Hi, Dear developer!
Thank you for your great work. I'm trying to run the inference code of SUPIR by below script.
CUDA_VISIBLE_DEVICES=0,1 python test.py --img_dir input_folder/ --save_dir ./results-F --SUPIR_sign F --upscale 4 --loading_half_params --load_8bit_llava
But the out of memory has occurred. So, How should I do to run the inference using two 3090 24G GPU?