metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
datasets:
- vivos
metrics:
- wer
model-index:
- name: working
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: 0.5176762726262825
working
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: 0.9086
- Wer: 0.5177
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: 500
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 5.7609 | 2.0 | 292 | 3.5829 | 1.0 |
| 3.2939 | 4.0 | 584 | 2.5400 | 0.9850 |
| 1.6184 | 6.0 | 876 | 1.2841 | 0.7316 |
| 1.1576 | 8.0 | 1168 | 1.1273 | 0.6652 |
| 0.9843 | 10.0 | 1460 | 1.0547 | 0.6139 |
| 0.885 | 12.0 | 1752 | 0.9854 | 0.5722 |
| 0.8103 | 14.0 | 2044 | 0.9524 | 0.5594 |
| 0.7457 | 16.0 | 2336 | 0.9294 | 0.5336 |
| 0.6929 | 18.0 | 2628 | 0.9186 | 0.5253 |
| 0.6589 | 20.0 | 2920 | 0.9086 | 0.5177 |
Framework versions
- Transformers 4.42.3
- Pytorch 2.1.2
- Datasets 2.20.0
- Tokenizers 0.19.1