README.md

metadata

language:
  - ru
license: apache-2.0
base_model: openai/whisper-small
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
datasets:
  - fleurs
metrics:
  - wer
model-index:
  - name: Whisper Small ru - Chee Li
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Google Fleurs
          type: fleurs
          config: ru_ru
          split: None
          args: 'config: ru split: test'
        metrics:
          - name: Wer
            type: wer
            value: 50.354088722608225

Whisper Small ru - Chee Li

This model is a fine-tuned version of openai/whisper-small on the Google Fleurs dataset. It achieves the following results on the evaluation set:

• Loss: 0.2500
• Wer: 50.3541

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
0.0049	5.4645	1000	0.2170	29.2090
0.0013	10.9290	2000	0.2340	43.3993
0.0006	16.3934	3000	0.2457	49.9800
0.0004	21.8579	4000	0.2500	50.3541

Framework versions

• Transformers 4.44.0
• Pytorch 2.3.1+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1