metadata
license: apache-2.0
tags:
- generated_from_trainer
base_model: facebook/wav2vec2-large-xlsr-53
datasets:
- common_voice_13_0
metrics:
- wer
model-index:
- name: wav2vec2-xlsr-53-CV-demo-google-colab-Ezra_William_Prod13
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: common_voice_13_0
type: common_voice_13_0
config: id
split: test
args: id
metrics:
- type: wer
value: 0.5518989675516224
name: Wer
wav2vec2-xlsr-53-CV-demo-google-colab-Ezra_William_Prod13
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.5399
- Wer: 0.5519
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: 3e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 9
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.7217 | 0.9 | 500 | 2.9517 | 1.0 |
| 2.9149 | 1.8 | 1000 | 2.8778 | 1.0 |
| 2.851 | 2.7 | 1500 | 2.6437 | 1.0 |
| 2.0653 | 3.6 | 2000 | 1.0367 | 0.8727 |
| 1.1893 | 4.5 | 2500 | 0.7226 | 0.7006 |
| 0.9685 | 5.4 | 3000 | 0.6301 | 0.6358 |
| 0.8742 | 6.29 | 3500 | 0.5778 | 0.5890 |
| 0.8076 | 7.19 | 4000 | 0.5576 | 0.5696 |
| 0.7624 | 8.09 | 4500 | 0.5412 | 0.5525 |
| 0.7604 | 8.99 | 5000 | 0.5399 | 0.5519 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2