>>106201478
>seems like you have real fp8 support
interestingly, little to no info confirming that this is supported on the XTX shows up on google

>>106201515
>find fp8 scaled instead of just fp8
Has anyone even uploaded a scaled fp8 for qwen yet? all I can find is these:
https://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI/blob/main/split_files/diffusion_models/qwen_image_fp8_e4m3fn.safetensors
https://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI/blob/main/non_official/diffusion_models/qwen_image_distill_full_fp8_e4m3fn.safetensors