README.md

metadata

language:
  - ru
license: apache-2.0
base_model: openai/whisper-base
tags:
  - generated_from_trainer
datasets:
  - aangry-mouse/stepik_ml_ru_2
metrics:
  - wer
model-index:
  - name: Whisper Base Ml Ru
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: ML датасет
          type: aangry-mouse/stepik_ml_ru_2
          args: 'config: ru, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 33.821550154382074

Whisper Base Ml Ru

This model is a fine-tuned version of openai/whisper-base on the ML датасет dataset. It achieves the following results on the evaluation set:

• Loss: 0.4592
• Wer: 33.8216

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: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6497	0.6649	250	0.6474	40.6539
0.4388	1.3298	500	0.5218	37.7009
0.4485	1.9947	750	0.4651	37.5030
0.296	2.6596	1000	0.4592	33.8216

Framework versions

• Transformers 4.41.0
• Pytorch 2.0.1+cu118
• Datasets 2.19.1
• Tokenizers 0.19.1