zh-CN-model - whucedar

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

• Loss: 0.3881
• Wer: 134.2494

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
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• training_steps: 500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.375	0.6897	100	0.4223	331.7557
0.2198	1.3793	200	0.3885	178.8295
0.0881	2.0690	300	0.3812	123.1552
0.0877	2.7586	400	0.3873	132.3155
0.0604	3.4483	500	0.3881	134.2494

Framework versions

• Transformers 4.42.3
• Pytorch 2.3.1+cu118
• Datasets 2.20.0
• Tokenizers 0.19.1