metadata
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: ''
results: []
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice dataset. It achieves the following results on the evaluation set:
- Loss: 0.5217
- Wer: 0.4698
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: 7e-05
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2000
- num_epochs: 50.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.3711 | 2.61 | 300 | 4.3122 | 1.0 |
| 3.1653 | 5.22 | 600 | 3.1156 | 1.0 |
| 2.8904 | 7.83 | 900 | 2.8421 | 0.9918 |
| 0.9207 | 10.43 | 1200 | 0.9895 | 0.8689 |
| 0.6384 | 13.04 | 1500 | 0.6994 | 0.7700 |
| 0.5215 | 15.65 | 1800 | 0.5628 | 0.6443 |
| 0.4573 | 18.26 | 2100 | 0.5316 | 0.6174 |
| 0.3875 | 20.87 | 2400 | 0.4932 | 0.5779 |
| 0.3562 | 23.48 | 2700 | 0.4972 | 0.5475 |
| 0.3218 | 26.09 | 3000 | 0.4895 | 0.5219 |
| 0.2954 | 28.7 | 3300 | 0.5226 | 0.5192 |
| 0.287 | 31.3 | 3600 | 0.4957 | 0.5146 |
| 0.2587 | 33.91 | 3900 | 0.4944 | 0.4893 |
| 0.2496 | 36.52 | 4200 | 0.4976 | 0.4895 |
| 0.2365 | 39.13 | 4500 | 0.5185 | 0.4819 |
| 0.2264 | 41.74 | 4800 | 0.5152 | 0.4776 |
| 0.2224 | 44.35 | 5100 | 0.5031 | 0.4746 |
| 0.2096 | 46.96 | 5400 | 0.5062 | 0.4708 |
| 0.2038 | 49.57 | 5700 | 0.5217 | 0.4698 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2.dev0
- Tokenizers 0.11.0