metadata
library_name: transformers
license: apache-2.0
base_model: openai/whisper-tiny
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-tiny-akan
results: []
whisper-tiny-akan
This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.0839
- Wer: 47.4674
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: 0.0001
- 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: 500
- training_steps: 2000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.4022 | 5.0 | 250 | 0.7692 | 58.4660 |
| 0.1035 | 10.0 | 500 | 0.8971 | 57.0912 |
| 0.0421 | 15.0 | 750 | 0.9763 | 51.6064 |
| 0.0213 | 20.0 | 1000 | 1.0427 | 50.8828 |
| 0.0091 | 25.0 | 1250 | 1.0521 | 50.2026 |
| 0.0018 | 30.0 | 1500 | 1.0755 | 48.0897 |
| 0.0004 | 35.0 | 1750 | 1.0794 | 47.9305 |
| 0.0004 | 40.0 | 2000 | 1.0839 | 47.4674 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.0+cu121
- Datasets 3.0.0
- Tokenizers 0.19.1