arXiv subjects

Chiyoung Kim

Publications and source records attributed to Chiyoung Kim.

3 recordsLinked to original sources

arXiv2026

A Tree-based RAG Framework for Evidence-Intensive QA via Adaptive Planning and Topology-Aware Evidence Gathering

Recent structured RAG methods leverage tree- or graph-based reasoning structures to improve multi-hop QA. However, they face key limitations in evidence-intensive QA, where answering a question requires synthesizing information scattered across dozens or even hundreds of documents: structural rigidity, which limits adaptive reasoning expansion, and topology-ignorant evidence gathering, which prevents effective integration of evidence across different reasoning nodes. To address these issues, we propose APT-RAG, an Adaptive Planning and Topology-aware evidence gathering RAG framework. Adaptive planning dynamically expands the reasoning structure based on question dependencies and evidence requirements, while topology-aware evidence gathering improves evidence coverage through sibling evidence reuse, direct retrieval, and evidence aggregation from child nodes. We further introduce evidence-guided batched answer generation to reduce significant generation overhead in evidence-intensive QA. In the experiments on evidence-intensive QA benchmarks, APT-RAG outperforms existing structured RAG methods. Our code is available at https://github.com/hyudsl/APT-RAG.

cs.AI↗

arXiv2026

CoTinyVLA: Chain-of-Thought Distillation for a Sub-Billion-Parameter Vision-Language-Action Model

Vision-Language-Action (VLA) models translate natural-language commands into robot action sequences, but leading systems on the LIBERO-Plus robustness benchmark use three- to seven-billion-parameter backbones whose memory demands can exceed embedded robotic budgets. We present CoTinyVLA, a 0.9B-parameter action model on a Qwen3.5-0.8B backbone that obtains that robustness by structuring supervision instead of enlarging the model. Three components target different axes of the problem: dual-view temporal input of 16 history frames per step with textual camera and time markers; hierarchical chain-of-thought (CoT) distillation from a 35B teacher into an episode-level Plan and a chunk-level Think span over task phase, gripper state and next subaction; and paraphrase augmentation expanding 40 base commands into 800 variants. On LIBERO-Plus, spanning 10,030 perturbed tasks across seven perturbation dimensions, CoTinyVLA reaches 90.8% on Spatial, 87.3% on Object, 86.6% on Goal and 80.7% on Long, leading the strongest 7B baseline on all four suites by 4.7, 2.8, 15.9 and 3.0 points, with every margin interval excluding zero. The gains concentrate on the hardest axes of the benchmark: across the eleven published baselines none exceeds 53.2% on Robot Initial States in any suite, whereas CoTinyVLA reaches 73.6% on Goal against 39.9% for the strongest baseline. Ablations show the three components to be separable by perturbation axis, and at a matched image budget how frames are divided between the two cameras and across time accounts for 8.6 points on its own. Closed-loop inference peaks at 2.25 GiB of allocated GPU memory, and paired interventions show the episode Plan to be load-bearing: replacing it with an empty or contradictory span costs 40 to 45 points of success. Structured supervision thus lets a 0.9B backbone exceed all of them. Code: https://github.com/BrainJellyPie/CoTinyVLA

cs.AI↗

arXiv2025

Effects of Yttrium Doping on Oxygen Conductivity in Ba(Fe, Co, Zr, Y)O_{3-δ} Cathode Materials for Proton Ceramic Fuel Cells

Proton ceramic fuel cells (PCFCs) achieve high efficiency at reduced operating temperatures, but their performance is often limited by slow oxygen reduction reaction (ORR) kinetics at the cathode. The BaCoFeZrY (BCFZY) perovskite family is a promising triple-conducting air-electrode material, yet the role of Y dopants in governing oxygen transport remains unclear. In this study, we examine the effect of Y content on oxygen conductivity in three compositions: BCFZ, BCFZY0.1, and BCFY. Oxygen conductivity was evaluated from the product of oxygen tracer diffusivity and oxygen defect concentration. Ab initio molecular dynamics simulations were used to determine tracer diffusivity and migration energies, while defect concentrations were estimated from reference data. Y doping slightly decreases oxygen conductivity from BCFZ to BCFZY0.1, from 337 to 203 mS/cm at 500 C, with activation energies of 0.155 and 0.172 eV. BCFY shows much lower conductivity (99 mS/cm) and a higher activation energy of 0.261 eV. Computed conductivities are higher and more Arrhenius-like than experimental values, suggesting that microstructural features such as grain boundaries strongly limit oxygen transport in real materials. A series-circuit model combining bulk conductivity and fitted grain-boundary parameters provides semi-quantitative agreement with experiment. These results clarify the role of Y doping in oxygen transport and provide insight for optimizing cathode performance in PCFCs.

cond-mat.mtrl-sci↗