和Monet提出的机制类似。 【🔧SFT】Sketch-in-Latents: Eliciting Unified Reasoning in MLLMs (Arxiv 2025.12) [paper] 用一个额外的encoder（siglip2+projector）+pooling 把aux img映射为gt visual embedding，和latent对齐；此外还有一个text的NTP ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results