RASA enhances MoE model safety by repairing safety-critical experts and maintaining routing consistency.

Jiacheng Liang, Yuhui Wang, Tanqiu Jiang et al.

EMA anchor and Top-k KL improve RL training for LLMs, achieving 53.9% accuracy on math reasoning.

Lunjun Zhang, Jimmy Ba

SparVAR accelerates VAR models by exploiting sparsity, generating 1024x1024 images in just 1 second.

Zekun Li, Ning Wang, Tongxin Bai et al.

ImpRIF formalizes implicit reasoning as verifiable graphs, boosting instruction following by over 10% on benchmarks.

Yuancheng Yang, Lin Yang, Xu Wang et al.

Proxy compression enhances language model efficiency, significantly outperforming byte-level baselines.

Lin Zheng, Xinyu Li, Qian Liu et al.

Agent-Omit enhances LLM agent efficiency by adaptively omitting context; experiments show comparable performance to frontier methods.

Yansong Ning, Jun Fang, Naiqiang Tan et al.

Hierarchical taxonomy of data agents from L0 to L5, detailing autonomous capabilities and evolution.

Yuyu Luo, Guoliang Li, Ju Fan et al.

Supervised learning as lossy compression using finite blocklength analysis to reveal generalization and sample complexity.

Kosuke Sugiyama, Masato Uchida

Proposes a large-scale simulation-based framework using LLMs for preference-aligned proactive assistants with device-level personalization.

Ziyi Xuan, Yiwen Wu, Zhaoyang Yan et al.

BridgeV2W integrates embodiment masks into pretrained video models via ControlNet, improving multi-view robustness and cross-robot generalization.

Yixiang Chen, Peiyan Li, Jiabing Yang et al.

QVLA reduces VRAM usage to 29.2% while retaining 98.9% performance in LIBERO using channel-wise quantization.

Yuhao Xu, Yantai Yang, Zhenyang Fan et al.

AsymEP and Dyadic EP recover exact gradients in non-conservative systems; AsymEP reaches 94.9% on highly asymmetric MNIST networks.

Antonino Emanuele Scurria, Dimitri Vanden Abeele, Bortolo Matteo Mognetti et al.

Proposes basis rotation to mitigate gradient staleness in asynchronous pipeline parallelism, reducing training iterations by 81.7%.

Hyunji Jung, Sungbin Shin, Namhoon Lee

WGRPO pairs rare successes and failures, raising AIME 2025 Pass@8 from 16.8 to 22.2.

Yujuan Pang, Jiaxin Li, Xin Sheng et al.

Introduces SalamaBench, a benchmark using MLCommons taxonomy, evaluating 12 safety categories across 8170 prompts for Arabic models.

Omar Abdelnasser, Fatemah Alharbi, Khaled Khasawneh et al.

GRU-based control of a dual-segment continuum robot achieved position/orientation RMSEs of 1.11mm/4.62°, outperforming LSTM and others.

Yuancheng Shao, Yao Zhang, Jia Gu et al.

CompTok employs diffusion-based visual tokenization with InfoGAN objectives, enhancing compositionality and predictability of the token space, achieving SOTA in image generation.

Bingchen Zhao, Qiushan Guo, Ye Wang et al.

Agentic Proposing enhances LLM reasoning via compositional skill synthesis, achieving 91.6% accuracy on AIME25 with a 30B model.

Zhengbo Jiao, Shaobo Wang, Zifan Zhang et al.

LiDAR introduces a reward-guided sampling method using marginal samples and forward kernels, achieving 9.5× speedup over gradient guidance without neural backpropagation.

Yeongmin Kim, Donghyeok Shin, Byeonghu Na et al.

Introduces Contrastive Concept-Tree Search (CCTS), leveraging hierarchical semantic concepts to improve LLM-assisted algorithm discovery efficiency.

Timothee Leleu, Sudeera Gunathilaka, Federico Ghimenti et al.