<!-- 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. -->
w2v2-libri-10min
This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 2.1788
- Wer: 0.6238
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
- lr_scheduler_warmup_steps: 500
- training_steps: 2500
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
5.5273 | 62.5 | 250 | 3.0307 | 1.0 |
1.5106 | 125.0 | 500 | 1.4811 | 0.6598 |
0.082 | 187.5 | 750 | 1.8040 | 0.6376 |
0.0396 | 250.0 | 1000 | 2.0617 | 0.6750 |
0.0211 | 312.5 | 1250 | 2.1518 | 0.6819 |
0.014 | 375.0 | 1500 | 2.2421 | 0.6445 |
0.0087 | 437.5 | 1750 | 2.2472 | 0.6196 |
0.0052 | 500.0 | 2000 | 2.1142 | 0.6307 |
0.0033 | 562.5 | 2250 | 2.2207 | 0.6335 |
0.0029 | 625.0 | 2500 | 2.1788 | 0.6238 |
Framework versions
- Transformers 4.29.2
- Pytorch 2.0.1+cu118
- Datasets 1.18.3
- Tokenizers 0.13.3