I am looking for a senior ML Engineer who can help me on multiple tasks. I installed the GUI version of Stable Diffusion and with that I was able to make 512 by 512 pixel images using my GeForce RTX 3070 GPU with 8 GB of memory. However when I try to do the same thing with the command line interface, I run out of memory. If I reduce the size of the image to 256 X 256, it gives a result, but obviously much lower quality.
So part 1 of this project is why do I run out of memory at 6.13 GiB when I have 8 GiB on the card, and part 2 is what does the GUI do differently to allow 512 by 512 output? It would be great if you have any idea to reduce the load on the GPU.
Job is closed
This job is already closed and no longer accepting applicants, sorry.
