<!-- 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. -->
wav2vec2-large-mms-1b-por
This model is a fine-tuned version of facebook/mms-1b-all on the mozilla-foundation/common_voice_13_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.1340
- Wer: 0.1141
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: 32
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 100
- training_steps: 2500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.3219 | 0.55 | 500 | 0.1743 | 0.1302 |
| 0.2443 | 1.1 | 1000 | 0.1480 | 0.1206 |
| 0.2358 | 1.65 | 1500 | 0.1402 | 0.1167 |
| 0.223 | 2.21 | 2000 | 0.1364 | 0.1159 |
| 0.2213 | 2.76 | 2500 | 0.1340 | 0.1141 |
Framework versions
- Transformers 4.35.0.dev0
- Pytorch 2.1.0+cu121
- Datasets 2.14.5
- Tokenizers 0.14.1
