Wav2vec2-xls-r-300m
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the KELONMYOSA/dusha_emotion_audio dataset. It achieves the following results on the evaluation set:
- Loss: 0.5633
- Accuracy: 0.7970
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: 0.003
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.7868 | 1.0 | 24170 | 0.7561 | 0.7318 |
| 0.7147 | 2.0 | 48340 | 0.6984 | 0.7459 |
| 0.669 | 3.0 | 72510 | 0.6263 | 0.7727 |
| 0.6362 | 4.0 | 96680 | 0.5832 | 0.7902 |
| 0.4476 | 5.0 | 120850 | 0.5633 | 0.7970 |
Framework versions
- Transformers 4.29.2
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3