metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: w2v2-base-pretrained_lr5e-5_at0.8_da0.6
results: []
w2v2-base-pretrained_lr5e-5_at0.8_da0.6
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: 2.3139
- Wer: 0.1773
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: 32
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 22.3022 | 8.93 | 250 | 14.1336 | 1.0 |
| 7.1613 | 17.86 | 500 | 12.5928 | 1.0 |
| 4.3974 | 26.79 | 750 | 12.3385 | 1.0 |
| 4.139 | 35.71 | 1000 | 9.3127 | 1.0 |
| 3.3971 | 44.64 | 1250 | 3.0328 | 1.0 |
| 1.9421 | 53.57 | 1500 | 1.4361 | 0.7381 |
| 0.2784 | 62.5 | 1750 | 1.4964 | 0.2896 |
| 0.1222 | 71.43 | 2000 | 1.7459 | 0.2559 |
| 0.0769 | 80.36 | 2250 | 2.0649 | 0.2255 |
| 0.0559 | 89.29 | 2500 | 2.2779 | 0.1892 |
| 0.0432 | 98.21 | 2750 | 2.2997 | 0.1901 |
| 0.036 | 107.14 | 3000 | 2.2112 | 0.1871 |
| 0.0309 | 116.07 | 3250 | 2.3675 | 0.1867 |
| 0.028 | 125.0 | 3500 | 2.3204 | 0.1820 |
| 0.0252 | 133.93 | 3750 | 2.3129 | 0.1803 |
| 0.0238 | 142.86 | 4000 | 2.3139 | 0.1773 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.0.0
- Datasets 2.14.6
- Tokenizers 0.14.1