10 papers
Budgeted KV-Cache Repair for Evolving Contexts
Han Yu, Wenhui Zhu, Jincheng Cao +15
Language-model agents revise their working context as information becomes obsolete, but deleting text does not erase its influence on cached computation. Later key-value (KV) state…
Maximum-modulus points on concentric circles
Yunpeng Li
We give a negative answer to the second question of Erdős Problem #1117: can the number of maximum-modulus points of a nonzero, non-monomial entire function on ten…
AvoKV-E: Payload-Aware KV Cache Eviction for Long Reasoning
Han Yu, Wenhui Zhu, Xiwen Chen +9
Long-output reasoning shifts the KV-cache bottleneck from the fixed prompt to the generated trace. Existing reasoning-cache eviction methods largely treat cached entries as routing…
Smaller Models, Better Rejects: Preference Distillation Scaling
Rui Cai, Wenhui Zhu, Xiwen Chen +15
Preference distillation typically treats a teacher response as preferred and the student's own response as rejected. This assumes that self-generated failures are the most informat…
LatentMT: Machine Translation with Latent Reasoning
Wei-Rui Chen, Samar M. Magdy, Chiyu Zhang +3
Latent-reasoning looped language models (LoopLMs) offer a different scaling path for machine translation (MT): instead of increasing parameter count or emitting explicit chain-of-t…
TREK: Distill to Explore, Reinforce to Refine
Yuanda Xu, Zhengze Zhou, Kayhan Behdin +10
Group Relative Policy Optimization (GRPO) is effective when the current policy already samples useful reasoning trajectories, but it stalls on hard prompts whose correct solution m…
S-SPPO: Semantic-Calibrated Self-Play Preference Optimization
Xiwen Chen, Wenhui Zhu, Jingjing Wang +13
Aligning Large Language Models (LLMs) with human preferences is often formulated via Direct Preference Optimization (DPO). However, the standard Bradley-Terry instantiation of DPO…
Sampling for Quality: Training-Free Reward-Guided LLM Decoding via Sequential Monte Carlo
Jelena Markovic-Voronov, Wenhui Zhu, Bo Long +5
We introduce a principled probabilistic framework for reward-guided decoding in large language models, addressing the limitations of standard decoding methods that optimize token-l…
EvoCurr: Self-evolving Curriculum with Behavior Code Generation for Complex Decision-making
Yang Cheng, Weiyu Ma, Zilai Wang +6
Complex decision-making often requires agents to progress through intermediate tasks rather than solve the final target directly. Existing LLM self-refinement methods typically ite…
SMAC-Hard: Enabling Mixed Opponent Strategy Script and Self-play on SMAC
Yue Deng, Yan Yu, Weiyu Ma +4
The availability of challenging simulation environments is pivotal for advancing the field of Multi-Agent Reinforcement Learning (MARL). In cooperative MARL settings, the StarCraft…