chinese-english-whisper-finetune-take4

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

• Loss: 1.9883
• Wer: 89.9066
• Mer: 67.2773
• Cer: 63.7665

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: 0.0001
• 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: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Mer	Cer
2.2342	0.2907	200	2.5602	99.5330	67.4210	76.1019
1.4461	0.5814	400	2.4321	96.6236	67.3132	74.7949
2.1937	0.8721	600	2.0802	97.8807	68.3908	67.5634
1.1723	1.1628	800	2.0404	94.9713	66.1279	65.5982
1.1299	1.4535	1000	1.9883	89.9066	67.2773	63.7665

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1