w2v-bert-2.0-swahili-V100-32GB-CV14.0
This model is a fine-tuned version of facebook/w2v-bert-2.0 on the common_voice_14_0 dataset. It achieves the following results on the evaluation set:
- Loss: inf
- Wer: 0.9282
- Cer: 0.3257
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: 5e-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
- lr_scheduler_warmup_steps: 500
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 1.5245 | 0.19 | 300 | inf | 0.2379 | 0.0709 |
| 0.2545 | 0.38 | 600 | inf | 0.2225 | 0.0672 |
| 0.2269 | 0.57 | 900 | inf | 0.2020 | 0.0621 |
| 0.2027 | 0.77 | 1200 | inf | 0.1941 | 0.0604 |
| 0.1866 | 0.96 | 1500 | inf | 0.1893 | 0.0591 |
| 0.1721 | 1.15 | 1800 | inf | 0.1747 | 0.0538 |
| 0.1689 | 1.34 | 2100 | inf | 0.1781 | 0.0543 |
| 0.1647 | 1.53 | 2400 | inf | 0.1795 | 0.0545 |
| 0.1652 | 1.72 | 2700 | inf | 0.1736 | 0.0541 |
| 0.1659 | 1.91 | 3000 | inf | 0.1733 | 0.0528 |
| 0.1653 | 2.1 | 3300 | inf | 0.1753 | 0.0532 |
| 0.1577 | 2.3 | 3600 | inf | 0.1762 | 0.0530 |
| 0.192 | 2.49 | 3900 | inf | 0.1876 | 0.0579 |
| 0.2557 | 2.68 | 4200 | inf | 0.2411 | 0.0619 |
| 0.3876 | 2.87 | 4500 | inf | 0.2376 | 0.0677 |
| 0.4498 | 3.06 | 4800 | inf | 0.2080 | 0.0622 |
| 0.4865 | 3.25 | 5100 | inf | 0.2706 | 0.0744 |
| 0.842 | 3.44 | 5400 | inf | 0.5120 | 0.1169 |
| 0.9809 | 3.64 | 5700 | inf | 0.6735 | 0.1610 |
| 1.0493 | 3.83 | 6000 | inf | 0.8517 | 0.2787 |
| 1.236 | 4.02 | 6300 | inf | 0.7717 | 0.1951 |
| 1.2051 | 4.21 | 6600 | inf | 0.7491 | 0.1868 |
| 1.1908 | 4.4 | 6900 | inf | 0.8410 | 0.2340 |
| 1.1987 | 4.59 | 7200 | inf | 0.9118 | 0.2833 |
| 1.2397 | 4.78 | 7500 | inf | 0.9282 | 0.3257 |
| 1.2443 | 4.97 | 7800 | inf | 0.9282 | 0.3257 |
| 1.2428 | 5.17 | 8100 | inf | 0.9282 | 0.3257 |
| 1.2422 | 5.36 | 8400 | inf | 0.9282 | 0.3257 |
| 1.249 | 5.55 | 8700 | inf | 0.9282 | 0.3257 |
| 1.2518 | 5.74 | 9000 | inf | 0.9282 | 0.3257 |
| 1.2374 | 5.93 | 9300 | inf | 0.9282 | 0.3257 |
| 1.2369 | 6.12 | 9600 | inf | 0.9282 | 0.3257 |
| 1.2454 | 6.31 | 9900 | inf | 0.9282 | 0.3257 |
Framework versions
- Transformers 4.38.1
- Pytorch 2.2.1
- Datasets 2.17.0
- Tokenizers 0.15.2