6x16/whisper-medium-yue-continuous

self-trained Whisper-Medium Yue model (to be fine-tuned by other datasets) #JL

This model is a fine-tuned version of openai/whisper-medium on the Common Voice 17.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2796
• Cer: 8.3321

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: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• 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	Cer
0.0187	5.5556	1000	0.2198	9.1962
0.0006	11.1111	2000	0.2458	8.8610
0.0001	16.6667	3000	0.2670	8.4587
0.0001	22.2222	4000	0.2757	8.3544
0.0001	27.7778	5000	0.2796	8.3321

Framework versions

• Transformers 4.52.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1