Hugging Face's logo

Rziane
/
whisper-medium-aeb

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
Tunisian Arabic
whisper
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
whisper-medium-aeb / README.md
Rziane's picture
Rziane
End of training
351f1c2 verified
preview code
|
raw
history blame contribute delete
1.91 kB
metadata
library_name: transformers
language:
  - aeb
license: apache-2.0
base_model: openai/whisper-medium
tags:
  - generated_from_trainer
datasets:
  - AT
metrics:
  - wer
model-index:
  - name: Whisper medium AT
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: AT
          type: AT
          args: 'config: aeb, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 65.98418372874012

Whisper medium AT

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

  • Loss: 0.9915
  • Wer: 65.9842

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: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 4
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
No log 1.0 293 1.3198 74.6073
1.7949 2.0 586 1.0108 70.6316
1.7949 3.0 879 0.9583 65.9517
0.5076 4.0 1172 0.9915 65.9842

Framework versions

  • Transformers 4.45.1
  • Pytorch 2.4.1+cu121
  • Datasets 3.0.1
  • Tokenizers 0.20.0