TransUNet icon indicating copy to clipboard operation
TransUNet copied to clipboard

Have you tried to train the model without pretraining on imagenet?

Open licaizi opened this issue 3 years ago • 1 comments

licaizi avatar Apr 20 '21 12:04 licaizi

Comment out the load pretrain in the train.py file like this:

# net.load_from(weights=np.load(config_vit.pretrained_path)) # load pretrain

liyiersan avatar Aug 04 '21 05:08 liyiersan