wav2vec2-large-xlsr-53-finetuned-ks
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.4923
- Accuracy: 0.7871
- F1: 0.7863
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: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
|---|---|---|---|---|---|
| 1.379 | 1.0 | 141 | 1.3767 | 0.2991 | 0.1377 |
| 1.3611 | 2.0 | 283 | 1.3600 | 0.2991 | 0.1377 |
| 1.3393 | 3.0 | 424 | 1.3515 | 0.2991 | 0.1377 |
| 1.2932 | 4.0 | 566 | 1.3306 | 0.3607 | 0.3098 |
| 1.2356 | 5.0 | 707 | 1.2202 | 0.4397 | 0.3926 |
| 1.2222 | 6.0 | 849 | 1.3719 | 0.3601 | 0.2778 |
| 1.036 | 7.0 | 990 | 1.2779 | 0.4290 | 0.3781 |
| 1.0348 | 8.0 | 1132 | 1.2845 | 0.4257 | 0.3824 |
| 0.9044 | 9.0 | 1273 | 1.2239 | 0.4927 | 0.4646 |
| 0.8557 | 10.0 | 1415 | 1.6261 | 0.3926 | 0.3253 |
| 0.804 | 11.0 | 1556 | 1.0748 | 0.5703 | 0.5558 |
| 0.6517 | 12.0 | 1698 | 1.2891 | 0.5471 | 0.5294 |
| 0.6063 | 13.0 | 1839 | 0.9921 | 0.6552 | 0.6514 |
| 0.5008 | 14.0 | 1981 | 1.4346 | 0.5391 | 0.5162 |
| 0.5425 | 15.0 | 2122 | 1.3406 | 0.5802 | 0.5573 |
| 0.3806 | 16.0 | 2264 | 1.2260 | 0.6353 | 0.6291 |
| 0.4022 | 17.0 | 2405 | 1.7530 | 0.5444 | 0.5197 |
| 0.3001 | 18.0 | 2547 | 1.3619 | 0.6247 | 0.6132 |
| 0.1921 | 19.0 | 2688 | 1.3687 | 0.6505 | 0.6443 |
| 0.2704 | 20.0 | 2830 | 1.2533 | 0.6810 | 0.6745 |
| 0.3145 | 21.0 | 2971 | 1.6079 | 0.6233 | 0.6133 |
| 0.2045 | 22.0 | 3113 | 1.1432 | 0.7215 | 0.7198 |
| 0.2444 | 23.0 | 3254 | 1.4012 | 0.6936 | 0.6861 |
| 0.2223 | 24.0 | 3396 | 1.5944 | 0.6585 | 0.6533 |
| 0.2415 | 25.0 | 3537 | 1.1057 | 0.7454 | 0.7420 |
| 0.2233 | 26.0 | 3679 | 1.4083 | 0.7036 | 0.6997 |
| 0.119 | 27.0 | 3820 | 1.3240 | 0.7341 | 0.7323 |
| 0.1125 | 28.0 | 3962 | 1.8332 | 0.6658 | 0.6590 |
| 0.1577 | 29.0 | 4103 | 1.8048 | 0.6764 | 0.6714 |
| 0.1169 | 30.0 | 4245 | 1.3329 | 0.7573 | 0.7563 |
| 0.1348 | 31.0 | 4386 | 2.0588 | 0.6485 | 0.6359 |
| 0.1203 | 32.0 | 4528 | 1.6487 | 0.7082 | 0.7012 |
| 0.1262 | 33.0 | 4669 | 1.5428 | 0.7261 | 0.7236 |
| 0.0679 | 34.0 | 4811 | 1.5458 | 0.7374 | 0.7357 |
| 0.0741 | 35.0 | 4952 | 1.4596 | 0.7546 | 0.7508 |
| 0.0913 | 36.0 | 5094 | 1.3710 | 0.7699 | 0.7702 |
| 0.2104 | 37.0 | 5235 | 1.6693 | 0.7367 | 0.7344 |
| 0.0856 | 38.0 | 5377 | 1.6339 | 0.75 | 0.7483 |
| 0.0931 | 39.0 | 5518 | 1.6512 | 0.7580 | 0.7571 |
| 0.0613 | 40.0 | 5660 | 1.6046 | 0.7646 | 0.7638 |
| 0.0713 | 41.0 | 5801 | 1.4553 | 0.7785 | 0.7779 |
| 0.025 | 42.0 | 5943 | 1.5725 | 0.7639 | 0.7625 |
| 0.0811 | 43.0 | 6084 | 1.7562 | 0.75 | 0.7474 |
| 0.0315 | 44.0 | 6226 | 1.4923 | 0.7871 | 0.7863 |
| 0.1026 | 45.0 | 6367 | 1.6013 | 0.7712 | 0.7706 |
| 0.0489 | 46.0 | 6509 | 1.7439 | 0.7533 | 0.7502 |
| 0.0248 | 47.0 | 6650 | 1.6019 | 0.7745 | 0.7730 |
| 0.0269 | 48.0 | 6792 | 1.6128 | 0.7679 | 0.7659 |
| 0.0114 | 49.0 | 6933 | 1.5737 | 0.7798 | 0.7788 |
| 0.0609 | 49.82 | 7050 | 1.6570 | 0.7712 | 0.7692 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.1+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0