<!-- 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-mecita-coraa-portuguese-all-clean-05
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: 0.1089
- Wer: 0.0833
- Cer: 0.0226
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 |
---|---|---|---|---|---|
28.3067 | 1.0 | 67 | 3.2489 | 1.0 | 1.0 |
7.0285 | 2.0 | 134 | 2.9648 | 1.0 | 1.0 |
3.012 | 3.0 | 201 | 2.9002 | 1.0 | 1.0 |
3.012 | 4.0 | 268 | 2.8732 | 1.0 | 1.0 |
2.91 | 5.0 | 335 | 2.7571 | 1.0 | 1.0 |
2.6396 | 6.0 | 402 | 1.4049 | 1.0 | 0.2939 |
2.6396 | 7.0 | 469 | 0.5538 | 0.2890 | 0.0712 |
1.2247 | 8.0 | 536 | 0.3766 | 0.2199 | 0.0541 |
0.7133 | 9.0 | 603 | 0.2976 | 0.1875 | 0.0479 |
0.7133 | 10.0 | 670 | 0.2570 | 0.1610 | 0.0424 |
0.5354 | 11.0 | 737 | 0.2390 | 0.1531 | 0.0403 |
0.4555 | 12.0 | 804 | 0.2139 | 0.1356 | 0.0363 |
0.4555 | 13.0 | 871 | 0.1968 | 0.1303 | 0.0340 |
0.4168 | 14.0 | 938 | 0.1846 | 0.1270 | 0.0341 |
0.3717 | 15.0 | 1005 | 0.1720 | 0.1118 | 0.0304 |
0.3717 | 16.0 | 1072 | 0.1735 | 0.1104 | 0.0311 |
0.3164 | 17.0 | 1139 | 0.1564 | 0.1121 | 0.0306 |
0.311 | 18.0 | 1206 | 0.1542 | 0.1055 | 0.0287 |
0.311 | 19.0 | 1273 | 0.1484 | 0.1071 | 0.0286 |
0.2748 | 20.0 | 1340 | 0.1460 | 0.1012 | 0.0271 |
0.2443 | 21.0 | 1407 | 0.1411 | 0.0979 | 0.0266 |
0.2443 | 22.0 | 1474 | 0.1397 | 0.0959 | 0.0267 |
0.2502 | 23.0 | 1541 | 0.1379 | 0.0923 | 0.0253 |
0.2362 | 24.0 | 1608 | 0.1346 | 0.0899 | 0.0251 |
0.2362 | 25.0 | 1675 | 0.1336 | 0.0989 | 0.0268 |
0.2399 | 26.0 | 1742 | 0.1290 | 0.0956 | 0.0255 |
0.2546 | 27.0 | 1809 | 0.1256 | 0.0906 | 0.0244 |
0.2546 | 28.0 | 1876 | 0.1254 | 0.0956 | 0.0256 |
0.2274 | 29.0 | 1943 | 0.1230 | 0.0903 | 0.0240 |
0.2035 | 30.0 | 2010 | 0.1199 | 0.0890 | 0.0239 |
0.2035 | 31.0 | 2077 | 0.1198 | 0.0893 | 0.0241 |
0.2137 | 32.0 | 2144 | 0.1185 | 0.0886 | 0.0246 |
0.1963 | 33.0 | 2211 | 0.1219 | 0.0883 | 0.0244 |
0.1963 | 34.0 | 2278 | 0.1193 | 0.0873 | 0.0241 |
0.1755 | 35.0 | 2345 | 0.1222 | 0.0853 | 0.0242 |
0.1956 | 36.0 | 2412 | 0.1172 | 0.0860 | 0.0240 |
0.1956 | 37.0 | 2479 | 0.1179 | 0.0850 | 0.0238 |
0.1777 | 38.0 | 2546 | 0.1175 | 0.0860 | 0.0234 |
0.2008 | 39.0 | 2613 | 0.1142 | 0.0837 | 0.0226 |
0.2008 | 40.0 | 2680 | 0.1124 | 0.0870 | 0.0234 |
0.1703 | 41.0 | 2747 | 0.1129 | 0.0804 | 0.0223 |
0.1713 | 42.0 | 2814 | 0.1128 | 0.0820 | 0.0227 |
0.1713 | 43.0 | 2881 | 0.1126 | 0.0797 | 0.0223 |
0.1827 | 44.0 | 2948 | 0.1126 | 0.0800 | 0.0225 |
0.1582 | 45.0 | 3015 | 0.1128 | 0.0807 | 0.0228 |
0.1582 | 46.0 | 3082 | 0.1109 | 0.0784 | 0.0220 |
0.1644 | 47.0 | 3149 | 0.1115 | 0.0840 | 0.0230 |
0.1598 | 48.0 | 3216 | 0.1119 | 0.0823 | 0.0231 |
0.1598 | 49.0 | 3283 | 0.1095 | 0.0820 | 0.0227 |
0.159 | 50.0 | 3350 | 0.1106 | 0.0790 | 0.0223 |
0.1645 | 51.0 | 3417 | 0.1089 | 0.0833 | 0.0226 |
0.1645 | 52.0 | 3484 | 0.1113 | 0.0787 | 0.0222 |
0.1518 | 53.0 | 3551 | 0.1106 | 0.0830 | 0.0230 |
0.137 | 54.0 | 3618 | 0.1092 | 0.0790 | 0.0222 |
0.137 | 55.0 | 3685 | 0.1100 | 0.0813 | 0.0227 |
0.1454 | 56.0 | 3752 | 0.1105 | 0.0794 | 0.0226 |
0.1405 | 57.0 | 3819 | 0.1112 | 0.0777 | 0.0223 |
0.1405 | 58.0 | 3886 | 0.1130 | 0.0800 | 0.0225 |
0.1369 | 59.0 | 3953 | 0.1111 | 0.0794 | 0.0223 |
0.1513 | 60.0 | 4020 | 0.1122 | 0.0797 | 0.0218 |
0.1513 | 61.0 | 4087 | 0.1106 | 0.0797 | 0.0218 |
0.1513 | 62.0 | 4154 | 0.1123 | 0.0777 | 0.0216 |
0.1285 | 63.0 | 4221 | 0.1097 | 0.0787 | 0.0223 |
0.1285 | 64.0 | 4288 | 0.1116 | 0.0784 | 0.0223 |
0.1538 | 65.0 | 4355 | 0.1118 | 0.0804 | 0.0225 |
0.1325 | 66.0 | 4422 | 0.1106 | 0.0800 | 0.0222 |
0.1325 | 67.0 | 4489 | 0.1125 | 0.0771 | 0.0219 |
0.138 | 68.0 | 4556 | 0.1123 | 0.0804 | 0.0225 |
0.1339 | 69.0 | 4623 | 0.1111 | 0.0813 | 0.0229 |
0.1339 | 70.0 | 4690 | 0.1115 | 0.0804 | 0.0232 |
0.1307 | 71.0 | 4757 | 0.1117 | 0.0820 | 0.0234 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.0+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3