Spaces:

lambdalabs

/

stable-diffusion-image-variations

like 447

Running on A10G

Hello kind stranger
Thanks a lot for the code and model

I was trying to run the code here locally in a docker container, everything is working perfectly but whenever I change sample count to anything other than "1" my PC freezes for a second then it gives an error with RuntimeError: CUDA error: unknown error.

I'm using a 3080ti and I know this might not have enough vram to run such thing but I've already ran some stable diffusion projects locally and set samples to multiple.

The following line is the only thing related I can see in the stacktrace

File "/app/lambda-diffusers/lambda_diffusers/pipelines/pipeline_stable_diffusion_im_embed.py", line 136, in __call__
noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=image_embeddings)["sample"]

Thanks again for model and knowledge