RAIN-Merging is a gradient-free method that enhances instruction-following in large reasoning models while preserving their thinking format, with over 8% improvement in benchmarks.

Zhehao Huang, Yuhang Liu, Baijiong Lin et al.

Introduces MBT framework to enhance multi-hop QA accuracy and efficiency through a five-phase structure.

Ik-hwan Kim, Hyeongrok Han, Mingi Jung et al.

VeRO framework employs versioning and structured feedback to optimize coding agents, achieving up to 8% performance gains across tasks.

Varun Ursekar, Apaar Shanker, Veronica Chatrath et al.

FlowCorrect uses VR interface for human corrections, boosting robotic manipulation success to 80%.

Edgar Welte, Yitian Shi, Rosa Wolf et al.

LARFT combines length-aware reinforcement learning and hindsight length awareness, improving length instruction following by +20.92 points on average.

Wei Zhang, Lintong Du, Yuanhe Zhang et al.

UpSkill enhances LLM multi-attempt accuracy via Mutual Information Skill Learning, achieving ~3% improvement.

Devan Shah, Owen Yang, Daniel Yang et al.

Transformer-based foundation models enable full-stack transfer in robotics, covering language, vision, and motor skills.

Freek Stulp, Samuel Bustamante, João Silvério et al.

This study evaluates 8 state-of-the-art LLMs' calibration in predicting causal discovery algorithm performance, revealing widespread 'algorithmic blindness'.

Sohan Venkatesh, Ashish Mahendran Kurapath, Tejas Melkote

ProactiveMobile is a comprehensive benchmark for enhancing proactive intelligence on mobile devices, achieving a 20.82% success rate with Qwen2.5-VL-7B-Instruct.

Dezhi Kong, Zhengzhao Feng, Qiliang Liang et al.

Kareto uses simulation-based Pareto optimization to dynamically configure multi-tier storage for KV caches, improving efficiency.

Xianzhe Zheng, Zhengheng Wang, Ruiyan Ma et al.

PhysicEdit models physical state transitions with 38K video trajectories, improving realism by 5.9%.

Liangbing Zhao, Le Zhuo, Sayak Paul et al.

Proposed TSD-KD improves reasoning by combining preference-based indirect and confidence-based direct distillation, outperforming baselines by up to 54.4%.

Minsang Kim, Seung Jun Baek

LessMimic leverages Distance Field (DF) for long-horizon humanoid interaction, enabling reference-free inference and skill composition with 80-100% success across object scales.

Yutang Lin, Jieming Cui, Yixuan Li et al.

Developed a systematic boundary data analysis and correlator computation for self-dual Yang-Mills (SDYM) within AdS/CFT, using Fefferman-Graham expansion and boundary conditions.

Evgeny Skvortsov, Richard Van Dongen

Introduces MIGRASCOPE, an MI-based framework to evaluate and optimize RAG retrievers, revealing redundancy and synergy with specific metrics.

Wenqing Zheng, Dmitri Kalaev, Noah Fatsi et al.

A ReAct agent with VES* and VCES shows that GPT-4o can trade about 7% accuracy for up to 12.16× speed.

Germán T. Eizaguirre, Lars Tissen, Marc Sánchez-Artigas

Pancake employs multi-layer index caching, cross-agent index coordination, and GPU-CPU collaboration, achieving 4.29× throughput in multi-agent LLM storage scenarios.

Zhengding Hu, Zaifeng Pan, Prabhleen Kaur et al.

Proposes RoPE-ID method, enhancing long input handling by applying RoPE at high frequency to a subset of channels.

Davis Wertheimer, Aozhong Zhang, Derrick Liu et al.

NoRD achieves efficient vision-language-action learning on Waymo and NAVSIM with <60% data, no reasoning.

Ishaan Rawal, Shubh Gupta, Yihan Hu et al.

Scaling SSMs with tensor parallelism on multiple GPUs improves inference throughput.

Anurag Dutt, Nimit Shah, Hazem Masarani et al.