not an issue (yet), question re aimdo

From your **CAVEATS**: "What you should generally do though, is completely flush the pytorch caching allocator before each new model run, which avoids completely un-used reservations from taking priority over the next models weights."

Who is the '**_you_**' in that sentence?

And..

When I start ComfyUI I have begun seeing this:

aimdo: src-win/cuda-detour.c:77:INFO:aimdo_setup_hooks: found driver at 00007FF9B1540000, installing 4 hooks
aimdo: src-win/cuda-detour.c:61:DEBUG:install_hook_entrys: hooks successfully installed
aimdo: src/control.c:69:INFO:comfy-aimdo inited for GPU: NVIDIA GeForce RTX 3090 (VRAM: 24575 MB)
DynamicVRAM support detected and enabled

I have no recollection of installing anything having to do with memory management. Or is this just built-in to ComfyUI now (version 0.18.2)

Thanks

Start with the CAVEATS text and the log references to src-win/cuda-detour.c and src/control.c. Check how comfy-aimdo is initialized and whether ComfyUI 0.18.2 enables it, then clarify who “you” refers to and document whether users must flush the PyTorch allocator themselves.

Written by the indexing model from the issue text.