metadata
base_model: facebook/wav2vec2-base
datasets:
- vivos
license: apache-2.0
metrics:
- wer
tags:
- generated_from_trainer
model-index:
- name: wav2vec2-vivos
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: 0.23636599442318915
name: Wer
wav2vec2-vivos
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: 0.4755
- Wer: 0.2364
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: 0.0002
- train_batch_size: 32
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.2
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.349 | 2.0 | 146 | 0.9904 | 0.6088 |
| 0.718 | 4.0 | 292 | 0.6959 | 0.4630 |
| 0.4692 | 6.0 | 438 | 0.5304 | 0.3414 |
| 0.3385 | 8.0 | 584 | 0.5078 | 0.3216 |
| 0.2627 | 10.0 | 730 | 0.4659 | 0.2788 |
| 0.2033 | 12.0 | 876 | 0.4751 | 0.2656 |
| 0.1699 | 14.0 | 1022 | 0.4659 | 0.2519 |
| 0.1688 | 16.0 | 1168 | 0.4662 | 0.2394 |
| 0.1269 | 18.0 | 1314 | 0.4707 | 0.2375 |
| 0.1162 | 20.0 | 1460 | 0.4755 | 0.2364 |
Framework versions
- Transformers 4.44.0
- Pytorch 2.4.0
- Datasets 2.21.0
- Tokenizers 0.19.1