-
Thanks for your work.
I do training on celeba-hq dataset, and after 110k steps, I find that the images seem to have color problem, is there something wrong i need to do with datasets?
![64a5ac5ea0…
-
### Is there an existing issue for this?
- [X] I have searched the existing issues and checked the recent builds/commits
### What happened?
Updated the repo (forked) with the current commit:
https…
-
Traceback (most recent call last):
File "scripts/stable_txt2img.py", line 294, in
main()
File "scripts/stable_txt2img.py", line 248, in main
samples_ddim, _ = sampler.sample(S=opt.ddi…
-
我看代码上gen_images是不同程度异常的拼接，然后 recon_images中的x0_det也涉及使用DDIM进行异常合成。这两部分返回值的区别是什么？如果我想合成最终的异常图像应该以哪个为主？期待您的回复！
-
Hi!
The code in this repository has helped me a lot!
I found that as the batch size increases, the training time increases dramatically. When I set the batch size to 4 (the dataset has 25k image…
-
I noticed that you've removed the DDIM option in the current version of the code, even though it didn't seem to work in the initial version. Sampling efficiency is one of the obstacles to practicality…
-
It seems that the fp16 setting is not effective. I tried to use fp16 manually, and offload the autoencoder and CLIP to cpu memory during ddim denoising, and can run a 45x512x768 video on 12G GPU memor…
-
We found that guidance scale is not work beside more than 1. it get entirely different image
-
Thank you for the project, it's definitely worth learning from~~ / 感谢大佬的项目，非常值得学习 ~~!
大老： 测试中发现这些问题，其它的都可以跑通没问题 / Da Lao: These issues were found during testing, but everything else can run smoothl…
-
Running tutorial_train.py (with fill50k dataset) on Tesla V100 16Gb results in out-of-memory during image logging (DDIM Sampler).
I am running with `batch_size=1` and `accumulate_grad_batches=4` (O…