metadata
language:
- ru
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- tbkazakova/even_speech_biblical
metrics:
- wer
model-index:
- name: Whisper Small Even - Sanchit Gandhi
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Even Speech Biblical
type: tbkazakova/even_speech_biblical
config: default
split: None
args: 'config: ru, split: train'
metrics:
- name: Wer
type: wer
value: 53.88015717092338
Whisper Small Even - Sanchit Gandhi
This model is a fine-tuned version of openai/whisper-small on the Even Speech Biblical dataset. It achieves the following results on the evaluation set:
- Loss: 0.4591
- Wer: 53.8802
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 300
- training_steps: 2000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.05 | 5.9880 | 500 | 0.3920 | 63.7525 |
| 0.0022 | 11.9760 | 1000 | 0.4307 | 57.3674 |
| 0.0003 | 17.9641 | 1500 | 0.4528 | 51.7682 |
| 0.0003 | 23.9521 | 2000 | 0.4591 | 53.8802 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1