-
Notifications
You must be signed in to change notification settings - Fork 36
Issues: FoundationVision/LlamaGen
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Author
Label
Projects
Milestones
Assignee
Sort
Issues list
Do you try class 2 Image generation with the image resolution of 512X512?
#42
opened Jul 7, 2024 by
OliverRensu
你好,vq_ds16_c2i_training.pt 在私有的通用图片上finetune效果变差了,就试了在imgenet数据集接着finetune,效果也变差了,想问一下是哪里出问题了?
#36
opened Jun 24, 2024 by
353xiong
Question about why not try using image tokenizer and a ready made llama3 etc LLM model with lora?
#31
opened Jun 19, 2024 by
lucasjinreal
Which parameters are trainable? Are the encoder and decoder in VQGAN fixed? Is the llama fixed?
#26
opened Jun 17, 2024 by
tanshuai0219
[Feature] ControlNet support via process similar to PixArt's ControlNet-Transformer
enhancement
New feature or request
#24
opened Jun 16, 2024 by
kabachuha
RuntimeError: shape '[1, 512, 1, 32, 2]' is invalid for input of size 16448
#21
opened Jun 14, 2024 by
beniz
ProTip!
Exclude everything labeled
bug
with -label:bug.