metadata
language:
- ml
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- malayalam
metrics:
- wer
model-index:
- name: Whisper Small ml
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: malayalam speech
type: malayalam
args: 'config: ml, split: test'
metrics:
- name: Wer
type: wer
value: 36.515601783060916
Whisper Small ml
This model is a fine-tuned version of openai/whisper-small on the malayalam speech dataset. It achieves the following results on the evaluation set:
- Loss: 0.0355
- Wer: 36.5156
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: 1000
- training_steps: 8000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0147 | 2.1716 | 2000 | 0.0248 | 43.0349 |
| 0.0051 | 4.3431 | 4000 | 0.0258 | 38.2522 |
| 0.0011 | 6.5147 | 6000 | 0.0317 | 37.6022 |
| 0.0009 | 8.6862 | 8000 | 0.0355 | 36.5156 |
Framework versions
- Transformers 4.42.4
- Pytorch 2.3.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1