xls-r-300m-sv-robust / trainer_state.json
marinone94's picture
sync
61752de
raw
history blame
29.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 49.99782451051487,
"global_step": 17200,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.29,
"learning_rate": 3.6375e-06,
"loss": 12.958,
"step": 100
},
{
"epoch": 0.58,
"learning_rate": 7.35e-06,
"loss": 10.9696,
"step": 200
},
{
"epoch": 0.87,
"learning_rate": 1.1099999999999999e-05,
"loss": 4.9039,
"step": 300
},
{
"epoch": 1.16,
"learning_rate": 1.485e-05,
"loss": 3.4263,
"step": 400
},
{
"epoch": 1.45,
"learning_rate": 1.8599999999999998e-05,
"loss": 3.1522,
"step": 500
},
{
"epoch": 1.45,
"eval_loss": 3.128952741622925,
"eval_runtime": 133.9857,
"eval_samples_per_second": 34.481,
"eval_steps_per_second": 4.314,
"eval_wer": 1.0,
"step": 500
},
{
"epoch": 1.74,
"learning_rate": 2.2349999999999998e-05,
"loss": 3.0702,
"step": 600
},
{
"epoch": 2.03,
"learning_rate": 2.6099999999999997e-05,
"loss": 3.0589,
"step": 700
},
{
"epoch": 2.32,
"learning_rate": 2.985e-05,
"loss": 3.0223,
"step": 800
},
{
"epoch": 2.61,
"learning_rate": 3.36e-05,
"loss": 3.0021,
"step": 900
},
{
"epoch": 2.91,
"learning_rate": 3.735e-05,
"loss": 2.9576,
"step": 1000
},
{
"epoch": 2.91,
"eval_loss": 2.963273763656616,
"eval_runtime": 130.7049,
"eval_samples_per_second": 35.347,
"eval_steps_per_second": 4.422,
"eval_wer": 1.0,
"step": 1000
},
{
"epoch": 3.2,
"learning_rate": 4.11e-05,
"loss": 2.9499,
"step": 1100
},
{
"epoch": 3.49,
"learning_rate": 4.484999999999999e-05,
"loss": 2.8923,
"step": 1200
},
{
"epoch": 3.78,
"learning_rate": 4.8599999999999995e-05,
"loss": 2.7406,
"step": 1300
},
{
"epoch": 4.07,
"learning_rate": 5.234999999999999e-05,
"loss": 2.3473,
"step": 1400
},
{
"epoch": 4.36,
"learning_rate": 5.6099999999999995e-05,
"loss": 1.9853,
"step": 1500
},
{
"epoch": 4.36,
"eval_loss": 0.8902471661567688,
"eval_runtime": 128.9711,
"eval_samples_per_second": 35.822,
"eval_steps_per_second": 4.482,
"eval_wer": 0.6104448387688919,
"step": 1500
},
{
"epoch": 4.65,
"learning_rate": 5.985e-05,
"loss": 1.8166,
"step": 1600
},
{
"epoch": 4.94,
"learning_rate": 6.359999999999999e-05,
"loss": 1.7352,
"step": 1700
},
{
"epoch": 5.23,
"learning_rate": 6.735e-05,
"loss": 1.6603,
"step": 1800
},
{
"epoch": 5.52,
"learning_rate": 7.11e-05,
"loss": 1.6183,
"step": 1900
},
{
"epoch": 5.81,
"learning_rate": 7.484999999999999e-05,
"loss": 1.5867,
"step": 2000
},
{
"epoch": 5.81,
"eval_loss": 0.47926634550094604,
"eval_runtime": 128.7387,
"eval_samples_per_second": 35.887,
"eval_steps_per_second": 4.49,
"eval_wer": 0.36639539864162024,
"step": 2000
},
{
"epoch": 6.1,
"learning_rate": 7.452631578947368e-05,
"loss": 1.539,
"step": 2100
},
{
"epoch": 6.39,
"learning_rate": 7.40328947368421e-05,
"loss": 1.4847,
"step": 2200
},
{
"epoch": 6.68,
"learning_rate": 7.353947368421052e-05,
"loss": 1.4785,
"step": 2300
},
{
"epoch": 6.97,
"learning_rate": 7.304605263157895e-05,
"loss": 1.4361,
"step": 2400
},
{
"epoch": 7.27,
"learning_rate": 7.255263157894736e-05,
"loss": 1.4608,
"step": 2500
},
{
"epoch": 7.27,
"eval_loss": 0.38162761926651,
"eval_runtime": 130.0928,
"eval_samples_per_second": 35.513,
"eval_steps_per_second": 4.443,
"eval_wer": 0.3094903016582023,
"step": 2500
},
{
"epoch": 7.56,
"learning_rate": 7.205921052631578e-05,
"loss": 1.4239,
"step": 2600
},
{
"epoch": 7.85,
"learning_rate": 7.15657894736842e-05,
"loss": 1.3778,
"step": 2700
},
{
"epoch": 8.14,
"learning_rate": 7.107236842105262e-05,
"loss": 1.414,
"step": 2800
},
{
"epoch": 8.43,
"learning_rate": 7.057894736842104e-05,
"loss": 1.3957,
"step": 2900
},
{
"epoch": 8.72,
"learning_rate": 7.008552631578947e-05,
"loss": 1.3496,
"step": 3000
},
{
"epoch": 8.72,
"eval_loss": 0.34154650568962097,
"eval_runtime": 129.9374,
"eval_samples_per_second": 35.556,
"eval_steps_per_second": 4.448,
"eval_wer": 0.27825368659364863,
"step": 3000
},
{
"epoch": 9.01,
"learning_rate": 6.959210526315788e-05,
"loss": 1.3661,
"step": 3100
},
{
"epoch": 9.3,
"learning_rate": 6.909868421052631e-05,
"loss": 1.3144,
"step": 3200
},
{
"epoch": 9.59,
"learning_rate": 6.860526315789474e-05,
"loss": 1.3253,
"step": 3300
},
{
"epoch": 9.88,
"learning_rate": 6.811184210526315e-05,
"loss": 1.3247,
"step": 3400
},
{
"epoch": 10.17,
"learning_rate": 6.761842105263158e-05,
"loss": 1.3058,
"step": 3500
},
{
"epoch": 10.17,
"eval_loss": 0.307181179523468,
"eval_runtime": 133.008,
"eval_samples_per_second": 34.735,
"eval_steps_per_second": 4.346,
"eval_wer": 0.2519121336351955,
"step": 3500
},
{
"epoch": 10.46,
"learning_rate": 6.712499999999999e-05,
"loss": 1.2943,
"step": 3600
},
{
"epoch": 10.75,
"learning_rate": 6.663157894736842e-05,
"loss": 1.2986,
"step": 3700
},
{
"epoch": 11.05,
"learning_rate": 6.613815789473683e-05,
"loss": 1.3045,
"step": 3800
},
{
"epoch": 11.34,
"learning_rate": 6.564473684210526e-05,
"loss": 1.281,
"step": 3900
},
{
"epoch": 11.63,
"learning_rate": 6.515131578947369e-05,
"loss": 1.2533,
"step": 4000
},
{
"epoch": 11.63,
"eval_loss": 0.2877204418182373,
"eval_runtime": 130.1763,
"eval_samples_per_second": 35.49,
"eval_steps_per_second": 4.44,
"eval_wer": 0.23814477146178792,
"step": 4000
},
{
"epoch": 11.92,
"learning_rate": 6.46578947368421e-05,
"loss": 1.268,
"step": 4100
},
{
"epoch": 12.21,
"learning_rate": 6.416447368421053e-05,
"loss": 1.2522,
"step": 4200
},
{
"epoch": 12.5,
"learning_rate": 6.367105263157894e-05,
"loss": 1.2473,
"step": 4300
},
{
"epoch": 12.79,
"learning_rate": 6.317763157894737e-05,
"loss": 1.2416,
"step": 4400
},
{
"epoch": 13.08,
"learning_rate": 6.268421052631578e-05,
"loss": 1.2535,
"step": 4500
},
{
"epoch": 13.08,
"eval_loss": 0.27913230657577515,
"eval_runtime": 130.449,
"eval_samples_per_second": 35.416,
"eval_steps_per_second": 4.431,
"eval_wer": 0.23202594382916233,
"step": 4500
},
{
"epoch": 13.37,
"learning_rate": 6.21907894736842e-05,
"loss": 1.2196,
"step": 4600
},
{
"epoch": 13.66,
"learning_rate": 6.169736842105262e-05,
"loss": 1.2203,
"step": 4700
},
{
"epoch": 13.95,
"learning_rate": 6.120394736842105e-05,
"loss": 1.2238,
"step": 4800
},
{
"epoch": 14.24,
"learning_rate": 6.0710526315789474e-05,
"loss": 1.2226,
"step": 4900
},
{
"epoch": 14.53,
"learning_rate": 6.021710526315789e-05,
"loss": 1.2273,
"step": 5000
},
{
"epoch": 14.53,
"eval_loss": 0.27258577942848206,
"eval_runtime": 129.7722,
"eval_samples_per_second": 35.601,
"eval_steps_per_second": 4.454,
"eval_wer": 0.22823227069693447,
"step": 5000
},
{
"epoch": 14.82,
"learning_rate": 5.972368421052631e-05,
"loss": 1.2016,
"step": 5100
},
{
"epoch": 15.12,
"learning_rate": 5.9230263157894734e-05,
"loss": 1.2156,
"step": 5200
},
{
"epoch": 15.41,
"learning_rate": 5.873684210526315e-05,
"loss": 1.2005,
"step": 5300
},
{
"epoch": 15.7,
"learning_rate": 5.8243421052631574e-05,
"loss": 1.187,
"step": 5400
},
{
"epoch": 15.99,
"learning_rate": 5.7749999999999994e-05,
"loss": 1.2083,
"step": 5500
},
{
"epoch": 15.99,
"eval_loss": 0.26375535130500793,
"eval_runtime": 128.7955,
"eval_samples_per_second": 35.871,
"eval_steps_per_second": 4.488,
"eval_wer": 0.22119561891941503,
"step": 5500
},
{
"epoch": 16.28,
"learning_rate": 5.725657894736842e-05,
"loss": 1.1987,
"step": 5600
},
{
"epoch": 16.57,
"learning_rate": 5.6763157894736834e-05,
"loss": 1.1832,
"step": 5700
},
{
"epoch": 16.86,
"learning_rate": 5.626973684210526e-05,
"loss": 1.142,
"step": 5800
},
{
"epoch": 17.15,
"learning_rate": 5.577631578947368e-05,
"loss": 1.1863,
"step": 5900
},
{
"epoch": 17.44,
"learning_rate": 5.52828947368421e-05,
"loss": 1.1606,
"step": 6000
},
{
"epoch": 17.44,
"eval_loss": 0.253081351518631,
"eval_runtime": 130.3816,
"eval_samples_per_second": 35.434,
"eval_steps_per_second": 4.433,
"eval_wer": 0.21740194578718716,
"step": 6000
},
{
"epoch": 17.73,
"learning_rate": 5.478947368421052e-05,
"loss": 1.154,
"step": 6100
},
{
"epoch": 18.02,
"learning_rate": 5.429605263157895e-05,
"loss": 1.1842,
"step": 6200
},
{
"epoch": 18.31,
"learning_rate": 5.380263157894736e-05,
"loss": 1.1603,
"step": 6300
},
{
"epoch": 18.6,
"learning_rate": 5.330921052631578e-05,
"loss": 1.1692,
"step": 6400
},
{
"epoch": 18.89,
"learning_rate": 5.281578947368421e-05,
"loss": 1.1545,
"step": 6500
},
{
"epoch": 18.89,
"eval_loss": 0.2468416541814804,
"eval_runtime": 130.515,
"eval_samples_per_second": 35.398,
"eval_steps_per_second": 4.429,
"eval_wer": 0.2108548002202778,
"step": 6500
},
{
"epoch": 19.19,
"learning_rate": 5.232236842105262e-05,
"loss": 1.1389,
"step": 6600
},
{
"epoch": 19.48,
"learning_rate": 5.182894736842105e-05,
"loss": 1.1635,
"step": 6700
},
{
"epoch": 19.77,
"learning_rate": 5.133552631578947e-05,
"loss": 1.1547,
"step": 6800
},
{
"epoch": 20.06,
"learning_rate": 5.084210526315789e-05,
"loss": 1.1323,
"step": 6900
},
{
"epoch": 20.35,
"learning_rate": 5.034868421052631e-05,
"loss": 1.1344,
"step": 7000
},
{
"epoch": 20.35,
"eval_loss": 0.2494264841079712,
"eval_runtime": 131.1616,
"eval_samples_per_second": 35.224,
"eval_steps_per_second": 4.407,
"eval_wer": 0.20504191396928348,
"step": 7000
},
{
"epoch": 20.64,
"learning_rate": 4.9855263157894736e-05,
"loss": 1.1423,
"step": 7100
},
{
"epoch": 20.93,
"learning_rate": 4.936184210526315e-05,
"loss": 1.1264,
"step": 7200
},
{
"epoch": 21.22,
"learning_rate": 4.886842105263158e-05,
"loss": 1.1362,
"step": 7300
},
{
"epoch": 21.51,
"learning_rate": 4.8375e-05,
"loss": 1.1156,
"step": 7400
},
{
"epoch": 21.8,
"learning_rate": 4.7881578947368424e-05,
"loss": 1.1173,
"step": 7500
},
{
"epoch": 21.8,
"eval_loss": 0.24467654526233673,
"eval_runtime": 161.8558,
"eval_samples_per_second": 28.544,
"eval_steps_per_second": 3.571,
"eval_wer": 0.19800526219176406,
"step": 7500
},
{
"epoch": 22.09,
"learning_rate": 4.739309210526315e-05,
"loss": 1.1319,
"step": 7600
},
{
"epoch": 22.38,
"learning_rate": 4.690460526315789e-05,
"loss": 1.1065,
"step": 7700
},
{
"epoch": 22.67,
"learning_rate": 4.641118421052631e-05,
"loss": 1.1189,
"step": 7800
},
{
"epoch": 22.96,
"learning_rate": 4.5917763157894736e-05,
"loss": 1.1204,
"step": 7900
},
{
"epoch": 23.26,
"learning_rate": 4.542434210526315e-05,
"loss": 1.1081,
"step": 8000
},
{
"epoch": 23.26,
"eval_loss": 0.24276763200759888,
"eval_runtime": 130.7165,
"eval_samples_per_second": 35.344,
"eval_steps_per_second": 4.422,
"eval_wer": 0.19977972220522547,
"step": 8000
},
{
"epoch": 23.55,
"learning_rate": 4.4930921052631576e-05,
"loss": 1.1097,
"step": 8100
},
{
"epoch": 23.84,
"learning_rate": 4.4437499999999996e-05,
"loss": 1.0912,
"step": 8200
},
{
"epoch": 24.13,
"learning_rate": 4.3944078947368416e-05,
"loss": 1.1305,
"step": 8300
},
{
"epoch": 24.42,
"learning_rate": 4.3450657894736836e-05,
"loss": 1.1219,
"step": 8400
},
{
"epoch": 24.71,
"learning_rate": 4.295723684210526e-05,
"loss": 1.1023,
"step": 8500
},
{
"epoch": 24.71,
"eval_loss": 0.23292256891727448,
"eval_runtime": 133.1697,
"eval_samples_per_second": 34.693,
"eval_steps_per_second": 4.34,
"eval_wer": 0.19512941320443003,
"step": 8500
},
{
"epoch": 25.0,
"learning_rate": 4.2463815789473676e-05,
"loss": 1.1049,
"step": 8600
},
{
"epoch": 25.29,
"learning_rate": 4.19703947368421e-05,
"loss": 1.1194,
"step": 8700
},
{
"epoch": 25.58,
"learning_rate": 4.1476973684210523e-05,
"loss": 1.1028,
"step": 8800
},
{
"epoch": 25.87,
"learning_rate": 4.098355263157895e-05,
"loss": 1.0614,
"step": 8900
},
{
"epoch": 26.16,
"learning_rate": 4.0490131578947364e-05,
"loss": 1.0923,
"step": 9000
},
{
"epoch": 26.16,
"eval_loss": 0.23875781893730164,
"eval_runtime": 132.625,
"eval_samples_per_second": 34.835,
"eval_steps_per_second": 4.358,
"eval_wer": 0.19616961390197638,
"step": 9000
},
{
"epoch": 26.45,
"learning_rate": 3.9996710526315784e-05,
"loss": 1.0789,
"step": 9100
},
{
"epoch": 26.74,
"learning_rate": 3.950328947368421e-05,
"loss": 1.0844,
"step": 9200
},
{
"epoch": 27.03,
"learning_rate": 3.9009868421052624e-05,
"loss": 1.0961,
"step": 9300
},
{
"epoch": 27.32,
"learning_rate": 3.851644736842105e-05,
"loss": 1.0691,
"step": 9400
},
{
"epoch": 27.61,
"learning_rate": 3.802302631578947e-05,
"loss": 1.0798,
"step": 9500
},
{
"epoch": 27.61,
"eval_loss": 0.23633554577827454,
"eval_runtime": 130.4004,
"eval_samples_per_second": 35.429,
"eval_steps_per_second": 4.433,
"eval_wer": 0.19442574802667809,
"step": 9500
},
{
"epoch": 27.91,
"learning_rate": 3.752960526315789e-05,
"loss": 1.0609,
"step": 9600
},
{
"epoch": 28.2,
"learning_rate": 3.703618421052631e-05,
"loss": 1.0779,
"step": 9700
},
{
"epoch": 28.49,
"learning_rate": 3.654276315789473e-05,
"loss": 1.0693,
"step": 9800
},
{
"epoch": 28.78,
"learning_rate": 3.604934210526316e-05,
"loss": 1.0786,
"step": 9900
},
{
"epoch": 29.07,
"learning_rate": 3.5560855263157894e-05,
"loss": 1.0769,
"step": 10000
},
{
"epoch": 29.07,
"eval_loss": 0.23424996435642242,
"eval_runtime": 129.7477,
"eval_samples_per_second": 35.608,
"eval_steps_per_second": 4.455,
"eval_wer": 0.19130514593403905,
"step": 10000
},
{
"epoch": 29.36,
"learning_rate": 3.5067434210526314e-05,
"loss": 1.0655,
"step": 10100
},
{
"epoch": 29.65,
"learning_rate": 3.4574013157894734e-05,
"loss": 1.0618,
"step": 10200
},
{
"epoch": 29.94,
"learning_rate": 3.4080592105263154e-05,
"loss": 1.071,
"step": 10300
},
{
"epoch": 30.23,
"learning_rate": 3.3587171052631575e-05,
"loss": 1.0618,
"step": 10400
},
{
"epoch": 30.52,
"learning_rate": 3.3093749999999995e-05,
"loss": 1.0672,
"step": 10500
},
{
"epoch": 30.52,
"eval_loss": 0.22497320175170898,
"eval_runtime": 129.3889,
"eval_samples_per_second": 35.706,
"eval_steps_per_second": 4.467,
"eval_wer": 0.18745028452548493,
"step": 10500
},
{
"epoch": 30.81,
"learning_rate": 3.260032894736842e-05,
"loss": 1.0642,
"step": 10600
},
{
"epoch": 31.1,
"learning_rate": 3.210690789473684e-05,
"loss": 1.0567,
"step": 10700
},
{
"epoch": 31.39,
"learning_rate": 3.161348684210526e-05,
"loss": 1.0566,
"step": 10800
},
{
"epoch": 31.68,
"learning_rate": 3.112006578947368e-05,
"loss": 1.0553,
"step": 10900
},
{
"epoch": 31.97,
"learning_rate": 3.06266447368421e-05,
"loss": 1.0735,
"step": 11000
},
{
"epoch": 31.97,
"eval_loss": 0.23052582144737244,
"eval_runtime": 130.8716,
"eval_samples_per_second": 35.302,
"eval_steps_per_second": 4.417,
"eval_wer": 0.18741969038732179,
"step": 11000
},
{
"epoch": 32.27,
"learning_rate": 3.0133223684210525e-05,
"loss": 1.0583,
"step": 11100
},
{
"epoch": 32.56,
"learning_rate": 2.9639802631578946e-05,
"loss": 1.0502,
"step": 11200
},
{
"epoch": 32.85,
"learning_rate": 2.9146381578947366e-05,
"loss": 1.0529,
"step": 11300
},
{
"epoch": 33.14,
"learning_rate": 2.865296052631579e-05,
"loss": 1.0436,
"step": 11400
},
{
"epoch": 33.43,
"learning_rate": 2.8159539473684206e-05,
"loss": 1.0628,
"step": 11500
},
{
"epoch": 33.43,
"eval_loss": 0.22911641001701355,
"eval_runtime": 129.7033,
"eval_samples_per_second": 35.62,
"eval_steps_per_second": 4.456,
"eval_wer": 0.18509453588692407,
"step": 11500
},
{
"epoch": 33.72,
"learning_rate": 2.7666118421052626e-05,
"loss": 1.0228,
"step": 11600
},
{
"epoch": 34.01,
"learning_rate": 2.717269736842105e-05,
"loss": 1.0556,
"step": 11700
},
{
"epoch": 34.3,
"learning_rate": 2.667927631578947e-05,
"loss": 1.0616,
"step": 11800
},
{
"epoch": 34.59,
"learning_rate": 2.6185855263157893e-05,
"loss": 1.0397,
"step": 11900
},
{
"epoch": 34.88,
"learning_rate": 2.5692434210526313e-05,
"loss": 1.0451,
"step": 12000
},
{
"epoch": 34.88,
"eval_loss": 0.22632543742656708,
"eval_runtime": 130.5561,
"eval_samples_per_second": 35.387,
"eval_steps_per_second": 4.427,
"eval_wer": 0.18558404209753412,
"step": 12000
},
{
"epoch": 35.17,
"learning_rate": 2.5199013157894733e-05,
"loss": 1.0387,
"step": 12100
},
{
"epoch": 35.46,
"learning_rate": 2.4705592105263157e-05,
"loss": 1.043,
"step": 12200
},
{
"epoch": 35.75,
"learning_rate": 2.4212171052631577e-05,
"loss": 1.019,
"step": 12300
},
{
"epoch": 36.05,
"learning_rate": 2.3718749999999997e-05,
"loss": 1.052,
"step": 12400
},
{
"epoch": 36.34,
"learning_rate": 2.322532894736842e-05,
"loss": 1.0299,
"step": 12500
},
{
"epoch": 36.34,
"eval_loss": 0.22569817304611206,
"eval_runtime": 131.0957,
"eval_samples_per_second": 35.241,
"eval_steps_per_second": 4.409,
"eval_wer": 0.1833812641497889,
"step": 12500
},
{
"epoch": 36.63,
"learning_rate": 2.273190789473684e-05,
"loss": 1.0492,
"step": 12600
},
{
"epoch": 36.92,
"learning_rate": 2.2238486842105264e-05,
"loss": 1.0405,
"step": 12700
},
{
"epoch": 37.21,
"learning_rate": 2.174506578947368e-05,
"loss": 1.0405,
"step": 12800
},
{
"epoch": 37.5,
"learning_rate": 2.12516447368421e-05,
"loss": 1.0186,
"step": 12900
},
{
"epoch": 37.79,
"learning_rate": 2.076315789473684e-05,
"loss": 1.0368,
"step": 13000
},
{
"epoch": 37.79,
"eval_loss": 0.2230287492275238,
"eval_runtime": 141.6123,
"eval_samples_per_second": 32.624,
"eval_steps_per_second": 4.082,
"eval_wer": 0.1808419506822493,
"step": 13000
},
{
"epoch": 38.08,
"learning_rate": 2.026973684210526e-05,
"loss": 1.0346,
"step": 13100
},
{
"epoch": 38.37,
"learning_rate": 1.9776315789473684e-05,
"loss": 1.0348,
"step": 13200
},
{
"epoch": 38.66,
"learning_rate": 1.9282894736842104e-05,
"loss": 1.016,
"step": 13300
},
{
"epoch": 38.95,
"learning_rate": 1.8789473684210528e-05,
"loss": 1.047,
"step": 13400
},
{
"epoch": 39.24,
"learning_rate": 1.8296052631578948e-05,
"loss": 1.0322,
"step": 13500
},
{
"epoch": 39.24,
"eval_loss": 0.22309929132461548,
"eval_runtime": 135.3243,
"eval_samples_per_second": 34.14,
"eval_steps_per_second": 4.271,
"eval_wer": 0.18332007587346263,
"step": 13500
},
{
"epoch": 39.53,
"learning_rate": 1.7802631578947368e-05,
"loss": 1.0091,
"step": 13600
},
{
"epoch": 39.82,
"learning_rate": 1.7309210526315788e-05,
"loss": 1.0311,
"step": 13700
},
{
"epoch": 40.12,
"learning_rate": 1.6815789473684208e-05,
"loss": 1.0436,
"step": 13800
},
{
"epoch": 40.41,
"learning_rate": 1.632236842105263e-05,
"loss": 1.0089,
"step": 13900
},
{
"epoch": 40.7,
"learning_rate": 1.582894736842105e-05,
"loss": 1.0451,
"step": 14000
},
{
"epoch": 40.7,
"eval_loss": 0.2196696549654007,
"eval_runtime": 131.6599,
"eval_samples_per_second": 35.09,
"eval_steps_per_second": 4.39,
"eval_wer": 0.18172918068898,
"step": 14000
},
{
"epoch": 40.99,
"learning_rate": 1.5335526315789472e-05,
"loss": 1.0268,
"step": 14100
},
{
"epoch": 41.28,
"learning_rate": 1.4842105263157894e-05,
"loss": 1.0357,
"step": 14200
},
{
"epoch": 41.57,
"learning_rate": 1.4348684210526315e-05,
"loss": 1.0211,
"step": 14300
},
{
"epoch": 41.86,
"learning_rate": 1.3855263157894737e-05,
"loss": 0.9896,
"step": 14400
},
{
"epoch": 42.15,
"learning_rate": 1.3361842105263156e-05,
"loss": 1.0304,
"step": 14500
},
{
"epoch": 42.15,
"eval_loss": 0.22407057881355286,
"eval_runtime": 130.69,
"eval_samples_per_second": 35.351,
"eval_steps_per_second": 4.423,
"eval_wer": 0.18127026861653306,
"step": 14500
},
{
"epoch": 42.44,
"learning_rate": 1.2868421052631577e-05,
"loss": 1.0092,
"step": 14600
},
{
"epoch": 42.73,
"learning_rate": 1.2375e-05,
"loss": 0.9983,
"step": 14700
},
{
"epoch": 43.02,
"learning_rate": 1.188157894736842e-05,
"loss": 1.0297,
"step": 14800
},
{
"epoch": 43.31,
"learning_rate": 1.1388157894736841e-05,
"loss": 1.0079,
"step": 14900
},
{
"epoch": 43.6,
"learning_rate": 1.0894736842105263e-05,
"loss": 1.0102,
"step": 15000
},
{
"epoch": 43.6,
"eval_loss": 0.22326770424842834,
"eval_runtime": 129.9475,
"eval_samples_per_second": 35.553,
"eval_steps_per_second": 4.448,
"eval_wer": 0.17949580860307165,
"step": 15000
},
{
"epoch": 43.89,
"learning_rate": 1.0401315789473685e-05,
"loss": 1.0303,
"step": 15100
},
{
"epoch": 44.19,
"learning_rate": 9.912828947368419e-06,
"loss": 1.0099,
"step": 15200
},
{
"epoch": 44.48,
"learning_rate": 9.41940789473684e-06,
"loss": 1.0098,
"step": 15300
},
{
"epoch": 44.77,
"learning_rate": 8.925986842105263e-06,
"loss": 1.0068,
"step": 15400
},
{
"epoch": 45.06,
"learning_rate": 8.432565789473683e-06,
"loss": 1.0135,
"step": 15500
},
{
"epoch": 45.06,
"eval_loss": 0.2199595868587494,
"eval_runtime": 129.6997,
"eval_samples_per_second": 35.621,
"eval_steps_per_second": 4.456,
"eval_wer": 0.1794346203267454,
"step": 15500
},
{
"epoch": 45.35,
"learning_rate": 7.939144736842104e-06,
"loss": 1.0279,
"step": 15600
},
{
"epoch": 45.64,
"learning_rate": 7.4457236842105255e-06,
"loss": 1.0116,
"step": 15700
},
{
"epoch": 45.93,
"learning_rate": 6.952302631578946e-06,
"loss": 1.0106,
"step": 15800
},
{
"epoch": 46.22,
"learning_rate": 6.458881578947368e-06,
"loss": 1.0132,
"step": 15900
},
{
"epoch": 46.51,
"learning_rate": 5.965460526315788e-06,
"loss": 1.014,
"step": 16000
},
{
"epoch": 46.51,
"eval_loss": 0.22073601186275482,
"eval_runtime": 130.9629,
"eval_samples_per_second": 35.277,
"eval_steps_per_second": 4.413,
"eval_wer": 0.177904913418589,
"step": 16000
},
{
"epoch": 46.8,
"learning_rate": 5.47203947368421e-06,
"loss": 1.0231,
"step": 16100
},
{
"epoch": 47.09,
"learning_rate": 4.978618421052631e-06,
"loss": 1.0161,
"step": 16200
},
{
"epoch": 47.38,
"learning_rate": 4.485197368421052e-06,
"loss": 0.9842,
"step": 16300
},
{
"epoch": 47.67,
"learning_rate": 3.991776315789474e-06,
"loss": 0.9901,
"step": 16400
},
{
"epoch": 47.96,
"learning_rate": 3.498355263157895e-06,
"loss": 1.0071,
"step": 16500
},
{
"epoch": 47.96,
"eval_loss": 0.2204526960849762,
"eval_runtime": 130.0967,
"eval_samples_per_second": 35.512,
"eval_steps_per_second": 4.443,
"eval_wer": 0.1783638254910359,
"step": 16500
},
{
"epoch": 48.26,
"learning_rate": 3.0049342105263154e-06,
"loss": 1.0114,
"step": 16600
},
{
"epoch": 48.55,
"learning_rate": 2.5115131578947363e-06,
"loss": 1.0035,
"step": 16700
},
{
"epoch": 48.84,
"learning_rate": 2.0180921052631577e-06,
"loss": 1.0091,
"step": 16800
},
{
"epoch": 49.13,
"learning_rate": 1.524671052631579e-06,
"loss": 1.0022,
"step": 16900
},
{
"epoch": 49.42,
"learning_rate": 1.0312499999999999e-06,
"loss": 0.9729,
"step": 17000
},
{
"epoch": 49.42,
"eval_loss": 0.22042310237884521,
"eval_runtime": 130.006,
"eval_samples_per_second": 35.537,
"eval_steps_per_second": 4.446,
"eval_wer": 0.17766016031328397,
"step": 17000
},
{
"epoch": 49.71,
"learning_rate": 5.378289473684209e-07,
"loss": 1.0161,
"step": 17100
},
{
"epoch": 50.0,
"learning_rate": 4.9342105263157887e-08,
"loss": 1.0062,
"step": 17200
},
{
"epoch": 50.0,
"step": 17200,
"total_flos": 5.074861809449127e+19,
"train_loss": 1.4085150359397711,
"train_runtime": 29405.2698,
"train_samples_per_second": 18.755,
"train_steps_per_second": 0.585
}
],
"max_steps": 17200,
"num_train_epochs": 50,
"total_flos": 5.074861809449127e+19,
"trial_name": null,
"trial_params": null
}