<!-- 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. -->
xls-r-300m-bemba-15hrs
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.2754
- Wer: 0.3481
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: 0.0003
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.5142 | 0.71 | 400 | 0.5585 | 0.7501 |
| 0.6351 | 1.43 | 800 | 0.3185 | 0.5058 |
| 0.4892 | 2.15 | 1200 | 0.2813 | 0.4655 |
| 0.4021 | 2.86 | 1600 | 0.2539 | 0.4159 |
| 0.3505 | 3.58 | 2000 | 0.2411 | 0.4000 |
| 0.3045 | 4.29 | 2400 | 0.2512 | 0.3951 |
| 0.274 | 5.01 | 2800 | 0.2402 | 0.3922 |
| 0.2335 | 5.72 | 3200 | 0.2403 | 0.3764 |
| 0.2032 | 6.44 | 3600 | 0.2383 | 0.3657 |
| 0.1783 | 7.16 | 4000 | 0.2603 | 0.3518 |
| 0.1487 | 7.87 | 4400 | 0.2479 | 0.3577 |
| 0.1281 | 8.59 | 4800 | 0.2638 | 0.3518 |
| 0.113 | 9.3 | 5200 | 0.2754 | 0.3481 |
Framework versions
- Transformers 4.19.0.dev0
- Pytorch 1.10.0+cu111
- Datasets 2.1.0
- Tokenizers 0.12.1
