metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
datasets:
- vivos
metrics:
- wer
model-index:
- name: wav2vec2-vivos-asr
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: vivos
type: vivos
config: default
split: None
args: default
metrics:
- name: Wer
type: wer
value: 0.46064565231179044
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: 0.8301
- Wer: 0.4606
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: linear
- lr_scheduler_warmup_steps: 400
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 5.7906 | 2.0 | 292 | 3.6543 | 1.0 |
| 3.4396 | 4.0 | 584 | 3.5033 | 1.0 |
| 3.4081 | 6.0 | 876 | 3.4360 | 1.0 |
| 2.4196 | 8.0 | 1168 | 1.5751 | 0.8002 |
| 1.3285 | 10.0 | 1460 | 1.1699 | 0.6628 |
| 1.0944 | 12.0 | 1752 | 1.0408 | 0.6051 |
| 0.9742 | 14.0 | 2044 | 0.9772 | 0.5657 |
| 0.9219 | 16.0 | 2336 | 0.9344 | 0.5515 |
| 0.817 | 18.0 | 2628 | 0.8871 | 0.5176 |
| 0.7636 | 20.0 | 2920 | 0.8734 | 0.5050 |
| 0.7192 | 22.0 | 3212 | 0.8556 | 0.4909 |
| 0.6904 | 24.0 | 3504 | 0.8471 | 0.4772 |
| 0.6703 | 26.0 | 3796 | 0.8489 | 0.4754 |
| 0.6343 | 28.0 | 4088 | 0.8364 | 0.4689 |
| 0.6161 | 30.0 | 4380 | 0.8301 | 0.4606 |
Framework versions
- Transformers 4.42.3
- Pytorch 2.1.2
- Datasets 2.20.0
- Tokenizers 0.19.1