AI RESEARCH
RL makes MLLMs see better than SFT
arXiv CS.LG
•
ArXi:2510.16333v2 Announce Type: replace-cross A dominant assumption in Multimodal Language Model (MLLM) research is that its performance is largely inherited from the LLM backbone, given its immense parameter scale and remarkable capabilities. This has created a void in the understanding of the vision encoder, which determines how MLLMs perceive images. The recent shift in MLLM