GenericSSL

GenericSSL copied to clipboard

Hello, the code is not modified, but the results are different and the results are very different, what is the way to avoid these unstable factors

2

Hello, the code is not modified, but the results are different and the results are very different, what is the way to avoid these unstable factors

hhjjaaa

Question about Training Results Discrepancy

2

Dear Author, I followed your setup exactly and preprocessed the LA dataset as instructed. Then, I ran the following command: `bash train.sh -c 0 -e diffusion -t la_0.05 -i 'experiment1'...

Halleyawoo

Hello! In the paper Section3.3, it is written as follows "we also transmit with knowledge distillation strategy: $\theta$ = ..." I think $\xi$ and $\psi$ are updated by EMA, as...

wh-isper

Diffusion denoising branches are useless in unlabelled data

6

I reproduced the code you provided in the Synapse dataset and got the same result as in the paper. However, I visualized the result of unlabeled data in the diffusion...

hhjjaaa

negative losses

3

Hi , I am finding negative losses while training with my own dataset. I would like to ask you what causes this. 

hhjjaaa

Dear author： synapse数据链接中npy和npy1有什么区别呢？

geek-APTX4869