metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: w2v2-base-pretrained_lr5e-5_at0.4_da1
results: []
w2v2-base-pretrained_lr5e-5_at0.4_da1
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: 1.4243
- Wer: 0.2593
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: 500
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 17.3727 | 4.03 | 250 | 8.4434 | 1.0 |
| 4.5158 | 8.06 | 500 | 5.6577 | 1.0 |
| 3.3968 | 12.1 | 750 | 5.5711 | 1.0 |
| 3.4087 | 16.13 | 1000 | 4.7248 | 1.0 |
| 3.2781 | 20.16 | 1250 | 4.1466 | 1.0004 |
| 3.0895 | 24.19 | 1500 | 4.0231 | 1.0004 |
| 2.7559 | 28.23 | 1750 | 2.6579 | 0.9735 |
| 1.8327 | 32.26 | 2000 | 1.5957 | 0.8979 |
| 0.7991 | 36.29 | 2250 | 1.1338 | 0.5656 |
| 0.4226 | 40.32 | 2500 | 1.1239 | 0.4088 |
| 0.2746 | 44.35 | 2750 | 1.2772 | 0.3430 |
| 0.1934 | 48.39 | 3000 | 1.2697 | 0.3187 |
| 0.1437 | 52.42 | 3250 | 1.3526 | 0.3033 |
| 0.1151 | 56.45 | 3500 | 1.3560 | 0.2777 |
| 0.0975 | 60.48 | 3750 | 1.3470 | 0.2606 |
| 0.0885 | 64.52 | 4000 | 1.4243 | 0.2593 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.0.0
- Datasets 2.14.6
- Tokenizers 0.14.1