zfx123zfx/

Bench-377

	Magic Bench: A Comprehensive Text-to-Image Generation Evaluation Dataset




	
	📖 Overview

Magic Bench is a comprehensive evaluation dataset designed for text-to-image generation models. It contains 377 carefully curated prompts with detailed annotations across multiple dimensions, providing both Chinese and English versions for cross-lingual evaluation.

377 evaluation prompts covering diverse scenarios Bilingual support: Both Chinese and English… See the full description on the dataset page: https://huggingface.co/datasets/zfx123zfx/Bench-377.