| 1%|▌ | 100/11800 [12:45<25:55:55, 7.98s/it] | |
| 2%|█ | 199/11800 [24:54<20:08:45, 6.25s/it] | |
| 3%|█▌ | 299/11800 [37:18<29:05:53, 9.11s/it] | |
| 3%|██ | 400/11800 [49:53<22:55:28, 7.24s/it] | |
| 4%|██▍ | 499/11800 [1:01:50<17:39:50, 5.63s/it] | |
| 4%|██▍ | 500/11800 [1:01:55<16:52:37, 5.38s/it]The following columns in the evaluation set don't have a corresponding argument in `Wav2Vec2ForCTC.forward` and have been ignored: input_length. | |
| ***** Running Evaluation ***** | |
| Num examples = 6463 | |
| Batch size = 72 | |
| Configuration saved in ./checkpoint-500/config.json | |
| Model weights saved in ./checkpoint-500/pytorch_model.bin | |
| Configuration saved in ./checkpoint-500/preprocessor_config.json | |
| {'eval_loss': 0.2469930201768875, 'eval_wer': 0.36629738582545746, 'eval_runtime': 294.3209, 'eval_samples_per_second': 21.959, 'eval_steps_per_second': 0.306, 'epoch': 4.24} | |
| Configuration saved in ./preprocessor_config.json | |
| 5%|██▉ | 599/11800 [1:19:42<27:05:31, 8.71s/it] | |
| 6%|███▍ | 699/11800 [1:31:54<19:48:30, 6.42s/it] | |
| 7%|███▉ | 800/11800 [1:44:25<27:46:25, 9.09s/it] | |
| 8%|████▍ | 899/11800 [1:56:44<23:11:00, 7.66s/it] | |
| 8%|████▊ | 1000/11800 [2:08:58<16:37:42, 5.54s/it]The following columns in the evaluation set don't have a corresponding argument in `Wav2Vec2ForCTC.forward` and have been ignored: input_length. | |
| ***** Running Evaluation ***** | |
| Num examples = 6463 | |
| Batch size = 72 | |
| {'loss': 0.1435, 'learning_rate': 0.00099519617359424, 'epoch': 8.47} | |
| Configuration saved in ./checkpoint-1000/config.json | |
| Model weights saved in ./checkpoint-1000/pytorch_model.bin | |
| Configuration saved in ./checkpoint-1000/preprocessor_config.json | |
| {'eval_loss': 0.20002000033855438, 'eval_wer': 0.2791095533162254, 'eval_runtime': 295.0975, 'eval_samples_per_second': 21.901, 'eval_steps_per_second': 0.305, 'epoch': 8.47} | |
| Configuration saved in ./preprocessor_config.json | |
| Deleting older checkpoint [checkpoint-500] due to args.save_total_limit | |
| 9%|█████▎ | 1099/11800 [2:27:28<26:00:03, 8.75s/it] | |
| 10%|█████▊ | 1199/11800 [2:39:47<20:07:56, 6.84s/it] | |
| 11%|██████▎ | 1300/11800 [2:52:03<27:51:31, 9.55s/it] | |
| 12%|██████▊ | 1400/11800 [3:04:32<21:51:09, 7.56s/it] | |
| 13%|███████▏ | 1499/11800 [3:16:35<16:58:53, 5.93s/it] | |
| 13%|███████▏ | 1500/11800 [3:16:41<16:33:01, 5.78s/it]The following columns in the evaluation set don't have a corresponding argument in `Wav2Vec2ForCTC.forward` and have been ignored: input_length. | |
| ***** Running Evaluation ***** | |
| Num examples = 6463 | |
| Batch size = 72 | |
| Configuration saved in ./checkpoint-1500/config.json | |
| {'eval_loss': 0.20303700864315033, 'eval_wer': 0.26521457929106423, 'eval_runtime': 301.1246, 'eval_samples_per_second': 21.463, 'eval_steps_per_second': 0.299, 'epoch': 12.71} | |
| Model weights saved in ./checkpoint-1500/pytorch_model.bin | |
| Configuration saved in ./checkpoint-1500/preprocessor_config.json | |
| Configuration saved in ./preprocessor_config.json | |
| Adding files tracked by Git LFS: ['wandb/run-20220205_233515-2f29fa6z/run-2f29fa6z.wandb']. This may take a bit of time if the files are large. | |