metadata
library_name: transformers
language:
- pmx
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- iitd-duk/paula
metrics:
- wer
model-index:
- name: Whisper-Small-paula
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Paula
type: iitd-duk/paula
metrics:
- name: Wer
type: wer
value: 97.75910364145658
Whisper-Small-paula
This model is a fine-tuned version of openai/whisper-small on the Paula dataset. It achieves the following results on the evaluation set:
- Loss: 2.7488
- Wer: 97.7591
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: 12
- eval_batch_size: 8
- seed: 42
- 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: 100
- training_steps: 500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.1 | 5.0 | 100 | 2.4980 | 131.0924 |
| 0.0249 | 10.0 | 200 | 2.5853 | 97.7591 |
| 0.0065 | 15.0 | 300 | 2.6842 | 98.0392 |
| 0.0026 | 20.0 | 400 | 2.7265 | 96.9188 |
| 0.0016 | 25.0 | 500 | 2.7488 | 97.7591 |
Framework versions
- Transformers 4.46.0.dev0
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.1