Hugging Face's logo

mohammadalihumayun
/
whisper-small_urdu_hazrat_v1

Automatic Speech Recognition
Transformers
Safetensors
Urdu
whisper
Generated from Trainer
Inference Endpoints
Model card Files Community
whisper-small_urdu_hazrat_v1 / README.md
mohammadalihumayun's picture
mohammadalihumayun
End of training
f69a484 verified
preview code
|
raw
history blame contribute delete
1.82 kB
metadata
library_name: transformers
language:
  - ur
license: apache-2.0
base_model: openai/whisper-small
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: Whisper Urdu (Hazrat v1 250 samples)
    results: []

Whisper Urdu (Hazrat v1 250 samples)

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

  • Loss: 0.9203
  • Wer: 53.0197

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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 25
  • num_epochs: 5
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
1.3285 0.9375 15 1.1243 73.4611
1.0515 1.875 30 0.9972 57.4332
0.8574 2.8125 45 0.9302 61.2660
0.7007 3.75 60 0.9257 56.0395
0.6162 4.6875 75 0.9203 53.0197

Framework versions

  • Transformers 4.47.0
  • Pytorch 2.5.1+cu121
  • Datasets 3.1.0
  • Tokenizers 0.21.0