<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
wav2vec2-large-xlsr-53-MIR_ST500_ASR
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the /WORKSPACE/DATASETS/DATASETS/MIR_ST500/MIR_ST500.PY - ASR dataset. It achieves the following results on the evaluation set:
- Loss: 0.5180
- Wer: 0.5824
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- total_train_batch_size: 8
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 15.0
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
56.764 | 0.13 | 100 | 24.4254 | 0.9990 |
7.5081 | 0.27 | 200 | 2.9111 | 1.0 |
3.4899 | 0.4 | 300 | 2.1361 | 1.0 |
2.4094 | 0.53 | 400 | 1.9088 | 1.0 |
2.6764 | 0.67 | 500 | 1.8543 | 1.0 |
3.3107 | 0.8 | 600 | 1.7979 | 1.0 |
2.2856 | 0.93 | 700 | 1.7571 | 1.0 |
1.856 | 1.07 | 800 | 1.7351 | 0.9648 |
1.8882 | 1.2 | 900 | 1.7181 | 0.9654 |
2.1731 | 1.33 | 1000 | 1.6736 | 0.9637 |
1.8252 | 1.46 | 1100 | 1.3468 | 0.9647 |
1.9092 | 1.6 | 1200 | 1.3302 | 0.9627 |
1.9435 | 1.73 | 1300 | 1.2428 | 0.9634 |
1.3027 | 1.86 | 1400 | 1.2133 | 0.9644 |
1.3438 | 2.0 | 1500 | 1.2002 | 0.9635 |
1.2161 | 2.13 | 1600 | 1.1901 | 0.9636 |
1.203 | 2.26 | 1700 | 1.1620 | 0.9616 |
1.1159 | 2.4 | 1800 | 1.1660 | 0.9598 |
1.1466 | 2.53 | 1900 | 1.2089 | 0.9605 |
1.0563 | 2.66 | 2000 | 1.1732 | 0.9603 |
1.1019 | 2.8 | 2100 | 1.1468 | 0.9612 |
1.029 | 2.93 | 2200 | 1.1188 | 0.9622 |
1.0079 | 3.06 | 2300 | 1.0604 | 0.9617 |
1.0483 | 3.2 | 2400 | 1.0499 | 0.9612 |
0.9892 | 3.33 | 2500 | 1.0292 | 0.9606 |
0.9556 | 3.46 | 2600 | 1.0228 | 0.9604 |
0.9626 | 3.6 | 2700 | 1.0028 | 0.9617 |
1.0537 | 3.73 | 2800 | 1.0051 | 0.9608 |
1.0648 | 3.86 | 2900 | 0.9723 | 0.9604 |
0.8657 | 3.99 | 3000 | 0.9620 | 0.9605 |
0.8964 | 4.13 | 3100 | 1.0432 | 0.9612 |
0.9639 | 4.26 | 3200 | 0.9322 | 0.9589 |
0.8965 | 4.39 | 3300 | 0.9091 | 0.9559 |
0.8257 | 4.53 | 3400 | 0.8999 | 0.9499 |
0.8002 | 4.66 | 3500 | 0.8754 | 0.9554 |
0.7335 | 4.79 | 3600 | 0.8608 | 0.9572 |
0.936 | 4.93 | 3700 | 0.8564 | 0.9510 |
0.8185 | 5.06 | 3800 | 0.8890 | 0.9517 |
0.7422 | 5.19 | 3900 | 0.8262 | 0.9392 |
0.7784 | 5.33 | 4000 | 0.8292 | 0.9259 |
0.8123 | 5.46 | 4100 | 0.8180 | 0.9374 |
0.7256 | 5.59 | 4200 | 0.8158 | 0.9077 |
0.7638 | 5.73 | 4300 | 0.8423 | 0.9170 |
0.6737 | 5.86 | 4400 | 0.7818 | 0.9182 |
0.7644 | 5.99 | 4500 | 0.7692 | 0.8934 |
0.7911 | 6.13 | 4600 | 0.7627 | 0.8978 |
0.6922 | 6.26 | 4700 | 0.7627 | 0.8906 |
0.7369 | 6.39 | 4800 | 0.7570 | 0.8838 |
0.6642 | 6.52 | 4900 | 0.9476 | 0.8953 |
0.7502 | 6.66 | 5000 | 0.7336 | 0.8955 |
0.6243 | 6.79 | 5100 | 0.7583 | 0.8896 |
0.6912 | 6.92 | 5200 | 0.7764 | 0.8761 |
0.7744 | 7.06 | 5300 | 0.7615 | 0.8790 |
0.6195 | 7.19 | 5400 | 0.7114 | 0.8712 |
0.7418 | 7.32 | 5500 | 0.8314 | 0.8864 |
0.7658 | 7.46 | 5600 | 0.8531 | 0.8718 |
0.6821 | 7.59 | 5700 | 0.9068 | 0.8786 |
0.6931 | 7.72 | 5800 | 0.7549 | 0.8645 |
0.6771 | 7.86 | 5900 | 0.7138 | 0.8442 |
0.6735 | 7.99 | 6000 | 0.6947 | 0.8493 |
0.6427 | 8.12 | 6100 | 0.6997 | 0.8475 |
0.6988 | 8.26 | 6200 | 0.6814 | 0.8098 |
0.6726 | 8.39 | 6300 | 0.6656 | 0.8259 |
0.6247 | 8.52 | 6400 | 0.6438 | 0.8314 |
0.5101 | 8.66 | 6500 | 0.6323 | 0.8446 |
0.5325 | 8.79 | 6600 | 0.6305 | 0.8413 |
0.5918 | 8.92 | 6700 | 0.6353 | 0.8076 |
0.617 | 9.05 | 6800 | 0.6544 | 0.8118 |
0.4861 | 9.19 | 6900 | 0.6174 | 0.8429 |
0.6396 | 9.32 | 7000 | 0.6140 | 0.8117 |
0.436 | 9.45 | 7100 | 0.6148 | 0.7887 |
0.6141 | 9.59 | 7200 | 0.6133 | 0.8007 |
0.5781 | 9.72 | 7300 | 0.6135 | 0.8211 |
0.52 | 9.85 | 7400 | 0.6155 | 0.8042 |
0.6681 | 9.99 | 7500 | 0.6074 | 0.7843 |
0.5004 | 10.12 | 7600 | 0.5950 | 0.8035 |
0.4993 | 10.25 | 7700 | 0.5888 | 0.7710 |
0.4768 | 10.39 | 7800 | 0.5922 | 0.7633 |
0.4535 | 10.52 | 7900 | 0.5906 | 0.8030 |
0.517 | 10.65 | 8000 | 0.5875 | 0.7823 |
0.5894 | 10.79 | 8100 | 0.5882 | 0.7932 |
0.6005 | 10.92 | 8200 | 0.5798 | 0.7922 |
0.4284 | 11.05 | 8300 | 0.5775 | 0.7701 |
0.5163 | 11.19 | 8400 | 0.5715 | 0.7592 |
0.4701 | 11.32 | 8500 | 0.5955 | 0.7485 |
0.5152 | 11.45 | 8600 | 0.6041 | 0.6914 |
0.4442 | 11.58 | 8700 | 0.5614 | 0.7439 |
0.4451 | 11.72 | 8800 | 0.5619 | 0.7033 |
0.4433 | 11.85 | 8900 | 0.5562 | 0.7246 |
0.4799 | 11.98 | 9000 | 0.5834 | 0.7040 |
0.4832 | 12.12 | 9100 | 0.5902 | 0.7349 |
0.523 | 12.25 | 9200 | 0.5562 | 0.7326 |
0.4419 | 12.38 | 9300 | 0.5472 | 0.7326 |
0.437 | 12.52 | 9400 | 0.5466 | 0.7100 |
0.4797 | 12.65 | 9500 | 0.5470 | 0.6698 |
0.3971 | 12.78 | 9600 | 0.5437 | 0.6835 |
0.5254 | 12.92 | 9700 | 0.5385 | 0.6747 |
0.5046 | 13.05 | 9800 | 0.5330 | 0.6554 |
0.4692 | 13.18 | 9900 | 0.5305 | 0.6527 |
0.4305 | 13.32 | 10000 | 0.5292 | 0.6314 |
0.6132 | 13.45 | 10100 | 0.5405 | 0.6028 |
0.4741 | 13.58 | 10200 | 0.5311 | 0.6207 |
0.398 | 13.72 | 10300 | 0.5320 | 0.6261 |
0.458 | 13.85 | 10400 | 0.5240 | 0.6242 |
0.4154 | 13.98 | 10500 | 0.5262 | 0.6215 |
0.3702 | 14.11 | 10600 | 0.5206 | 0.6136 |
0.427 | 14.25 | 10700 | 0.5231 | 0.6289 |
0.4307 | 14.38 | 10800 | 0.5210 | 0.5908 |
0.4738 | 14.51 | 10900 | 0.5211 | 0.5826 |
0.5522 | 14.65 | 11000 | 0.5193 | 0.5886 |
0.4717 | 14.78 | 11100 | 0.5194 | 0.5907 |
0.4819 | 14.91 | 11200 | 0.5178 | 0.5870 |
Framework versions
- Transformers 4.18.0
- Pytorch 1.9.1+cu102
- Datasets 2.0.0
- Tokenizers 0.11.6