Add MixedPrecisionOps compatible fp8mixed qwen_image_edit_2509 with sensitive layers kept in high precision and using full_precision_matrix_mult

#11

Sign up or log in to comment