jethrowang/vanilla-whisper-medium

Whisper Medium Hakka Condenser

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

• Loss: 0.0401
• Cer: 1.8101

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: 32
• 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: 1521
• training_steps: 15215
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.0328	0.9997	3043	0.0681	4.6235
0.0135	1.9993	6086	0.0515	2.8839
0.0045	2.9990	9129	0.0440	1.9904
0.0028	3.9987	12172	0.0403	2.0760
0.0007	4.9984	15215	0.0401	1.8101

Framework versions

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