metadata
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
- generated_from_trainer
datasets:
- common_voice_13_0
metrics:
- wer
model-index:
- name: wav2vec2-large-xls-r-300m-bretonwelsh-colab
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: common_voice_13_0
type: common_voice_13_0
config: cy
split: test
args: cy
metrics:
- name: Wer
type: wer
value: 0.29761332022507164
wav2vec2-large-xls-r-300m-bretonwelsh-colab
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.4506
- Wer: 0.2976
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: 0.0004
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.9503 | 0.98 | 800 | 0.8330 | 0.7296 |
| 0.6531 | 1.95 | 1600 | 0.5592 | 0.5470 |
| 0.4637 | 2.93 | 2400 | 0.4711 | 0.4539 |
| 0.3449 | 3.91 | 3200 | 0.4484 | 0.4116 |
| 0.2694 | 4.88 | 4000 | 0.4313 | 0.3860 |
| 0.2087 | 5.86 | 4800 | 0.4115 | 0.3616 |
| 0.1649 | 6.84 | 5600 | 0.4105 | 0.3378 |
| 0.1313 | 7.81 | 6400 | 0.4409 | 0.3236 |
| 0.1079 | 8.79 | 7200 | 0.4402 | 0.3093 |
| 0.0897 | 9.77 | 8000 | 0.4506 | 0.2976 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.0.1+cu118
- Datasets 2.14.4
- Tokenizers 0.13.3