README.md

metadata

language:
  - kr
license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - Jungwonchang/ksponspeech_partial
metrics:
  - wer
base_model: openai/whisper-large-v2
model-index:
  - name: Whisper large-v2, KsponSpeech Partial 10 epochs
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: KsponSpeech
          type: Jungwonchang/ksponspeech_partial
          config: eval
          split: test
          args: eval
        metrics:
          - type: wer
            value: 25.714073744343054
            name: Wer
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: Jungwonchang/ksponspeech
          type: Jungwonchang/ksponspeech
          config: eval
          split: test
        metrics:
          - type: wer
            value: 25.54
            name: WER

Whisper large-v2, KsponSpeech Partial 10 epochs

This model is a fine-tuned version of openai/whisper-large-v2 on the KsponSpeech dataset. It achieves the following results on the evaluation set:

• Loss: 0.0194
• Wer: 25.7141

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: 50
• training_steps: 300

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2225	1.15	100	0.1394	27.9769
0.0507	3.11	200	0.0449	14.9640
0.0114	5.07	300	0.0194	25.7141

Framework versions

• Transformers 4.31.0
• Pytorch 1.12.1+cu116
• Datasets 2.14.0
• Tokenizers 0.12.1