metadata

library_name: transformers
language:
  - en
license: apache-2.0
base_model: openai/whisper-small
tags:
  - generated_from_trainer
datasets:
  - babs/nigerian-accented-english
metrics:
  - wer
model-index:
  - name: Whisper Small english - Nigerian accent
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Nigerian accented english
          type: babs/nigerian-accented-english
          args: 'config: en, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 48.805954719321825

Whisper Small english - Nigerian accent

This model is a fine-tuned version of openai/whisper-small on the Nigerian accented english dataset. It achieves the following results on the evaluation set:

Loss: 1.3324
Wer: 48.8060

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: 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
0.3217	4.6948	1000	0.8267	55.4637
0.0653	9.3897	2000	1.0964	53.9440
0.0161	14.0845	3000	1.2460	51.9901
0.0044	18.7793	4000	1.3324	48.8060

Framework versions

Transformers 4.45.1
Pytorch 2.4.1+cu121
Datasets 3.0.1
Tokenizers 0.20.0