metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: results
results: []
results
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: 0.0159
- Wer: 0.4194
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: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.456 | 2.8571 | 500 | 2.8521 | 1.0 |
| 1.7436 | 5.7143 | 1000 | 0.0950 | 0.5740 |
| 0.0812 | 8.5714 | 1500 | 0.0351 | 0.4816 |
| 0.0377 | 11.4286 | 2000 | 0.0277 | 0.4739 |
| 0.024 | 14.2857 | 2500 | 0.0198 | 0.4397 |
| 0.0181 | 17.1429 | 3000 | 0.0169 | 0.4276 |
| 0.0135 | 20.0 | 3500 | 0.0189 | 0.4276 |
| 0.0102 | 22.8571 | 4000 | 0.0176 | 0.4238 |
| 0.0086 | 25.7143 | 4500 | 0.0153 | 0.4221 |
| 0.0074 | 28.5714 | 5000 | 0.0159 | 0.4194 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.3.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1