wav2vec2_prosodic_minimal
This model is a fine-tuned version of facebook/wav2vec2-base on the MatsRooth/prosodic_minimal dataset. It achieves the following results on the evaluation set:
- Loss: 0.1205
- Accuracy: 0.9802
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: 3e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 0
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.3769 | 1.0 | 2151 | 0.2877 | 0.9318 |
| 0.4347 | 2.0 | 4302 | 0.1720 | 0.9608 |
| 0.3185 | 3.0 | 6453 | 0.1805 | 0.9625 |
| 0.3362 | 4.0 | 8604 | 0.1409 | 0.9720 |
| 0.2105 | 5.0 | 10755 | 0.1173 | 0.9753 |
| 0.2106 | 6.0 | 12906 | 0.1460 | 0.9733 |
| 0.1819 | 7.0 | 15057 | 0.1431 | 0.9743 |
| 0.2946 | 8.0 | 17208 | 0.1245 | 0.9766 |
| 0.0416 | 9.0 | 19359 | 0.1241 | 0.9786 |
| 0.1006 | 10.0 | 21510 | 0.1205 | 0.9802 |
Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.9.0+cu128
- Datasets 2.13.1
- Tokenizers 0.15.0