metadata
library_name: transformers
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-1b
tags:
- generated_from_trainer
model-index:
- name: wav2vec2-1b-E50_freq_pause
results: []
wav2vec2-1b-E50_freq_pause
This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.8597
- Cer: 24.3832
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: 2
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 8
- 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: 50
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 15.9037 | 0.2580 | 200 | 5.1166 | 92.5458 |
| 11.4762 | 0.5160 | 400 | 11.0693 | 93.5092 |
| 5.6392 | 0.7741 | 600 | 4.8196 | 92.3990 |
| 4.5741 | 1.0321 | 800 | 4.5280 | 92.4988 |
| 4.5003 | 1.2901 | 1000 | 4.4228 | 91.0421 |
| 4.4138 | 1.5481 | 1200 | 4.3815 | 91.8233 |
| 4.3121 | 1.8062 | 1400 | 4.2013 | 81.1560 |
| 4.2003 | 2.0642 | 1600 | 4.5168 | 89.6617 |
| 4.0984 | 2.3222 | 1800 | 4.3367 | 87.5059 |
| 4.0011 | 2.5802 | 2000 | 4.0720 | 77.6668 |
| 3.8311 | 2.8383 | 2200 | 4.1406 | 80.9269 |
| 3.425 | 3.0963 | 2400 | 3.0046 | 62.9347 |
| 2.1444 | 3.3543 | 2600 | 1.8194 | 41.1948 |
| 1.4272 | 3.6123 | 2800 | 1.4620 | 36.2371 |
| 1.1442 | 3.8703 | 3000 | 1.1785 | 30.7272 |
| 0.9262 | 4.1284 | 3200 | 1.1745 | 32.0841 |
| 0.7623 | 4.3864 | 3400 | 0.9678 | 27.3026 |
| 0.7033 | 4.6444 | 3600 | 0.9814 | 27.2968 |
| 0.6478 | 4.9024 | 3800 | 0.8597 | 24.3832 |
Framework versions
- Transformers 4.45.2
- Pytorch 2.3.1.post100
- Datasets 2.19.1
- Tokenizers 0.20.1