mehran's picture
Uploading check points
93b868d
raw
history blame
25.5 kB
{
"best_metric": 33.46883689165677,
"best_model_checkpoint": "./whisper-small-fa-aug/checkpoint-5000",
"epoch": 2.03745,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.2000000000000006e-07,
"loss": 3.8,
"step": 25
},
{
"epoch": 0.0,
"learning_rate": 9.200000000000001e-07,
"loss": 3.0497,
"step": 50
},
{
"epoch": 0.0,
"learning_rate": 1.42e-06,
"loss": 2.0102,
"step": 75
},
{
"epoch": 0.0,
"learning_rate": 1.9200000000000003e-06,
"loss": 1.5948,
"step": 100
},
{
"epoch": 0.0,
"learning_rate": 2.42e-06,
"loss": 1.4094,
"step": 125
},
{
"epoch": 0.0,
"learning_rate": 2.92e-06,
"loss": 1.2611,
"step": 150
},
{
"epoch": 0.0,
"learning_rate": 3.4200000000000007e-06,
"loss": 1.1366,
"step": 175
},
{
"epoch": 0.01,
"learning_rate": 3.920000000000001e-06,
"loss": 1.0817,
"step": 200
},
{
"epoch": 0.01,
"learning_rate": 4.42e-06,
"loss": 1.0324,
"step": 225
},
{
"epoch": 0.01,
"learning_rate": 4.92e-06,
"loss": 0.9176,
"step": 250
},
{
"epoch": 0.01,
"learning_rate": 5.420000000000001e-06,
"loss": 0.7975,
"step": 275
},
{
"epoch": 0.01,
"learning_rate": 5.92e-06,
"loss": 0.7079,
"step": 300
},
{
"epoch": 0.01,
"learning_rate": 6.42e-06,
"loss": 0.5832,
"step": 325
},
{
"epoch": 0.01,
"learning_rate": 6.92e-06,
"loss": 0.4469,
"step": 350
},
{
"epoch": 0.01,
"learning_rate": 7.420000000000001e-06,
"loss": 0.3913,
"step": 375
},
{
"epoch": 0.01,
"learning_rate": 7.92e-06,
"loss": 0.4189,
"step": 400
},
{
"epoch": 0.01,
"learning_rate": 8.42e-06,
"loss": 0.3756,
"step": 425
},
{
"epoch": 0.01,
"learning_rate": 8.920000000000001e-06,
"loss": 0.3876,
"step": 450
},
{
"epoch": 0.01,
"learning_rate": 9.42e-06,
"loss": 0.3802,
"step": 475
},
{
"epoch": 0.01,
"learning_rate": 9.920000000000002e-06,
"loss": 0.4061,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 9.994683544303798e-06,
"loss": 0.3885,
"step": 525
},
{
"epoch": 0.01,
"learning_rate": 9.988354430379748e-06,
"loss": 0.3449,
"step": 550
},
{
"epoch": 0.01,
"learning_rate": 9.982025316455697e-06,
"loss": 0.3292,
"step": 575
},
{
"epoch": 0.01,
"learning_rate": 9.975696202531647e-06,
"loss": 0.3439,
"step": 600
},
{
"epoch": 0.02,
"learning_rate": 9.969367088607596e-06,
"loss": 0.3031,
"step": 625
},
{
"epoch": 0.02,
"learning_rate": 9.963037974683545e-06,
"loss": 0.3159,
"step": 650
},
{
"epoch": 0.02,
"learning_rate": 9.956708860759495e-06,
"loss": 0.3304,
"step": 675
},
{
"epoch": 0.02,
"learning_rate": 9.950379746835444e-06,
"loss": 0.2918,
"step": 700
},
{
"epoch": 0.02,
"learning_rate": 9.944050632911392e-06,
"loss": 0.2785,
"step": 725
},
{
"epoch": 0.02,
"learning_rate": 9.937721518987341e-06,
"loss": 0.2618,
"step": 750
},
{
"epoch": 0.02,
"learning_rate": 9.931392405063291e-06,
"loss": 0.2862,
"step": 775
},
{
"epoch": 0.02,
"learning_rate": 9.92506329113924e-06,
"loss": 0.2962,
"step": 800
},
{
"epoch": 0.02,
"learning_rate": 9.91873417721519e-06,
"loss": 0.2935,
"step": 825
},
{
"epoch": 0.02,
"learning_rate": 9.912405063291141e-06,
"loss": 0.2827,
"step": 850
},
{
"epoch": 0.02,
"learning_rate": 9.90607594936709e-06,
"loss": 0.2702,
"step": 875
},
{
"epoch": 0.02,
"learning_rate": 9.89974683544304e-06,
"loss": 0.2987,
"step": 900
},
{
"epoch": 0.02,
"learning_rate": 9.89341772151899e-06,
"loss": 0.3067,
"step": 925
},
{
"epoch": 0.02,
"learning_rate": 9.887088607594939e-06,
"loss": 0.2962,
"step": 950
},
{
"epoch": 0.02,
"learning_rate": 9.880759493670888e-06,
"loss": 0.2807,
"step": 975
},
{
"epoch": 0.03,
"learning_rate": 9.874430379746836e-06,
"loss": 0.2533,
"step": 1000
},
{
"epoch": 0.03,
"eval_loss": 0.3274388611316681,
"eval_runtime": 1481.9106,
"eval_samples_per_second": 7.025,
"eval_steps_per_second": 0.439,
"eval_wer": 40.657571667832,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 9.868101265822785e-06,
"loss": 0.21,
"step": 1025
},
{
"epoch": 0.03,
"learning_rate": 9.861772151898735e-06,
"loss": 0.2463,
"step": 1050
},
{
"epoch": 0.03,
"learning_rate": 9.855443037974684e-06,
"loss": 0.2688,
"step": 1075
},
{
"epoch": 0.03,
"learning_rate": 9.849113924050633e-06,
"loss": 0.2293,
"step": 1100
},
{
"epoch": 0.03,
"learning_rate": 9.842784810126583e-06,
"loss": 0.2283,
"step": 1125
},
{
"epoch": 0.03,
"learning_rate": 9.836455696202532e-06,
"loss": 0.228,
"step": 1150
},
{
"epoch": 0.03,
"learning_rate": 9.830126582278482e-06,
"loss": 0.1919,
"step": 1175
},
{
"epoch": 0.03,
"learning_rate": 9.823797468354431e-06,
"loss": 0.2173,
"step": 1200
},
{
"epoch": 0.03,
"learning_rate": 9.81746835443038e-06,
"loss": 0.2066,
"step": 1225
},
{
"epoch": 0.03,
"learning_rate": 9.81113924050633e-06,
"loss": 0.1826,
"step": 1250
},
{
"epoch": 0.03,
"learning_rate": 9.80481012658228e-06,
"loss": 0.1855,
"step": 1275
},
{
"epoch": 0.03,
"learning_rate": 9.798481012658229e-06,
"loss": 0.1702,
"step": 1300
},
{
"epoch": 0.03,
"learning_rate": 9.792151898734178e-06,
"loss": 0.1714,
"step": 1325
},
{
"epoch": 0.03,
"learning_rate": 9.785822784810128e-06,
"loss": 0.1686,
"step": 1350
},
{
"epoch": 0.03,
"learning_rate": 9.779493670886077e-06,
"loss": 0.1762,
"step": 1375
},
{
"epoch": 0.04,
"learning_rate": 9.773164556962027e-06,
"loss": 0.1722,
"step": 1400
},
{
"epoch": 0.04,
"learning_rate": 9.766835443037976e-06,
"loss": 0.2051,
"step": 1425
},
{
"epoch": 0.04,
"learning_rate": 9.760506329113924e-06,
"loss": 0.2071,
"step": 1450
},
{
"epoch": 0.04,
"learning_rate": 9.754177215189873e-06,
"loss": 0.1664,
"step": 1475
},
{
"epoch": 0.04,
"learning_rate": 9.747848101265823e-06,
"loss": 0.1871,
"step": 1500
},
{
"epoch": 0.04,
"learning_rate": 9.741518987341772e-06,
"loss": 0.1893,
"step": 1525
},
{
"epoch": 0.04,
"learning_rate": 9.735189873417721e-06,
"loss": 0.1756,
"step": 1550
},
{
"epoch": 0.04,
"learning_rate": 9.728860759493671e-06,
"loss": 0.1687,
"step": 1575
},
{
"epoch": 0.04,
"learning_rate": 9.72253164556962e-06,
"loss": 0.1685,
"step": 1600
},
{
"epoch": 0.04,
"learning_rate": 9.71620253164557e-06,
"loss": 0.1805,
"step": 1625
},
{
"epoch": 0.04,
"learning_rate": 9.70987341772152e-06,
"loss": 0.1868,
"step": 1650
},
{
"epoch": 0.04,
"learning_rate": 9.70354430379747e-06,
"loss": 0.212,
"step": 1675
},
{
"epoch": 0.04,
"learning_rate": 9.69721518987342e-06,
"loss": 0.2019,
"step": 1700
},
{
"epoch": 0.04,
"learning_rate": 9.69088607594937e-06,
"loss": 0.2076,
"step": 1725
},
{
"epoch": 0.04,
"learning_rate": 9.684556962025317e-06,
"loss": 0.1971,
"step": 1750
},
{
"epoch": 1.0,
"learning_rate": 9.678227848101266e-06,
"loss": 0.2424,
"step": 1775
},
{
"epoch": 1.0,
"learning_rate": 9.671898734177216e-06,
"loss": 0.2346,
"step": 1800
},
{
"epoch": 1.0,
"learning_rate": 9.665569620253165e-06,
"loss": 0.2436,
"step": 1825
},
{
"epoch": 1.0,
"learning_rate": 9.659240506329115e-06,
"loss": 0.2613,
"step": 1850
},
{
"epoch": 1.0,
"learning_rate": 9.652911392405064e-06,
"loss": 0.2781,
"step": 1875
},
{
"epoch": 1.0,
"learning_rate": 9.646582278481013e-06,
"loss": 0.2465,
"step": 1900
},
{
"epoch": 1.0,
"learning_rate": 9.640253164556963e-06,
"loss": 0.2386,
"step": 1925
},
{
"epoch": 1.0,
"learning_rate": 9.633924050632912e-06,
"loss": 0.2298,
"step": 1950
},
{
"epoch": 1.01,
"learning_rate": 9.627594936708862e-06,
"loss": 0.264,
"step": 1975
},
{
"epoch": 1.01,
"learning_rate": 9.621265822784811e-06,
"loss": 0.2077,
"step": 2000
},
{
"epoch": 1.01,
"eval_loss": 0.25397753715515137,
"eval_runtime": 1477.9319,
"eval_samples_per_second": 7.044,
"eval_steps_per_second": 0.44,
"eval_wer": 35.12252231743606,
"step": 2000
},
{
"epoch": 1.01,
"learning_rate": 9.61493670886076e-06,
"loss": 0.192,
"step": 2025
},
{
"epoch": 1.01,
"learning_rate": 9.60860759493671e-06,
"loss": 0.1903,
"step": 2050
},
{
"epoch": 1.01,
"learning_rate": 9.60227848101266e-06,
"loss": 0.2135,
"step": 2075
},
{
"epoch": 1.01,
"learning_rate": 9.595949367088609e-06,
"loss": 0.1917,
"step": 2100
},
{
"epoch": 1.01,
"learning_rate": 9.589620253164558e-06,
"loss": 0.1579,
"step": 2125
},
{
"epoch": 1.01,
"learning_rate": 9.583291139240508e-06,
"loss": 0.1779,
"step": 2150
},
{
"epoch": 1.01,
"learning_rate": 9.576962025316457e-06,
"loss": 0.1504,
"step": 2175
},
{
"epoch": 1.01,
"learning_rate": 9.570632911392405e-06,
"loss": 0.1574,
"step": 2200
},
{
"epoch": 1.01,
"learning_rate": 9.564303797468354e-06,
"loss": 0.1757,
"step": 2225
},
{
"epoch": 1.01,
"learning_rate": 9.557974683544304e-06,
"loss": 0.1996,
"step": 2250
},
{
"epoch": 1.01,
"learning_rate": 9.551645569620253e-06,
"loss": 0.1716,
"step": 2275
},
{
"epoch": 1.01,
"learning_rate": 9.545316455696203e-06,
"loss": 0.1509,
"step": 2300
},
{
"epoch": 1.01,
"learning_rate": 9.538987341772152e-06,
"loss": 0.1402,
"step": 2325
},
{
"epoch": 1.01,
"learning_rate": 9.532658227848101e-06,
"loss": 0.1612,
"step": 2350
},
{
"epoch": 1.02,
"learning_rate": 9.526329113924051e-06,
"loss": 0.1357,
"step": 2375
},
{
"epoch": 1.02,
"learning_rate": 9.52e-06,
"loss": 0.1426,
"step": 2400
},
{
"epoch": 1.02,
"learning_rate": 9.51367088607595e-06,
"loss": 0.1644,
"step": 2425
},
{
"epoch": 1.02,
"learning_rate": 9.5073417721519e-06,
"loss": 0.1196,
"step": 2450
},
{
"epoch": 1.02,
"learning_rate": 9.501012658227849e-06,
"loss": 0.1332,
"step": 2475
},
{
"epoch": 1.02,
"learning_rate": 9.494683544303798e-06,
"loss": 0.1096,
"step": 2500
},
{
"epoch": 1.02,
"learning_rate": 9.488354430379747e-06,
"loss": 0.1263,
"step": 2525
},
{
"epoch": 1.02,
"learning_rate": 9.482025316455697e-06,
"loss": 0.1467,
"step": 2550
},
{
"epoch": 1.02,
"learning_rate": 9.475696202531646e-06,
"loss": 0.1372,
"step": 2575
},
{
"epoch": 1.02,
"learning_rate": 9.469367088607596e-06,
"loss": 0.1256,
"step": 2600
},
{
"epoch": 1.02,
"learning_rate": 9.463037974683545e-06,
"loss": 0.1319,
"step": 2625
},
{
"epoch": 1.02,
"learning_rate": 9.456708860759495e-06,
"loss": 0.1385,
"step": 2650
},
{
"epoch": 1.02,
"learning_rate": 9.450379746835444e-06,
"loss": 0.147,
"step": 2675
},
{
"epoch": 1.02,
"learning_rate": 9.444050632911393e-06,
"loss": 0.1527,
"step": 2700
},
{
"epoch": 1.02,
"learning_rate": 9.437721518987343e-06,
"loss": 0.1357,
"step": 2725
},
{
"epoch": 1.02,
"learning_rate": 9.431392405063292e-06,
"loss": 0.122,
"step": 2750
},
{
"epoch": 1.03,
"learning_rate": 9.425063291139242e-06,
"loss": 0.0918,
"step": 2775
},
{
"epoch": 1.03,
"learning_rate": 9.418734177215191e-06,
"loss": 0.1218,
"step": 2800
},
{
"epoch": 1.03,
"learning_rate": 9.41240506329114e-06,
"loss": 0.139,
"step": 2825
},
{
"epoch": 1.03,
"learning_rate": 9.40607594936709e-06,
"loss": 0.1205,
"step": 2850
},
{
"epoch": 1.03,
"learning_rate": 9.39974683544304e-06,
"loss": 0.1158,
"step": 2875
},
{
"epoch": 1.03,
"learning_rate": 9.393417721518989e-06,
"loss": 0.1156,
"step": 2900
},
{
"epoch": 1.03,
"learning_rate": 9.387088607594937e-06,
"loss": 0.1044,
"step": 2925
},
{
"epoch": 1.03,
"learning_rate": 9.380759493670886e-06,
"loss": 0.112,
"step": 2950
},
{
"epoch": 1.03,
"learning_rate": 9.374430379746835e-06,
"loss": 0.1023,
"step": 2975
},
{
"epoch": 1.03,
"learning_rate": 9.368101265822785e-06,
"loss": 0.0966,
"step": 3000
},
{
"epoch": 1.03,
"eval_loss": 0.2638459801673889,
"eval_runtime": 1500.936,
"eval_samples_per_second": 6.936,
"eval_steps_per_second": 0.434,
"eval_wer": 37.93069806013106,
"step": 3000
},
{
"epoch": 1.03,
"learning_rate": 9.361772151898734e-06,
"loss": 0.0942,
"step": 3025
},
{
"epoch": 1.03,
"learning_rate": 9.355443037974684e-06,
"loss": 0.0826,
"step": 3050
},
{
"epoch": 1.03,
"learning_rate": 9.349113924050633e-06,
"loss": 0.086,
"step": 3075
},
{
"epoch": 1.03,
"learning_rate": 9.342784810126583e-06,
"loss": 0.0956,
"step": 3100
},
{
"epoch": 1.03,
"learning_rate": 9.336455696202532e-06,
"loss": 0.0935,
"step": 3125
},
{
"epoch": 1.03,
"learning_rate": 9.330126582278481e-06,
"loss": 0.0923,
"step": 3150
},
{
"epoch": 1.04,
"learning_rate": 9.323797468354431e-06,
"loss": 0.1163,
"step": 3175
},
{
"epoch": 1.04,
"learning_rate": 9.31746835443038e-06,
"loss": 0.1099,
"step": 3200
},
{
"epoch": 1.04,
"learning_rate": 9.31113924050633e-06,
"loss": 0.0879,
"step": 3225
},
{
"epoch": 1.04,
"learning_rate": 9.30481012658228e-06,
"loss": 0.0994,
"step": 3250
},
{
"epoch": 1.04,
"learning_rate": 9.298481012658229e-06,
"loss": 0.1069,
"step": 3275
},
{
"epoch": 1.04,
"learning_rate": 9.292151898734178e-06,
"loss": 0.0957,
"step": 3300
},
{
"epoch": 1.04,
"learning_rate": 9.285822784810127e-06,
"loss": 0.0866,
"step": 3325
},
{
"epoch": 1.04,
"learning_rate": 9.279493670886077e-06,
"loss": 0.0965,
"step": 3350
},
{
"epoch": 1.04,
"learning_rate": 9.273164556962026e-06,
"loss": 0.0952,
"step": 3375
},
{
"epoch": 1.04,
"learning_rate": 9.266835443037976e-06,
"loss": 0.1178,
"step": 3400
},
{
"epoch": 1.04,
"learning_rate": 9.260506329113925e-06,
"loss": 0.1154,
"step": 3425
},
{
"epoch": 1.04,
"learning_rate": 9.254177215189875e-06,
"loss": 0.1075,
"step": 3450
},
{
"epoch": 1.04,
"learning_rate": 9.247848101265824e-06,
"loss": 0.1161,
"step": 3475
},
{
"epoch": 1.04,
"learning_rate": 9.241518987341773e-06,
"loss": 0.1153,
"step": 3500
},
{
"epoch": 2.0,
"learning_rate": 9.235189873417723e-06,
"loss": 0.1343,
"step": 3525
},
{
"epoch": 2.0,
"learning_rate": 9.228860759493672e-06,
"loss": 0.1374,
"step": 3550
},
{
"epoch": 2.0,
"learning_rate": 9.222531645569622e-06,
"loss": 0.1436,
"step": 3575
},
{
"epoch": 2.0,
"learning_rate": 9.216202531645571e-06,
"loss": 0.1561,
"step": 3600
},
{
"epoch": 2.0,
"learning_rate": 9.20987341772152e-06,
"loss": 0.1709,
"step": 3625
},
{
"epoch": 2.0,
"learning_rate": 9.20354430379747e-06,
"loss": 0.142,
"step": 3650
},
{
"epoch": 2.0,
"learning_rate": 9.197215189873418e-06,
"loss": 0.1389,
"step": 3675
},
{
"epoch": 2.0,
"learning_rate": 9.190886075949367e-06,
"loss": 0.1398,
"step": 3700
},
{
"epoch": 2.01,
"learning_rate": 9.184556962025317e-06,
"loss": 0.1526,
"step": 3725
},
{
"epoch": 2.01,
"learning_rate": 9.178227848101266e-06,
"loss": 0.1273,
"step": 3750
},
{
"epoch": 2.01,
"learning_rate": 9.171898734177215e-06,
"loss": 0.1119,
"step": 3775
},
{
"epoch": 2.01,
"learning_rate": 9.165569620253165e-06,
"loss": 0.1088,
"step": 3800
},
{
"epoch": 2.01,
"learning_rate": 9.159493670886076e-06,
"loss": 0.1362,
"step": 3825
},
{
"epoch": 2.01,
"learning_rate": 9.153164556962026e-06,
"loss": 0.1247,
"step": 3850
},
{
"epoch": 2.01,
"learning_rate": 9.146835443037975e-06,
"loss": 0.0974,
"step": 3875
},
{
"epoch": 2.01,
"learning_rate": 9.140506329113925e-06,
"loss": 0.1218,
"step": 3900
},
{
"epoch": 2.01,
"learning_rate": 9.134177215189874e-06,
"loss": 0.0846,
"step": 3925
},
{
"epoch": 2.01,
"learning_rate": 9.127848101265823e-06,
"loss": 0.0964,
"step": 3950
},
{
"epoch": 2.01,
"learning_rate": 9.121518987341773e-06,
"loss": 0.1166,
"step": 3975
},
{
"epoch": 2.01,
"learning_rate": 9.115189873417722e-06,
"loss": 0.1212,
"step": 4000
},
{
"epoch": 2.01,
"eval_loss": 0.22517943382263184,
"eval_runtime": 1537.556,
"eval_samples_per_second": 6.771,
"eval_steps_per_second": 0.423,
"eval_wer": 69.04990325045935,
"step": 4000
},
{
"epoch": 2.01,
"learning_rate": 9.108860759493672e-06,
"loss": 0.1076,
"step": 4025
},
{
"epoch": 2.01,
"learning_rate": 9.102531645569621e-06,
"loss": 0.086,
"step": 4050
},
{
"epoch": 2.01,
"learning_rate": 9.09620253164557e-06,
"loss": 0.0773,
"step": 4075
},
{
"epoch": 2.01,
"learning_rate": 9.08987341772152e-06,
"loss": 0.098,
"step": 4100
},
{
"epoch": 2.02,
"learning_rate": 9.08354430379747e-06,
"loss": 0.0747,
"step": 4125
},
{
"epoch": 2.02,
"learning_rate": 9.077215189873419e-06,
"loss": 0.0895,
"step": 4150
},
{
"epoch": 2.02,
"learning_rate": 9.070886075949368e-06,
"loss": 0.0957,
"step": 4175
},
{
"epoch": 2.02,
"learning_rate": 9.064556962025318e-06,
"loss": 0.0789,
"step": 4200
},
{
"epoch": 2.02,
"learning_rate": 9.058227848101267e-06,
"loss": 0.0718,
"step": 4225
},
{
"epoch": 2.02,
"learning_rate": 9.051898734177215e-06,
"loss": 0.0737,
"step": 4250
},
{
"epoch": 2.02,
"learning_rate": 9.045569620253164e-06,
"loss": 0.0731,
"step": 4275
},
{
"epoch": 2.02,
"learning_rate": 9.039240506329114e-06,
"loss": 0.087,
"step": 4300
},
{
"epoch": 2.02,
"learning_rate": 9.032911392405063e-06,
"loss": 0.0875,
"step": 4325
},
{
"epoch": 2.02,
"learning_rate": 9.026582278481013e-06,
"loss": 0.0768,
"step": 4350
},
{
"epoch": 2.02,
"learning_rate": 9.020253164556962e-06,
"loss": 0.0842,
"step": 4375
},
{
"epoch": 2.02,
"learning_rate": 9.013924050632911e-06,
"loss": 0.092,
"step": 4400
},
{
"epoch": 2.02,
"learning_rate": 9.007594936708861e-06,
"loss": 0.0949,
"step": 4425
},
{
"epoch": 2.02,
"learning_rate": 9.00126582278481e-06,
"loss": 0.0989,
"step": 4450
},
{
"epoch": 2.02,
"learning_rate": 8.99493670886076e-06,
"loss": 0.0908,
"step": 4475
},
{
"epoch": 2.02,
"learning_rate": 8.98860759493671e-06,
"loss": 0.0783,
"step": 4500
},
{
"epoch": 2.03,
"learning_rate": 8.982278481012659e-06,
"loss": 0.0651,
"step": 4525
},
{
"epoch": 2.03,
"learning_rate": 8.975949367088608e-06,
"loss": 0.0791,
"step": 4550
},
{
"epoch": 2.03,
"learning_rate": 8.969620253164557e-06,
"loss": 0.0829,
"step": 4575
},
{
"epoch": 2.03,
"learning_rate": 8.963291139240507e-06,
"loss": 0.0797,
"step": 4600
},
{
"epoch": 2.03,
"learning_rate": 8.956962025316456e-06,
"loss": 0.0743,
"step": 4625
},
{
"epoch": 2.03,
"learning_rate": 8.950632911392406e-06,
"loss": 0.0814,
"step": 4650
},
{
"epoch": 2.03,
"learning_rate": 8.944303797468355e-06,
"loss": 0.0594,
"step": 4675
},
{
"epoch": 2.03,
"learning_rate": 8.937974683544305e-06,
"loss": 0.0709,
"step": 4700
},
{
"epoch": 2.03,
"learning_rate": 8.931645569620254e-06,
"loss": 0.0621,
"step": 4725
},
{
"epoch": 2.03,
"learning_rate": 8.925316455696203e-06,
"loss": 0.0599,
"step": 4750
},
{
"epoch": 2.03,
"learning_rate": 8.918987341772153e-06,
"loss": 0.0625,
"step": 4775
},
{
"epoch": 2.03,
"learning_rate": 8.912658227848102e-06,
"loss": 0.0541,
"step": 4800
},
{
"epoch": 2.03,
"learning_rate": 8.906329113924052e-06,
"loss": 0.059,
"step": 4825
},
{
"epoch": 2.03,
"learning_rate": 8.900000000000001e-06,
"loss": 0.0576,
"step": 4850
},
{
"epoch": 2.03,
"learning_rate": 8.89367088607595e-06,
"loss": 0.0623,
"step": 4875
},
{
"epoch": 2.03,
"learning_rate": 8.8873417721519e-06,
"loss": 0.0592,
"step": 4900
},
{
"epoch": 2.04,
"learning_rate": 8.88101265822785e-06,
"loss": 0.0692,
"step": 4925
},
{
"epoch": 2.04,
"learning_rate": 8.874683544303799e-06,
"loss": 0.077,
"step": 4950
},
{
"epoch": 2.04,
"learning_rate": 8.868354430379748e-06,
"loss": 0.0662,
"step": 4975
},
{
"epoch": 2.04,
"learning_rate": 8.862025316455696e-06,
"loss": 0.0605,
"step": 5000
},
{
"epoch": 2.04,
"eval_loss": 0.2468421906232834,
"eval_runtime": 1442.9116,
"eval_samples_per_second": 7.215,
"eval_steps_per_second": 0.451,
"eval_wer": 33.46883689165677,
"step": 5000
}
],
"max_steps": 40000,
"num_train_epochs": 9223372036854775807,
"total_flos": 2.30810603175936e+19,
"trial_name": null,
"trial_params": null
}