wav2vec2-cls-r-300m-es
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the COMMON_VOICE - ES dataset. It achieves the following results on the evaluation set:
- Loss: 0.5160
- Wer: 0.4016
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: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 8.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.1277 | 1.14 | 500 | 2.0259 | 0.9999 |
| 1.4111 | 2.28 | 1000 | 1.1251 | 0.8894 |
| 0.8461 | 3.42 | 1500 | 0.8205 | 0.7244 |
| 0.5042 | 4.57 | 2000 | 0.6116 | 0.5463 |
| 0.3072 | 5.71 | 2500 | 0.5507 | 0.4506 |
| 0.2181 | 6.85 | 3000 | 0.5213 | 0.4177 |
| 0.1608 | 7.99 | 3500 | 0.5161 | 0.4019 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2.dev0
- Tokenizers 0.11.0
Evaluation Commands
- To evaluate on
mozilla-foundation/common_voice_7_0with splittest
python eval.py --model_id samitizerxu/wav2vec2-xls-r-300m-es --dataset mozilla-foundation/common_voice_7_0 --config es --split test
- To evaluate on
speech-recognition-community-v2/dev_data
python eval.py --model_id samitizerxu/wav2vec2-xls-r-300m-es --dataset speech-recognition-community-v2/dev_data --config es --split validation --chunk_length_s 5.0 --stride_length_s 1.0