metadata
license: mit
tags:
- generated_from_trainer
metrics:
- wer
base_model: facebook/w2v-bert-2.0
model-index:
- name: w2v-bert-2.0-nonstudio_and_studioRecords
results: []
w2v-bert-2.0-nonstudio_and_studioRecords
This model is a fine-tuned version of facebook/w2v-bert-2.0 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1642
- Wer: 0.1271
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.0788 | 0.46 | 600 | 0.3546 | 0.4640 |
| 0.1732 | 0.92 | 1200 | 0.2307 | 0.3431 |
| 0.1247 | 1.38 | 1800 | 0.2097 | 0.2916 |
| 0.1092 | 1.84 | 2400 | 0.2121 | 0.2742 |
| 0.0896 | 2.3 | 3000 | 0.1817 | 0.2565 |
| 0.0776 | 2.76 | 3600 | 0.1715 | 0.2249 |
| 0.0695 | 3.22 | 4200 | 0.1684 | 0.2142 |
| 0.0581 | 3.68 | 4800 | 0.1667 | 0.2050 |
| 0.0521 | 4.14 | 5400 | 0.1629 | 0.1876 |
| 0.0426 | 4.6 | 6000 | 0.1553 | 0.1819 |
| 0.0389 | 5.06 | 6600 | 0.1485 | 0.1692 |
| 0.03 | 5.52 | 7200 | 0.1388 | 0.1667 |
| 0.03 | 5.98 | 7800 | 0.1441 | 0.1607 |
| 0.0208 | 6.44 | 8400 | 0.1444 | 0.1520 |
| 0.0209 | 6.9 | 9000 | 0.1339 | 0.1498 |
| 0.0145 | 7.36 | 9600 | 0.1418 | 0.1403 |
| 0.013 | 7.82 | 10200 | 0.1387 | 0.1351 |
| 0.0103 | 8.28 | 10800 | 0.1504 | 0.1321 |
| 0.0079 | 8.74 | 11400 | 0.1593 | 0.1331 |
| 0.0066 | 9.2 | 12000 | 0.1551 | 0.1222 |
| 0.004 | 9.66 | 12600 | 0.1642 | 0.1271 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.1.1+cu121
- Datasets 2.16.1
- Tokenizers 0.15.1