Closed jaymefosa closed 9 months ago
I haven't had these problems yet
Is it possible to give an inference script? I'm not sure about some of the details.
Is it possible to give an inference script? I'm not sure about some of the details.
I've written a simple untested inference script that I can share with you if you want
Is it possible to give an inference script? I'm not sure about some of the details.
I've written a simple untested inference script that I can share with you if you want
If possible, please send it to zju.shen2017@gmail.com.
I also wrote a test script with reference to magic-animate to test the results of the first training phase. The denoised image only has pose information and no texture information. I don’t know if it is normal.
After debugging, The update function of ReferenceNetAttention does not work in the first training stage because reader_attn_modules is [].
Is it possible to give an inference script? I'm not sure about some of the details.
I've written a simple untested inference script that I can share with you if you want
If possible, I also need the inference code for more details. (yxpanglight@gmail.com)
Is it possible to give an inference script? I'm not sure about some of the details.
I've written a simple untested inference script that I can share with you if you want
If possible, I also need the inference code for more details. Thanks a lot ! (popaslan@sina.com)
If possible, I hope I can get one. Thanks a lot.(zhanzhan4399@gmail.com) @guoqincode
(654227280@qq.com),test scripts,thanks @guoqincode
Can anybody share the infer code? THANKS a lot~ (1124015396@qq.com)
Per the title I've been a little perplexed to see that what was denoised well at 30 inference timesteps @ 60k training steps, requires 70 steps @ 100k training steps.
My implementation is slightly different than yours so there could be quite a few things going on. Just curious if you noticed any similar behaviors since you're in the middle of training these days.
Thank you