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。
Adrian Kosowski, Przemysław Uznański, Jan Chorowski +2 authors
BDH, a biologically inspired Large Language Model, combines scale-free network architecture with Hebbian learning to achieve Transformer-like performance while maintaining interpretability.
50 篇论文 · 按点赞排序
Shuai Yang, Wei Huang, Ruihang Chu +9 authors
LongLive is a frame-level autoregressive framework for real-time and interactive long video generation, addressing efficiency and quality challenges through causal attention, KV-recache, streaming long tuning, and short window attention.
Zijian Wu, Xiangyan Liu, Xinyuan Zhang +12 authors
MCPMark is a comprehensive benchmark for evaluating MCP use in real-world workflows, featuring diverse tasks that require richer interactions with the environment, and reveals that current LLMs perform poorly on these tasks.
Junbo Niu, Zheng Liu, Zhuangcheng Gu +58 authors
MinerU2.5, a 1.2B-parameter document parsing vision-language model, achieves state-of-the-art recognition accuracy with computational efficiency through a coarse-to-fine parsing strategy.
Fang Wu, Weihao Xuan, Heli Qi +4 authors
DeepSearch integrates Monte Carlo Tree Search into RLVR training to enhance exploration and credit assignment, achieving state-of-the-art performance with reduced computational cost.
Qinsi Wang, Bo Liu, Tianyi Zhou +6 authors
Vision-Zero is a domain-agnostic framework that enhances vision-language models through self-improvement in competitive visual games, using Iterative Self-Play Policy Optimization and achieving state-of-the-art performance without human annotation.
Xu Wujiang, Wentian Zhao, Zhenting Wang +6 authors
Entropy-regularized Policy Optimization (EPO) addresses exploration-exploitation challenges in multi-turn environments with sparse rewards, improving performance in tasks like ScienceWorld and ALFWorld.
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.
Yuling Shi, Yichun Qian, Hongyu Zhang +2 authors
LongCodeZip is a code compression framework for LLMs that uses dual-stage compression to reduce context size without degrading performance, improving efficiency in code intelligence applications.
Justin Cui, Jie Wu, Ming Li +6 authors
A method is proposed to enhance long-horizon video generation by using sampled segments from self-generated long videos to guide student models, maintaining quality and consistency without additional supervision or retraining.
Zichen Liu, Anya Sims, Keyu Duan +16 authors
GEM, an open-source environment simulator, facilitates experience-based learning for large language models by providing a standardized framework and diverse environments for training and benchmarking reinforcement learning algorithms.
Runzhe Zhan, Yafu Li, Zhi Wang +5 authors
ExGRPO, a framework that prioritizes valuable reasoning experiences, improves and stabilizes reinforcement learning from verifiable rewards for large language models.
Xinyu Tian, Shu Zou, Zhaoyuan Yang +5 authors
VAPO-Thinker-7B enhances multimodal reasoning by anchoring the process to visual information, improving performance on visual tasks while maintaining logical inference.
Lorenz K. Müller, Philippe Bich, Jiawei Zhuang +3 authors
SINQ enhances post-training quantization by introducing a second-axis scale factor and Sinkhorn-Knopp-style algorithm to minimize matrix imbalance, improving perplexity on large language models.
Renjie Luo, Zichen Liu, Xiangyan Liu +5 authors
Feedback-conditional policy (FCP) enables LLMs to learn from verbal feedback by treating it as a conditioning signal, improving expressiveness over scalar rewards.
Xiangxin Zhou, Zichen Liu, Haonan Wang +5 authors
A variational reasoning framework treats thinking traces as latent variables, optimizing them through variational inference to improve language model reasoning.
Hengtao Li, Pengxiang Ding, Runze Suo +8 authors
VLA-RFT uses a data-driven world model to fine-tune VLA models efficiently, reducing sample requirements and improving robustness under perturbations.
Shaobo Wang, Jiaming Wang, Jiajun Zhang +8 authors
Quadrant-based Tuning (Q-Tuning) optimizes both sample and token pruning in supervised fine-tuning of large language models, achieving superior performance with reduced data.
Yuhan Song, Linhao Zhang, Chuhan Wu +4 authors
StableToken, a multi-branch consensus-driven tokenizer, enhances token stability and robustness in speech processing, improving SpeechLLMs' performance under noisy conditions.
Hyun Ryu, Doohyuk Jang, Hyemin S. Lee +16 authors
An automated engine evaluates the factuality of review points in AI conference papers, demonstrating moderate agreement with human experts and higher accuracy at the premise level.
Fang Wu, Xu Huang, Weihao Xuan +8 authors
Multiplayer Nash Preference Optimization (MNPO) extends Nash learning from human feedback to handle complex, non-transitive human preferences by formulating alignment as an n-player game.
Bo-Hsu Ke, You-Zhe Xie, Yu-Lun Liu +1 authors
A novel density-guided poisoning method for 3D Gaussian Splatting enhances attack effectiveness by strategically injecting Gaussian points and disrupting multi-view consistency.
Yanxu Chen, Zijun Yao, Yantao Liu +4 authors
StockBench evaluates large language models in realistic stock trading environments, revealing challenges and opportunities in developing LLM-powered financial agents.
Zhepei Wei, Xiao Yang, Kai Sun +12 authors
TruthRL, a reinforcement learning framework, enhances the truthfulness of large language models by balancing accuracy and abstention, significantly reducing hallucinations and improving performance across benchmarks.
Junsong Chen, Yuyang Zhao, Jincheng Yu +17 authors
SANA-Video, a small diffusion model, efficiently generates high-resolution, high-quality videos with strong text-video alignment using linear attention and a constant-memory KV cache, achieving competitive performance at a lower cost and faster speed.
Zhihong Chen, Xuehai Bai, Yang Shi +9 authors
OpenGPT-4o-Image, a large-scale dataset with hierarchical task taxonomy and automated generation, significantly improves performance in image generation and editing tasks.
Thanh-Long V. Le, Myeongho Jeon, Kim Vu +2 authors
RL-ZVP, a novel reinforcement learning algorithm, leverages zero-variance prompts to improve the accuracy and pass rate of Large Language Models in math reasoning tasks.
Xiang An, Yin Xie, Kaicheng Yang +19 authors
LLaVA-OneVision-1.5 is a family of large multimodal models that achieves state-of-the-art performance with reduced costs through efficient training and high-quality datasets.
Ziniu Li, Congliang Chen, Tianyun Yang +5 authors
An adaptive exploration budget allocation method for reinforcement learning in Large Language Models improves training efficiency and performance on mathematical reasoning benchmarks.
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