README.md

metadata

language:
  - ru
license: apache-2.0
base_model: openai/whisper-tiny
tags:
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_14_0
metrics:
  - wer
model-index:
  - name: Whisper Tiny Ru
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 14.0
          type: mozilla-foundation/common_voice_14_0
          config: ru
          split: None
          args: 'config: ru, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 188.79573968349942

Whisper Tiny Ru

This model is a fine-tuned version of openai/whisper-tiny on the Common Voice 14.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4739
• Wer: 188.7957

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
0.464	0.61	1000	0.5444	201.0197
0.3774	1.22	2000	0.5003	180.8949
0.3566	1.82	3000	0.4796	195.6722
0.2962	2.43	4000	0.4739	188.7957

Framework versions

• Transformers 4.37.2
• Pytorch 2.2.2
• Datasets 2.18.0
• Tokenizers 0.15.1