CheeLi03's picture
Upload tokenizer
a485ae6 verified
metadata
base_model: openai/whisper-base
datasets:
  - fleurs
language:
  - pl
library_name: transformers
license: apache-2.0
metrics:
  - wer
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
model-index:
  - name: Whisper Base Polish Punctuation 5k - Chee Li
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: Google Fleurs
          type: fleurs
          config: pl_pl
          split: None
          args: 'config: pl split: test'
        metrics:
          - type: wer
            value: 36.63131454088752
            name: Wer

Whisper Base Polish Punctuation 5k - Chee Li

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

  • Loss: 0.6252
  • Wer: 36.6313

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: Use 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: 4000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.0545 5.0251 1000 0.5089 36.1010
0.0052 10.0503 2000 0.5784 35.7731
0.0027 15.0754 3000 0.6118 36.4359
0.0021 20.1005 4000 0.6252 36.6313

Framework versions

  • Transformers 4.46.2
  • Pytorch 2.3.1+cu121
  • Datasets 2.20.0
  • Tokenizers 0.20.3