metadata
library_name: transformers
language:
- hu
license: apache-2.0
base_model: openai/whisper-base
tags:
- generated_from_trainer
datasets:
- fleurs
metrics:
- wer
model-index:
- name: Whisper Base Hu 1944
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: sarpba/big_audio_data_hun_v2
type: fleurs
config: hu_hu
split: None
args: hu_hu
metrics:
- name: Wer
type: wer
value: 29.48142356294297
Whisper Base Hu 1944
This model is a fine-tuned version of openai/whisper-base on the sarpba/big_audio_data_hun_v2 dataset. It achieves the following results on the evaluation set:
- Loss: 0.7999
- Wer Ortho: 33.8788
- Wer: 29.4814
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: 0.0003
- train_batch_size: 64
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- training_steps: 8000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|---|---|---|---|---|---|
| 0.2523 | 0.3770 | 1000 | 0.9703 | 50.8988 | 46.7185 |
| 0.1859 | 0.7539 | 2000 | 0.8605 | 43.4345 | 39.4103 |
| 0.127 | 1.1309 | 3000 | 0.8378 | 40.6107 | 36.0040 |
| 0.1226 | 1.5079 | 4000 | 0.8153 | 38.9189 | 34.1842 |
| 0.1105 | 1.8848 | 5000 | 0.7847 | 36.6018 | 32.1979 |
| 0.0659 | 2.2618 | 6000 | 0.8298 | 35.3752 | 30.6379 |
| 0.0594 | 2.6388 | 7000 | 0.8132 | 34.8255 | 30.2280 |
| 0.0316 | 3.0157 | 8000 | 0.7999 | 33.8788 | 29.4814 |
Framework versions
- Transformers 4.45.2
- Pytorch 2.3.0+cu121
- Datasets 3.0.1
- Tokenizers 0.20.1