whisper-medium-cs / README.md
sgangireddy's picture
Update README.md
f87b739
metadata
license: apache-2.0
tags:
  - generated_from_trainer
  - whisper-event
datasets:
  - mozilla-foundation/common_voice_11_0
metrics:
  - wer
model-index:
  - name: openai/whisper-medium
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: mozilla-foundation/common_voice_11_0
          type: mozilla-foundation/common_voice_11_0
          config: cs
          split: test
          args: cs
        metrics:
          - name: Wer
            type: wer
            value: 11.835877792305851

openai/whisper-medium

This model is a fine-tuned version of openai/whisper-medium on the common_voice_11_0 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1805
  • Wer: 11.8358

Model description

The Model is fine-tuned for 1000 steps/updates on CV11 Czech train+valiation data.

  • Zero-shot - 18.80 (CV9 test data, even on CV11 the WER is closer OR a bit higher than this)
  • Fine-tuned - 11.83 (CV11 test data)

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: 64
  • eval_batch_size: 32
  • 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
0.0076 4.06 1000 0.1805 11.8358

Framework versions

  • Transformers 4.26.0.dev0
  • Pytorch 1.13.0+cu117
  • Datasets 2.7.1.dev0
  • Tokenizers 0.13.2