SakuraJ/whisper-small-Chinese

Whisper Small chinese - sakura

This model is a fine-tuned version of openai/whisper-small on the thchs30 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2458
• Cer: 9.7819

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• 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: 10
• training_steps: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.304	0.32	20	0.3021	12.5776
0.2813	0.64	40	0.2758	10.9862
0.2852	0.96	60	0.2584	10.3349
0.1476	1.272	80	0.2497	9.9908
0.1371	1.592	100	0.2458	9.7819

Framework versions

• Transformers 4.50.3
• Pytorch 2.5.1+cu124
• Datasets 3.5.0
• Tokenizers 0.21.0