Quo Vadis, World Modeling?
Yu Yang, Xuemeng Yang, Licheng Wen +17 authors
This work proposes agent-centric interactive world proxies that provide diverse feedback for continual agent improvement across inference, training, and co-evolution stages.
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。
686 篇论文 · 按点赞排序
Wan-Streamer v0.2: Higher Resolution, Same Latency
Lianghua Huang, Zhi-Fan Wu, Yupeng Shi +23 authors
Wan-Streamer v0.2 enhances audio-visual interaction by increasing visual resolution while maintaining low latency through optimized thinker-performer architecture with multi-GPU parallel processing.
Multi-Resolution Flow Matching: Training-Free Diffusion Acceleration via Staged Sampling
Xingyu Zheng, Xianglong Liu, Yifu Ding +4 authors
MrFlow accelerates text-to-image diffusion by combining low-resolution generation with pixel-space super-resolution and noise injection, achieving up to 25x speedup without training or runtime modifications.
HelloWorld: Enabling Socially Interactive Characters in Video World Models
Liangyang Ouyang, Ruicong Liu, Xuangeng Chu +2 authors
HelloWorld is a video world model that enables users to trigger natural social responses from on-screen characters via self-distilled training and inference-time cross-attention modulation.
GigaWorld-1: A Roadmap to Build World Models for Robot Policy Evaluation
GigaWorld Team, Angyuan Ma, Boyuan Wang +24 authors
World models for robotic policy evaluation are systematically studied through a new benchmark, revealing that long-horizon rollout consistency and robot-specific controllability are more important than short-term visual realism for reliable policy assessment.
AISPA: User-Centric System Prompt Auditing for Large Language Model Applications
Xiangning Lin, Shenzhe Zhu, Shu Yang +23 authors
System prompts are instructions configured by developers to govern the behaviors of foundation models in AI applications. They are used throughout commercial AI products, but are rarely disclosed to the public or regulators, creating a serious trust and accountability gap in the wide deployment of AI systems. In this paper, we introduce Artificial Intelligence System Prompt Assurance (AISPA), a user-centric framework for systematically auditing system prompts in AI systems. AISPA examines specific parts of a system prompt and evaluates them along eight dimensions that matter to users. We then use this framework to review 3,249 instructions from system prompts in 88 commercial AI products, classifying each instruction as either protective (of users) or problematic. Our audit surfaces four core findings. First, system prompt design varies substantially across products and developers, with some organizations averaging over 60 protective instructions per product while others average fewer than 5. Second, protective instructions are widely adopted but shallow in scope: 98.9% of products contain at least one, yet only 24% cover all eight dimensions of the AISPA taxonomy. Third, system prompts have grown steadily longer and more protective of users, suggesting that user protection is becoming a more visible concern in commercial prompt design. Fourth, despite this progress, problematic instructions remain pervasive: roughly 40% of products contain at least one instruction that works against user interests, and protective and problematic instructions frequently coexist within the same prompt. Our findings highlight the need for greater transparency, standardization, and independent oversight for system prompts in commercial AI products.
DiffusionGemma Technical Report
DiffusionGemma Team, Adrien Ali Taïga, James Assiene +41 authors
DiffusionGemma is a fine-tuned mixture-of-experts language model that uses discrete diffusion to generate text blocks in parallel, achieving high speed while preserving capabilities like multimodal inputs and reasoning.
AgenticDataBench: A Comprehensive Benchmark for Data Agents
Zhaoyan Sun, Shan Zhong, Daizhou Wen +10 authors
A comprehensive benchmark named AgenticDataBench is introduced to evaluate data agents across diverse domains with fine-grained task annotations and skill-based coverage metrics.
TurboServe: Serving Streaming Video Generation Efficiently and Economically
Youhe Jiang, Haoxu Wang, Haotong Bao +5 authors
TurboServe is a specialized serving system for streaming video generation that addresses session state management and dynamic resource allocation challenges through integrated scheduling, autoscaling, and migration mechanisms.
WorldExam: Benchmarking World Models from Apparent Appearance to Inherent Reactivity
Yuxue Yang, Shuyao Shang, Jiahe Wang +13 authors
Controllable video generation models are increasingly being developed as world models. Accordingly, evaluating them in this role extends beyond the apparent appearance of generated videos to the inherent reactivity of the worlds they depict: the ability to infer from the scene state how the world should react and to generate plausible consequences not explicitly described in the input. Yet existing benchmarks mainly assess visual quality or explicit instruction fulfillment by checking whether requested actions and interaction outcomes are realized, leaving inherent reactivity underexamined. We introduce WorldExam, a hierarchical diagnostic benchmark spanning four levels: Visual Quality, Control Adherence, Spatial Consistency, and World Reactivity. It comprises 1,474 cases across eight dedicated tasks and supports unified evaluation of camera-, action-, and language-driven model paradigms. The World Reactivity level evaluates scene-conditioned reactions and goal-directed behaviors beyond what is explicitly specified in the input. Evaluation of 20 representative models reveals a clear capability split. Camera-driven models excel at camera control, but their interfaces do not support dynamic interaction; action-driven models control subjects more precisely but often leave the world unresponsive; and language-driven models perform better on interaction but follow complex controls less faithfully. No model combines broad task coverage with consistently strong performance, showing that high visual quality and explicit instruction fulfillment do not guarantee inherent reactivity.
SAF-OPD: Stable Advantage Fusion for On-Policy Distillation
Yifan Ding, Xincheng Wei, Yoshua Y. Li +7 authors
Reinforcement learning with verifiable rewards (RLVR) broadcasts a single response-level reward to every token, while on-policy distillation (OPD) scores each token against a stronger teacher for a dense advantage but caps performance at teacher quality and discourages exploration beyond it. Their complementarity makes combining RLVR and OPD promising, but we find that fusing the two advantages with a fixed coefficient triggers entropy collapse from two miscalibrations: a magnitude mismatch, where token-level OPD advantages can spike far beyond the bounded RLVR advantage and erase its signal, and a temporal mismatch, where sustained full-strength OPD keeps pulling the student toward the teacher and limits exploration needed to surpass it. We propose SAF, a Stable Advantage Fusion framework that resolves both issues via a lightweight, four-stage pipeline applied only to the OPD advantage: a sparsify-then-compress mechanism for magnitude control paired with a warm-up-then-anneal mechanism for temporal control, with each stage independently switchable and adding negligible overhead. Instantiating RLVR with GRPO, we evaluate SAF across seven mathematical reasoning and code generation benchmarks with Qwen3-1.7B/4B/8B: SAF avoids entropy collapse and consistently outperforms fixed-coefficient GRPO+OPD fusion, improving the aggregate score by 0.51-2.70% across all six model-domain settings while achieving more stable training.
OneDayAgent: Towards a Long-Horizon Harness for Autonomous Agents
Jingsheng Zheng, Xinyuan Fang, Jintian Zhang +3 authors
LLM agents are increasingly applied to open-ended everyday requests that span work, study, and life. These tasks are long-horizon, cross-environment, and multimodal, forcing the agent to preserve goals and constraints across many steps while navigating heterogeneous tools and attachments. While prior work has addressed individual failure modes such as goals drift, states loss, and context overflow, whether a single harness can manage them jointly and remain effective across backends has received less study. We present OneDayAgent, a long-horizon harness for autonomous agents. OneDayAgent turns an open-ended request into a managed execution process that decomposes tasks into bounded subtasks, maintains execution memory under context pressure, and verifies and repairs the final deliverable. We evaluate OneDayAgent on AgentIF-OneDay across 104 tasks. With the GLM-5.2 backend, OneDayAgent sets a new state of the art with an overall score of 0.821. The same harness runs across five backend LLMs from three model families, indicating the harness generalizes across backends without tuning, even as different models induce distinct execution styles under the same workflow.
Hi-Q: Hierarchical Evidence-guided Query Refinement for Multi-Hop Question Answering
Jueun Kim, Sungho Park, Wook-Shin Han
Hi-Q is an evidence-conditioned framework that dynamically refines multi-hop queries into hierarchical trees guided by corpus support signals, improving retrieval and answer accuracy without fixed graphs.
From Production Traffic to Post-Training: Building a Self-Hosted LLM That Covers the Corporate Request Mix
Olga Tsymboi, Dmitrii Stoianov, Ramil Latypov +11 authors
A smaller self-hosted LLM trained with separate GRPO experts merged via SLERP outperforms a much larger baseline on instruction following, function-calling, and internal tasks while serving half of platform traffic at lower cost.
PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents
Shuhan Xue, Zixin Ding, Yichen Shen +6 authors
Recursive self-improvement requires agents to turn accumulated experience into better future behavior. Personal AI agents offer a concrete setting for studying this capability because they retain preferences, task histories, tool routines, and learned skills across sessions. Yet whether retained experience actually improves them over time has not been systematically tested. We introduce PAST-Bench, a benchmark designed to isolate this question. Each agent runs through ordered sequences of fresh-session tasks under matched conditions that turn retained experience on and off. It spans 26 scenarios and 204 episodes across memory, procedural reuse, information gathering, and update. We report both later-task gains and whether those gains follow the intended save, retrieve, and update pathway. Across seven base models and four agent frameworks, improvement is real but uneven across capabilities. Agents with the same headline gain can differ markedly in whether that gain is supported by evidence of the intended pathway. Guided by these findings, we develop Hermes+, which extends Hermes with five targeted interventions across stages of the agent loop. Hermes+ raises the average gain from retained experience and provides clearer pathway evidence, with its strongest improvement on tasks requiring outdated state to be replaced, although the effect remains capability- and model-dependent. Together, PAST-Bench and Hermes+ provide an evaluation and diagnostic foundation for studying how persistent agents can progress from retaining experience to systematically improving through it. Code: https://github.com/Gen-Verse/PAST-Bench
From Economic Agents to Agentic Economies: A Systems Blueprint for Economic World Models
Jiale Han, Xiang Li, Jing Qian +7 authors
Economic World Models (EWMs) are generative economic models that simulate how economies evolve from within by modeling heterogeneous agents, their beliefs and actions, and the market and institutional mechanisms through which their interactions produce aggregate outcomes. This paper develops an implementation roadmap for building economic world models as generative engines in which heterogeneous agents act, interact, adapt, and co-evolve with markets and institutions, thereby producing economic dynamics from the inside. We organize EWM systems into a six-level capability ladder, from fixed rule-based agent worlds to adaptive and LLM-based agent worlds, self-evolving agents, evolving institutional worlds, and sim-to-real economic twins aligned with real observations. A systematic literature survey across these levels reveals that existing work remains concentrated in lower-level agent and simulation environments, while systems with self-evolving agents, endogenous institutions, persistent empirical alignment, and validated economic mechanisms remain rare. By translating the EWM agenda into an implementation blueprint, this paper aims to accelerate the development of the next generation of economic simulation environments that can serve as high-fidelity sandboxes for human decision-makers and as training, planning, evaluation, and safety substrates for AI agents. We release a curated paper list and related resources to support future research.
HarnessOpt-Bench: Evaluating LLMs at Harness Optimization
Varun Ursekar, Apaar Shanker, Yash Maurya +4 authors
As LLMs are increasingly deployed within agentic systems, their capabilities depend not only on the model weights but also on the harness: the prompts, tools, control flow, memory, and orchestration code surrounding them. This makes automated harness optimization -- the iterative and evaluation-guided improvement of a harness by an AI system -- both an important route to improving AI systems and a demanding capability for AI systems themselves. Yet the community lacks a common protocol for measuring how well frontier LLMs perform at this task. We introduce HarnessOpt-Bench, a benchmark for end-to-end harness optimization under expensive and stochastic evaluation. An optimizer, an LLM paired with a coding harness, receives a target agent's seed harness, graded evaluation feedback, and a fixed target-evaluation budget. It edits the harness and nominates a final candidate, which is scored by its normalized gain over the seed on a held-out test partition that remains inaccessible throughout search. A trusted execution environment enforces the evaluation boundary, meters target-agent resource use, and preserves candidate versions for audit. We evaluate 5 frontier LLMs as optimizers both under a shared coding harness and under their native harnesses across 4 downstream tasks, over 111 scored runs. Experiment results show that optimizer models separate more than the coding harnesses they act through, native harnesses are not consistently superior, and gains vary substantially across tasks and seed regimes. These results establish harness optimization as a measurable and discriminative capability with large space for improvement.
The Missing Temporal Link: Temporal Context Routing for Script-Driven Audio-Video Generation
Yichen Liu, Quanwei Zhang, Haozhe Wang +7 authors
Temporal Context Routing improves script-aligned timing of shots and dialogue in joint audio-video generation by mapping structured script timing onto shared video-audio temporal axes.
Fewer Clarifications, Better Code: Benchmarking Cross-Session Personalized Ambiguity Adaptation in Coding Assistants
Zijian Xu, Wenshuo Zhang, Zisen Qin +4 authors
AI-assisted coding increasingly translates informal user intent into executable software, yet coding requests often contain ambiguities that recur in user-specific ways across tasks and sessions. Existing disambiguation methods typically address each ambiguous request in isolation within the current coding session, often through eliciting additional clarification. However, whether resolved session history from the same user can serve as memory for resolving recurring personalized ambiguity in a newly opened session remains underexplored. We formulate personalized ambiguity adaptation as a new task: given a user's previously resolved coding sessions and a new ambiguous request, an assistant should identify the recurring ambiguity pattern, produce the intended executable solution, and minimize clarification. To benchmark this task, we introduce CAPA, which characterizes personalized coding ambiguity through six mechanisms and injects these mechanisms into unambiguous executable tasks using a controlled three-stage generation pipeline. CAPA contains 600 coding sessions across 60 balanced user--ambiguity cells, including 300 held-out evaluation sessions. We evaluate 12 recent LLMs under no-history and same-user-history conditions using executable success, first-turn success, and turns-to-completion. Our analyses examine task difficulty, user identity, and memory-based history use, and we further propose same-user history gating as a lightweight inference-time method. CAPA provides a foundation for developing long-term coding assistants that better align generated code with user intent while reducing repeated clarification.
Ideas Have Genomes: Benchmarking Scientific Lineage Reasoning and Lineage-Grounded Idea Generation
Yifan Zhou, Qihao Yang, Yan Li +14 authors
A benchmark for scientific lineage reasoning and idea generation is introduced, organizing scientific works as genetic-like Idea Genome objects and evaluating both reasoning and generation capabilities.
WHALE: A Simple Recipe for Joint Harness-Weight Optimization
Haechan Kim, Yoonho Lee, Gisang Lee +2 authors
WHALE alternates model weight updates and harness search to jointly optimize agent performance across reasoning tasks.
DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation
Xin Cheng, Xingkai Yu, Chenze Shao +30 authors
DSpark enhances LLM inference speed by combining parallel draft generation with adaptive verification that reduces waste and improves throughput in high-concurrency settings.
WorldDirector: Building Controllable World Simulators with Persistent Dynamic Memory
Hanlin Wang, Hao Ouyang, Qiuyu Wang +10 authors
WorldDirector enables controllable video generation with persistent object memory by decoupling semantic motion planning from visual rendering through LLM coordination of 3D trajectories and camera movements.
On the Design Fundamentals of Pixel Text Representation Learning
Chaohao Yuan, Ruifeng Yuan, Zhuoxu Huang +4 authors
Pixel Linguist II improves visual text encoding through variable resolution training, natural image-text grounding, layout-aware rendering, and multilingual curricula, achieving state-of-the-art results and strong compression robustness.
LightNav-0: Eliciting VLM Spatial Intelligence for Generalist Embodied Navigation
Shaoan Wang, Aocheng Luo, Fei Huang +17 authors
LightNav-0 is a compact generalist navigation model that leverages a pretrained vision-language model’s spatial reasoning via unified pointing tokens and action tokenization to achieve state-of-the-art embodied navigation across diverse tasks and robots.
SHAPE of Chain-of-Thought in Math Reasoning
Jonghyun Song, Sangjun Song, Minjae Oh +3 authors
SHAPE analyzes chain-of-thought reasoning via semantic spaces and heuristics to diagnose LLM mathematical reasoning and improve post-training.
ELDR: Expert-Locality-Aware Decode Routing for PD-Disaggregated MoE Serving
Sangjin Choi, Sukmin Cho, Yifan Xiong +3 authors
ELDR is an expert-locality-aware decode router for prefill-decode disaggregated Mixture-of-Experts serving that improves performance by predicting expert activations and routing requests accordingly.
Beyond Retrieval: Progressive Latent Memory Evolution for Streaming Video Understanding
Hongyu Qu, Guangming Yao, Ling Xing +7 authors
LatentStream introduces a progressive latent working memory framework that internalizes streaming visual evidence into compact evolving tokens for continuous reasoning.
SKT: Skill-Use Training at Scale via Verified Synthetic Data Generation
Zelin Tan, Yiqun Zhang, Hao Li +11 authors
Agent skills have become an important mechanism for equipping language-model agents with reusable procedural knowledge. However, providing skills alone does not guarantee that current models can effectively identify, apply, and coordinate them. To improve skill-use capabilities, we introduce SKT, a verified data synthesis pipeline that constructs skill-grounded tasks and executable trajectories from large collections of agent skills. SKT selects suitable single-skill and multi-skill configurations, synthesizes tasks through rule-based and agent-based verification with feedback-guided repair, and retains only successful trajectories that substantially use every required skill. Using 2,000 public skills, SKT produces 4,000 task packages and 27,164 verified trajectories. Based on the same pipeline and a disjoint test pool, we further construct SkillEval, a held-out executable benchmark for evaluating skill use. Experiments across diverse models, benchmarks, and agent harnesses show that supervised fine-tuning on SKT-generated trajectories consistently improves skill-use performance. Verification ablations, cross-harness evaluation, and scaling experiments further demonstrate that these gains depend on high-quality supervision, extend beyond a single agent interface, and increase with broader skill coverage. Together, these results establish verified data synthesis as an effective and scalable approach for skill-use training.
UniClawBench: A Universal Benchmark for Proactive Agents on Real-World Tasks
Zhekai Chen, Chengqi Duan, Kaiyue Sun +4 authors
UniClawBench introduces a capability-driven benchmark for evaluating proactive agents in real-world environments using live Docker container evaluation and closed-loop assessment with multiple agent roles.