Monet: Enabling multimodal LLMs to reason in latent visual space

Monet：赋予多模态大模型如人类一般的抽象视觉思考能力

Monet trains Qwen2.5-VL-7B into Monet-7B to reason with continuous latent visual embeddings instead of external tools; the work is accepted by CVPR 2026 and releases paper, code, model, and a 125K SFT dataset. The method uses three-stage SFT plus VLPO reinforcement learning; the post reports 3% to 9.75% gains on in-distribution tasks and 2.31% on out-of-distribution abstract visual reasoning versus the base model. The key detail is the VLPO mechanism and dataset construction; the post does not disclose one unified table of absolute headline scores.

Why it matters: This hits HKR-H and HKR-K: the angle is abstract visual reasoning, and the post includes 125K SFT data, a 3-stage SFT setup, VLPO, and 3%–9.75% / 2.31% gains. HKR-R is weaker because full absolute leaderboard scores and real deployment evidence are not disclosed, so it lands as a