metadata
language:
- pt
license: apache-2.0
base_model: openai/whisper-base
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- mozilla-foundation/common_voice_16_0
metrics:
- wer
model-index:
- name: Whisper Base using Common Voice 16 (pt)
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Mozilla Common Voices - 16.0 - Portuguese
type: mozilla-foundation/common_voice_16_0
config: pt
split: test[0:5400]
args: pt
metrics:
- name: Wer
type: wer
value: 25.542580301884676
Whisper Base using Common Voice 16 (pt)
This model is a fine-tuned version of openai/whisper-base on the Mozilla Common Voices - 16.0 - Portuguese dataset. It achieves the following results on the evaluation set:
- Loss: 0.3952
- Wer: 25.5426
- Wer Normalized: 19.7098
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 400
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Wer Normalized |
|---|---|---|---|---|---|
| 0.5344 | 0.37 | 500 | 0.5264 | 35.9965 | 30.1234 |
| 0.438 | 0.74 | 1000 | 0.4904 | 33.4453 | 28.3776 |
| 0.1871 | 1.11 | 1500 | 0.4595 | 30.3929 | 24.5163 |
| 0.1955 | 1.48 | 2000 | 0.4342 | 28.6566 | 22.9762 |
| 0.1754 | 1.85 | 2500 | 0.4199 | 28.2674 | 22.4147 |
| 0.0649 | 2.22 | 3000 | 0.4090 | 26.7860 | 20.7689 |
| 0.0595 | 2.59 | 3500 | 0.4026 | 26.1839 | 20.2018 |
| 0.0626 | 2.96 | 4000 | 0.3952 | 25.5426 | 19.7098 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.1
- Datasets 2.16.1
- Tokenizers 0.15.0