Searcharxiv⌕ Search

arXiv subjects

Huan Zhu

Publications and source records attributed to Huan Zhu.

4 recordsLinked to original sources
arXiv2026

Think Through a Bottleneck: Hourglass Reasoning for Rigorous Induction

Self-refinement often fails to strengthen few-shot inductive reasoning in large language models. Prompting a model to explicitly state its inferred rule does little on its own. What actually matters is a structurally enforced isolation between reasoning stages, so that information can only pass between them as a compressed symbolic state. We introduce \textbf{Hourglass reasoning}, which enforces strict context isolation between reasoning stages. The frozen LLM acts as a meta-constructor, building for each task a symbolic encoder--decoder: an Induction module compresses the support examples into a schema $ϕ$ (encoder) and a transient scaffold $z$; a Deduction module derives rule $T$ (decoder) from these and discards $z$; an Implementer compiles $(ϕ, T)$ into artifacts; an error-driven Refiner revises $(ϕ, T)$ and regenerates artifacts from scratch. Only $(ϕ, T)$ crosses stage boundaries, so all refinement stays anchored to the rule. We evaluate Hourglass across three benchmarks spanning visual abstraction, hardware synthesis, and textual rule induction, using GPT-5.5 and Gemini 3.1 Pro. On ARC-AGI-2, it raises best-of-5 accuracy by up to 14 points over an iterative-refinement baseline. On ChipBench, it nearly doubles Verilog synthesis accuracy with GPT-5.5, from 31\% to 58\%. BBEH-Linguini draws on puzzles from the International Linguistics Olympiad, a setting where prior work has shown that explicit verbalization can hurt performance. Hourglass mitigates this tendency, and on Gemini 3.1 Pro, it reverses the effect entirely. Ablations confirm that these gains come from the isolation between stages and the quality of the initial induction, not from prompt wording or the particular symbolic form used. It is how information flows through the reasoning process, rather than the language used to express it, that drives inductive reasoning in frozen LLMs.

cs.AI↗
arXiv2026

IRPM: Intergroup Relative Preference Modeling for Pointwise Generative Reward Models

Generative Reward Models (GRMs) have demonstrated strong performance in reward modeling, due to their interpretability and potential for refinement through reinforcement learning (RL). However, widely used pairwise GRMs create a computational bottleneck in reinforcement learning from human feedback (RLHF), when calibrating or aggregating preference signals over n candidates, often incurring O(n^2) pairwise judgments. To address this issue, we propose Intergroup Relative Preference Modeling (IRPM), an RL-based method that extends the Bradley--Terry preference-learning paradigm via intergroup comparisons to train pointwise GRMs from pairwise preference data. IRPM derives pointwise reward for each response by contrasting groups of chosen vs. rejected samples, enabling pointwise scores comparable across candidate sets and O(n) reward evaluation for a variable number of candidates during RL training, while preserving interpretability and scalability. Experiments show that IRPM achieves state-of-the-art performance among pointwise GRMs on RM-Bench, JudgeBench and RewardBench, and approaches the performance of leading pairwise GRMs. In addition, IRPM achieves substantial gains in post-training evaluations, demonstrating its effectiveness.

cs.LG↗
arXiv2020

Embedding Lithium-ion Battery Scrapping Criterion and Degradation Model in Optimal Operation of Peak-shaving Energy Storage

Lithium-ion battery systems have been used in practical power systems for peak-shaving, demand response, and frequency regulation. However, a lithium-ion battery is degrading while cycling and would be scrapped when the capacity reduces to a certain threshold (e.g. 80%). Such scrapping criterion may not explore the maximum benefit from the battery storage. In this paper, we propose a novel scrapping criterion for peak-shaving energy storage based on battery efficiency, time-of-use price, and arbitrage benefit. A new battery life model with scrapping parameters is then derived using this criterion. Embedded with the life model, an optimal operation method for peak-shaving energy storage system is presented. The results of case study show that the operation method could maximize the benefits of peak-shaving energy storage while delaying battery degradation. Compared with the traditional 80% capacity-based scrapping criterion, our efficiency-based scrapping criterion can significantly improve the lifetime benefit of the battery.

eess.SP↗