metadata
language:
- hr
license: apache-2.0
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- google/fleurs
metrics:
- wer
base_model: openai/whisper-small
model-index:
- name: melita1mu
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: common_voice_hr_fleurs
type: google/fleurs
config: hr_hr
split: test
args: 'config: hr, split: test'
metrics:
- type: wer
value: 45.596060228687875
name: Wer
melita1mu
This model is a fine-tuned version of openai/whisper-small on the common_voice_hr_fleurs dataset. It achieves the following results on the evaluation set:
- Loss: 0.5013
- Wer: 45.5961
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: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0204 | 4.17 | 1000 | 0.4216 | 36.3580 |
| 0.0017 | 8.33 | 2000 | 0.4697 | 37.7222 |
| 0.0008 | 12.5 | 3000 | 0.4922 | 39.6015 |
| 0.0006 | 16.67 | 4000 | 0.5013 | 45.5961 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1