ReflecTool framework enhances clinical agents' tool usage in the medical domain, outperforming pure LLMs by over 10 points in experiments.

Yusheng Liao, Shuyang Jiang, Yanfeng Wang et al.

LongVU employs spatiotemporal adaptive compression using DINOv2 and cross-modal queries, enabling efficient long video understanding within limited context.

Xiaoqian Shen, Yunyang Xiong, Changsheng Zhao et al.

Constructed a large-scale dataset with 1.9 million music versions using Discogs metadata, improving version identification performance.

R. Oguz Araz, Xavier Serra, Dmitry Bogdanov

PyramidDrop reduces visual redundancy layer-wise, achieving 40% training time and 55% FLOPs inference acceleration with minimal performance loss.

Long Xing, Qidong Huang, Xiaoyi Dong et al.

VoiceBench benchmark evaluates LLM-based voice assistants across diverse real-world scenarios, revealing performance gaps in robustness and safety, with detailed multi-metric analysis.

Yiming Chen, Xianghu Yue, Chen Zhang et al.

Interchangeable Token Embeddings combine shared semantics with random identity, reaching 95.94% LTL correctness and extending vocabularies.

İlker Işık, Ramazan Gokberk Cinbis, Ebru Aydin Gol

Incorporating known group invariances into Gaussian process kernels enhances Bayesian optimization sample efficiency, with theoretical bounds and practical validation.

Theodore Brown, Alexandru Cioba, Ilija Bogunovic

Proposes Scene Language, a multimodal, program-based scene representation inferred from pre-trained models, enabling high-fidelity 3D/4D scene generation and editing without training.

Yunzhi Zhang, Zizhang Li, Matt Zhou et al.

TIPS combines spatial awareness via synthetic captions and self-supervised masked modeling, boosting dense image understanding.

Kevis-Kokitsi Maninis, Kaifeng Chen, Soham Ghosh et al.

Proposed API-calling and Hybrid Agents, API Agent improved by 24.0% on WebArena, achieving 38.9% success rate.

Yueqi Song, Frank Xu, Shuyan Zhou et al.

Reflection-Bench, inspired by cognitive psychology, evaluates LLMs' epistemic agency across seven dimensions, revealing significant limitations in meta-reflection with scores below 50 in many models.

Lingyu Li, Yixu Wang, Haiquan Zhao et al.

Using ALTI, this study analyzes layer-wise token contributions in LLM-based MT, revealing source dominance and positional biases.

Emmanouil Zaranis, Nuno M. Guerreiro, André F. T. Martins

Proposed TMCHT task and ARCJ method achieve 52.93% attack success in multi-agent systems.

Tianyi Men, Pengfei Cao, Zhuoran Jin et al.

Systematic psycholinguistic evaluation of language models reveals partial sensitivity to argument roles, with models distinguishing plausible from implausible verbs but lacking human-like selective patterns.

Eun-Kyoung Rosa Lee, Sathvik Nair, Naomi Feldman

Differentiable simulation accelerates quadrotor control training, achieving recovery in seconds with minimal samples, combining simple models and visual features.

Johannes Heeg, Yunlong Song, Davide Scaramuzza

Enhance Dense Passage Retrieval with Entailment Tuning, achieving 3% improvement on NQ dataset.

Lu Dai, Hao Liu, Hui Xiong

MiCEval introduces a fine-grained, multi-metric framework for automated evaluation of multimodal chain-of-thought, outperforming cosine similarity and fine-tuning metrics.

Xiongtao Zhou, Jie He, Lanyu Chen et al.

GenEOL leverages LLMs to generate diverse sentence variations, improving embeddings by averaging, with an average gain of 2.85 points on STS benchmarks.

Raghuveer Thirukovalluru, Bhuwan Dhingra

JAMUN employs walk-jump sampling in a noised space to generate protein conformations 10x faster than MD, with strong transferability.

Ameya Daigavane, Bodhi P. Vani, Darcy Davidson et al.

A Gaussian process-based approach for causal intervention function uncertainty quantification, improving posterior coverage and calibration.

Hugh Dance, Peter Orbanz, Arthur Gretton