metadata
library_name: transformers
language:
- es
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- ylacombe/google-chilean-spanish
metrics:
- wer
model-index:
- name: Whisper Small ES-CL - Roberto Castro-Vexler - wintrained
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: OpenSLR Chilean Spanish
type: ylacombe/google-chilean-spanish
args: 'config: es-cl, split: test'
metrics:
- name: Wer
type: wer
value: 4.804410606458388
Whisper Small ES-CL - Roberto Castro-Vexler - wintrained
This model is a fine-tuned version of openai/whisper-small on the OpenSLR Chilean Spanish dataset. It achieves the following results on the evaluation set:
- Loss: 0.1276
- Wer: 4.8044
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: 1e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0009 | 8.6207 | 1000 | 0.1144 | 5.1588 |
| 0.0002 | 17.2414 | 2000 | 0.1197 | 4.8569 |
| 0.0001 | 25.8621 | 3000 | 0.1254 | 4.8307 |
| 0.0001 | 34.4828 | 4000 | 0.1276 | 4.8044 |
Framework versions
- Transformers 4.48.3
- Pytorch 2.9.1+cpu
- Datasets 3.6.0
- Tokenizers 0.21.4