yasserben/clouds

[CVPR 2024] Official Implementation of Collaborating Foundation models for Domain Generalized Semantic Segmentation

clouds's Issues

请问训练该模型需要什么样的显卡？

Numbers of generated images

Hi there,

Shout out to your great work!

I'm wondering how many generated images you used for training? I saw there are 100 prompts in the provided txt file, does it mean the model trained with 100 generated images?

Cityscapes training and joint training

Hi,

I'm running the training code for training cityscapes and eval on acdc. I encountered some problems:

1.For warm up training, i found the training is not corrected conducted, it directly jump to inference step(same for gta5 training):

After training, i will get a saved pth file like this:

I'm wondering is it normal and what's objective of warmup training?

2.For joint training, i found it requires load pretrained weight in config file:

I didn't find it in the repo. So i assume it may be the warmup checkpoint and set it accordingly. But it throw out error like this:

Can you help me fix it? Much appreciated!

Training time

Thanks for releasing the code. How long does your method need to train?

How to generate images by the diffusion model

Excellent job!
Recently, I want to reproduce your results. But I don't know how to run the generate_txt_im.py. Could you provide more imformation about this process? Thanks a lot and Looking for your reply!