metadata
language:
- ko
license: apache-2.0
base_model: openai/whisper-small
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- arrow
metrics:
- wer
model-index:
- name: whisper-kor3_de_3
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: whisper-kor3_de_3
type: arrow
config: default
split: train
args: 'config: ko, split: valid'
metrics:
- name: Wer
type: wer
value: 23.377308707124012
whisper-kor3_de_3
This model is a fine-tuned version of openai/whisper-small on the whisper-kor3_de_3 dataset. It achieves the following results on the evaluation set:
- Loss: 0.3579
- Wer: 23.3773
- Cer: 10.7523
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: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- training_steps: 400
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.1425 | 0.21 | 50 | 0.3326 | 24.2480 | 11.1910 |
| 0.1145 | 0.42 | 100 | 0.3466 | 23.6148 | 11.5938 |
| 0.1082 | 0.64 | 150 | 0.3518 | 29.3668 | 15.6717 |
| 0.0986 | 0.85 | 200 | 0.3485 | 24.3008 | 11.0256 |
| 0.0771 | 1.06 | 250 | 0.3536 | 23.5620 | 10.8314 |
| 0.0513 | 1.27 | 300 | 0.3576 | 23.3509 | 10.5437 |
| 0.0424 | 1.48 | 350 | 0.3587 | 23.6148 | 10.7811 |
| 0.0526 | 1.69 | 400 | 0.3579 | 23.3773 | 10.7523 |
Framework versions
- Transformers 4.33.2
- Pytorch 2.0.1+cu117
- Datasets 2.14.5
- Tokenizers 0.13.3