<!-- 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-mms-1b-turkish
This model is a fine-tuned version of facebook/mms-1b-all on the common_voice_6_1 dataset. It achieves the following results on the evaluation set:
- Loss: 0.1443
- Wer: 0.2098
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: 0.001
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 10
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
4.2036 | 0.46 | 100 | 0.1980 | 0.2614 |
0.3 | 0.92 | 200 | 0.1918 | 0.2725 |
0.2735 | 1.38 | 300 | 0.1672 | 0.2346 |
0.2672 | 1.83 | 400 | 0.1671 | 0.2312 |
0.2641 | 2.29 | 500 | 0.1598 | 0.2248 |
0.2541 | 2.75 | 600 | 0.1587 | 0.2270 |
0.2696 | 3.21 | 700 | 0.1546 | 0.2235 |
0.2315 | 3.67 | 800 | 0.1559 | 0.2259 |
0.2396 | 4.13 | 900 | 0.1534 | 0.2172 |
0.2284 | 4.59 | 1000 | 0.1521 | 0.2172 |
0.2342 | 5.05 | 1100 | 0.1523 | 0.2178 |
0.2163 | 5.5 | 1200 | 0.1520 | 0.2184 |
0.2272 | 5.96 | 1300 | 0.1504 | 0.2182 |
0.2122 | 6.42 | 1400 | 0.1483 | 0.2149 |
0.2162 | 6.88 | 1500 | 0.1472 | 0.2100 |
0.2104 | 7.34 | 1600 | 0.1466 | 0.2104 |
0.2004 | 7.8 | 1700 | 0.1457 | 0.2110 |
0.2156 | 8.26 | 1800 | 0.1455 | 0.2134 |
0.1981 | 8.72 | 1900 | 0.1451 | 0.2103 |
0.1921 | 9.17 | 2000 | 0.1452 | 0.2105 |
0.19 | 9.63 | 2100 | 0.1443 | 0.2098 |
Framework versions
- Transformers 4.31.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3