<!-- 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. -->
comp_3090
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: 4.0263
- Cer: 0.9947
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-06
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 981
- num_epochs: 30
Training results
Training Loss | Epoch | Step | Validation Loss | Cer |
---|---|---|---|---|
46.2038 | 0.31 | 100 | 52.3651 | 1.0232 |
46.7609 | 0.61 | 200 | 52.2540 | 1.0257 |
45.0071 | 0.92 | 300 | 52.0587 | 1.0351 |
41.9868 | 1.22 | 400 | 51.7005 | 1.1262 |
38.7309 | 1.53 | 500 | 49.4417 | 1.0934 |
33.823 | 1.83 | 600 | 40.2333 | 0.9995 |
29.1305 | 2.14 | 700 | 31.4724 | 1.0 |
21.7575 | 2.45 | 800 | 23.1356 | 1.0 |
16.4239 | 2.75 | 900 | 16.3337 | 1.0 |
11.9535 | 3.06 | 1000 | 12.2005 | 1.0 |
9.914 | 3.36 | 1100 | 9.9800 | 1.0 |
8.6519 | 3.67 | 1200 | 8.5962 | 1.0 |
7.7254 | 3.98 | 1300 | 7.6834 | 1.0 |
6.9742 | 4.28 | 1400 | 7.0192 | 1.0 |
6.5832 | 4.59 | 1500 | 6.5116 | 1.0 |
6.0663 | 4.89 | 1600 | 6.1000 | 1.0 |
5.6379 | 5.2 | 1700 | 5.7657 | 1.0 |
5.431 | 5.5 | 1800 | 5.4872 | 1.0 |
5.0887 | 5.81 | 1900 | 5.2548 | 1.0 |
4.9387 | 6.12 | 2000 | 5.0602 | 1.0 |
4.8131 | 6.42 | 2100 | 4.8975 | 1.0 |
4.6514 | 6.73 | 2200 | 4.7620 | 1.0 |
4.5532 | 7.03 | 2300 | 4.6492 | 1.0 |
4.3926 | 7.34 | 2400 | 4.5561 | 1.0 |
4.4024 | 7.65 | 2500 | 4.4806 | 1.0 |
4.3113 | 7.95 | 2600 | 4.4201 | 1.0 |
4.244 | 8.26 | 2700 | 4.3734 | 1.0 |
4.2026 | 8.56 | 2800 | 4.3384 | 1.0 |
4.1714 | 8.87 | 2900 | 4.3123 | 1.0 |
4.1611 | 9.17 | 3000 | 4.2899 | 1.0 |
4.1734 | 9.48 | 3100 | 4.2741 | 1.0 |
4.1349 | 9.79 | 3200 | 4.2645 | 1.0 |
4.1354 | 10.09 | 3300 | 4.2542 | 1.0 |
4.1337 | 10.4 | 3400 | 4.2453 | 1.0 |
4.1247 | 10.7 | 3500 | 4.2393 | 1.0 |
4.0781 | 11.01 | 3600 | 4.2325 | 1.0 |
4.1031 | 11.31 | 3700 | 4.2268 | 1.0 |
4.079 | 11.62 | 3800 | 4.2226 | 1.0 |
4.0746 | 11.93 | 3900 | 4.2180 | 1.0 |
4.134 | 12.23 | 4000 | 4.2137 | 1.0 |
4.106 | 12.54 | 4100 | 4.2081 | 1.0 |
4.1156 | 12.84 | 4200 | 4.2042 | 1.0 |
4.0492 | 13.15 | 4300 | 4.1965 | 1.0 |
4.0673 | 13.46 | 4400 | 4.1889 | 1.0 |
4.0102 | 13.76 | 4500 | 4.1760 | 1.0 |
4.0961 | 14.07 | 4600 | 4.1631 | 1.0 |
4.1059 | 14.37 | 4700 | 4.1521 | 1.0 |
4.0817 | 14.68 | 4800 | 4.1426 | 1.0 |
4.0876 | 14.98 | 4900 | 4.1357 | 1.0 |
3.9945 | 15.29 | 5000 | 4.1224 | 1.0 |
3.9328 | 15.6 | 5100 | 4.1149 | 1.0 |
3.9507 | 15.9 | 5200 | 4.1073 | 1.0 |
4.0651 | 16.21 | 5300 | 4.1012 | 1.0 |
4.0564 | 16.51 | 5400 | 4.0928 | 1.0 |
4.0383 | 16.82 | 5500 | 4.0893 | 1.0 |
3.9363 | 17.13 | 5600 | 4.0848 | 1.0 |
3.95 | 17.43 | 5700 | 4.0801 | 1.0 |
3.9208 | 17.74 | 5800 | 4.0754 | 1.0 |
3.9867 | 18.04 | 5900 | 4.0712 | 1.0 |
4.038 | 18.35 | 6000 | 4.0693 | 1.0 |
3.9908 | 18.65 | 6100 | 4.0680 | 1.0 |
3.9808 | 18.96 | 6200 | 4.0669 | 1.0 |
3.9243 | 19.27 | 6300 | 4.0592 | 1.0 |
3.9499 | 19.57 | 6400 | 4.0558 | 1.0 |
3.9115 | 19.88 | 6500 | 4.0568 | 1.0 |
3.9691 | 20.18 | 6600 | 4.0549 | 1.0 |
3.9523 | 20.49 | 6700 | 4.0502 | 1.0 |
3.978 | 20.8 | 6800 | 4.0488 | 1.0 |
3.9714 | 21.1 | 6900 | 4.0443 | 1.0 |
3.943 | 21.41 | 7000 | 4.0432 | 1.0 |
3.9466 | 21.71 | 7100 | 4.0411 | 1.0 |
3.9541 | 22.02 | 7200 | 4.0386 | 1.0 |
3.9479 | 22.32 | 7300 | 4.0398 | 1.0 |
3.9056 | 22.63 | 7400 | 4.0398 | 1.0 |
3.9059 | 22.94 | 7500 | 4.0393 | 1.0 |
3.9854 | 23.24 | 7600 | 4.0347 | 0.9999 |
3.9454 | 23.55 | 7700 | 4.0324 | 0.9998 |
3.989 | 23.85 | 7800 | 4.0325 | 0.9993 |
3.874 | 24.16 | 7900 | 4.0322 | 0.9984 |
3.8832 | 24.46 | 8000 | 4.0300 | 0.9980 |
3.8495 | 24.77 | 8100 | 4.0313 | 0.9969 |
4.0393 | 25.08 | 8200 | 4.0286 | 0.9975 |
4.0241 | 25.38 | 8300 | 4.0290 | 0.9966 |
3.9938 | 25.69 | 8400 | 4.0288 | 0.9960 |
3.9031 | 25.99 | 8500 | 4.0285 | 0.9957 |
3.8455 | 26.3 | 8600 | 4.0275 | 0.9957 |
3.8827 | 26.61 | 8700 | 4.0285 | 0.9951 |
3.8677 | 26.91 | 8800 | 4.0272 | 0.9952 |
4.0282 | 27.22 | 8900 | 4.0269 | 0.9951 |
3.9954 | 27.52 | 9000 | 4.0271 | 0.9949 |
3.9914 | 27.83 | 9100 | 4.0264 | 0.9950 |
3.8657 | 28.13 | 9200 | 4.0265 | 0.9948 |
3.8687 | 28.44 | 9300 | 4.0266 | 0.9947 |
3.8746 | 28.75 | 9400 | 4.0264 | 0.9947 |
3.9765 | 29.05 | 9500 | 4.0263 | 0.9947 |
4.0199 | 29.36 | 9600 | 4.0263 | 0.9947 |
3.9756 | 29.66 | 9700 | 4.0263 | 0.9947 |
3.9395 | 29.97 | 9800 | 4.0263 | 0.9947 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.0.1
- Datasets 2.12.0
- Tokenizers 0.13.2