wav2vec2-xlsr-1b-mecita-portuguese-all-text-a_coisa-os_morcegos
This model is a fine-tuned version of jonatasgrosman/wav2vec2-xls-r-1b-portuguese on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1774
- Wer: 0.0844
- Cer: 0.0266
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: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 100
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
---|---|---|---|---|---|
25.5905 | 1.0 | 79 | 0.4495 | 0.2580 | 0.0734 |
3.1482 | 2.0 | 158 | 0.2479 | 0.1204 | 0.0380 |
0.4247 | 3.0 | 237 | 0.2347 | 0.1025 | 0.0345 |
0.3136 | 4.0 | 316 | 0.2044 | 0.1017 | 0.0322 |
0.3136 | 5.0 | 395 | 0.1906 | 0.0930 | 0.0296 |
0.2985 | 6.0 | 474 | 0.2050 | 0.0963 | 0.0311 |
0.2413 | 7.0 | 553 | 0.2025 | 0.0971 | 0.0309 |
0.2267 | 8.0 | 632 | 0.2006 | 0.0885 | 0.0291 |
0.224 | 9.0 | 711 | 0.1991 | 0.0917 | 0.0291 |
0.224 | 10.0 | 790 | 0.1881 | 0.0885 | 0.0281 |
0.1864 | 11.0 | 869 | 0.1841 | 0.0893 | 0.0278 |
0.1951 | 12.0 | 948 | 0.1809 | 0.0895 | 0.0282 |
0.1794 | 13.0 | 1027 | 0.1923 | 0.0833 | 0.0280 |
0.1621 | 14.0 | 1106 | 0.1949 | 0.0857 | 0.0277 |
0.1621 | 15.0 | 1185 | 0.1929 | 0.0817 | 0.0266 |
0.1695 | 16.0 | 1264 | 0.1907 | 0.0839 | 0.0270 |
0.1528 | 17.0 | 1343 | 0.1839 | 0.0906 | 0.0286 |
0.1592 | 18.0 | 1422 | 0.1866 | 0.0903 | 0.0281 |
0.1519 | 19.0 | 1501 | 0.2031 | 0.0857 | 0.0275 |
0.1519 | 20.0 | 1580 | 0.1948 | 0.0860 | 0.0278 |
0.1257 | 21.0 | 1659 | 0.1850 | 0.0860 | 0.0262 |
0.1288 | 22.0 | 1738 | 0.1774 | 0.0844 | 0.0266 |
0.115 | 23.0 | 1817 | 0.1960 | 0.0844 | 0.0265 |
0.115 | 24.0 | 1896 | 0.1832 | 0.0825 | 0.0258 |
0.1223 | 25.0 | 1975 | 0.1920 | 0.0828 | 0.0261 |
0.1175 | 26.0 | 2054 | 0.1951 | 0.0803 | 0.0260 |
0.1051 | 27.0 | 2133 | 0.1996 | 0.0825 | 0.0266 |
0.1033 | 28.0 | 2212 | 0.2152 | 0.0847 | 0.0274 |
0.1033 | 29.0 | 2291 | 0.2082 | 0.0879 | 0.0277 |
0.0961 | 30.0 | 2370 | 0.2153 | 0.0855 | 0.0274 |
0.1003 | 31.0 | 2449 | 0.2044 | 0.0903 | 0.0288 |
0.1129 | 32.0 | 2528 | 0.2050 | 0.0855 | 0.0268 |
0.0939 | 33.0 | 2607 | 0.2028 | 0.0860 | 0.0271 |
0.0939 | 34.0 | 2686 | 0.2031 | 0.0847 | 0.0274 |
0.0846 | 35.0 | 2765 | 0.2046 | 0.0822 | 0.0269 |
0.083 | 36.0 | 2844 | 0.2094 | 0.0825 | 0.0265 |
0.0844 | 37.0 | 2923 | 0.2176 | 0.0820 | 0.0268 |
0.0829 | 38.0 | 3002 | 0.2082 | 0.0817 | 0.0267 |
0.0829 | 39.0 | 3081 | 0.2200 | 0.0893 | 0.0286 |
0.103 | 40.0 | 3160 | 0.2102 | 0.0841 | 0.0276 |
0.0728 | 41.0 | 3239 | 0.2143 | 0.0817 | 0.0271 |
0.079 | 42.0 | 3318 | 0.2131 | 0.0825 | 0.0265 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.2.1+cu121
- Datasets 2.17.0
- Tokenizers 0.13.3
- Downloads last month
- 4
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.