xls-r-300m-bemba-20hrs
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: 0.2815
- Wer: 0.3435
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 |
---|---|---|---|---|
3.3301 | 0.54 | 400 | 0.5177 | 0.7570 |
0.6437 | 1.08 | 800 | 0.3580 | 0.5658 |
0.5149 | 1.61 | 1200 | 0.2953 | 0.5004 |
0.4547 | 2.15 | 1600 | 0.2701 | 0.4464 |
0.4084 | 2.69 | 2000 | 0.2743 | 0.4383 |
0.3606 | 3.23 | 2400 | 0.2482 | 0.3952 |
0.3227 | 3.76 | 2800 | 0.2461 | 0.3965 |
0.3025 | 4.3 | 3200 | 0.2484 | 0.4015 |
0.2697 | 4.84 | 3600 | 0.2357 | 0.3838 |
0.2443 | 5.38 | 4000 | 0.2385 | 0.3822 |
0.2287 | 5.91 | 4400 | 0.2353 | 0.3747 |
0.1977 | 6.45 | 4800 | 0.2337 | 0.3624 |
0.1895 | 6.99 | 5200 | 0.2319 | 0.3568 |
0.1561 | 7.53 | 5600 | 0.2540 | 0.3561 |
0.1448 | 8.06 | 6000 | 0.2772 | 0.3612 |
0.1221 | 8.6 | 6400 | 0.2755 | 0.3596 |
0.1133 | 9.14 | 6800 | 0.2733 | 0.3495 |
0.0969 | 9.68 | 7200 | 0.2815 | 0.3435 |
Framework versions
- Transformers 4.19.0.dev0
- Pytorch 1.10.0+cu111
- Datasets 2.1.0
- Tokenizers 0.12.1
- Downloads last month
- 2