README.md

metadata

license: apache-2.0
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: openai/whisper-medium
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: rishabhjain16/infer_myst
          type: rishabhjain16/infer_myst
          config: en
          split: test
        metrics:
          - type: wer
            value: 11.96
            name: WER
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: rishabhjain16/infer_pfs
          type: rishabhjain16/infer_pfs
          config: en
          split: test
        metrics:
          - type: wer
            value: 3.12
            name: WER
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: rishabhjain16/infer_cmu
          type: rishabhjain16/infer_cmu
          config: en
          split: test
        metrics:
          - type: wer
            value: 8.92
            name: WER
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: rishabhjain16/libritts_dev_clean
          type: rishabhjain16/libritts_dev_clean
          config: en
          split: test
        metrics:
          - type: wer
            value: 5.39
            name: WER
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: rishabhjain16/infer_pf_swedish
          type: rishabhjain16/infer_pf_swedish
          config: en
          split: test
        metrics:
          - type: wer
            value: 7.74
            name: WER
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: rishabhjain16/infer_pf_german
          type: rishabhjain16/infer_pf_german
          config: en
          split: test
        metrics:
          - type: wer
            value: 36.21
            name: WER
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: rishabhjain16/infer_pf_italian
          type: rishabhjain16/infer_pf_italian
          config: en
          split: test
        metrics:
          - type: wer
            value: 4.16
            name: WER
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: rishabhjain16/infer_so_chinese
          type: rishabhjain16/infer_so_chinese
          config: en
          split: test
        metrics:
          - type: wer
            value: 14.4
            name: WER

openai/whisper-medium

This model is a fine-tuned version of openai/whisper-medium on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3622
• Wer: 10.6091

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: 32
• 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5314	0.12	500	0.2762	13.2758
0.1619	1.09	1000	0.2541	11.6909
0.069	2.05	1500	0.2768	10.2892
0.0756	3.02	2000	0.2756	11.6142
0.0324	3.14	2500	0.2961	11.1800
0.0171	4.11	3000	0.3322	11.1689
0.0046	5.07	3500	0.3653	10.5858
0.0091	6.03	4000	0.3622	10.6091

Framework versions

• Transformers 4.29.0
• Pytorch 1.14.0a0+44dac51
• Datasets 2.12.0
• Tokenizers 0.13.3