Skip to content
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

pixelated generations #23

Open
jashwanth0712 opened this issue Nov 18, 2024 · 4 comments
Open

pixelated generations #23

jashwanth0712 opened this issue Nov 18, 2024 · 4 comments

Comments

@jashwanth0712
Copy link

When i am trying to generate the orbit motion even with the example i am getting pixelated generation , is there any mistake that i am doing while running
image

@chenshuo20
Copy link
Collaborator

Are you ensuring the scale is correct? The scale for LoRA should be set to 1/rank.

lora_rank = 256
pipe.load_lora_weights(lora_path, weight_name="pytorch_lora_weights.safetensors", adapter_name="test_1")
pipe.fuse_lora(lora_scale=1 / lora_rank)
pipe.to("cuda")

@jashwanth0712
Copy link
Author

yes i am using 1/rank , still getting the pixelated results

pipe_image.load_lora_weights(lora_path, weight_name="pytorch_lora_weights.safetensors", adapter_name="test_1")
pipe.fuse_lora(components=["transformer"], lora_scale=1/lora_rank)
pipe_image = pipe_image.to(device)

@chenshuo20
Copy link
Collaborator

Could you provide more details? For example, your code and the information about your environment.

@jashwanth0712
Copy link
Author

I am currently running it in runpod , A100 . and did not alter the code much from the git clone
only change currently present is that i have added

components=["transformer"] in the fuse_lora method

pipe_image.load_lora_weights(lora_path, weight_name="pytorch_lora_weights.safetensors", adapter_name="test_1")
pipe.fuse_lora(components=["transformer"], lora_scale=1/lora_rank)
pipe_image = pipe_image.to(device)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants