metadata
library_name: transformers
language:
- hi
license: apache-2.0
base_model: openai/whisper-small
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- aihub_adult_baseline
model-index:
- name: whisper-small-Yfreq_speed_pause
results: []
whisper-small-Yfreq_speed_pause
This model is a fine-tuned version of openai/whisper-small on the aihub old50 adult freq speed pause changed dataset. It achieves the following results on the evaluation set:
- Loss: 0.1747
- Cer: 4.9107
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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.4171 | 0.1289 | 100 | 0.2695 | 6.4497 |
| 0.2579 | 0.2579 | 200 | 0.2163 | 5.8094 |
| 0.2428 | 0.3868 | 300 | 0.2107 | 5.8329 |
| 0.2151 | 0.5158 | 400 | 0.2013 | 5.6391 |
| 0.1904 | 0.6447 | 500 | 0.1946 | 5.4805 |
| 0.1687 | 0.7737 | 600 | 0.1862 | 5.0576 |
| 0.1739 | 0.9026 | 700 | 0.1814 | 5.3102 |
| 0.0887 | 1.0316 | 800 | 0.1809 | 4.9988 |
| 0.0749 | 1.1605 | 900 | 0.1834 | 5.0341 |
| 0.078 | 1.2895 | 1000 | 0.1778 | 4.9988 |
| 0.0754 | 1.4184 | 1100 | 0.1786 | 5.1457 |
| 0.066 | 1.5474 | 1200 | 0.1748 | 5.0576 |
| 0.0662 | 1.6763 | 1300 | 0.1749 | 4.8872 |
| 0.0604 | 1.8053 | 1400 | 0.1747 | 4.9636 |
| 0.0626 | 1.9342 | 1500 | 0.1747 | 4.9107 |
Framework versions
- Transformers 4.46.0.dev0
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.1