Memory requirements

[skydam]

Does anybody else have these ridiculous memory requirements? I'm running on a 3090 for more than an hour per generation. 150.20s/it.

[GraftingRayman]

any chance you can throw a screen shot of your InfuseNetModel directory?

[skydam]

Yes of course! It's using 41.6GB GPU memory which means that it's swapping like crazy.

[GraftingRayman]

Thank you, for some reason when i try to run it, its trying to join the shards and its running out of vram

Wow, 41.6GB vram is huge, no chance with my 16gb

[EndlessSora]

Thank you guys for your suggestion. We will try to improve memory usage for users' convenience. We also welcome community contributions.

[skydam]

That would be great. The results are really promising, but >45 minutes per item is a bit much if you know what i mean! ;-)

[EndlessSora]

We have released our Hugging Face online demo. Please feel free to try it: https://huggingface.co/spaces/ByteDance/InfiniteYou-FLUX

[niftyflora]

i am on a 3090 24gb vram and i cannot even run it? just rebooted the computer to make sure its fresh and nothing else is running but still nothing. here is the full error: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.00 MiB. GPU 0 has a total capacity of 23.68 GiB of which 12.88 MiB is free. Process 2212 has 394.45 MiB memory in use. Including non-PyTorch memory, this process has 23.14 GiB memory in use. Of the allocated memory 22.89 GiB is allocated by PyTorch, and 8.64 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

[niknah]

Is it possible to use flux fp8 instead?

[EndlessSora]

Please consider following some tips at https://github.com/bytedance/InfiniteYou?#memory-requirements first.