metadata
base_model: facebook/wav2vec2-base
datasets:
- vivos
license: apache-2.0
metrics:
- wer
tags:
- generated_from_trainer
model-index:
- name: wav2vec2-vivos-asr
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: vivos
type: vivos
config: default
split: None
args: default
metrics:
- type: wer
value: 0.3726759841005257
name: Wer
wav2vec2-vivos-asr
This model is a fine-tuned version of facebook/wav2vec2-base on the vivos dataset. It achieves the following results on the evaluation set:
- Loss: 0.7912
- Wer: 0.3727
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
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 400
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 5.8168 | 2.0 | 292 | 3.6240 | 1.0 |
| 3.4344 | 4.0 | 584 | 3.4785 | 1.0 |
| 3.0271 | 6.0 | 876 | 1.8947 | 0.9142 |
| 1.2453 | 8.0 | 1168 | 1.0293 | 0.6091 |
| 0.7876 | 10.0 | 1460 | 0.8472 | 0.5229 |
| 0.6062 | 12.0 | 1752 | 0.7675 | 0.4748 |
| 0.4929 | 14.0 | 2044 | 0.7494 | 0.4303 |
| 0.4376 | 16.0 | 2336 | 0.7481 | 0.4063 |
| 0.3523 | 18.0 | 2628 | 0.7580 | 0.4007 |
| 0.309 | 20.0 | 2920 | 0.7676 | 0.3851 |
| 0.2694 | 22.0 | 3212 | 0.7631 | 0.3819 |
| 0.2531 | 24.0 | 3504 | 0.7717 | 0.3761 |
| 0.2472 | 26.0 | 3796 | 0.7825 | 0.3710 |
| 0.2223 | 28.0 | 4088 | 0.7905 | 0.3732 |
| 0.2183 | 30.0 | 4380 | 0.7912 | 0.3727 |
Framework versions
- Transformers 4.42.3
- Pytorch 2.1.2
- Datasets 2.20.0
- Tokenizers 0.19.1