Out of Memory about clip-interrogator HOT 5 OPEN

Oh maybe append new line doesn't work for prompts, so append file with leading comma.

from clip-interrogator.

Are you running with the --lowvram option? In theory should be able to work on 6GB no problem (CLIP Interrogator just needs 2.5GB) but I didn't sort out yet how to also kick the diffusion model out of VRAM and load it back in so the rest of the UI keeps working.
With lowvram option it is indeed only loading one model at a time on the CLIP Interrogator side and swapping it out for the next one. The main diffusion model is eating up a lot though which is the issue currently.

from clip-interrogator.

oh I'll have to try that on the next batch of x thousands images.
Where do I put the --lowvram option?
I was using A1111 didn't realize there was a different git for that.

from clip-interrogator.

Is it possible to run this on actually low vram? My card only has 2 and it just fails to work, even if I run webui with --lowvram flag

from clip-interrogator.

torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 6.76 GiB (GPU 0; 23.99 GiB total capacity; 18.50 GiB already allocated; 2.45 GiB free; 18.56 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

Is there a bug in the code's memory allocation management? Even running the program on a 4090 graphics card results in an out-of-memory error

from clip-interrogator.