README.md

metadata

language:
  - tel
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
  - generated_from_trainer
datasets:
  - jayasuryajsk/google-fleurs-te-romanized
metrics:
  - wer
model-index:
  - name: Wishper-Large-V3-Telugu_Romanized
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Telugu Romanized 1.0
          type: jayasuryajsk/google-fleurs-te-romanized
        metrics:
          - name: Wer
            type: wer
            value: 67.44785563627842

Wishper-Large-V3-Telugu_Romanized

This model is a fine-tuned version of openai/whisper-large-v3 on the Telugu Romanized 1.0 dataset. It achieves the following results on the evaluation set:

• Loss: 1.5824
• Wer: 67.4479

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: 20
• 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: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0079	8.6207	1000	1.4603	64.9051
0.0007	17.2414	2000	1.5824	67.4479

Framework versions

• Transformers 4.40.1
• Pytorch 2.2.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1