whisper-small-hi / README.md
Taqwa's picture
update model card README.md
cc8941d
metadata
language:
  - hi
license: apache-2.0
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_11_0
metrics:
  - wer
model-index:
  - name: Whisper Small Hi - Sanchit Gandhi
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 11.0
          type: mozilla-foundation/common_voice_11_0
          args: 'config: hi, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 35.74028612545501

Whisper Small Hi - Sanchit Gandhi

This model is a fine-tuned version of Taqwa/whisper-small-hiTaqwa on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3353
  • Wer: 35.7403

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: 500
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.0762 0.31 125 0.2818 33.3573
0.0653 0.61 250 0.2930 33.9584
0.062 0.92 375 0.3060 34.7456
0.0518 1.22 500 0.3353 35.7403

Framework versions

  • Transformers 4.25.0.dev0
  • Pytorch 1.12.1+cu113
  • Datasets 2.7.1
  • Tokenizers 0.13.2