metadata

license: apache-2.0
base_model: openai/whisper-small
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
datasets:
  - arrow
metrics:
  - wer
model-index:
  - name: whisper-small-indian_eng
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: arrow
          type: arrow
          config: default
          split: validation
          args: default
        metrics:
          - name: Wer
            type: wer
            value: 16.853932584269664

whisper-small-indian_eng

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

Loss: 0.7659
Wer: 16.8539

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: 8
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 25
training_steps: 200
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.515	10.0	50	1.0232	16.8539
0.1717	20.0	100	0.7364	19.1011
0.0006	30.0	150	0.7300	15.7303
0.0003	40.0	200	0.7659	16.8539

Framework versions

Transformers 4.42.0.dev0
Pytorch 2.2.0+cu121
Datasets 2.16.1
Tokenizers 0.19.1