metadata
license: apache-2.0
tags:
- generated_from_trainer
base_model: openai/whisper-large-v2
model-index:
- name: whisper-large-dog
results: []
whisper-large-dog
This model is a fine-tuned version of openai/whisper-large-v2 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.2858
- Cer: 8.8124
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: 4e-06
- 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: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.2143 | 1.38 | 500 | 0.2300 | 9.4167 |
| 0.0891 | 2.76 | 1000 | 0.2233 | 9.1356 |
| 0.032 | 4.14 | 1500 | 0.2425 | 8.7702 |
| 0.0132 | 5.52 | 2000 | 0.2495 | 8.7702 |
| 0.0074 | 6.91 | 2500 | 0.2674 | 8.9670 |
| 0.0016 | 8.29 | 3000 | 0.2781 | 8.6718 |
| 0.0013 | 9.67 | 3500 | 0.2837 | 8.7280 |
| 0.0016 | 11.05 | 4000 | 0.2858 | 8.8124 |
Framework versions
- Transformers 4.38.2
- Pytorch 2.0.0
- Datasets 2.18.0
- Tokenizers 0.15.2