jethrowang/android_baseline-whisper-tiny_tat

Whisper Tiny Taiwanese Simulated Android

This model is a fine-tuned version of openai/whisper-tiny on the TAT ASR Aligned dataset. It achieves the following results on the evaluation set:

• Loss: 0.6140
• Cer: 11.1506

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: 0.0001
• train_batch_size: 64
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 681
• training_steps: 6810
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.345	0.9985	681	0.4722	17.2090
0.2022	1.9971	1362	0.4060	13.0439
0.1228	2.9956	2043	0.4379	13.2219
0.0721	3.9941	2724	0.4696	12.3827
0.0406	4.9927	3405	0.5141	12.5748
0.021	5.9912	4086	0.5437	12.2795
0.0107	6.9897	4767	0.5696	11.8216
0.0034	7.9883	5448	0.5935	11.4186
0.0011	8.9868	6129	0.6080	11.2588
0.0006	9.9853	6810	0.6140	11.1506

Framework versions

• Transformers 4.42.3
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1