11 papers
From Concentration to Differentiation and Back: Routing Effective Rank in MoE Reasoning Cohorts
Kang Chen, Sihan Zhao, Yixin Cao +1
Test-time scaling produces cohorts of reasoning rollouts, yet there is no standard label-free account of how their internal computation reorganizes as inference unfolds. We introdu…
Beyond the Trace: Coupling an Interpretable Reasoning-State Readout to Native MoE Routing
Kang Chen, Sihan Zhao, Yixin Cao +1
What a reasoning model writes is only a partial record of the process that produces it. We introduce a two-level internal readout for mixture-of-experts reasoning. We first distill…
Disagree to Explore, Agree to Commit: Routing-Guided Test-Time Scaling for Software Agents
Kang Chen, Junjie Nian, Yixin Cao +1
Software-engineering agents solve repository-level tasks through long, stochastic tool-use trajectories, and repeated attempts often find fixes missed by one run. Test-time scaling…
Does the Same Token Mean the Same State? MoE Routing as Signal for Reasoning Control
Kang Chen, Minshen Yu, Junjie Nian +3
In sparse Mixture-of-Experts language models, does the same token id imply the same router state and the same experts producing it? Holding the emitted token id fixed at repeated a…
TraceGraph: Shared Decision Landscapes for Diagnosing and Improving Agent Trajectories
Junjie Nian, Kang Chen, Ge Zhang +2
Agent benchmarks increasingly record rich interaction trajectories, yet evaluation often reduces each rollout to a pass rate or reward score. We introduce TraceGraph, a graph-based…
SliceGraph: Mapping Process Isomers in Multi-Run Chain-of-Thought Reasoning
Kang Chen, Junjie Nian, Yixin Cao +1
Multi-run chain-of-thought reasoning is usually collapsed to final-answer aggregates, which discard howsampled trajectories share, split, and rejoin through intermediate computatio…
NEX: Neuron Explore-Exploit Scoring for Label-Free Chain-of-Thought Selection and Model Ranking
Kang Chen, Zhuoka Feng, Sihan Zhao +5
Large language models increasingly spend inference compute sampling multiple chain-of-thought traces or searching over merged checkpoints. This shifts the bottleneck from generatio…
Thinking Traps in Long Chain-of-Thought: A Measurable Study and Trap-Aware Adaptive Restart
Kang Chen, Fan Yu, Junjie Nian +6
Scaling test-time compute via Long Chain-of-Thought (Long-CoT) significantly enhances reasoning capabilities, yet extended generation does not guarantee correctness: after an early…
ARM: Role-Conditioned Neuron Transplantation for Training-Free Generalist LLM Agent Merging
Zhuoka Feng, Kang Chen, Sihan Zhao +7
Interactive large language model agents have advanced rapidly, but most remain specialized to a single environment and fail to adapt robustly to other environments. Model merging o…
Do LLMs Signal When They're Right? Evidence from Neuron Agreement
Kang Chen, Yaoning Wang, Kai Xiong +4
Large language models (LLMs) commonly boost reasoning via sample-evaluate-ensemble decoders, achieving label free gains without ground truth. However, prevailing strategies score c…
Less Data Less Tokens: Multilingual Unification Learning for Efficient Test-Time Reasoning in LLMs
Kang Chen, Mengdi Zhang, Yixin Cao
This paper explores the challenges of test-time scaling of large language models (LLMs), regarding both the data and inference efficiency. We highlight the diversity of multi-lingu…