<!-- 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: 2.2362
- Wer: 0.5975
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 |
---|---|---|---|---|
5.0838 | 62.5 | 250 | 2.8997 | 1.0 |
0.8742 | 125.0 | 500 | 1.4743 | 0.6459 |
0.0778 | 187.5 | 750 | 1.7923 | 0.6473 |
0.0335 | 250.0 | 1000 | 1.7897 | 0.6183 |
0.0198 | 312.5 | 1250 | 1.9338 | 0.6266 |
0.0127 | 375.0 | 1500 | 2.2423 | 0.6279 |
0.0103 | 437.5 | 1750 | 2.2283 | 0.6252 |
0.0067 | 500.0 | 2000 | 2.0829 | 0.6349 |
0.0037 | 562.5 | 2250 | 2.2445 | 0.6169 |
0.0025 | 625.0 | 2500 | 2.2362 | 0.5975 |
Framework versions
- Transformers 4.29.2
- Pytorch 2.0.1+cu118
- Datasets 1.18.3
- Tokenizers 0.13.3