10 papers
MetaBench-Harness: Unlocking End-to-End Optimization of Benchmark Harnesses
Xuanjun Chen, Hua-Hsuan Chen, Wei-Chung Lu +3
Rapid progress in Large Language Models (LLMs) is saturating static benchmarks faster than they can be designed. While existing automated evolution frameworks attempt to generate h…
DominoTree: Conditional Tree-Structured Drafting with Domino for Speculative Decoding
Saw S. Lin, Jyh-Shing Roger Jang
Speculative decoding speeds up large language model inference: a small drafter proposes several tokens, and the target model verifies them in one parallel pass. Drafting a tree let…
Only Ask What You Don't Know: Grounded Delta Planning for Efficient Multi-step RAG
Wei-Chieh Chou, Xuanjun Chen, Jian-Ren Lin +3
Multi-hop question answering remains challenging for Retrieval-Augmented Generation (RAG) because existing approaches either propagate errors across iterative retrieval rounds or o…
Mitigating Proxy-to-Wild Domain Gap in Deepfake Speech
Xuanjun Chen, Yun-Shing Wu, Wei-Chung Lu +4
Recent neural audio codec-based speech generation (CodecFake) produces highly realistic audio, posing a challenge to existing deepfake countermeasure models. While using codec resy…
Equilibrium Residuals Expose Three Regimes of Matrix-Game Strategic Reasoning in Language Models
Wenhua Nie, Binhan Luo, Zijie Meng +2
Large language models can score well on named game-theory benchmarks while failing on the same strategic computation once semantic cues are removed. We show this gap with procedura…
Identified-Set Geometry of Distributional Model Extraction under Top- Censored API Access
Wenhua Nie, ZiCheng Zhu, Jianan Wu +3
Modern LLM APIs often reveal only top- logit scores and censor the remaining vocabulary. We study the per-position distribution-recovery limits of this access model. For censori…
Future Validity is the Missing Statistic: From Impossibility to -Estimation for Grammar-Faithful Speculative Decoding
Wenhua Nie, Zijie Meng, Kun Zou +5
Grammar-constrained generation is often combined with local vocabulary masking and speculative decoding, but the resulting sampling law is not the grammar-conditional distribution…
Gradient Starvation in Binary-Reward GRPO: Why Group-Mean Centering Fails and Why the Simplest Fix Works
Wenhua Nie, Jianan Wu, Junlin Liu +6
Group Relative Policy Optimization (GRPO) is a standard algorithm for reinforcement learning from verifiable rewards, but its group-mean-centered advantage can fail under binary re…
The Coupling Tax: How Shared Token Budgets Undermine Visible Chain-of-Thought Under Fixed Output Limits
Wenhua Nie, Junlin Liu, Jianan Wu +5
Chain-of-thought reasoning is often treated as a monotone way to improve language-model accuracy by letting a model think longer. We identify a countervailing effect, the coupling…
CodaRAG: Connecting the Dots with Associativity Inspired by Complementary Learning
Cheng-Yen Li, Xuanjun Chen, Claire Lin +4
Large Language Models (LLMs) struggle with knowledge-intensive tasks due to hallucinations and fragmented reasoning over dispersed information. While Retrieval-Augmented Generation…