metadata
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- fsicoli/cv16-fleurs
metrics:
- wer
model-index:
- name: whisper-small-pt-cv16-fleurs
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: fsicoli/cv16-fleurs default
type: fsicoli/cv16-fleurs
args: default
metrics:
- name: Wer
type: wer
value: 0.14422069232381535
whisper-small-pt-cv16-fleurs
This model is a fine-tuned version of openai/whisper-small on the fsicoli/cv16-fleurs default dataset. It achieves the following results on the evaluation set:
- Loss: 0.2154
- Wer: 0.1442
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-06
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- total_train_batch_size: 16
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 5000
- training_steps: 5000
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.3915 | 0.47 | 1000 | 0.2978 | 0.1858 |
| 0.2942 | 0.93 | 2000 | 0.2500 | 0.1626 |
| 0.2877 | 1.4 | 3000 | 0.2336 | 0.1536 |
| 0.2303 | 1.87 | 4000 | 0.2225 | 0.1482 |
| 0.2192 | 2.33 | 5000 | 0.2154 | 0.1442 |
Framework versions
- Transformers 4.39.0.dev0
- Pytorch 2.2.1
- Datasets 2.16.1
- Tokenizers 0.15.2