": 0.0007,
"step": 5475
},
{
"epoch": 733.0,
"learning_rate": 4.524623115577889e-07,
"loss": 0.0007,
"step": 5500
},
{
"epoch": 736.0,
"learning_rate": 4.4994974874371855e-07,
"loss": 0.0007,
"step": 5525
},
{
"epoch": 739.0,
"learning_rate": 4.4743718592964825e-07,
"loss": 0.0007,
"step": 5550
},
{
"epoch": 743.0,
"learning_rate": 4.449246231155779e-07,
"loss": 0.0007,
"step": 5575
},
{
"epoch": 746.0,
"learning_rate": 4.4241206030150753e-07,
"loss": 0.0007,
"step": 5600
},
{
"epoch": 749.0,
"learning_rate": 4.3989949748743717e-07,
"loss": 0.0007,
"step": 5625
},
{
"epoch": 753.0,
"learning_rate": 4.373869346733668e-07,
"loss": 0.0007,
"step": 5650
},
{
"epoch": 756.0,
"learning_rate": 4.348743718592965e-07,
"loss": 0.0007,
"step": 5675
},
{
"epoch": 759.0,
"learning_rate": 4.3236180904522615e-07,
"loss": 0.0007,
"step": 5700
},
{
"epoch": 763.0,
"learning_rate": 4.2984924623115573e-07,
"loss": 0.0007,
"step": 5725
},
{
"epoch": 766.0,
"learning_rate": 4.273366834170854e-07,
"loss": 0.0007,
"step": 5750
},
{
"epoch": 769.0,
"learning_rate": 4.24824120603015e-07,
"loss": 0.0007,
"step": 5775
},
{
"epoch": 773.0,
"learning_rate": 4.223115577889447e-07,
"loss": 0.0007,
"step": 5800
},
{
"epoch": 776.0,
"learning_rate": 4.1979899497487435e-07,
"loss": 0.0007,
"step": 5825
},
{
"epoch": 779.0,
"learning_rate": 4.17286432160804e-07,
"loss": 0.0006,
"step": 5850
},
{
"epoch": 783.0,
"learning_rate": 4.1477386934673363e-07,
"loss": 0.0006,
"step": 5875
},
{
"epoch": 786.0,
"learning_rate": 4.1226130653266327e-07,
"loss": 0.0006,
"step": 5900
},
{
"epoch": 789.0,
"learning_rate": 4.0974874371859297e-07,
"loss": 0.0006,
"step": 5925
},
{
"epoch": 793.0,
"learning_rate": 4.072361809045226e-07,
"loss": 0.0006,
"step": 5950
},
{
"epoch": 796.0,
"learning_rate": 4.0472361809045225e-07,
"loss": 0.0006,
"step": 5975
},
{
"epoch": 799.0,
"learning_rate": 4.022110552763819e-07,
"loss": 0.0006,
"step": 6000
},
{
"epoch": 799.0,
"eval_loss": 0.8268894553184509,
"eval_runtime": 21.3407,
"eval_samples_per_second": 12.699,
"eval_steps_per_second": 0.422,
"eval_wer": 49.44684944684944,
"step": 6000
},
{
"epoch": 803.0,
"learning_rate": 3.9969849246231153e-07,
"loss": 0.0006,
"step": 6025
},
{
"epoch": 806.0,
"learning_rate": 3.971859296482412e-07,
"loss": 0.0006,
"step": 6050
},
{
"epoch": 809.0,
"learning_rate": 3.9467336683417087e-07,
"loss": 0.0006,
"step": 6075
},
{
"epoch": 813.0,
"learning_rate": 3.921608040201005e-07,
"loss": 0.0006,
"step": 6100
},
{
"epoch": 816.0,
"learning_rate": 3.8964824120603015e-07,
"loss": 0.0006,
"step": 6125
},
{
"epoch": 819.0,
"learning_rate": 3.871356783919598e-07,
"loss": 0.0006,
"step": 6150
},
{
"epoch": 823.0,
"learning_rate": 3.846231155778895e-07,
"loss": 0.0006,
"step": 6175
},
{
"epoch": 826.0,
"learning_rate": 3.8211055276381907e-07,
"loss": 0.0006,
"step": 6200
},
{
"epoch": 829.0,
"learning_rate": 3.795979899497487e-07,
"loss": 0.0006,
"step": 6225
},
{
"epoch": 833.0,
"learning_rate": 3.7708542713567835e-07,
"loss": 0.0006,
"step": 6250
},
{
"epoch": 836.0,
"learning_rate": 3.74572864321608e-07,
"loss": 0.0006,
"step": 6275
},
{
"epoch": 839.0,
"learning_rate": 3.7206030150753763e-07,
"loss": 0.0006,
"step": 6300
},
{
"epoch": 843.0,
"learning_rate": 3.6954773869346733e-07,
"loss": 0.0005,
"step": 6325
},
{
"epoch": 846.0,
"learning_rate": 3.6703517587939697e-07,
"loss": 0.0005,
"step": 6350
},
{
"epoch": 849.0,
"learning_rate": 3.645226130653266e-07,
"loss": 0.0005,
"step": 6375
},
{
"epoch": 853.0,
"learning_rate": 3.6201005025125625e-07,
"loss": 0.0005,
"step": 6400
},
{
"epoch": 856.0,
"learning_rate": 3.594974874371859e-07,
"loss": 0.0005,
"step": 6425
},
{
"epoch": 859.0,
"learning_rate": 3.569849246231156e-07,
"loss": 0.0005,
"step": 6450
},
{
"epoch": 863.0,
"learning_rate": 3.544723618090452e-07,
"loss": 0.0005,
"step": 6475
},
{
"epoch": 866.0,
"learning_rate": 3.5195979899497487e-07,
"loss": 0.0005,
"step": 6500
},
{
"epoch": 869.0,
"learning_rate": 3.494472361809045e-07,
"loss": 0.0005,
"step": 6525
},
{
"epoch": 873.0,
"learning_rate": 3.4693467336683415e-07,
"loss": 0.0005,
"step": 6550
},
{
"epoch": 876.0,
"learning_rate": 3.4442211055276384e-07,
"loss": 0.0005,
"step": 6575
},
{
"epoch": 879.0,
"learning_rate": 3.419095477386935e-07,
"loss": 0.0005,
"step": 6600
},
{
"epoch": 883.0,
"learning_rate": 3.393969849246231e-07,
"loss": 0.0005,
"step": 6625
},
{
"epoch": 886.0,
"learning_rate": 3.368844221105527e-07,
"loss": 0.0005,
"step": 6650
},
{
"epoch": 889.0,
"learning_rate": 3.3437185929648235e-07,
"loss": 0.0005,
"step": 6675
},
{
"epoch": 893.0,
"learning_rate": 3.3185929648241205e-07,
"loss": 0.0005,
"step": 6700
},
{
"epoch": 896.0,
"learning_rate": 3.293467336683417e-07,
"loss": 0.0005,
"step": 6725
},
{
"epoch": 899.0,
"learning_rate": 3.2683417085427133e-07,
"loss": 0.0005,
"step": 6750
},
{
"epoch": 903.0,
"learning_rate": 3.2432160804020097e-07,
"loss": 0.0005,
"step": 6775
},
{
"epoch": 906.0,
"learning_rate": 3.218090452261306e-07,
"loss": 0.0005,
"step": 6800
},
{
"epoch": 909.0,
"learning_rate": 3.192964824120603e-07,
"loss": 0.0005,
"step": 6825
},
{
"epoch": 913.0,
"learning_rate": 3.1678391959798994e-07,
"loss": 0.0005,
"step": 6850
},
{
"epoch": 916.0,
"learning_rate": 3.142713567839196e-07,
"loss": 0.0005,
"step": 6875
},
{
"epoch": 919.0,
"learning_rate": 3.1175879396984923e-07,
"loss": 0.0005,
"step": 6900
},
{
"epoch": 923.0,
"learning_rate": 3.0924623115577887e-07,
"loss": 0.0005,
"step": 6925
},
{
"epoch": 926.0,
"learning_rate": 3.0673366834170856e-07,
"loss": 0.0004,
"step": 6950
},
{
"epoch": 929.0,
"learning_rate": 3.042211055276382e-07,
"loss": 0.0004,
"step": 6975
},
{
"epoch": 933.0,
"learning_rate": 3.0170854271356784e-07,
"loss": 0.0004,
"step": 7000
},
{
"epoch": 933.0,
"eval_loss": 0.8481872081756592,
"eval_runtime": 21.2103,
"eval_samples_per_second": 12.777,
"eval_steps_per_second": 0.424,
"eval_wer": 49.398749398749395,
"step": 7000
},
{
"epoch": 936.0,
"learning_rate": 2.991959798994975e-07,
"loss": 0.0004,
"step": 7025
},
{
"epoch": 939.0,
"learning_rate": 2.966834170854271e-07,
"loss": 0.0004,
"step": 7050
},
{
"epoch": 943.0,
"learning_rate": 2.941708542713568e-07,
"loss": 0.0004,
"step": 7075
},
{
"epoch": 946.0,
"learning_rate": 2.916582914572864e-07,
"loss": 0.0004,
"step": 7100
},
{
"epoch": 949.0,
"learning_rate": 2.8914572864321605e-07,
"loss": 0.0004,
"step": 7125
},
{
"epoch": 953.0,
"learning_rate": 2.866331658291457e-07,
"loss": 0.0004,
"step": 7150
},
{
"epoch": 956.0,
"learning_rate": 2.8412060301507533e-07,
"loss": 0.0004,
"step": 7175
},
{
"epoch": 959.0,
"learning_rate": 2.81608040201005e-07,
"loss": 0.0004,
"step": 7200
},
{
"epoch": 963.0,
"learning_rate": 2.7909547738693466e-07,
"loss": 0.0004,
"step": 7225
},
{
"epoch": 966.0,
"learning_rate": 2.765829145728643e-07,
"loss": 0.0004,
"step": 7250
},
{
"epoch": 969.0,
"learning_rate": 2.7407035175879395e-07,
"loss": 0.0004,
"step": 7275
},
{
"epoch": 973.0,
"learning_rate": 2.715577889447236e-07,
"loss": 0.0004,
"step": 7300
},
{
"epoch": 976.0,
"learning_rate": 2.690452261306533e-07,
"loss": 0.0004,
"step": 7325
},
{
"epoch": 979.0,
"learning_rate": 2.665326633165829e-07,
"loss": 0.0004,
"step": 7350
},
{
"epoch": 983.0,
"learning_rate": 2.6402010050251256e-07,
"loss": 0.0004,
"step": 7375
},
{
"epoch": 986.0,
"learning_rate": 2.615075376884422e-07,
"loss": 0.0004,
"step": 7400
},
{
"epoch": 989.0,
"learning_rate": 2.5899497487437184e-07,
"loss": 0.0004,
"step": 7425
},
{
"epoch": 993.0,
"learning_rate": 2.5648241206030154e-07,
"loss": 0.0004,
"step": 7450
},
{
"epoch": 996.0,
"learning_rate": 2.539698492462312e-07,
"loss": 0.0004,
"step": 7475
},
{
"epoch": 999.0,
"learning_rate": 2.514572864321608e-07,
"loss": 0.0004,
"step": 7500
},
{
"epoch": 1003.0,
"learning_rate": 2.4894472361809046e-07,
"loss": 0.0004,
"step": 7525
},
{
"epoch": 1006.0,
"learning_rate": 2.464321608040201e-07,
"loss": 0.0004,
"step": 7550
},
{
"epoch": 1009.0,
"learning_rate": 2.4391959798994974e-07,
"loss": 0.0004,
"step": 7575
},
{
"epoch": 1013.0,
"learning_rate": 2.414070351758794e-07,
"loss": 0.0004,
"step": 7600
},
{
"epoch": 1016.0,
"learning_rate": 2.38894472361809e-07,
"loss": 0.0004,
"step": 7625
},
{
"epoch": 1019.0,
"learning_rate": 2.3638190954773867e-07,
"loss": 0.0004,
"step": 7650
},
{
"epoch": 1023.0,
"learning_rate": 2.3386934673366833e-07,
"loss": 0.0004,
"step": 7675
},
{
"epoch": 1026.0,
"learning_rate": 2.3135678391959797e-07,
"loss": 0.0004,
"step": 7700
},
{
"epoch": 1029.0,
"learning_rate": 2.2884422110552761e-07,
"loss": 0.0004,
"step": 7725
},
{
"epoch": 1033.0,
"learning_rate": 2.2633165829145728e-07,
"loss": 0.0004,
"step": 7750
},
{
"epoch": 1036.0,
"learning_rate": 2.2381909547738692e-07,
"loss": 0.0004,
"step": 7775
},
{
"epoch": 1039.0,
"learning_rate": 2.213065326633166e-07,
"loss": 0.0003,
"step": 7800
},
{
"epoch": 1043.0,
"learning_rate": 2.1879396984924623e-07,
"loss": 0.0004,
"step": 7825
},
{
"epoch": 1046.0,
"learning_rate": 2.1628140703517587e-07,
"loss": 0.0003,
"step": 7850
},
{
"epoch": 1049.0,
"learning_rate": 2.1376884422110554e-07,
"loss": 0.0003,
"step": 7875
},
{
"epoch": 1053.0,
"learning_rate": 2.1125628140703515e-07,
"loss": 0.0003,
"step": 7900
},
{
"epoch": 1056.0,
"learning_rate": 2.087437185929648e-07,
"loss": 0.0003,
"step": 7925
},
{
"epoch": 1059.0,
"learning_rate": 2.0623115577889446e-07,
"loss": 0.0003,
"step": 7950
},
{
"epoch": 1063.0,
"learning_rate": 2.037185929648241e-07,
"loss": 0.0003,
"step": 7975
},
{
"epoch": 1066.0,
"learning_rate": 2.0120603015075377e-07,
"loss": 0.0003,
"step": 8000
},
{
"epoch": 1066.0,
"eval_loss": 0.8661959171295166,
"eval_runtime": 25.6861,
"eval_samples_per_second": 10.55,
"eval_steps_per_second": 0.35,
"eval_wer": 54.64165464165465,
"step": 8000
},
{
"epoch": 1069.0,
"learning_rate": 1.986934673366834e-07,
"loss": 0.0003,
"step": 8025
},
{
"epoch": 1073.0,
"learning_rate": 1.9618090452261305e-07,
"loss": 0.0003,
"step": 8050
},
{
"epoch": 1076.0,
"learning_rate": 1.9366834170854272e-07,
"loss": 0.0003,
"step": 8075
},
{
"epoch": 1079.0,
"learning_rate": 1.9115577889447233e-07,
"loss": 0.0003,
"step": 8100
},
{
"epoch": 1083.0,
"learning_rate": 1.88643216080402e-07,
"loss": 0.0003,
"step": 8125
},
{
"epoch": 1086.0,
"learning_rate": 1.8613065326633164e-07,
"loss": 0.0003,
"step": 8150
},
{
"epoch": 1089.0,
"learning_rate": 1.8361809045226128e-07,
"loss": 0.0003,
"step": 8175
},
{
"epoch": 1093.0,
"learning_rate": 1.8110552763819095e-07,
"loss": 0.0003,
"step": 8200
},
{
"epoch": 1096.0,
"learning_rate": 1.785929648241206e-07,
"loss": 0.0003,
"step": 8225
},
{
"epoch": 1099.0,
"learning_rate": 1.7608040201005026e-07,
"loss": 0.0003,
"step": 8250
},
{
"epoch": 1103.0,
"learning_rate": 1.735678391959799e-07,
"loss": 0.0003,
"step": 8275
},
{
"epoch": 1106.0,
"learning_rate": 1.7105527638190954e-07,
"loss": 0.0003,
"step": 8300
},
{
"epoch": 1109.0,
"learning_rate": 1.685427135678392e-07,
"loss": 0.0003,
"step": 8325
},
{
"epoch": 1113.0,
"learning_rate": 1.6603015075376882e-07,
"loss": 0.0003,
"step": 8350
},
{
"epoch": 1116.0,
"learning_rate": 1.635175879396985e-07,
"loss": 0.0003,
"step": 8375
},
{
"epoch": 1119.0,
"learning_rate": 1.6100502512562813e-07,
"loss": 0.0003,
"step": 8400
},
{
"epoch": 1123.0,
"learning_rate": 1.5849246231155777e-07,
"loss": 0.0003,
"step": 8425
},
{
"epoch": 1126.0,
"learning_rate": 1.5597989949748744e-07,
"loss": 0.0003,
"step": 8450
},
{
"epoch": 1129.0,
"learning_rate": 1.5346733668341708e-07,
"loss": 0.0003,
"step": 8475
},
{
"epoch": 1133.0,
"learning_rate": 1.5095477386934672e-07,
"loss": 0.0003,
"step": 8500
},
{
"epoch": 1136.0,
"learning_rate": 1.484422110552764e-07,
"loss": 0.0003,
"step": 8525
},
{
"epoch": 1139.0,
"learning_rate": 1.45929648241206e-07,
"loss": 0.0003,
"step": 8550
},
{
"epoch": 1143.0,
"learning_rate": 1.4341708542713567e-07,
"loss": 0.0003,
"step": 8575
},
{
"epoch": 1146.0,
"learning_rate": 1.409045226130653e-07,
"loss": 0.0003,
"step": 8600
},
{
"epoch": 1149.0,
"learning_rate": 1.3839195979899495e-07,
"loss": 0.0003,
"step": 8625
},
{
"epoch": 1153.0,
"learning_rate": 1.3587939698492462e-07,
"loss": 0.0003,
"step": 8650
},
{
"epoch": 1156.0,
"learning_rate": 1.3336683417085426e-07,
"loss": 0.0003,
"step": 8675
},
{
"epoch": 1159.0,
"learning_rate": 1.3085427135678393e-07,
"loss": 0.0003,
"step": 8700
},
{
"epoch": 1163.0,
"learning_rate": 1.2834170854271357e-07,
"loss": 0.0003,
"step": 8725
},
{
"epoch": 1166.0,
"learning_rate": 1.258291457286432e-07,
"loss": 0.0003,
"step": 8750
},
{
"epoch": 1169.0,
"learning_rate": 1.2331658291457288e-07,
"loss": 0.0003,
"step": 8775
},
{
"epoch": 1173.0,
"learning_rate": 1.208040201005025e-07,
"loss": 0.0003,
"step": 8800
},
{
"epoch": 1176.0,
"learning_rate": 1.1829145728643216e-07,
"loss": 0.0003,
"step": 8825
},
{
"epoch": 1179.0,
"learning_rate": 1.157788944723618e-07,
"loss": 0.0003,
"step": 8850
},
{
"epoch": 1183.0,
"learning_rate": 1.1326633165829145e-07,
"loss": 0.0003,
"step": 8875
},
{
"epoch": 1186.0,
"learning_rate": 1.1075376884422111e-07,
"loss": 0.0003,
"step": 8900
},
{
"epoch": 1189.0,
"learning_rate": 1.0824120603015075e-07,
"loss": 0.0003,
"step": 8925
},
{
"epoch": 1193.0,
"learning_rate": 1.0572864321608039e-07,
"loss": 0.0003,
"step": 8950
},
{
"epoch": 1196.0,
"learning_rate": 1.0321608040201004e-07,
"loss": 0.0003,
"step": 8975
},
{
"epoch": 1199.0,
"learning_rate": 1.007035175879397e-07,
"loss": 0.0003,
"step": 9000
},
{
"epoch": 1199.0,
"eval_loss": 0.8799856305122375,
"eval_runtime": 21.6525,
"eval_samples_per_second": 12.516,
"eval_steps_per_second": 0.416,
"eval_wer": 49.92784992784993,
"step": 9000
},
{
"epoch": 1203.0,
"learning_rate": 9.819095477386934e-08,
"loss": 0.0003,
"step": 9025
},
{
"epoch": 1206.0,
"learning_rate": 9.567839195979899e-08,
"loss": 0.0003,
"step": 9050
},
{
"epoch": 1209.0,
"learning_rate": 9.316582914572863e-08,
"loss": 0.0003,
"step": 9075
},
{
"epoch": 1213.0,
"learning_rate": 9.065326633165829e-08,
"loss": 0.0003,
"step": 9100
},
{
"epoch": 1216.0,
"learning_rate": 8.814070351758794e-08,
"loss": 0.0003,
"step": 9125
},
{
"epoch": 1219.0,
"learning_rate": 8.562814070351758e-08,
"loss": 0.0003,
"step": 9150
},
{
"epoch": 1223.0,
"learning_rate": 8.311557788944722e-08,
"loss": 0.0003,
"step": 9175
},
{
"epoch": 1226.0,
"learning_rate": 8.060301507537688e-08,
"loss": 0.0003,
"step": 9200
},
{
"epoch": 1229.0,
"learning_rate": 7.809045226130653e-08,
"loss": 0.0003,
"step": 9225
},
{
"epoch": 1233.0,
"learning_rate": 7.557788944723617e-08,
"loss": 0.0003,
"step": 9250
},
{
"epoch": 1236.0,
"learning_rate": 7.306532663316583e-08,
"loss": 0.0003,
"step": 9275
},
{
"epoch": 1239.0,
"learning_rate": 7.055276381909547e-08,
"loss": 0.0003,
"step": 9300
},
{
"epoch": 1243.0,
"learning_rate": 6.804020100502512e-08,
"loss": 0.0003,
"step": 9325
},
{
"epoch": 1246.0,
"learning_rate": 6.552763819095478e-08,
"loss": 0.0003,
"step": 9350
},
{
"epoch": 1249.0,
"learning_rate": 6.301507537688442e-08,
"loss": 0.0003,
"step": 9375
},
{
"epoch": 1253.0,
"learning_rate": 6.050251256281406e-08,
"loss": 0.0003,
"step": 9400
},
{
"epoch": 1256.0,
"learning_rate": 5.798994974874372e-08,
"loss": 0.0003,
"step": 9425
},
{
"epoch": 1259.0,
"learning_rate": 5.547738693467337e-08,
"loss": 0.0003,
"step": 9450
},
{
"epoch": 1263.0,
"learning_rate": 5.2964824120603014e-08,
"loss": 0.0003,
"step": 9475
},
{
"epoch": 1266.0,
"learning_rate": 5.045226130653266e-08,
"loss": 0.0003,
"step": 9500
},
{
"epoch": 1269.0,
"learning_rate": 4.793969849246231e-08,
"loss": 0.0003,
"step": 9525
},
{
"epoch": 1273.0,
"learning_rate": 4.542713567839196e-08,
"loss": 0.0003,
"step": 9550
},
{
"epoch": 1276.0,
"learning_rate": 4.2914572864321604e-08,
"loss": 0.0003,
"step": 9575
},
{
"epoch": 1279.0,
"learning_rate": 4.040201005025125e-08,
"loss": 0.0003,
"step": 9600
},
{
"epoch": 1283.0,
"learning_rate": 3.7889447236180906e-08,
"loss": 0.0003,
"step": 9625
},
{
"epoch": 1286.0,
"learning_rate": 3.5376884422110554e-08,
"loss": 0.0003,
"step": 9650
},
{
"epoch": 1289.0,
"learning_rate": 3.28643216080402e-08,
"loss": 0.0003,
"step": 9675
},
{
"epoch": 1293.0,
"learning_rate": 3.035175879396985e-08,
"loss": 0.0003,
"step": 9700
},
{
"epoch": 1296.0,
"learning_rate": 2.7839195979899496e-08,
"loss": 0.0003,
"step": 9725
},
{
"epoch": 1299.0,
"learning_rate": 2.5326633165829147e-08,
"loss": 0.0003,
"step": 9750
},
{
"epoch": 1303.0,
"learning_rate": 2.2814070351758795e-08,
"loss": 0.0003,
"step": 9775
},
{
"epoch": 1306.0,
"learning_rate": 2.0301507537688442e-08,
"loss": 0.0003,
"step": 9800
},
{
"epoch": 1309.0,
"learning_rate": 1.778894472361809e-08,
"loss": 0.0003,
"step": 9825
},
{
"epoch": 1313.0,
"learning_rate": 1.527638190954774e-08,
"loss": 0.0003,
"step": 9850
},
{
"epoch": 1316.0,
"learning_rate": 1.2763819095477386e-08,
"loss": 0.0003,
"step": 9875
},
{
"epoch": 1319.0,
"learning_rate": 1.0251256281407034e-08,
"loss": 0.0003,
"step": 9900
},
{
"epoch": 1323.0,
"learning_rate": 7.738693467336683e-09,
"loss": 0.0003,
"step": 9925
},
{
"epoch": 1326.0,
"learning_rate": 5.226130653266331e-09,
"loss": 0.0003,
"step": 9950
},
{
"epoch": 1329.0,
"learning_rate": 2.7135678391959797e-09,
"loss": 0.0003,
"step": 9975
},
{
"epoch": 1333.0,
"learning_rate": 2.0100502512562814e-10,
"loss": 0.0003,
"step": 10000
},
{
"epoch": 1333.0,
"eval_loss": 0.885601282119751,
"eval_runtime": 21.3869,
"eval_samples_per_second": 12.671,
"eval_steps_per_second": 0.421,
"eval_wer": 49.831649831649834,
"step": 10000
},
{
"epoch": 1333.0,
"step": 10000,
"total_flos": 3.926247979548672e+19,
"train_loss": 0.017836267329892144,
"train_runtime": 30544.6776,
"train_samples_per_second": 20.953,
"train_steps_per_second": 0.327
}
],
"logging_steps": 25,
"max_steps": 10000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 1000,
"total_flos": 3.926247979548672e+19,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}