rngzhi/cs3264-project

Whipser Small - Singlish

This model is a fine-tuned version of openai/whisper-small on the National Speech Corpus(partial) dataset. It achieves the following results on the evaluation set:

• Loss: 0.2020
• Wer: 5.3795

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
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0068	5.01	500	0.1508	5.4137
0.001	11.01	1000	0.1691	5.0832
0.0003	16.02	1500	0.1769	5.1060
0.0006	22.01	2000	0.1840	5.0946
0.0005	28.0	2500	0.1891	5.1174
0.0003	33.02	3000	0.1933	5.2086
0.0005	39.01	3500	0.1962	5.2997
0.0002	45.0	4000	0.1991	5.3339
0.0002	50.02	4500	0.2010	5.3681
0.0003	56.01	5000	0.2020	5.3795

Framework versions

• Transformers 4.40.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.1.dev0
• Tokenizers 0.15.2