metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: w2v2-base-pretrained_lr1e-4_at0.8_da0.1
results: []
w2v2-base-pretrained_lr1e-4_at0.8_da0.1
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: 3.0817
- Wer: 1.0
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: 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: 100
- num_epochs: 200
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 37.6559 | 20.0 | 100 | 4.2522 | 1.0 |
| 3.3119 | 40.0 | 200 | 3.2132 | 1.0 |
| 3.0946 | 60.0 | 300 | 3.1024 | 1.0 |
| 3.0574 | 80.0 | 400 | 3.1054 | 1.0 |
| 3.0388 | 100.0 | 500 | 3.0969 | 1.0 |
| 3.0221 | 120.0 | 600 | 3.0894 | 1.0 |
| 3.0007 | 140.0 | 700 | 3.0842 | 1.0 |
| 2.9741 | 160.0 | 800 | 3.0784 | 1.0 |
| 2.9538 | 180.0 | 900 | 3.0814 | 1.0 |
| 2.9384 | 200.0 | 1000 | 3.0817 | 1.0 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.0.0
- Datasets 2.14.6
- Tokenizers 0.14.1