Introduces PhraseStereo, a stereo dataset for open-vocabulary phrase segmentation leveraging depth cues, with SSIM=0.601 and LPIPS=0.352 at optimal scale.

Thomas Campagnolo, Ezio Malis, Philippe Martinet et al.

EvolProver enhances automated theorem proving by evolving problems via symmetry and difficulty, achieving 53.8% pass@32.

Yuchen Tian, Ruiyuan Huang, Xuanwu Wang et al.

LAKAN integrates facial landmarks to dynamically modulate Kolmogorov-Arnold networks, boosting deepfake detection accuracy.

Jiayao Jiang, Bin Liu, Qi Chu et al.

Agent-ScanKit reveals multimodal agents' memory and reasoning via sensitivity perturbations, showing memory often outweighs reasoning.

Pengzhou Cheng, Lingzhong Dong, Zeng Wu et al.

This paper compares four reward models (dORM, dPRM, gORM, gPRM) across 14 multi-domain tasks, finding gORM most robust and effective.

Dong Bok Lee, Seanie Lee, Sangwoo Park et al.

This paper provides the first theoretical analysis of one-layer Mamba's training dynamics and robustness to outliers in in-context learning.

Hongkang Li, Songtao Lu, Xiaodong Cui et al.

ToV method quickly selects data by reversing train-validation roles, significantly reducing test loss.

Ayush Jain, Andrea Montanari, Eren Sasoglu

CWM: An open-weight large language model with world modeling, achieving 65.8% pass@1 on SWE-bench Verified.

FAIR CodeGen team, Jade Copet, Quentin Carbonneaux et al.

Ovi employs twin DiT backbones with blockwise bidirectional cross-attention and scaled-RoPE to unify audio-video generation, trained on hundreds of thousands of hours of raw data.

Chetwin Low, Weimin Wang, Calder Katyal

BC-MPPI enhances safety in path planning by integrating a Bayesian constraint layer, ensuring constraint adherence in complex environments.

Odichimnma Ezeji, Michael Ziegltrum, Giulio Turrisi et al.

AgentFlux uses decoupled fine-tuning with LoRA adapters, boosting local tool-calling accuracy by 46% on MCP-Bench.

Rohan Kadekodi, Zhan Jin, Keisuke Kamahori et al.

TGPO method improves task success rate by 31.6% in Signal Temporal Logic tasks.

Yue Meng, Fei Chen, Chuchu Fan

CRM links step rewards to outcomes via conditional hazards, reaching 43.3% on AIME24 in verifier-free RL, 16.7 points above PURE.

Zheng Zhang, Ziwei Shan, Kaitao Song et al.

fev-bench introduces a benchmark with 100 tasks across 7 domains, emphasizing covariates, using bootstrap confidence intervals for robust evaluation.

Oleksandr Shchur, Abdul Fatir Ansari, Caner Turkmen et al.

RE-Searcher combines goal-oriented planning and self-reflection, achieving state-of-the-art robustness and accuracy in complex search environments.

Daocheng Fu, Jianbiao Mei, Licheng Wen et al.

Mem-α optimizes memory construction via reinforcement learning, enhancing long-sequence processing.

Yu Wang, Ryuichi Takanobu, Zhiqi Liang et al.

Proposes SafeMind with a three-layer safety architecture, boosting safety rate by over 30% in embodied LLM agents evaluated on 5558 multimodal samples.

Ruolin Chen, Yinqian Sun, Jihang Wang et al.

Proposes RA3, a mid-training algorithm that learns action abstractions via temporal ELBO, improving code generation by 8 points on average.

Shenao Zhang, Donghan Yu, Yihao Feng et al.

Proposes SAC Flow, reparameterizing velocity with modern sequence models to stabilize flow policies, achieving state-of-the-art results.

Yixian Zhang, Shu'ang Yu, Tonghe Zhang et al.

Combining test-time training (TTT) with in-context learning (ICL), this work achieves low prediction risk for nonlinear single-index models, with error approaching noise levels as data grows.

Kento Kuwataka, Taiji Suzuki