metadata
library_name: transformers
language:
- en
license: apache-2.0
base_model: openai/whisper-tiny
tags:
- generated_from_trainer
datasets:
- FreeSound
metrics:
- wer
model-index:
- name: Whisper Tiny En - FreeSound based captions
results: []
Whisper Tiny En - FreeSound based captions
This model is a fine-tuned version of openai/whisper-tiny on the FreeSound Audio dataset. It achieves the following results on the evaluation set:
- Loss: 5.5085
- Wer: 91.7867
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: 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: 7000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.8757 | 24.3902 | 1000 | 4.1235 | 97.8963 |
| 0.0518 | 48.7805 | 2000 | 4.8741 | 94.9280 |
| 0.0234 | 73.1707 | 3000 | 5.1544 | 93.1124 |
| 0.0148 | 97.5610 | 4000 | 5.3503 | 93.4294 |
| 0.0141 | 121.9512 | 5000 | 5.4099 | 92.3631 |
| 0.0112 | 146.3415 | 6000 | 5.4837 | 92.4496 |
| 0.0104 | 170.7317 | 7000 | 5.5085 | 91.7867 |
Framework versions
- Transformers 4.45.2
- Pytorch 2.1.0+cu118
- Datasets 3.0.1
- Tokenizers 0.20.1