I’m very excited by this new benchmark dataset for visual generation that is suitable for the modern era of large scale generative models!
GPIC is a new visual generation benchmark dataset with 100M VLM-captioned image-text pairs for training, 1M for evaluation, ~28 trillion pixels, and full permissiveness for research/commercial use—but the post is only a tweet with no technical details or validation.
入选理由:GPIC 提供 1 亿 VLM-captioned 图文对用于训练,100 万用于 benchmark 评测。
