search

JCruan519 / VM-UNet

(ARXIV24) This is the official code repository for "VM-UNet: Vision Mamba UNet for Medical Image Segmentation".
Apache License 2.0
393 stars 12 forks source link

About saving weights #29

Open PURSUETHESUN opened 3 months ago

PURSUETHESUN commented 3 months ago

Thank you for your excellent work. I have some questions about saving weights. For example, the weight with the smallest loss is saved in your source code for testing. However, this may not be the highest value of Dice. Why don't you save Dice's best weights? If the best Dice weights are saved, higher accuracy can be achieved. Is there a difference between these two?

JCruan519 commented 3 months ago

@PURSUETHESUN Hello, testing the Dice metric on the test set at every step will lead to a significant increase in training time.