Holistic, Reliable and Scalable Benchmark for Text-to-Image Models:

1. Holistic skills evaluation. Rather than focus on isolated metrics such as accuracy, we measure 13 skills, which could be categorized into five critical skills; accuracy, robustness, generalization, fairness, and bias.
   
2. Broad scenarios coverage. HRS-Bench covers 50 applications, e.g., fashion, animals, transportation, food, and clothes.
   
3. Standardization. We propose a unified benchmark, where we fairly evaluate the existing models across a wide range of metrics.
   
4. Holistic prompts generation.