README.md

metadata

library_name: transformers
language:
  - ba
license: apache-2.0
base_model: openai/whisper-small
tags:
  - generated_from_trainer
datasets:
  - stdbug/common-voice-17-ba
metrics:
  - wer
model-index:
  - name: Whisper Small Bashkir
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: Common Voice 17.0 (ba)
          type: stdbug/common-voice-17-ba
          args: 'config: ba, split: test'
        metrics:
          - type: wer
            value: 24.55759888965996
            name: Wer

Whisper Small Bashkir

This model is a fine-tuned version of openai/whisper-small on the Common Voice 17.0 (ba) dataset. It achieves the following results on the evaluation set:

• Loss: 0.1616
• Wer: 24.5576

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 19576
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0861	0.9999	19576	0.1616	24.5576

Framework versions

• Transformers 4.48.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0