Explore · 每周精选
发现最受关注的研究论文,追踪研究趋势,订阅感兴趣的期刊与关键词。
50 篇论文 · 按点赞排序
Bogdan Zagribelnyy, Ivan Ilin, Nikita Bondarev +5 authors
Top-K prompting and plausibility-aware training improve diverse reaction prediction in single-step retrosynthesis, yielding state-of-the-art results on a large verified reaction dataset and motivating ensemble systems.
Zhiyuan He, Siwei Zhang, Zhiwen Zhou +7 authors
Agent Lightning v1.0 enables reproducible reinforcement learning for arbitrary agent harnesses, substantially improving coding-agent performance with minimal data and compute.
Haonan He, Haodi Lei, Yun Luo +13 authors
On-policy distillation from a long-context reasoning teacher to short-context students improves mathematical proof reasoning and generalizes to science benchmarks by aligning token spans, constraining length growth, and stabilizing training.
Mengru Wang, Haozhe Luo, Zhenqian Xu +6 authors
Retrieved memories can induce reasoning errors and belief distortions in large language models, and an inference-time strategy helps avoid these cognitive traps while maintaining benchmark performance.
Kai Chen, Jifeng Ding, Ning Ding +44 authors
Mobius-v0 separates global memory storage from iterative reasoning modules to improve knowledge compression and inference efficiency, yielding comparable performance with less training data and faster inference.
Zian Meng, Zhen Li, Chuanhao Li +2 authors
Marionette predicts explicit 3D articulated world states for interactive games, uses a fixed renderer for geometry, and synthesizes video via diffusion, enabling direct state-level control and long-horizon consistency repair.
Peter Schneider-Kamp, Jacob Nielsen, Gianluca Barmina +2 authors
Mimir v1 is a 1-billion-parameter Hierarchical Reasoning Model trained solely on permissible data that achieves competitive English results and state-of-the-art Danish performance across multiple benchmarks.
Qianxi Yan, Chunrong Chen, Jiuzhou Zhao +3 authors
SkillEvo improves agent skills through multi-turn feedback and active governance to sustain evolution gradients.
Adrian Hayler, Klemens Flöge, Alan Arazi +44 authors
Prior Labs released open-source tools including a unified relational benchmark framework, a TabPFN-based relational model, and a model-agnostic predictive interface to advance reproducible relational learning.
Minghui Guo, Shengqiong Wu, Hao Fei
V-RAE constructs semantically organized video latents from frozen vision representations to improve generation quality, convergence speed, and predictive modeling.
Yuhao Zhan, Bingxiang He, Zecong Tang +1 authors
PACE-Bench evaluates self-evolving agents on physics adaptation tasks requiring iterative code redesign after environmental mutations, revealing that simulator-grounded reflection outperforms unverified self-revision but mechanism redesign remains a major bottleneck.
Harold Haodong Chen, Zhiyu Hou, Wen-Jie Shu +4 authors
GenRouter is a unified routing framework that adaptively directs prompts to optimal agentic image-generation workflows, cutting costs and latency while improving visual alignment and enabling continuous self-evolution.
Zhaoyan Sun, Xiaoxiao Wang, Guoliang Li
An ACID-compliant framework for agentic transactions introduces semantic guarantees to ensure reliable, isolated, and durable execution of long-horizon LLM agent workflows.
Xinle Deng, Yida Xue, Xiangyuan Ru +14 authors
MobileMem is a benchmark and framework for evaluating on-device long-term memory through year-scale, multimodal mobile experience trajectories that require temporal reasoning, knowledge updating, and preference inference.
Yiming Du, Yuxin Jiang, Tao Yuan +9 authors
LEGO-RL connects native coding-agent harnesses to scalable policy-gradient training via in-process LLM proxying, sandbox orchestration, and integrated monitoring, improving sparse MoE model performance across multiple harnesses.
Guibin Zhang, Jiayang Lyu, Ran Sun +4 authors
Latent On-Policy Self-Distillation learns privileged teaching context end-to-end from experience to provide dense token-level supervision, improving agent performance and efficiency.
Xinye Li, Lingshuai Lin, Lei Wang +8 authors
ForgeWM progressively distills bidirectional video generators into efficient few-step interactive world models with aligned discrete and continuous controls, supporting low-latency interaction and replay-time refinement.
Jiayi Song, Shijie Huang, Fangtai Wu +5 authors
EditBridge enables efficient ultra high-resolution image editing via a diffusion bridge that translates low-resolution edits to high-resolution outputs while preserving source details through sparse attention.
Silin Chen, Haoyi Teng, Xiaodong Gu +5 authors
Repo0 uses a dual-graph architectural state and modularity-guided structural evolution to generate complete software repositories from natural-language requirements with high functionality coverage.
Andrei Cristian Popescu, Haitz Sáez de Ocáriz Borde, Pietro Liò
Looped language models improve compositional, multi-step tool use through recurrent computation, with adaptive inference balancing accuracy and compute cost.
北京市昌平区探索星信息技术及软件开发工作室
京ICP备2026059466号