4 papers · 1 filter
A Control Architecture for Training-Free Memory Use
Yanzhen Lu, Muchen Jiang, Zhicheng Qian +1
Prompt-injected memory can improve reasoning without updating model weights, but it also creates a control problem: retrieved content helps only when it is applied in the right sta…
State Transfer Reveals Reuse in Controlled Routing
Yanzhen Lu, Zhicheng Qian, Muchen Jiang +1
Prompt-based interventions can change model behavior, but trained success alone does not identify where the behaviorally relevant state is represented. We study this question in co…
CSPO: Alleviating Reward Ambiguity for Structured Table-to-LaTeX Generation
Yunfan Yang, Cuiling Lan, Jitao Sang +1
Tables contain rich structured information, yet when stored as images their contents remain "locked" within pixels. Converting table images into LaTeX code enables faithful digitiz…
Temperature as a Meta-Policy: Adaptive Temperature in LLM Reinforcement Learning
Haoran Dang, Cuiling Lan, Hai Wan +2
Temperature is a crucial hyperparameter in large language models (LLMs), controlling the trade-off between exploration and exploitation during text generation. High temperatures en…