논문 리뷰 Reinforcement Learning rl 티스토리챌린지 오블완 diffusion model 논문리뷰 multi-modal 강화학습 offline rl Image generation LLM vlm llava Large Language Model Muzero EXplOration RND supervised fine tuning steptool tool learning visual sketchpad vision language model bcq world model text decision transformer reward design talking face kolors sampled muzero character consistency munchausen rl m-rl random network distillation agac ape-x myvlm spatialvlm siglip IP-Adapter 이미지 생성 모델 Chain of Thought Multi Modal RLHF dreambooth 생성모델 generative model value function 이미지 생성 inpainting COT sft Bear planning R2D2 탐험 lipsync classification