README.md

metadata

language:
  - hi
license: apache-2.0
base_model: openai/whisper-base
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_11_0
metrics:
  - wer
model-index:
  - name: Whisper small withaq - T5SA
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 11.0 - Arabic
          type: mozilla-foundation/common_voice_11_0
          args: 'config: hi, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 64.01799100449776

Whisper small withaq - T5SA

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

• Loss: 1.6060
• Wer: 64.0180

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: 6
• 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0051	22.22	1000	1.3485	66.5792
0.0003	44.44	2000	1.5180	62.4313
0.0002	66.67	3000	1.5829	63.3683
0.0001	88.89	4000	1.6060	64.0180

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Tokenizers 0.15.1