Cannot generate images as python3 is eating all the memory

[LordMilutin]

I have installed everything according to the manual on Linux.
However, I cannot generate images as it shows:
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 26.00 MiB (GPU 0; 7.78 GiB total capacity; 6.86 GiB already allocated; 17.69 MiB free; 7.04 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

I have NVIDIA GeForce RTX 3050 with 8GB of RAM.

When this is turned on, python3 eats 7326MiB of memory, but when I turn it off, I only have about 500MiB in total system usage.

Is there any option for me to reduce in half the python3 memory consumption?

I passed --medvram env, and I tried --xformers, but it is unable to load, probably something with wrong versions, as I am running CUDA 12.2 and XFormers is 12.1.

Any info would be greatly appreciated!