xls-r-300m-sv-robust / trainer_state.json
marinone94's picture
End of training
4b40d1e
raw
history blame
14.2 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 99.9974025974026,
"global_step": 9600,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.04,
"learning_rate": 3.828125e-06,
"loss": 15.104,
"step": 100
},
{
"epoch": 2.08,
"learning_rate": 7.6953125e-06,
"loss": 12.2149,
"step": 200
},
{
"epoch": 3.12,
"learning_rate": 1.16015625e-05,
"loss": 4.9885,
"step": 300
},
{
"epoch": 4.17,
"learning_rate": 1.55078125e-05,
"loss": 3.3788,
"step": 400
},
{
"epoch": 5.21,
"learning_rate": 1.94140625e-05,
"loss": 3.1705,
"step": 500
},
{
"epoch": 6.25,
"learning_rate": 2.3320312499999995e-05,
"loss": 3.11,
"step": 600
},
{
"epoch": 7.29,
"learning_rate": 2.72265625e-05,
"loss": 3.0526,
"step": 700
},
{
"epoch": 8.33,
"learning_rate": 3.11328125e-05,
"loss": 3.0113,
"step": 800
},
{
"epoch": 9.37,
"learning_rate": 3.5039062499999995e-05,
"loss": 2.9601,
"step": 900
},
{
"epoch": 10.42,
"learning_rate": 3.89453125e-05,
"loss": 2.9099,
"step": 1000
},
{
"epoch": 10.42,
"eval_loss": 2.8369038105010986,
"eval_runtime": 198.5866,
"eval_samples_per_second": 25.405,
"eval_steps_per_second": 0.796,
"eval_wer": 1.0,
"step": 1000
},
{
"epoch": 11.46,
"learning_rate": 4.28515625e-05,
"loss": 2.6653,
"step": 1100
},
{
"epoch": 12.5,
"learning_rate": 4.675781249999999e-05,
"loss": 1.8958,
"step": 1200
},
{
"epoch": 13.54,
"learning_rate": 5.0664062499999996e-05,
"loss": 1.4812,
"step": 1300
},
{
"epoch": 14.58,
"learning_rate": 5.45703125e-05,
"loss": 1.3358,
"step": 1400
},
{
"epoch": 15.62,
"learning_rate": 5.8476562499999996e-05,
"loss": 1.2522,
"step": 1500
},
{
"epoch": 16.66,
"learning_rate": 6.238281249999999e-05,
"loss": 1.1919,
"step": 1600
},
{
"epoch": 17.71,
"learning_rate": 6.62890625e-05,
"loss": 1.1663,
"step": 1700
},
{
"epoch": 18.75,
"learning_rate": 7.019531249999999e-05,
"loss": 1.1289,
"step": 1800
},
{
"epoch": 19.79,
"learning_rate": 7.410156249999999e-05,
"loss": 1.0902,
"step": 1900
},
{
"epoch": 20.83,
"learning_rate": 7.4248046875e-05,
"loss": 1.0745,
"step": 2000
},
{
"epoch": 20.83,
"eval_loss": 0.19572903215885162,
"eval_runtime": 202.06,
"eval_samples_per_second": 24.968,
"eval_steps_per_second": 0.782,
"eval_wer": 0.16725371193237237,
"step": 2000
},
{
"epoch": 21.87,
"learning_rate": 7.327148437499999e-05,
"loss": 1.0485,
"step": 2100
},
{
"epoch": 22.91,
"learning_rate": 7.2294921875e-05,
"loss": 1.0291,
"step": 2200
},
{
"epoch": 23.96,
"learning_rate": 7.1318359375e-05,
"loss": 1.007,
"step": 2300
},
{
"epoch": 25.0,
"learning_rate": 7.0341796875e-05,
"loss": 1.0008,
"step": 2400
},
{
"epoch": 26.04,
"learning_rate": 6.9365234375e-05,
"loss": 0.988,
"step": 2500
},
{
"epoch": 27.08,
"learning_rate": 6.8388671875e-05,
"loss": 0.9766,
"step": 2600
},
{
"epoch": 28.12,
"learning_rate": 6.7412109375e-05,
"loss": 0.9663,
"step": 2700
},
{
"epoch": 29.17,
"learning_rate": 6.6435546875e-05,
"loss": 0.9539,
"step": 2800
},
{
"epoch": 30.21,
"learning_rate": 6.545898437499999e-05,
"loss": 0.9479,
"step": 2900
},
{
"epoch": 31.25,
"learning_rate": 6.448242187499999e-05,
"loss": 0.934,
"step": 3000
},
{
"epoch": 31.25,
"eval_loss": 0.1579357087612152,
"eval_runtime": 198.7733,
"eval_samples_per_second": 25.381,
"eval_steps_per_second": 0.795,
"eval_wer": 0.1388858784003468,
"step": 3000
},
{
"epoch": 32.29,
"learning_rate": 6.3505859375e-05,
"loss": 0.9285,
"step": 3100
},
{
"epoch": 33.33,
"learning_rate": 6.252929687499999e-05,
"loss": 0.9121,
"step": 3200
},
{
"epoch": 34.37,
"learning_rate": 6.155273437499999e-05,
"loss": 0.9016,
"step": 3300
},
{
"epoch": 35.42,
"learning_rate": 6.0576171875e-05,
"loss": 0.9023,
"step": 3400
},
{
"epoch": 36.46,
"learning_rate": 5.9599609374999994e-05,
"loss": 0.9004,
"step": 3500
},
{
"epoch": 37.5,
"learning_rate": 5.862304687499999e-05,
"loss": 0.8844,
"step": 3600
},
{
"epoch": 38.54,
"learning_rate": 5.7646484375e-05,
"loss": 0.8771,
"step": 3700
},
{
"epoch": 39.58,
"learning_rate": 5.6669921875e-05,
"loss": 0.876,
"step": 3800
},
{
"epoch": 40.62,
"learning_rate": 5.569335937499999e-05,
"loss": 0.8708,
"step": 3900
},
{
"epoch": 41.66,
"learning_rate": 5.4716796874999997e-05,
"loss": 0.8691,
"step": 4000
},
{
"epoch": 41.66,
"eval_loss": 0.14571049809455872,
"eval_runtime": 195.237,
"eval_samples_per_second": 25.84,
"eval_steps_per_second": 0.809,
"eval_wer": 0.12899642353961202,
"step": 4000
},
{
"epoch": 42.71,
"learning_rate": 5.3740234374999996e-05,
"loss": 0.8624,
"step": 4100
},
{
"epoch": 43.75,
"learning_rate": 5.2763671874999995e-05,
"loss": 0.8556,
"step": 4200
},
{
"epoch": 44.79,
"learning_rate": 5.1787109375e-05,
"loss": 0.8607,
"step": 4300
},
{
"epoch": 45.83,
"learning_rate": 5.0810546875e-05,
"loss": 0.8536,
"step": 4400
},
{
"epoch": 46.87,
"learning_rate": 4.983398437499999e-05,
"loss": 0.8493,
"step": 4500
},
{
"epoch": 47.91,
"learning_rate": 4.8857421875e-05,
"loss": 0.8456,
"step": 4600
},
{
"epoch": 48.96,
"learning_rate": 4.7880859375e-05,
"loss": 0.8333,
"step": 4700
},
{
"epoch": 50.0,
"learning_rate": 4.6904296874999996e-05,
"loss": 0.8346,
"step": 4800
},
{
"epoch": 51.04,
"learning_rate": 4.5927734375e-05,
"loss": 0.8403,
"step": 4900
},
{
"epoch": 52.08,
"learning_rate": 4.4951171874999995e-05,
"loss": 0.8328,
"step": 5000
},
{
"epoch": 52.08,
"eval_loss": 0.14348936080932617,
"eval_runtime": 197.7739,
"eval_samples_per_second": 25.509,
"eval_steps_per_second": 0.799,
"eval_wer": 0.12054297171344966,
"step": 5000
},
{
"epoch": 53.12,
"learning_rate": 4.3974609374999994e-05,
"loss": 0.8275,
"step": 5100
},
{
"epoch": 54.17,
"learning_rate": 4.2998046875e-05,
"loss": 0.8262,
"step": 5200
},
{
"epoch": 55.21,
"learning_rate": 4.2021484375e-05,
"loss": 0.8167,
"step": 5300
},
{
"epoch": 56.25,
"learning_rate": 4.1044921875e-05,
"loss": 0.8194,
"step": 5400
},
{
"epoch": 57.29,
"learning_rate": 4.0068359375e-05,
"loss": 0.8192,
"step": 5500
},
{
"epoch": 58.33,
"learning_rate": 3.9091796874999996e-05,
"loss": 0.8176,
"step": 5600
},
{
"epoch": 59.37,
"learning_rate": 3.8115234374999995e-05,
"loss": 0.8115,
"step": 5700
},
{
"epoch": 60.42,
"learning_rate": 3.7138671874999994e-05,
"loss": 0.8129,
"step": 5800
},
{
"epoch": 61.46,
"learning_rate": 3.6162109375e-05,
"loss": 0.8052,
"step": 5900
},
{
"epoch": 62.5,
"learning_rate": 3.518554687499999e-05,
"loss": 0.8068,
"step": 6000
},
{
"epoch": 62.5,
"eval_loss": 0.13501569628715515,
"eval_runtime": 207.5786,
"eval_samples_per_second": 24.304,
"eval_steps_per_second": 0.761,
"eval_wer": 0.11913406307575593,
"step": 6000
},
{
"epoch": 63.54,
"learning_rate": 3.4208984375e-05,
"loss": 0.8004,
"step": 6100
},
{
"epoch": 64.58,
"learning_rate": 3.32421875e-05,
"loss": 0.7904,
"step": 6200
},
{
"epoch": 65.62,
"learning_rate": 3.2265625e-05,
"loss": 0.7947,
"step": 6300
},
{
"epoch": 66.66,
"learning_rate": 3.12890625e-05,
"loss": 0.7981,
"step": 6400
},
{
"epoch": 67.71,
"learning_rate": 3.0312499999999998e-05,
"loss": 0.8018,
"step": 6500
},
{
"epoch": 68.75,
"learning_rate": 2.93359375e-05,
"loss": 0.7922,
"step": 6600
},
{
"epoch": 69.79,
"learning_rate": 2.8359374999999996e-05,
"loss": 0.796,
"step": 6700
},
{
"epoch": 70.83,
"learning_rate": 2.73828125e-05,
"loss": 0.7887,
"step": 6800
},
{
"epoch": 71.87,
"learning_rate": 2.6406249999999998e-05,
"loss": 0.7796,
"step": 6900
},
{
"epoch": 72.91,
"learning_rate": 2.5429687499999997e-05,
"loss": 0.7822,
"step": 7000
},
{
"epoch": 72.91,
"eval_loss": 0.1346774846315384,
"eval_runtime": 195.6804,
"eval_samples_per_second": 25.782,
"eval_steps_per_second": 0.807,
"eval_wer": 0.11550341389400672,
"step": 7000
},
{
"epoch": 73.96,
"learning_rate": 2.4453125e-05,
"loss": 0.7882,
"step": 7100
},
{
"epoch": 75.0,
"learning_rate": 2.34765625e-05,
"loss": 0.7821,
"step": 7200
},
{
"epoch": 76.04,
"learning_rate": 2.2499999999999998e-05,
"loss": 0.7814,
"step": 7300
},
{
"epoch": 77.08,
"learning_rate": 2.1523437499999997e-05,
"loss": 0.784,
"step": 7400
},
{
"epoch": 78.12,
"learning_rate": 2.0546875e-05,
"loss": 0.7772,
"step": 7500
},
{
"epoch": 79.17,
"learning_rate": 1.95703125e-05,
"loss": 0.7785,
"step": 7600
},
{
"epoch": 80.21,
"learning_rate": 1.8593749999999998e-05,
"loss": 0.7731,
"step": 7700
},
{
"epoch": 81.25,
"learning_rate": 1.7617187499999997e-05,
"loss": 0.7726,
"step": 7800
},
{
"epoch": 82.29,
"learning_rate": 1.6640624999999996e-05,
"loss": 0.7765,
"step": 7900
},
{
"epoch": 83.33,
"learning_rate": 1.56640625e-05,
"loss": 0.7769,
"step": 8000
},
{
"epoch": 83.33,
"eval_loss": 0.1320926398038864,
"eval_runtime": 197.3103,
"eval_samples_per_second": 25.569,
"eval_steps_per_second": 0.801,
"eval_wer": 0.11306491817492142,
"step": 8000
},
{
"epoch": 84.37,
"learning_rate": 1.4687499999999998e-05,
"loss": 0.7713,
"step": 8100
},
{
"epoch": 85.42,
"learning_rate": 1.3710937499999999e-05,
"loss": 0.773,
"step": 8200
},
{
"epoch": 86.46,
"learning_rate": 1.2734375e-05,
"loss": 0.7668,
"step": 8300
},
{
"epoch": 87.5,
"learning_rate": 1.1757812499999999e-05,
"loss": 0.7766,
"step": 8400
},
{
"epoch": 88.54,
"learning_rate": 1.0781249999999998e-05,
"loss": 0.7596,
"step": 8500
},
{
"epoch": 89.58,
"learning_rate": 9.8046875e-06,
"loss": 0.7561,
"step": 8600
},
{
"epoch": 90.62,
"learning_rate": 8.837890624999999e-06,
"loss": 0.7585,
"step": 8700
},
{
"epoch": 91.66,
"learning_rate": 7.861328124999998e-06,
"loss": 0.7524,
"step": 8800
},
{
"epoch": 92.71,
"learning_rate": 6.884765624999999e-06,
"loss": 0.7591,
"step": 8900
},
{
"epoch": 93.75,
"learning_rate": 5.908203125e-06,
"loss": 0.7678,
"step": 9000
},
{
"epoch": 93.75,
"eval_loss": 0.13205420970916748,
"eval_runtime": 199.4472,
"eval_samples_per_second": 25.295,
"eval_steps_per_second": 0.792,
"eval_wer": 0.11146634875907663,
"step": 9000
},
{
"epoch": 94.79,
"learning_rate": 4.931640624999999e-06,
"loss": 0.7609,
"step": 9100
},
{
"epoch": 95.83,
"learning_rate": 3.955078125e-06,
"loss": 0.7611,
"step": 9200
},
{
"epoch": 96.87,
"learning_rate": 2.9785156249999998e-06,
"loss": 0.7692,
"step": 9300
},
{
"epoch": 97.91,
"learning_rate": 2.0019531249999997e-06,
"loss": 0.7576,
"step": 9400
},
{
"epoch": 98.96,
"learning_rate": 1.025390625e-06,
"loss": 0.7552,
"step": 9500
},
{
"epoch": 100.0,
"learning_rate": 4.8828124999999996e-08,
"loss": 0.771,
"step": 9600
},
{
"epoch": 100.0,
"step": 9600,
"total_flos": 1.4910750173005185e+20,
"train_loss": 1.3699857131640116,
"train_runtime": 54887.9396,
"train_samples_per_second": 22.422,
"train_steps_per_second": 0.175
}
],
"max_steps": 9600,
"num_train_epochs": 100,
"total_flos": 1.4910750173005185e+20,
"trial_name": null,
"trial_params": null
}