lab9_whisper-tiny-zh-tw

This model is a fine-tuned version of Wellyowo/whisper-tiny-zh-tw on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6336
• Wer Ortho: 64.0
• Wer: 62.1359

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.0088	0.6882	500	0.5502	60.0	61.1650
0.0051	1.3765	1000	0.5735	65.0	64.0777
0.0068	2.0647	1500	0.5820	63.0	63.1068
0.0021	2.7529	2000	0.5955	62.0	61.1650
0.0039	3.4412	2500	0.5858	62.0	61.1650
0.0018	4.1294	3000	0.5981	63.0	61.1650
0.0019	4.8176	3500	0.6322	63.0	61.1650
0.0102	5.5058	4000	0.6336	64.0	62.1359

Framework versions

• Transformers 4.40.1
• Pytorch 2.2.0+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1