arXiv subjects

Minsu Oh

Publications and source records attributed to Minsu Oh.

5 recordsLinked to original sources

arXiv2026

Less but Sufficient: Relation Knowledge Selects Compact Knowledge-Graph Contexts, and What External Checks Can and Cannot Catch

Large language models (LLMs) read knowledge-graph facts ranked by relevance, leaving open whether they suffice or conflict. Our checks use relation knowledge, written by hand or learned from training questions labelled with their relation paths. It selects a compact context of facts yielding an answer, and derives it; a derivability check tests whether an answer follows, a consistency check whether facts conflict, and a validation-chosen routing gate decides where the derived answer replaces the LLM's. Seven open LLMs read the compact context better in exact match than relevance-ranked facts; on ComplexWebQuestions (CWQ) they match SubgraphRAG's facts at a quarter of the prompt tokens, and stronger LLMs read SubgraphRAG's better. Where the routing gate passes, ours leads SubgraphRAG's in exact match in all 53 LLM, benchmark and prompt settings. A precision gate, returning the derived answer only when the check passes with exactly one derived entity, answered 18 and 5 percent of held-out WebQSP and CWQ questions in a pre-specified test, right for 90 and 96 percent. The consistency check judges facts by path evidence only where another path joins their entities, true of 2.5 percent of named-entity facts in our compact WebQSP and CWQ contexts.

cs.CL↗

arXiv2026

Collection-Date Assignment Redirects Answers in Temporal QA

A collection date records when a source revision was acquired, not when its contents changed. Can assigning this metadata redirect temporal question answering despite unchanged evidence? Swapping the question day and following day between two revisions lowers later-reference exact match (EM) by 37.6-51.4 percentage points across five open-weight LLMs on 595 Wikipedia pages, replicated on 128 disjoint pages. Previous-day swaps retain 29.0-39.6-point losses on 497 pages. Seven prompt variants, including explicit ignore-collection instructions and archived-revision provenance, retain the effect. Fictional answers and English month-name question dates show that remembered answers and identical date strings are unnecessary. Matched thinking reduces Qwen3.5's penalty from 30.1 to 11.1 points, while native reasoners vary. Two API LLMs show small Wikipedia swap effects and rarely shift to earlier answers on fictional records. Large swap-induced losses concentrate around the question date; swaps between distant dates have much smaller effects. Adding fields containing one observed acquisition date lowers EM relative to omission in three of five LLMs after correction. The diagnostic isolates metadata effects from changing evidence or missing the relevant revision, and provides a practical audit for temporal QA inputs.

cs.CL↗

arXiv2026

Evaluating Compressed Memory: Dependence, Utility, and Content Benefit

Does a language model answer better because its compressed memory contains useful document information? Replacing memory vectors with zeros can lower a trained model's score without showing that it outperforms a model trained without memory. We combine zeroing with separately trained models that cannot access stored text or receive the same constant input for every document. Across nine systems, some models rely on memory yet perform worse than the trained alternative. For example, in experiments that predict states in generated stories, zeroing memory worsens a model's prediction loss, yet a model trained without memory achieves lower loss than the original model with intact memory. On TriviaQA, giving xRAG memory from another question's passage raises F1, which measures word overlap with accepted answers, from 19.38 without a document to 24.78. Its effect on how often outputs contain an accepted answer is statistically uncertain. The question's retrieved passage scores higher still (29.88 F1). Thus a gain over no document alone does not establish that the gain requires the question's passage. Evaluations must therefore specify the comparison model, absolute performance, and generation and scoring conditions.

cs.CL↗

arXiv2026

Shared Stopping Decisions Change Answers in HQQ Cache Quantization

Language-model systems batch questions for throughput, but unrelated questions should not change a target's answer when its input and numerical execution are fixed. We study compression of the key and value cache, which stores attention representations reused during generation. With request-local groups, Transformers' Half-Quadratic Quantization (HQQ) backend updates compression parameters separately but uses a shared average error to decide when all updates stop. Replacing only the question batched with the target changes four-bit HQQ answers in 170/384 test comparisons across two models. Replaying the other execution's update counts reproduces its complete answer and cache fingerprints in every changed pair, in both directions. Computing the stopping mean in FP32 reduces cache differences but leaves answer changes. Native HQQ also changes confirmed numerical correctness in eight arithmetic pairs. Fixed iterations and request-local stopping remove observed companion dependence under matched controls. Request-local stopping remains sensitive to synthetic padding changes at the tensor level. Fixing the original iteration budget removes this decision path without tuning. Neither repair has an established quality advantage, and natural rebatching still changes answers. Request-independence audits must cover stopping decisions as well as quantization groups.

cs.CL↗

arXiv2023

Optical Structures for Thermophotovoltaic Emitters: Power Generation from Waste Heat

Heat is an inevitable outcome in energy consumption processes, as more than 65% of input energy is wasted as heat. If we can generate electricity from waste heat, it will help minimize the needs for fossil fuels in power plants and can reduce carbon emissions. One way to generate power from heat is through the use of thermophotovoltaics (TPVs), where photons radiated from thermal emitters are converted into electricity. To optimize TPV performance, it is crucial to design emitters such that their emissivity spectrum matches their operating temperature. For example, higher emissivity is needed at shorter (longer) wavelengths at higher (lower) temperatures. Thus, having the ability to create wavelength-selective emitters can enable TPV applications for a wider range of temperatures. This research focuses on utilizing metamaterials (2D emitters) and planar thin films (1D emitters) to create those emitters. Simulation, fabrication, material property analysis, and radiation measurements were used to characterize the emitters. Based on simulation, metamaterial emitters exhibit engineerable emissivity due to various mechanisms of their optical resonance. Also, large-area fabrication of 1D emitters (78 cm2) was achieved owing to their simple structure, which is required to produce higher TPV power output. Incorporating the characteristics of emitters of each type, their advantages and challenges are discussed. Therefore, the comprehensive results of this research help realize practical implementation of TPV applications.

physics.optics↗