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。
606 篇论文 · 按点赞排序
Yibin Wang, Yuhang Zang, Hao Li +2 authors
A unified reward model for multimodal understanding and generation improves the assessment and preference alignment in image and video tasks through joint learning and direct preference optimization.
Taewoong Kang, Kinam Kim, Dohyeon Kim +3 authors
EgoX framework generates egocentric videos from exocentric inputs using video diffusion models with LoRA adaptation, unified conditioning, and geometry-guided self-attention for coherence and visual fidelity.
Yonghao Zhuang, Junda Chen, Bo Pang +6 authors
CAD, a technique for long-context large language model training, improves throughput and balance by decoupling and distributing core attention computations.
Minju Seo, Jinheon Baek, Seongyun Lee +1 authors
PaperCoder is a multi-agent LLM framework that converts machine learning papers into functional code repositories through planning, analysis, and generation stages.
Yuxin Zuo, Kaiyan Zhang, Shang Qu +7 authors
Test-Time Reinforcement Learning (TTRL) enhances Large Language Models (LLMs) using unlabeled data through reinforcement learning, improving performance across tasks.
Zhaoyang Yu, Jiayi Zhang, Huixue Su +10 authors
ReCode, a recursive code generation paradigm, unifies high-level planning and low-level action in a single representation, enhancing decision granularity and data efficiency in LLM-based agents.
Guochao Jiang, Wenfeng Feng, Guofeng Quan +4 authors
A curriculum reinforcement learning framework dynamically adjusts training sample difficulty based on reward variance, improving LLM performance on mathematical reasoning tasks.
Valentin Khrulkov, Andrey Galichin, Denis Bashkirov +5 authors
GigaEvo is an open-source framework for LLM-guided evolutionary computation, offering modular and concurrent tools for research and experimentation in solving complex optimization problems.
Andrey Galichin, Alexey Dontsov, Polina Druzhinina +4 authors
A study uses Sparse Autoencoders to identify and enhance reasoning features in Large Language Models, providing insights into the mechanisms driving their reasoning capabilities.
Vera Soboleva, Aibek Alanov, Andrey Kuznetsov +1 authors
T-LoRA, a timestep-dependent low-rank adaptation framework, enhances diffusion model personalization with a dynamic fine-tuning strategy and orthogonal initialization, improving concept fidelity and text alignment in data-limited settings.
Zefan Cai, Haoyi Qiu, Tianyi Ma +9 authors
MMGR evaluates video and image models across reasoning abilities in multiple domains, revealing performance gaps and highlighting limitations in perceptual data and causal correctness.
NovelSeek Team, Bo Zhang, Shiyang Feng +22 authors
Artificial Intelligence (AI) is accelerating the transformation of scientific research paradigms, not only enhancing research efficiency but also driving innovation. We introduce NovelSeek, a unified closed-loop multi-agent framework to conduct Autonomous Scientific Research (ASR) across various scientific research fields, enabling researchers to tackle complicated problems in these fields with unprecedented speed and precision. NovelSeek highlights three key advantages: 1) Scalability: NovelSeek has demonstrated its versatility across 12 scientific research tasks, capable of generating innovative ideas to enhance the performance of baseline code. 2) Interactivity: NovelSeek provides an interface for human expert feedback and multi-agent interaction in automated end-to-end processes, allowing for the seamless integration of domain expert knowledge. 3) Efficiency: NovelSeek has achieved promising performance gains in several scientific fields with significantly less time cost compared to human efforts. For instance, in reaction yield prediction, it increased from 27.6% to 35.4% in just 12 hours; in enhancer activity prediction, accuracy rose from 0.52 to 0.79 with only 4 hours of processing; and in 2D semantic segmentation, precision advanced from 78.8% to 81.0% in a mere 30 hours.
Kaitao Song, Xiaohua Wang, Xu Tan +14 authors
A novel Chain-of-Model framework introduces hierarchical hidden state chains in Transformers to improve scaling efficiency and inference flexibility for language models.
Khalil Hennara, Sara Chrouf, Mohamed Motaism Hamed +3 authors
A method for integrating a new language into a large language model enhances performance in the target language while preserving existing knowledge.
Zhichao Wang, Dongyang Ma, Xinting Huang +6 authors
AutoDeco, a novel architecture, enables end-to-end generation by dynamically predicting decoding parameters, outperforming default strategies and achieving performance close to an oracle-tuned baseline.
Wanghan Xu, Yuhao Zhou, Yifan Zhou +104 authors
A framework for Scientific General Intelligence (SGI) is presented, evaluated using SGI-Bench, and improved with Test-Time Reinforcement Learning, highlighting gaps in existing models' scientific capabilities.
Dmitrii Stoianov, Danil Taranets, Olga Tsymboi +12 authors
T-pro 2.0 is an open-weight Russian LLM for hybrid reasoning and efficient inference, using a Cyrillic-dense tokenizer and EAGLE speculative-decoding pipeline.
Ning Shang, Yifei Liu, Yi Zhu +12 authors
rStar2-Agent, a 14B math reasoning model trained with agentic reinforcement learning, achieves state-of-the-art performance by efficiently handling complex problem-solving with advanced cognitive behaviors and minimal computational resources.
Zeyu Zhu, Kevin Qinghong Lin, Mike Zheng Shou
PaperTalker is a multi-agent framework that automates academic presentation video generation by integrating slide generation, layout refinement, subtitling, speech synthesis, and talking-head rendering, outperforming existing methods.
Jintao Zhang, Haoxu Wang, Kai Jiang +10 authors
SLA, a trainable attention method combining sparse and linear attention, accelerates Diffusion Transformer models for video generation with minimal quality loss.
Junkang Wu, Kexin Huang, Jiancan Wu +3 authors
Quantile Advantage Estimation stabilizes reinforcement learning with verifiable rewards by addressing entropy issues and improving performance on large language models.
Xiao Liang, Zhongzhi Li, Yeyun Gong +4 authors
An online self-play strategy with variational problem synthesis for RLVR training maintains policy entropy and improves Pass@k performance on reasoning benchmarks.
Yuchen Shi, Yuzheng Cai, Siqi Cai +15 authors
Existing Large Language Model (LLM) agent frameworks face two significant challenges: high configuration costs and static capabilities. Building a high-quality agent often requires extensive manual effort in tool integration and prompt engineering, while deployed agents struggle to adapt to dynamic environments without expensive fine-tuning. To address these issues, we propose Youtu-Agent, a modular framework designed for the automated generation and continuous evolution of LLM agents. Youtu-Agent features a structured configuration system that decouples execution environments, toolkits, and context management, enabling flexible reuse and automated synthesis. We introduce two generation paradigms: a Workflow mode for standard tasks and a Meta-Agent mode for complex, non-standard requirements, capable of automatically generating tool code, prompts, and configurations. Furthermore, Youtu-Agent establishes a hybrid policy optimization system: (1) an Agent Practice module that enables agents to accumulate experience and improve performance through in-context optimization without parameter updates; and (2) an Agent RL module that integrates with distributed training frameworks to enable scalable and stable reinforcement learning of any Youtu-Agents in an end-to-end, large-scale manner. Experiments demonstrate that Youtu-Agent achieves state-of-the-art performance on WebWalkerQA (71.47\%) and GAIA (72.8\%) using open-weight models. Our automated generation pipeline achieves over 81\% tool synthesis success rate, while the Practice module improves performance on AIME 2024/2025 by +2.7\% and +5.4\% respectively. Moreover, our Agent RL training achieves 40\% speedup with steady performance improvement on 7B LLMs, enhancing coding/reasoning and searching capabilities respectively up to 35\% and 21\% on Maths and general/multi-hop QA benchmarks.
Zhiyuan Hu, Yibo Wang, Hanze Dong +5 authors
Explicit alignment of large reasoning models with deduction, induction, and abduction through a three-stage pipeline improves scalability and reliability in reasoning tasks.
Xuekai Zhu, Daixuan Cheng, Dinghuai Zhang +20 authors
FlowRL enhances LLM reinforcement learning by matching the full reward distribution through flow balancing, improving diversity and performance over reward-maximizing methods.
Liangcai Su, Zhen Zhang, Guangyu Li +19 authors
AgentFounder, a deep research agent model incorporating Agentic Continual Pre-training, achieves state-of-the-art performance in agentic tasks while maintaining strong tool-use ability.
Elisei Rykov, Kseniia Petrushina, Maksim Savkin +6 authors
PsiloQA, a multilingual dataset with span-level hallucinations, enhances hallucination detection in large language models across 14 languages using an automated pipeline and encoder-based models.
Wenhan Liu, Xinyu Ma, Weiwei Sun +4 authors
A reasoning-intensive reranker, ReasonRank, achieves state-of-the-art performance in passage ranking tasks by using synthesized training data and a two-stage post-training approach with reinforcement learning.
Yufeng Cui, Honghao Chen, Haoge Deng +20 authors
Emu3.5, a large-scale multimodal world model, predicts next states in vision and language, enhanced with reinforcement learning and Discrete Diffusion Adaptation for efficient inference, achieving strong performance in various multimodal tasks.
SAM 3D Team, Xingyu Chen, Fu-Jen Chu +20 authors
SAM 3D is a generative model that reconstructs 3D objects from single images using a multi-stage training framework that includes synthetic pretraining and real-world alignment, achieving high performance in human preference tests.
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