Introduces O-Voxel and a 4B-parameter flow model for high-fidelity 3D generation with complex topology and detailed appearance, achieving fast inference and superior quality.

Jianfeng Xiang, Xiaoxue Chen, Sicheng Xu et al.

ART employs a category-agnostic transformer to reconstruct complete 3D articulated objects from sparse multi-state RGB images, predicting geometry, texture, and articulation parameters.

Zizhang Li, Cheng Zhang, Zhengqin Li et al.

ParaFormer integrates PageRank-enhanced attention to mitigate over-smoothing in deep graph transformers, achieving superior results on large-scale datasets.

Chaohao Yuan, Zhenjie Song, Ercan Engin Kuruoglu et al.

Multi-agent system with role-specific LLMs uses Kendall's W for consensus, enhanced by reinforcement learning for optimal oncology treatment recommendations.

Xudong Han, Xianglun Gao, Xiaoyi Qu et al.

FLAME combines Legendre Memory and normalization flow for lightweight, probabilistic time series forecasting with state-of-the-art zero-shot performance.

Xingjian Wu, Hanyin Cheng, Xiangfei Qiu et al.

RoboTracer employs 3D-aware VLM with reinforcement fine-tuning, achieving 79.1% success in spatial reasoning tasks.

Enshen Zhou, Yibo Li, Jingkun An et al.

Proposes a unified 'forms-functions-dynamics' framework for agent memory, covering token-level, parametric, and latent types with detailed functional and dynamic analysis.

Yuyang Hu, Shichun Liu, Yanwei Yue et al.

A GPT-2-style Transformer plus genetic search found at least six new F8 champion codes, with 92.4% test accuracy within ±3 distance units.

Yang-Hui He, Alexander Kasprzyk, Q Le et al.

BlossomRec enhances recommendation efficiency with block-level sparse attention, reducing memory usage.

Mengyang Ma, Xiaopeng Li, Wanyu Wang et al.

FIN-bench-v2 evaluates Finnish LLMs using a unified benchmark suite, validated with a 2.15B parameter model.

Joona Kytöniemi, Jousia Piha, Akseli Reunamo et al.

MedInsightBench evaluates multimodal medical data analysis; MedInsightAgent enhances LMM performance.

Zhenghao Zhu, Chuxue Cao, Sirui Han et al.

LINA predicts prompt-specific interventions to improve causal consistency and out-of-distribution instruction following in diffusion models, outperforming SOTA.

Shu Yu, Chaochao Lu

AutoTool employs a dual-phase optimization with PL ranking to enable dynamic tool selection, improving reasoning accuracy by 6.4% across ten benchmarks.

Jiaru Zou, Ling Yang, Yunzhe Qi et al.

Introduced Next Scene Prediction task using Qwen-VL and LTX, achieving 73% causal consistency.

Xinjie Li, Zhimin Chen, Rui Zhao et al.

Using formal logic and Kolmogorov complexity, the paper proves current LLMs lack consciousness, emphasizing the necessity of continual learning to overcome the falsifiability barrier.

Erik Hoel

DrivePI: A spatial-aware 4D MLLM that enhances autonomous driving understanding, perception, prediction, and planning, surpassing existing models.

Zhe Liu, Runhui Huang, Rui Yang et al.

JointAVBench introduces a multi-dimensional, automated benchmark for evaluating joint audio-visual reasoning of large models, covering five cognitive dimensions and four audio types.

Jianghan Chao, Jianzhang Gao, Wenhui Tan et al.

CoRe3D introduces a unified 3D reasoning framework combining semantic Chain-of-Thought and spatial geometric reasoning, significantly improving 3D understanding and generation metrics.

Tianjiao Yu, Xinzhuo Li, Yifan Shen et al.

FysicsWorld offers a full-modality benchmark supporting bidirectional I/O across image, video, audio, and text, covering 16 tasks.

Yue Jiang, Dingkang Yang, Minghao Han et al.

NL2Repo-Bench evaluates long-horizon repository generation of coding agents, with average pass rates below 40%, highlighting significant challenges.

Jingzhe Ding, Shengda Long, Changxin Pu et al.