metadata
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
- generated_from_trainer
datasets:
- p6_moderate
metrics:
- wer
model-index:
- name: p6_moderate
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: p6_moderate
type: p6_moderate
config: cs
split: test
args: cs
metrics:
- name: Wer
type: wer
value: 0.2500697155605131
p6_moderate
This model is a fine-tuned version of openai/whisper-large-v3 on the p6_moderate dataset. It achieves the following results on the evaluation set:
- Loss: 0.8973
- Wer: 0.2501
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: 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
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0015 | 27.78 | 1000 | 0.7347 | 0.2520 |
| 0.0008 | 55.56 | 2000 | 0.7309 | 0.2523 |
| 0.0001 | 83.33 | 3000 | 0.8548 | 0.2531 |
| 0.0 | 111.11 | 4000 | 0.8869 | 0.2503 |
| 0.0 | 138.89 | 5000 | 0.8973 | 0.2501 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.1.2+cu121
- Datasets 2.16.1
- Tokenizers 0.15.1