Research Paper Cockpit

Daily Digest - 2026-09-22

Papers first seen in this daily snapshot.

Daily Archives

Quick jump into generated daily digests.

Research Workflow

Latest digest: 2026-09-22.

Papers

58 visible entries

arxiv Score 26.5

Muon Can Outperform Dedicated Continual Learning Methods

2026-09-21 · Sebastian George Sincari, Bogdan Alexandru Gheorghe, Antonio Barbalau

Research Track A · General AI

Continual learning with Low-Rank Adapters (LoRA) typically mitigates forgetting by penalizing the overlap between a new update and the accumulated past weights, which discourages certain update directions without controlling how an update distributes its energy over the ones that remain. We ask whether that restriction…

Review
pending
Role
unreviewed
Read
now
arxiv Score 26.0

LoRA-generating hypernetworks for efficient on-device LLM generative personalization

2026-09-21 · Sean Augenstein, Li Ding, Jihwan Lee, Keith Rush, Andrey Zhmoginov

Research Track A · General AI

On-device large language models (`LLMs'), e.g. running on mobile phones, are ripe for improvement via personalization. The limited compute resources of mobile devices impose limits on model scale and thus model quality, making any realizable quality gains highly impactful. At the same time, their personal nature (i.e.,…

Review
pending
Role
unreviewed
Read
now
arxiv Score 25.3

ME-VLM:A Unified VLM for Embodied Cognition and Agent Coordination

2026-09-21 · Foundation Model, Li Auto Inc

General AI

Physical AI requires models to ground visual and linguistic understanding in real-world environments while accounting for environmental constraints and execution feedback. We introduce MachEmbodied-VLM (ME-VLM), a unified vision-language model with two variants, 4B and 35B-A3B, that brings together embodied cognition a…

Review
pending
Role
unreviewed
Read
now
arxiv Score 24.5

iSDFT: Information-Proximal Self-Distillation for Continual Learning in LLMs

2026-09-21 · Ahmed Khaled Khamis, Xiaotong Ji, Hassan Jaber, Rasul Tutunov, Matthieu Zimmer, Jun Wang, Haitham Bou-Ammar

Research Track A · General AI

On-policy self-distillation fine-tuning (SDFT) learns new skills from demonstrations while reducing forgetting, but it always distils toward the full demonstration-conditioned teacher. This fixes teacher influence at the full-teacher endpoint, providing no control over how much demonstration information should be trans…

Review
pending
Role
unreviewed
Read
now
arxiv Score 24.3

Bilinear Optimization Divergence: Diagnosing Factor-Constrained LoRA Continual Learning

2026-09-20 · YongShun Wang, JianLin Su, Yong Ma

Research Track A

Orthogonality in a LoRA factor does not by itself specify what the composed update protects: the answer depends on the task-start state, the parameterization, and the realized optimizer displacement. We formalize this question through Bilinear Optimization Divergence (BOD), an anchor-relative diagnostic of effective-up…

Review
pending
Role
unreviewed
Read
now
arxiv Score 23.8

RPMem: Learning Long-Term Recurrent Parametric Memory Across Sessions for LLM Agents

2026-09-20 · Fanyu Zhao, Ruike Cao, Liang Dong, Fugen Yao, Jian Xu, Guanjun Jiang, Han Zhang, Yifei Zhao, Yinsheng Li

Research Track A · General AI

Long-running LLM agents require memory that persists and evolves across sessions. Text-based memory retrieves and reconstructs past interactions at every query, making long-horizon performance increasingly dependent on retrieval quality and contextual reasoning as histories grow. Parametric memory encodes experience di…

Review
pending
Role
unreviewed
Read
now
huggingface Score 23.5

Document Retrieval-Aware Chunking (D-RAC): Universal Retrieval-Aware Ingestion of Enterprise Documents via PDF Normalization and Multimodal Markdown Conversion

2026-09-21 · Uday Allu, Abhivanth Sivaprakash, Pratik Singh, Aman Manocha

General AI

Retrieval-Augmented Generation (RAG) systems over enterprise knowledge bases must ingest heterogeneous document formats -- PDFs, Word documents, presentations, and scans -- whose content is locked inside complex visual layouts, multi-column pages, and dense tables. Rule-based extraction and OCR destroy reading order, f…

Review
pending
Role
unreviewed
Read
now
arxiv Score 23.3

VideoGen-Agent: Reinforcing Video Generation Agents

2026-09-21 · Binxu Li, Haoyi Duan, Yuhui Zhang, Yaohui Zhang, Zihao Lin, Kaituo Feng, Suozhi Huang, Xiangyi Li, Yu Li, Chunyuan Li, Shilong Liu, Mengdi Wang

General AI

Recent advances in video generative models have enabled high-fidelity, temporally coherent video generation. However, these models often struggle to satisfy prompts requiring specialized knowledge, specific identities, physical consistency, or ordered events. In this paper, we present VideoGen-Agent, a multimodal agent…

Review
pending
Role
unreviewed
Read
now
arxiv Score 22.3

SLICEChat: Progressive In-Encoder Token Pruning for Whole-Slide Pathology Language Models

2026-09-21 · Ali Kerem Bozkurt, Baris Cem Bakay, Ibrahim Kulac, Cigdem Gunduz-Demir, Erkut Erdem, Aykut Erdem

General AI

Whole-slide pathology images (WSIs) contain gigapixel-scale visual content, creating a major scalability challenge for slide-level multimodal large language models (MLLMs). Existing approaches process thousands of patch tokens and typically apply compression only after slide encoding, leaving multimodal attention compu…

Review
pending
Role
unreviewed
Read
now
arxiv Score 22.0

ACLArena: Agent Continue Learning in Multi-stage Post-training

2026-09-21 · Haixin Wang, Xiaoxuan Wang, Junkai Zhang, Han Zhang, Renliang Sun, Alexander K Taylor, Yidan Shi, Haoran Deng, Chenguang Wang, Jason Cong, Yizhou Sun, Wei Wang

Research Track A · General AI

Building general-purpose agents for industrial deployment requires integrating multiple capabilities, each typically acquired at a distinct stage of training. Yet there is currently no well-established recipe for Agent Continual Learning (ACL), with little understanding of the trade-offs among existing integration para…

Review
pending
Role
unreviewed
Read
now
arxiv Score 21.3

Time-Incremental Continued Pretraining of LLMs: Knowledge Updates Without Catastrophic Forgetting

2026-09-20 · Fırat Öncel, Salman Hussain Ali, Mirco Ravanelli, Cem Subakan, Çağatay Yıldız

Research Track A · General AI

Large language models (LLMs) drift out of date the moment their pretraining ends, yet retraining from scratch is prohibitively expensive. Continued pretraining (CPT) is the natural remedy, but it is typically evaluated through a continual learning lens that assumes disjoint data streams. This is a poor fit for time-inc…

Review
pending
Role
unreviewed
Read
now
arxiv Score 21.3

GRUET: Quantifying Uncertainty of Agentic Reasoning-and-Acting Processes

2026-09-21 · Shuang Liang, Xin-Yu Hu, Shao-Qun Zhang

General AI

Agents have attracted considerably increasing attention due to the power of executing both Reasoning and Acting (ReAct) in open and dynamic environments. The ReAct process typically exhibits a multi-turn trajectory in which one drives Large Language Models (LLMs) to generate both reasoning chains and task-specific acti…

Review
pending
Role
unreviewed
Read
now
huggingface Score 20.5

Jev-Mem: System-One-Controlled Agentic Memory for Efficient AI Agents

2026-09-21 · Dongming Jiang, Yi Li, Bingzhe Li

General AI

Agentic memory is becoming essential for long-horizon AI agents, yet many existing systems rely on autoregressive LLMs to control how memories are organized, retrieved, and used, placing expensive generation on the critical path of memory operations. We introduce \method, a new agentic memory architecture inspired by S…

Review
pending
Role
unreviewed
Read
now
arxiv Score 19.3

MedRSI: Recursive Self-Improvement for Medical Agents via Clinically Aligned Self-Evolution

2026-09-21 · Junde Wu, Jiayuan Zhu, Minghao Hu, Fenglin Liu, Jiazhen Pan

General AI

Medical agents increasingly combine general reasoning models with specialized clinical tools, yet their capabilities remain largely fixed by what clinicians and engineers design before deployment. Recursive self-improvement (RSI) offers a different paradigm in which agents learn from their own failures and autonomously…

Review
pending
Role
unreviewed
Read
now
arxiv Score 18.8

Seeing Through Conflicts: Improving Instruction Hierarchy Alignment in Vision-Language Models

2026-09-03 · Nicholas Sansoterra, Zishuo Zheng, Sachin Kumar

General AI

Instruction hierarchy (IH) alignment teaches language models to prioritize higher-level instructions when inputs conflict. While studied primarily in text-only settings, vision-language models (VLMs) introduce new challenges for IH: instructions may be embedded in images, split across modalities, visually transformed, …

Review
pending
Role
unreviewed
Read
now
arxiv Score 18.3

ME-Brain-1.0: Memory, Cognition and Action for Evolving Embodied Intelligence

2026-09-21 · Wei He, Hengtao Li, Zhongrui Yu, Xuhan Zhu, Maokui He, Zide Liu, Xiyue Zhang, Xianwei Mao, Chunpeng Zhou, Jia Shi, Yanze Xin, Jingwen Li, Jingxie Zheng, Sijie Zeng, Chenfeng Wang, Fan Lu, Zeyu Zhang, Shuai Guo, Hengxuan Zhang, Pengfei Yu, Yu Liu, Kun Zhan, Yan Xie

General AI

Current embodied systems largely rely on pretrained capabilities that remain fixed after deployment, limiting their ability to learn from physical interaction. We introduce MachEmbodied-Brain (ME-Brain), a self-evolving embodied system organized around a closed loop of action execution, experience acquisition, experien…

Review
pending
Role
unreviewed
Read
now
arxiv Score 18.3

When Tomorrow Becomes Today: Self-Evolving Policies for Agentic Time-Series Forecasting

2026-09-21 · Yifan Hu, Xilin Dai, Zhiyuan Qu, Yiding Liu, Zewei Dong, Jiang-ming Yang, Qiang Xu

General AI

Agentic time series forecasting concerns systems whose underlying mechanisms evolve, making the relative effectiveness of numerical models, reasoning strategies, and intervention rules inherently time-varying. Consequently, a time series agent must adapt the forecasts it produces and the orchestration policy that deter…

Review
pending
Role
unreviewed
Read
now
arxiv Score 17.3

DolphinBench: Mapping the Pareto Frontier of Agent Memory

2026-09-21 · Soumil Rathi, Deshraj Yadav, Taranjeet Singh

General AI

Agents today often take real-world actions that depend on long-term memory and context recall over time. However, most current memory benchmarks are built for a conversational question-answer format, where the question itself signals that some fact must be retrieved, and often which one. Moreover, benchmarks rarely req…

Review
pending
Role
unreviewed
Read
now
arxiv Score 16.3

MIGU: Multimodal Instruction Grounding under Uncertainty for Manipulation Planning

2026-09-21 · Mingke Lu, Anxing Xiao, David Hsu

General AI

Understanding natural human instructions is crucial for deploying robots in human-centric environments. We study multimodal instruction grounding, where language and gesture provide complementary but uncertain cues. We present MIGU, a modular framework that combines semantic and geometric evidence into a unified ground…

Review
pending
Role
unreviewed
Read
now
arxiv Score 16.3

RRSI: Regularized Recursive Self-Improvement of Agent Harnesses

2026-09-21 · Peng Xia, Rujun Han, Zifeng Wang, Yanfei Chen, Yufan Zhang, Yoonho Lee, Chengsong Huang, Han Yu, Zhongying CuiZhu, Yifei Ming, Huaxiu Yao, Burak Gokturk, Tomas Pfister, Chen-Yu Lee

General AI

An LLM agent's capability is largely magnified by its harness, namely the prompts, control flow, tooling, memory, and context management surrounding the frozen backbone model. Recent methods increasingly automate this process by iteratively proposing and selecting component-wise edits of an agent harness, practically e…

Review
pending
Role
unreviewed
Read
now
arxiv Score 15.3

CE$^4$L: Continual Ego, Exo, and Ego-Exo Learning

2026-09-20 · Hongwei Yan, Kanglei Zhou, Yuchen Liu, Qingyu Shi, Yi Zhong, Liyuan Wang

Research Track A · General AI

Perception for embodied agents is video-based, often multi-view (ego, exo, or both), and inherently continual, with simultaneous task and viewpoint shifts. Yet continual learning (CL) remains dominated by exo-only recognition tasks, obscuring behavior under these real-world coupled shifts. We introduce Continual Ego, E…

Review
pending
Role
unreviewed
Read
now
arxiv Score 15.3

A Distributional Optimisation Perspective on Combining Models in Deep Learning

2026-09-21 · Congye Wang, Yan Lin, Zheyang Shen, Matthew A. Fisher, Chris. J. Oates

General AI

Combining predictions from different models can improve performance at machine learning tasks, but the training of the individual models and the rule used to combine them are typically chosen separately, and by ad hoc means. Recent advances in distributional optimisation (i.e. where the optimisation occurs over the set…

Review
pending
Role
unreviewed
Read
now
arxiv Score 15.3

FinInteract: Benchmarking Clarification and Intent Integration in Ambiguous Financial Question Answering

2026-09-21 · Xinyu Wang, Tung Sum Thomas Kwok, Zhenghan Tai, Guang Cheng

General AI

Large language model agents increasingly answer financial questions by searching regulatory filings. Such questions are often deceptively under-specified: Meta Platforms' "operating income" is $46.75B consolidated but $62.87B for the Family of Apps segment, and each reading is exactly verifiable against the filing. A c…

Review
pending
Role
unreviewed
Read
now
huggingface Score 15.0

Deep Persona: A Psychologically Grounded Architecture and Evaluation Framework for Role-Playing Agents and Simulations

2026-09-06 · Rotem Dror, Zohar Elyoseph, Yuval Haber, Elad Refoua, Oshrat Ayalon, Adir Solomon

General AI

Existing approaches to persona simulation with Large Language Models (LLMs) mostly rely on shallow character descriptions that fail to sustain coherent character behavior across extended interactions. We introduce Deep Persona, a psychologically grounded, three-layered architecture that organizes personas into hierarch…

Review
pending
Role
unreviewed
Read
now
arxiv Score 13.6

One to More, More to One: Category-Aware Iterative Expert Training for Software Engineering Agents

2026-09-20 · Jie Zhao, Ziyu Jiang, Suhang Zheng, Minghui Shan, Xiaoxiao Xu, Lin Qu

Research Track A · General AI

Repository-level software engineering (SWE) comprises heterogeneous task categories, whose progress under pooled agentic reinforcement learning can be uneven: gains in some categories coincide with regressions in others, while aggregate resolution obscures these changes. Motivated by this category see-saw, we develop a…

Review
pending
Role
unreviewed
Read
now
arxiv Score 13.5

Few-Shot Demonstrations Elicit the Use of In-Context World Representations in LLMs

2026-09-21 · Kohsei Matsutani, Gouki Minegishi, Core Francisco Park, Takeshi Kojima, Yusuke Iwasawa, Yutaka Matsuo

Research Track B · General AI

Large language models (LLMs), when acting as agents, are expected to take observed data in context, infer the latent state space underlying the world, and leverage it for downstream prediction. However, prior work demonstrated that LLMs struggle to use representations learned in context on a graph tracking task, where …

Review
pending
Role
unreviewed
Read
now
arxiv Score 13.3

GameHorizon Suite: Multi-Horizon Data and Evaluation in Gameplay

2026-09-21 · Yiran Wang, Xingyilang Yin, Junfu Pu, Guangzhi Wang, Kaifeng Li, Mingyu Ouyang, Huiqiang Sun, Lingen Li, Cheng Cheng, Wangbo Yu, Honghao Chen, Xiaodong Cun, Chi-Man Pun, Zhiguo Cao, Ying Shan

General AI

Modern video games provide a measurable testbed for AI models, combining abilities of visual understanding, instruction decomposition, goal planning, and precise action control over multiple temporal horizons. Existing datasets and benchmarks, however, either cover a narrow range of games, lack language instructions, o…

Review
pending
Role
unreviewed
Read
now
arxiv Score 13.3

The Copy Ceiling: An Input-Exposure Control for Ontology-Grounded Generation over Curated Corpora

2026-09-21 · John J. O'Hare

General AI

When a language model answers from a curated corpus via graph-based retrieval, a large grounding uplift does not establish reasoning over the retrieved structure: the context may already expose the gold answers. We propose exposure accounting, which classifies each gold item by whether the shown context exposes it and …

Review
pending
Role
unreviewed
Read
now
huggingface Score 12.8

OmniEdu: Open Foundation Models for Learning and Teaching

2026-09-19 · Hao Liang, Qihan Lin, Meiyi Qiang, Linzhuang Sun, Hengyi Feng, Mingrui Chen, Sizhe Qiu, Wentao Zhang

General AI

Educational foundation models must solve problems, understand curriculum structure, diagnose learner difficulties, and provide appropriate instructional support. Existing educational language models often focus on either problem solving or tutoring, with training mixtures organized by source or task rather than capabil…

Review
pending
Role
unreviewed
Read
now
huggingface Score 12.5

CARE: Experience-Guided Atomic Corrective Execution for Vision-Language-Action Policies

2026-09-21 · Junlan Xiao, Junwei Jiang, Zaibin Zhang, Yifan Wang, Zhongbo Zhang, Huchuan Lu, Lijun Wang

General AI

Vision-Language-Action (VLA) policies achieve strong performance in robotic manipulation but remain brittle once execution deviates from nominal trajectories. We propose CARE (Corrective Atomic Robotic Execution), a framework that improves recovery by learning from failures encountered during execution. Instead of gene…

Review
pending
Role
unreviewed
Read
now
arxiv Score 12.3

Extracting Arguments, Not Just Classifying Them: Instruction-Tuned LLMs for Generative Component Detection

2026-09-21 · Sofiane Elguendouze, Erwan Hain, Elena Cabrio, Serena Villata

General AI

Argumentative component detection (ACD) is a core subtask of Argument(ation) Mining (AM) and one of its most challenging aspects, as it requires jointly delimiting argumentative spans and classifying them into components such as claims and premises. While research on this subtask remains relatively limited compared to …

Review
pending
Role
unreviewed
Read
now
arxiv Score 12.3

Harness-Zero: Harness Distillation via Agent-as-Harness

2026-09-21 · Haoran Ye, Yuxing Lu, Haonan Dong, Zhaochen Su, Guojie Song

General AI

Agent harnesses, the external systems that mediate model-environment interaction, can substantially improve agent performance, but their gains remain tied to the harness at deployment. Because the best harness varies across domains, instances, and models, a general-purpose agent must either settle for a suboptimal shar…

Review
pending
Role
unreviewed
Read
now
arxiv Score 12.3

onPanda: Efficient Annotation of On-Policy Alignment Data for LLMs and Agents via Token-Level Correction

2026-09-21 · Lei Yang, Mengyin Liu, Jia Wang, Hangyu Guo, Liang Zhao, Zheng Ge, Kang An, Binxing Jiao, Qi Han, Daxin Jiang, Siqi Shen, Xiangyu Zhang

General AI

We present onPanda, an interactive tool for efficiently annotating LLM alignment data and agent trajectories. onPanda adopts token-level correction as its core interaction: while reading a model response, the annotator locates the first inappropriate token and either picks a substitute from the model's candidate tokens…

Review
pending
Role
unreviewed
Read
now
arxiv Score 12.0

Canonical Procedural Actions: An Auditable Annotation Protocol for Tool-Use Agent Traces

2026-09-21 · Songqi Li, Dongqing Li, Zheqiao Cheng

Research Track A · General AI

Tool-use agent traces identify messages and API calls, but procedural analyses also need explicit units of action and inspectable links to their evidence. We present Canonical Procedural Actions (CPAs), an annotation protocol that records a procedural function, its first agent-event anchor, the agent events that realiz…

Review
pending
Role
unreviewed
Read
now
arxiv Score 11.3

Critical-State RL: Diagnosing Trainable States for Multi-Turn Tool Use

2026-09-21 · Zixiang Chen, Wenting Zhao, Zhepeng Cen, Akshara Prabhakar, Jielin Qiu, Jianguo Zhang, Zhiwei Liu, Tulika Manoj Awalgaonkar, Liangwei Yang, Shelby Heinecke, Silvio Savarese, Huan Wang

General AI

Multi-turn tool-use failures can hinge on a single model call, yet reward variation alone does not reveal which call would benefit from training. When rewards depend on later interactions, their variation can reflect downstream randomness rather than differences between the current actions. We introduce Critical-State …

Review
pending
Role
unreviewed
Read
soon
arxiv Score 11.3

Feedback Coding Enables Inference-Time Covert Agentic Communication

2026-09-21 · Sidong Guo, Sajani Vithana, Atefeh Gilani, Lalitha Sankar, Oliver Kosut, Flavio P. Calmon

General AI

As large language models (LLMs) are increasingly used to automate digital interactions, users can leverage LLM-generated text as cover for covert communication within seemingly benign conversations. Existing LLM steganography, however, is predominantly white-box, requiring the sender and receiver to share the cover sta…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 11.3

GradAgent: A Knowledge-Guided Multi-Agent System for Structure-Preserving Gradient-Flow Computation with an Application to Multicomponent Vesicle Dynamics

2026-09-21 · Zhenlin Guo, Jiale Meng, Shuqi Tang, Haiyan Su, Maosheng Jiang, Kaiwen Shi, Meng Zhao

General AI

High-order differential operators and nonlinear coupling make it challenging to construct conservative and energy-stable schemes for coupled gradient-flow systems. We present GradAgent, a knowledge-guided multi-agent system that coordinates three agents across model analysis, algorithm design and proofs, and numerical …

Review
pending
Role
unreviewed
Read
soon
arxiv Score 11.3

Rare Event Estimation via Iterative Unalignment

2026-09-21 · Hanming Yang, Daksh Mittal, Jing Dong, Hongseok Namkoong

General AI

As agents are deployed with increased autonomy, even extremely rare events along their stochastic output trajectories can occur and prove catastrophic. Safe deployment therefore does not depend on whether these events can occur, but on how often they might. We study the problem of estimating the probability of rare eve…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 11.3

SPHQuant: Efficient extreme low bit weight quantization for Vision-Language Models

2026-09-21 · Kewei Zhang, Zheng Chen, Haotong Qin, Yulun Zhang

General AI

Recent foundation models are moving toward native multimodal Vision-Language Models (VLMs), making VLMs a central form of next-generation foundation models. However, their large language backbones make edge deployment difficult due to high memory footprint and memory-bound autoregressive decoding. Weight-only post-trai…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 11.3

Steerable and Reactive Grasping Through Modular Design with a Three-Point Interface

2026-09-21 · Andrew Nguyen, Yonghyeon Lee, Sangbae Kim

General AI

Dexterous grasping requires deciding where to grasp, reaching the target, and maintaining stable contact. We connect these stages through a compact three-point interface that separates global geometric reasoning from local contact control. Given object geometry and optional language commands, our framework samples cont…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 11.3

When Quantization Preserves Accuracy but Not Evidence: Explanation-Aware Post-Training Quantization for Medical LLMs

2026-09-21 · Yeji Kim, Mi-Young Kim, Randy Goebel

General AI

Post-training quantization (PTQ) enables efficient deployment of large language models, and PTQ methods are usually optimized and evaluated with generic reconstruction, perplexity, or answer accuracy. But in explanation-critical domains, preserving only the final answer may be insufficient, since users may also inspect…

Review
pending
Role
unreviewed
Read
soon
huggingface Score 10.8

Mira-Scene: Pixel-Aligned Layouts for Generative 3D Scene

2026-09-20 · Yang-Tian Sun, Tianjia Liu, Zehuan Huang, Yi-Hua Huang, Xiaoyang Lyu, Ziyi Yang, Zi-Xin Zou, Yuan-Chen Guo, Yan-Pei Cao, Xiaojuan Qi

General AI

Single-image 3D object generation can now produce high-fidelity assets, yet accurately placing them into a coherent scene layout remains an open challenge. A central difficulty lies in how object layout is represented. Holistic methods absorb placement into a scene-level generation process, sacrificing object-level det…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 10.8

Circuit Hypernetworks for Quantum-Augmented Diffusion Language Models

2026-09-21 · Xiaoqiang Wang, Mengyang Xiong, Jun Dai, Bang Liu

General AI

Language models can be adapted by changing the computations applied to individual tokens. Quantum circuits offer one such approach, but evaluating wider circuits inside a large model can be computationally demanding. Here we introduce HyperQ, which adds token-conditioned quantum residual branches to a frozen masked-dif…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 10.3

ATCion: Exploring the Design of Icon-based Visual Aids for Enhancing In-cockpit Air Traffic Control Communication

2026-09-21 · Yue Lyu, Xizi Wang, Hanlu Ma, Yalong Yang, Jian Zhao

General AI

Effective communication between pilots and air traffic control (ATC) is essential for aviation safety, but verbal exchanges over radios are prone to miscommunication, especially under high workload conditions. While cockpit-embedded visual aids offer the potential to enhance ATC communication, little is known about how…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 10.3

Decoding Guardrails: XAI-Guided Perturbation Analysis of Prompt Injection Detection

2026-09-21 · Fernando Outeda, Gustavo Betarte, Juan Diego Campo, Fiorella Cravero

General AI

Large language models (LLMs) are increasingly deployed in production systems, raising concerns about their exposure to adversarial manipulation through prompt injection and jailbreak attacks. Classifier-based guardrails, such as Prompt Guard 2, are widely used as a first line of defense against such attacks, but their …

Review
pending
Role
unreviewed
Read
soon
arxiv Score 9.3

DexTacWAM: A Visuo-Tactile World-Action Model for Dexterous Manipulation

2026-09-21 · Haoran Yuan, Zekai Wang, Boning Shao, Haoran Lu, Trevor Darrell, Ismini Lourentzou, Wei Zhan

General AI

Dexterous manipulation depends on contact dynamics that are often only partially observable from vision. Recent World-Action Models (WAMs) couple predictive video world modeling with action generation, but remain largely vision-centric and therefore cannot directly model these contact dynamics. We present DexTacWAM, a …

Review
pending
Role
unreviewed
Read
soon
arxiv Score 9.3

Passthrough Rigidity: The Behavioral and Visuomotor Costs of Mediated Perception

2026-09-21 · Markus D. Solbach, Mohit Goyal, Sakar Khattar, Jayant Varma, Bjorn Vlaskamp, John K. Tsotsos, Konstantine Tsotsos

General AI

Broad public adoption of head-mounted displays using video passthrough remains elusive despite significant market investment. A precise understanding of why users experience persistent discomfort even as hardware factors such as resolution and latency have dramatically improved remains an open issue. This paper investi…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 9.3

Who Does What in AI Auditing? Designing Human-AI Collaboration for Auditing Generative AI

2026-09-21 · Eunkyu Park, Markelle Roesti, Wesley Hanwen Deng, Renata Barreto, Mohammad Tahaei, Kenneth Holstein, Jason Hong, Motahhare Eslami

General AI

AI auditing increasingly incorporates AI agents to expand the scale and breadth of audit coverage, yet little is known about how auditing work should be divided without displacing human judgment. We introduce Human-Agent Audit Collaboration (HAAC), a workflow and system for structuring human-AI collaboration in AI audi…

Review
pending
Role
unreviewed
Read
soon
huggingface Score 8.5

1% of Tokens Can Be Enough: On Gradient Estimation in On-Policy Distillation

2026-09-21 · Huanxin Sheng, Zhiling Ye, Haonan Wang, Jian Wang, Jinjie Gu, Jian Kang

General AI

Sparse on-policy distillation (OPD) allocates teacher supervision to a small subset of tokens in student-generated trajectories. However, useful teacher guidance can yield a noisy update when its gradient is estimated from a sampled next token. We study this estimation problem at a fixed prefix in information geometry …

Review
pending
Role
unreviewed
Read
soon
arxiv Score 8.3

Learning to Drive on Mars: Visual Multimodal Traversability Estimation for Off-World Navigation

2026-09-21 · Darren Chiu, Cole Wilson, Andrei Tumbar, Gaurav S. Sukhatme, Steven Myint

General AI

Autonomous navigation on Mars requires vehicles to distinguish between traversable terrains across diverse and visually challenging environments. However, progress in learning-based navigation for off-world environments has been limited by the lack of large-scale datasets. Since landing in Jezero Crater, the Mars 2020 …

Review
pending
Role
unreviewed
Read
soon
arxiv Score 8.3

Pinocchio: Fast Uncertainty Estimates for Black-Box Language Models

2026-09-21 · Kevin David Hayes, Arka Pal, Haosong Zhang, Tom Goldstein, Micah Goldblum

General AI

In high-stakes decision-making applications of large language models (LLMs), practitioners require not only accurate LLMs but also uncertainty estimates for their predictions. Existing approaches to uncertainty estimation for LLMs require access to log-probabilities output by the model or require fine-tuning access. Ho…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 8.3

WorldCrafter: Consistent Video World Model with Implicit 3D-aware Memory

2026-09-21 · Wangbo Yu, Kunhao Liu, Wenbo Hu, Shenghai Yuan, Chaoran Feng, Haiyang Zhou, Yukun Huang, Yiran Wang, Wang Zhao, Yingmin Luo, Ying Shan

General AI

Video world models enable interactive exploration of dynamic environments, yet struggle to respect prior observations over long horizons and across viewpoints. We present WorldCrafter, a video world model that learns a camera-queryable implicit 3D-aware memory for this purpose. The key insight is to let the requested v…

Review
pending
Role
unreviewed
Read
soon
arxiv Score 7.6

From Ranked Documents to Reliable Contexts: An Answer-Oriented Context Construct Framework for AI Search

2026-09-20 · Yunfei Zhong, Yinqiong Cai, Lixin Su, Haosheng Qian, Lixin Zou, Yixing Fan, Sheng Xu, Jiafeng Guo, Daiting Shi, Jingzhou He

General AI

Traditional Web search follows a human-facing paradigm in which users inspect ranked documents and synthesize information themselves. In AI Search, retrieved documents instead serve as inputs to a generation model, shifting the retrieval objective from ranking documents by Search Satisfaction to constructing reliable c…

Review
pending
Role
unreviewed
Read
later
huggingface Score 7.5

Streaming Video Editing with Easy Adaptation

2026-09-21 · Yujia Hu, Jiajun Li, Zihao He, Songhua Liu

General AI

In this paper, we propose SVEET, a framework that requires merely training on a pretrained bidirectional video diffusion model but supports high-quality streaming video editing in an auto-regressive fashion. To tackle this problem, we first systematically revisit existing video-to-video diffusion approaches and identif…

Review
pending
Role
unreviewed
Read
later
arxiv Score 7.3

PixelDiT2: Representation-Grounded Pixel Diffusion Transformers

2026-09-21 · Yongsheng Yu, Wei Xiong, Yichen Sheng, Shiqiu Liu, Jiebo Luo

General AI

Recent advances in pixel-space diffusion models have narrowed the image quality gap with latent-space diffusion, but still converge more slowly and lag behind in final image quality. We argue that a key reason is the lack of an explicit representation prior: unlike latent diffusion, which usually denoises in a compact …

Review
pending
Role
unreviewed
Read
later
huggingface Score 6.8

Grounded Action Model: 3D Grounding as a Foundation for Robotics

2026-09-20 · Gehao Zhang, Weikai Huang, Shailesh Shailesh, Yiyan Peng, Jiafei Duan, Ranjay Krishna

General AI

Manipulation policies must know which objects matter and where they are, yet the pretrained backbones that current robot foundation models build on, from language in vision-language-action models (VLAs) to video generation in world-action models (WAMs), do not directly require this metric grounding, leaving it to be le…

Review
pending
Role
unreviewed
Read
later
arxiv Score 6.6

BrainIAC: Interactive 3D Brain Lesion Segmentation across Heterogeneous MRI Modalities with Online Adaptation

2026-09-19 · Wentian Xu, Anthony P Addison, Ziyun Liang, Harry Anthony, Guang Yang, Konstantinos Kamnitsas

General AI

Brain lesion segmentation is a fundamental task in medical image analysis, playing a critical role in diagnosis, treatment planning, and longitudinal disease monitoring. Yet existing models still struggle to meet the demands of real clinical use, where deployments contain data distribution shifts, arising from differen…

Review
pending
Role
unreviewed
Read
later
huggingface Score 5.0

Measuring the Checker: Mutation Analysis for GPU-Kernel Benchmark Oracles

2026-09-02 · Mingzhe Du, Anh Tuan Luu, Dong Huang, See-Kiong Ng

General AI

Benchmarks for LLM-generated GPU kernels decide correctness with a few random inputs and a loose floating-point tolerance, and their verdicts now feed leaderboards and reinforcement-learning rewards. Recent work agrees these checkers are weak and patches them by hand---extra input distributions, fuzzing recipes, tighte…

Review
pending
Role
unreviewed
Read
later