@Oluwaphilemon1 Or you can run the int8 version inside comfyui even on 8gb cards and generate in less time; you dont have to keep the text encoder and vae loaded in vram at all times lol
@a4lg Running here on 3070ti+64gb ddr4 3600mhz, you have to make sure before you run strata only 600-700mb used vram; vision disabled obviously; IQ3_XXS 60-70tps with 128k context; still optimizing trying to find best balance