Model: alinerodrigues/wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-04 Source: Original Platform
license, tags, metrics, model-index
| license | tags | metrics | model-index | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| apache-2.0 |
|
|
|
wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-04
This model is a fine-tuned version of Edresson/wav2vec2-large-xlsr-coraa-portuguese on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1832
- Wer: 0.0896
- Cer: 0.0286
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 |
|---|---|---|---|---|---|
| 28.4874 | 1.0 | 86 | 3.3048 | 1.0 | 1.0 |
| 7.6481 | 2.0 | 172 | 2.9770 | 1.0 | 1.0 |
| 3.0297 | 3.0 | 258 | 2.9361 | 1.0 | 1.0 |
| 2.9192 | 4.0 | 344 | 2.9206 | 1.0 | 1.0 |
| 2.8826 | 5.0 | 430 | 2.6372 | 0.9990 | 0.9993 |
| 2.3512 | 6.0 | 516 | 0.8687 | 0.4358 | 0.1113 |
| 1.0373 | 7.0 | 602 | 0.5095 | 0.2419 | 0.0665 |
| 1.0373 | 8.0 | 688 | 0.4001 | 0.2101 | 0.0584 |
| 0.6758 | 9.0 | 774 | 0.3354 | 0.1799 | 0.0502 |
| 0.5667 | 10.0 | 860 | 0.2962 | 0.1617 | 0.0468 |
| 0.4908 | 11.0 | 946 | 0.2820 | 0.1471 | 0.0430 |
| 0.4084 | 12.0 | 1032 | 0.2581 | 0.1351 | 0.0399 |
| 0.3728 | 13.0 | 1118 | 0.2444 | 0.1230 | 0.0374 |
| 0.3577 | 14.0 | 1204 | 0.2408 | 0.1144 | 0.0355 |
| 0.3577 | 15.0 | 1290 | 0.2353 | 0.1100 | 0.0340 |
| 0.3237 | 16.0 | 1376 | 0.2227 | 0.1036 | 0.0332 |
| 0.3043 | 17.0 | 1462 | 0.2241 | 0.1004 | 0.0322 |
| 0.2876 | 18.0 | 1548 | 0.2260 | 0.1048 | 0.0329 |
| 0.2842 | 19.0 | 1634 | 0.2086 | 0.1070 | 0.0325 |
| 0.269 | 20.0 | 1720 | 0.2156 | 0.1006 | 0.0324 |
| 0.2396 | 21.0 | 1806 | 0.2065 | 0.0952 | 0.0307 |
| 0.2396 | 22.0 | 1892 | 0.2144 | 0.0960 | 0.0312 |
| 0.2411 | 23.0 | 1978 | 0.2018 | 0.0933 | 0.0307 |
| 0.2357 | 24.0 | 2064 | 0.2055 | 0.0940 | 0.0305 |
| 0.2268 | 25.0 | 2150 | 0.2099 | 0.0928 | 0.0307 |
| 0.2053 | 26.0 | 2236 | 0.2044 | 0.0933 | 0.0302 |
| 0.2188 | 27.0 | 2322 | 0.2037 | 0.0933 | 0.0301 |
| 0.2152 | 28.0 | 2408 | 0.2068 | 0.0933 | 0.0298 |
| 0.2152 | 29.0 | 2494 | 0.2044 | 0.0898 | 0.0283 |
| 0.1998 | 30.0 | 2580 | 0.2039 | 0.0888 | 0.0294 |
| 0.1934 | 31.0 | 2666 | 0.1965 | 0.0915 | 0.0296 |
| 0.19 | 32.0 | 2752 | 0.1994 | 0.0908 | 0.0291 |
| 0.2108 | 33.0 | 2838 | 0.2026 | 0.0888 | 0.0292 |
| 0.1922 | 34.0 | 2924 | 0.1957 | 0.0906 | 0.0292 |
| 0.1827 | 35.0 | 3010 | 0.1935 | 0.0893 | 0.0287 |
| 0.1827 | 36.0 | 3096 | 0.1933 | 0.0906 | 0.0295 |
| 0.1848 | 37.0 | 3182 | 0.1999 | 0.0883 | 0.0287 |
| 0.1807 | 38.0 | 3268 | 0.1949 | 0.0881 | 0.0287 |
| 0.1695 | 39.0 | 3354 | 0.1968 | 0.0906 | 0.0295 |
| 0.1764 | 40.0 | 3440 | 0.1894 | 0.0864 | 0.0284 |
| 0.1812 | 41.0 | 3526 | 0.1932 | 0.0871 | 0.0286 |
| 0.1566 | 42.0 | 3612 | 0.1889 | 0.0869 | 0.0281 |
| 0.1566 | 43.0 | 3698 | 0.1965 | 0.0864 | 0.0282 |
| 0.161 | 44.0 | 3784 | 0.1832 | 0.0896 | 0.0286 |
| 0.167 | 45.0 | 3870 | 0.1968 | 0.0886 | 0.0282 |
| 0.1556 | 46.0 | 3956 | 0.1835 | 0.0881 | 0.0283 |
| 0.157 | 47.0 | 4042 | 0.1912 | 0.0864 | 0.0279 |
| 0.1588 | 48.0 | 4128 | 0.1926 | 0.0883 | 0.0279 |
| 0.1543 | 49.0 | 4214 | 0.1961 | 0.0869 | 0.0282 |
| 0.1528 | 50.0 | 4300 | 0.1959 | 0.0886 | 0.0286 |
| 0.1528 | 51.0 | 4386 | 0.1938 | 0.0869 | 0.0277 |
| 0.1477 | 52.0 | 4472 | 0.1984 | 0.0839 | 0.0272 |
| 0.1533 | 53.0 | 4558 | 0.1969 | 0.0864 | 0.0274 |
| 0.1351 | 54.0 | 4644 | 0.1983 | 0.0856 | 0.0275 |
| 0.1468 | 55.0 | 4730 | 0.1970 | 0.0834 | 0.0272 |
| 0.1406 | 56.0 | 4816 | 0.1940 | 0.0817 | 0.0270 |
| 0.1412 | 57.0 | 4902 | 0.1894 | 0.0859 | 0.0277 |
| 0.1412 | 58.0 | 4988 | 0.1929 | 0.0856 | 0.0275 |
| 0.1435 | 59.0 | 5074 | 0.1977 | 0.0859 | 0.0278 |
| 0.1389 | 60.0 | 5160 | 0.1986 | 0.0849 | 0.0277 |
| 0.1366 | 61.0 | 5246 | 0.1977 | 0.0856 | 0.0272 |
| 0.1367 | 62.0 | 5332 | 0.1948 | 0.0851 | 0.0275 |
| 0.1268 | 63.0 | 5418 | 0.1947 | 0.0856 | 0.0275 |
| 0.1441 | 64.0 | 5504 | 0.1956 | 0.0856 | 0.0277 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.1.1+cu121
- Datasets 2.15.0
- Tokenizers 0.13.3
Description
Model synced from source: alinerodrigues/wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-04