Searcharxiv⌕ Search

arXiv subjects

Junzhe Wang

Publications and source records attributed to Junzhe Wang.

At least 19 recordsLinked to original sources

DriveFoundry: Co-Evolving Data and Model for Efficient Closed-Loop Iteration in Autonomous Driving

A driving foundation model must learn reusable decision knowledge from rare critical events and apply it under strict runtime constraints. We introduce DriveFoundry, a closed-loop framework improving iteration efficiency under data and computation budgets by coupling structured causal data refinement with adaptive inference. The Data Engine builds traceable scene-candidate records linking scene evidence and factor-decision relations to candidate trajectories and outcome assessments. Candidate comparisons and counterfactual annotations clarify how scene factors affect behavioral feasibility and preference. Verified model failures guide VLM-assisted annotation and expert correction, refining training data and annotation systems. The model engine, Xiaomi-Drive-o1, unifies Perception, Causation, Action, World, and Value within a shared vision-language backbone to enable cross-task knowledge transfer. Block-wise latent refill passes candidate-specific evidence and predicted visual consequences between task blocks, connecting scene understanding, planning, and evaluation. A learned value-of-computation head selects tasks and allocates latent tokens based on expected benefit and cost, adapting inference to each decision. On PhysicalAI, the 9B model achieves the lowest ADE@1 across evaluated splits and horizons among tested models. In AlpaSim, it outperforms Alpamayo 1.5 across all metrics and achieves lower close-encounter and off-road rates than the 34B Alpamayo 2 Super, while trailing in progress and AlpaSim scores. For the 0.8B variant, joint training reduces 5 s minADE@6 by about 54% without altering the inference path or token budget. Adaptive inference matches full reasoning performance using only 28% of tokens. Real-vehicle tests confirm runtime feasibility, supporting a unified paradigm for learning from critical experience and dynamically allocating computation.

cs.AI↗

Trajectory Evaluation via Rollout-Free World Model for End-to-end Autonomous Driving

Although existing candidate trajectory evaluators have substantially improved end-to-end planning accuracy, their underlying mechanism remains suboptimal. Model-based evaluators typically make scoring future-aware by explicitly predicting candidate-conditioned future states over the planning horizon, which we term rollout. However, rollout incurs high inference cost and accumulates prediction errors. We argue that planning does not require reconstructing a full future scene for each candidate, but only future-relevant information for reliable comparison. We therefore propose RFWorld, a Rollout-Free World model for end-to-end trajectory evaluation. RFWorld constructs a shared scene memory and shapes it into a future-readable representation via lightweight training-time temporal readout with semantic supervision, keeping training overhead low. At inference time, candidate trajectories directly query the shared memory via sparse spatial readout for scoring, reducing computation by roughly the planning horizon times the number of candidate trajectories, while avoiding recursive rollout errors. Experiments on NAVSIM and Bench2Drive show that RFWorld learns a more planning-aligned world representation and achieves state-of-the-art performance.

cs.RO↗

QREF: Improving DeepSearch Agents by Aligning Search with Reasoning via Query Refinement and Evidence Filtering

DeepSearch agents achieve strong performance on complex information-seeking tasks through iterative search and reasoning. However, they often issue intent-misaligned, under-constrained, or redundant search queries, resulting in poor evidence coverage and accumulating retrieval noise. To mitigate these limitations, we propose QREF, a plug-and-play Query Refiner and Evidence Filter. Specifically, the Query Refiner rewrites search queries according to the current reasoning intent while ensuring complementary coverage across each search batch. The Evidence Filter retains evidence relevant to both the current search batch and the reasoning objective, reducing distracting retrieval results. We initialize the Query Refiner with supervised fine-tuning and optimize it through reinforcement learning based on interactions with a real-world web search environment, reducing ineffective queries. Comprehensive experiments demonstrate that our approach improves the performance of DeepSearch agents across four challenging benchmarks, achieving average relative Pass@1 improvements of 33.55% on WebSailor and 11.04% on TongyiDeepResearch without modifying architectures or retraining their reasoning policies. Compared with the original DeepSearch backbone without QREF, the RL-trained refiner reduces the fraction of query batches that lack discriminative retrieval anchors by 95%. Compared with the SFT-trained refiner, it also lowers the filter-based ineffective query ratio.

cs.AI↗

GLIO2: A GPU-Parallelized Tightly-Coupled LiDAR-Inertial-GNSS System for Robust and Real-Time Global Localization and Mapping

Globally consistent, real-time state estimation in large-scale, perceptually degraded environments is essential for autonomous vehicles and aerial robots, and requires fusing LiDAR, inertial, and GNSS measurements. Existing fusion methods, however, share a scan-to-map front-end with two failure modes. First, each scan is aligned to an incrementally built map that drifts under degeneracy, and once the estimate diverges the error is irrecoverable. Second, even without divergence, a registration biased by dynamic objects or wrong correspondences is propagated as a single pose constraint with an over-confident covariance, leaving its correspondences unavailable for GNSS to re-weight or relinearize. We propose GLIO2, a tightly-coupled LiDAR-Inertial-GNSS system whose GPU-parallel front-end jointly optimizes scan-to-multiscan LiDAR, IMU pre-integration, and raw GNSS measurements in a single sliding-window factor graph, sustaining real-time operation on edge hardware. A complementary offline back-end reuses the same cached factors to refine the entire trajectory in batch, completing the 30-min, 4.51-km UrbanNav Whampoa sequence in about 24 s. Across three public benchmarks (UrbanNav, MARS-LVIG, M3DGR) and self-collected UAV and vehicle data, GLIO2 attains the best overall accuracy among evaluated systems. On a 5.66-km bridge traversed at up to 96 km/h, where every competing baseline diverges under LiDAR degeneracy, it maintains 1.6 m horizontal accuracy. On an NVIDIA Jetson Orin NX, the full pipeline runs at about 25 Hz (39.60 ms per scan). The source code and datasets will be released.

cs.RO↗

ReDraft, Don't Just Distill: Reference-Driven Revision for Continual VLLM Post-Training

Continual post-training of large multimodal models should add new capabilities while preserving those from pre-training, and the two goals pull in opposite directions. SFT gives explicit target supervision that learns a task from near-zero accuracy, but its off-policy targets move the model far enough to cause forgetting; on-policy methods such as RLVR and self-distillation preserve policy proximity yet supply little signal when the policy cannot yet solve the task. We introduce ReDraft (Reference-Driven Revision and Fine-Tuning), which obtains both from the model's own failures: using an expert response only as a reference, it has the model revise its own incorrect rollout, keeps the revision only if a verifier accepts it, and fine-tunes on what survives. Each retained target is therefore explicit, yet still close to the current policy. Across Counting, Clock Reading, and Jigsaw on Qwen2.5-VL-3B/7B, two of them with near-zero accuracy, ReDraft gains 56.9 points on the target task against SFT's 52.9 while cutting prior-task loss from 16.6 to 1.5 points (11.3x less forgetting), and improves on OPSD along both axes (19.3 gain, 6.2 loss). Data- and parameter-space analyses match the design: revised targets are more probable under the base model, and the updates they induce stay compact and follow SFT's direction more closely than OPSD's. Together, these results show that revising the model's own rollout rather than directly imitating an expert trajectory can reconcile cold-start acquisition with prior-capability retention.

cs.AI↗

From Gameplay to Policy: Towards Scalable Robot Data Collection via Gamified Robot-Free Interaction

Learning generalizable robot manipulation policies requires large-scale and diverse interaction data, yet collecting real-world demonstrations remains costly and difficult to scale. Existing approaches to data collection are either dependent on specific robot hardware that limits crowdsourcing and transferability, or suffer from incomplete annotation and limited behavioral diversity. Inspired by how games sustain long-term human engagement, we explore an alternative paradigm that turns data collection into an engaging gameplay experience and transfers the resulting human manipulation experience to real robots. We present Project Kitchen, a VR-based gamified egocentric data collection platform that elicits diverse, goal-directed manipulation while remaining independent of specific robot embodiments and hardware, making it applicable to broader and potentially large-scale deployment. To bridge the game-to-real gap, we further introduce Game2Policy, which extracts embodiment-invariant affordance cues, including contact points and sub-goal states, from gameplay trajectories. An affordance model is pre-trained on game-collected data and then jointly fine-tuned with downstream policies using only a handful of real-robot demonstrations. Experiments show that Game2Policy improves average success rates by 10.0 points in simulation and 18.3 points on real robots in the few-shot setting. User studies and quantitative analyses further show that Project Kitchen promotes diverse manipulation behaviors and provides an engaging data collection experience. These results demonstrate the potential of gamified virtual environments as a scalable source of manipulation knowledge. The platform and code will be released upon acceptance.

cs.RO↗

Teaching Foundation Models to Read mmWave: Pose-Guided Kinematic Representation for Human Behavior Understanding

Large language model agents need to perceive human behavior in physical environments. Millimeter-wave (mmWave) radar provides a privacy-friendly and contactless sensing modality, but radar observations are difficult to align with language. Existing radar-language methods often rely on synthetic data or lack explicit supervision for human body structure and motion. We present mmMind, a radar-language model that uses synchronized 3D pose as training-only supervision. A spatio-temporal radar encoder is pretrained to capture body configuration and motion dynamics, after which the pose head is removed so that inference requires radar alone. The learned radar representations are then aligned with an LLM for behavior captioning and spatio-temporal question answering. We also introduce mmMind-Bench, a real-world mmWave-language benchmark containing 17.9 hours of recordings from 23 participants across seven indoor environments. Experiments on captioning, question answering, and unseen-action generalization show that mmMind consistently outperforms existing radar-language baselines, while ablations confirm the importance of pose-guided pretraining.

cs.CV↗

LLM-Based Scientific Equation Discovery via Physics-Informed Token-Regularized Policy Optimization

Symbolic regression aims to distill mathematical equations from observational data. Recent approaches have successfully leveraged Large Language Models (LLMs) to generate equation hypotheses, capitalizing on their vast pre-trained scientific priors. However, existing frameworks predominantly treat the LLM as a static generator, relying on prompt-level guidance to steer exploration. This paradigm fails to update the model's internal representations based on search feedback, often yielding physically inconsistent or mathematically redundant expressions. In this work, we propose PiT-PO (Physics-informed Token-regularized Policy Optimization), a unified framework that evolves the LLM into an adaptive generator via reinforcement learning. Central to PiT-PO is a dual-constraint mechanism that rigorously enforces hierarchical physical validity while simultaneously applying fine-grained, token-level penalties to suppress redundant structures. Consequently, PiT-PO aligns LLM to produce equations that are both scientifically consistent and structurally parsimonious. Empirically, PiT-PO achieves state-of-the-art performance on standard benchmarks and successfully discovers novel turbulence models for challenging fluid dynamics problems. We also demonstrate that PiT-PO empowers small-scale models to outperform closed-source giants, democratizing access to high-performance scientific discovery.

cs.LG↗

AbICL: In-Context Learning for Antigen-Specific Antibody Affinity Ranking

Accurate ranking of antibody candidates according to their binding affinity is essential for therapeutic antibody discovery. However, existing methods treat affinity comparisons independently and ignore the contextual information encoded in other labeled comparisons, limiting their ability to capture antigen-specific binding landscapes. For many target antigens, a small number of experimentally characterized affinity comparisons are often available. An important question is whether the model can exploit these existing comparisons to infer antigen-specific ranking patterns that facilitate subsequent affinity ranking. This form of learning from labeled demonstrations closely resembles the paradigm of In-Context Learning, motivating us to revisit antibody affinity ranking from an ICL perspective. To this end, we propose AbICL, an ICL framework for antigen-specific antibody affinity ranking. AbICL combines a pretrained structural encoder with a context ranking head and is trained with an episodic meta-training strategy that enables the model to leverage support demonstrations for test-time adaptation without gradient updates. Experiments on the AbRank benchmark demonstrate that AbICL consistently outperforms existing ranking baselines across almost all data splits and evaluation benchmarks. Further analysis shows that the value of contextual demonstrations depends on how well they match the target inference task, and becomes increasingly pronounced under distribution shift and fine-grained affinity discrimination. These findings highlight the potential of ICL as an effective paradigm for antigen-specific antibody affinity ranking, particularly in challenging settings where a single global ranking function is insufficient.

cs.LG↗

AgentGym2: Benchmarking Large Language Model Agents in De-Idealized Real-World Environments

Language agents, i.e., LLM agents, progress rapidly and are increasingly deployed in production environments. This trend underscores the urgent need for rigorous and realistic evaluations. However, most existing benchmarks evaluate agents in simplified, idealized settings. They typically rely on pre-packaged tool interfaces, overlook critical steps, and assume inputs are clean and fully specified. Consequently, they understate the difficulty of real deployments, where uncertainty and noise are ubiquitous and agents must proactively explore the environment to uncover new tools. To bridge this gap, we present AgentGym2, a new evaluation framework with task instances grounded in real-world end-to-end working demands. Beyond reasoning and planning, it measures agents' ability to execute end-to-end procedures, discover tools via exploration, compose tools for unseen tasks, and remain robust to noisy and underspecified information. Experiments on 15 proprietary and open-source models show that even SOTA systems like Gemini and GPT-5 struggle on AgentGym2, revealing a substantial gap between the capability of current agents and the demands of real-world applications.

cs.AI↗

TacSE3: Equivariant SE(3) Motion Estimation from Low-Texture Visuotactile Images for In-Gripper Tracking and Compensation

Robotic in-hand manipulation requires reliable object-motion tracking under frequent visual occlusion, yet low-texture visuotactile images provide few stable correspondences for conventional image- or geometry-matching methods. This paper presents TacSE3, a tactile motion-estimation pipeline that converts low-texture visuotactile observations into a decoupled three-dimensional force field and estimates incremental rigid-body motion on SE(3). The method derives planar translation from contact-centroid motion and estimates rotation primarily from shear-related tactile responses, yielding a physically interpretable signal for in-gripper tracking and compensation. Experiments with paired DM-Tac fingertip sensors show that dual-sensor sensing reduces translation-rotation ambiguity, supports rotation tracking across axes and object geometries, and provides a lightweight compensation signal that improves disturbance tolerance in downstream manipulation tasks without retraining the base policy.

cs.RO↗

Faithfulness-QA: A Counterfactual Entity Substitution Dataset for Training Context-Faithful RAG Models

Retrieval-Augmented Generation (RAG) models frequently produce answers grounded in parametric memory rather than the retrieved context, undermining the core promise of retrieval augmentation. A fundamental obstacle to fixing this unfaithfulness is the lack of training data that explicitly requires models to prefer context over internal knowledge. We introduce Faithfulness-QA, a large-scale dataset of 99,094 samples constructed through counterfactual entity substitution. Starting from two established extractive QA benchmarks--SQuAD and TriviaQA--we automatically identify answer-bearing named entities in each context, replace them with type-consistent alternatives drawn from a curated bank of 76,953 entities, and thereby manufacture controlled knowledge conflicts between context and parametric memory. Rigorous quality filtering ensures 100% pass rates across four automated checks on random 200-sample audits. We release the full dataset, the construction pipeline, and a typed entity bank covering eight named entity categories. Faithfulness-QA is designed as a training resource for attention-based faithfulness objectives and as an evaluation benchmark for measuring context-grounding behavior in RAG systems. Data and code are available at https://github.com/qzhangFDU/faithfulness-qa-dataset.

cs.CL↗

CL-bench Life: Can Language Models Learn from Real-Life Context?

Today's AI assistants such as OpenClaw are designed to handle context effectively, making context learning an increasingly important capability for models. As these systems move beyond professional settings into everyday life, the nature of the contexts they must handle also shifts. Real-life contexts are often messy, fragmented, and deeply tied to personal and social experience, such as multi-party conversations, personal archives, and behavioral traces. Yet it remains unclear whether current frontier language models can reliably learn from such contexts and solve tasks grounded in them. To this end, we introduce CL-bench Life, a fully human-curated benchmark comprising 405 context-task pairs and 5,348 verification rubrics, covering common real-life scenarios. Solving tasks in CL-bench Life requires models to reason over complex, messy real-life contexts, calling for strong real-life context learning abilities that go far beyond those evaluated in existing benchmarks. We evaluate ten frontier LMs and find that real-life context learning remains highly challenging: even the best-performing model achieves only 19.3% task solving rate, while the average performance across models is only 13.8%. Models still struggle to reason over contexts such as messy group chat histories and fragmented behavioral records from everyday life. CL-bench Life provides a crucial testbed for advancing real-life context learning, and progress on it can enable more intelligent and reliable AI assistants in everyday life.

cs.CL↗

Enhancing LLM-based Search Agents via Contribution Weighted Group Relative Policy Optimization

Search agents extend Large Language Models (LLMs) beyond static parametric knowledge by enabling access to up-to-date and long-tail information unavailable during pretraining. While reinforcement learning has been widely adopted for training such agents, existing approaches face key limitations: process supervision often suffers from unstable value estimation, whereas outcome supervision struggles with credit assignment due to sparse, trajectory-level rewards. To bridge this gap, we propose Contribution-Weighted GRPO (CW-GRPO), a framework that integrates process supervision into group relative policy optimization. Instead of directly optimizing process rewards, CW-GRPO employs an LLM judge to assess the retrieval utility and reasoning correctness at each search round, producing per-round contribution scores. These scores are used to rescale outcome-based advantages along the trajectory, enabling fine-grained credit assignment without sacrificing optimization stability. Experiments on multiple knowledge-intensive benchmarks show that CW-GRPO outperforms standard GRPO by 5.0% on Qwen3-8B and 6.3% on Qwen3-1.7B, leading to more effective search behaviors. Additional analysis reveals that successful trajectories exhibit concentrated contributions in specific rounds, providing empirical insight into search agent tasks.

cs.LG↗

LLMEval-Fair: A Large-Scale Longitudinal Study on Robust and Fair Evaluation of Large Language Models

Existing evaluation of Large Language Models (LLMs) on static benchmarks is vulnerable to data contamination and leaderboard overfitting, critical issues that obscure true model capabilities. To address this, we introduce LLMEval-Fair, a framework for dynamic evaluation of LLMs. LLMEval-Fair is built on a proprietary bank of 220k graduate-level questions, from which it dynamically samples unseen test sets for each evaluation run. Its automated pipeline ensures integrity via contamination-resistant data curation, a novel anti-cheating architecture, and a calibrated LLM-as-a-judge process achieving 90% agreement with human experts, complemented by a relative ranking system for fair comparison. A 30-month longitudinal study of nearly 60 leading models reveals a performance ceiling on knowledge memorization and exposes data contamination vulnerabilities undetectable by static benchmarks. The framework demonstrates exceptional robustness in ranking stability and consistency, providing strong empirical validation for the dynamic evaluation paradigm. LLMEval-Fair offers a robust and credible methodology for assessing the true capabilities of LLMs beyond leaderboard scores, promoting the development of more trustworthy evaluation standards. Our code and data are publicly available at https://github.com/llmeval/LLMEval-Fair.

cs.CL↗

MM-Doc-R1: Training Agents for Long Document Visual Question Answering through Multi-turn Reinforcement Learning

Conventional Retrieval-Augmented Generation (RAG) systems often struggle with complex multi-hop queries over long documents due to their single-pass retrieval. We introduce MM-Doc-R1, a novel framework that employs an agentic, vision-aware workflow to address long document visual question answering through iterative information discovery and synthesis. To incentivize the information seeking capabilities of our agents, we propose Similarity-based Policy Optimization (SPO), addressing baseline estimation bias in existing multi-turn reinforcement learning (RL) algorithms like GRPO. Our core insight is that in multi-turn RL, the more semantically similar two trajectories are, the more accurate their shared baseline estimation becomes. Leveraging this, SPO calculates a more precise baseline by similarity-weighted averaging of rewards across multiple trajectories, unlike GRPO which inappropriately applies the initial state's baseline to all intermediate states. This provides a more stable and accurate learning signal for our agents, leading to superior training performance that surpasses GRPO. Our experiments on the MMLongbench-Doc benchmark show that MM-Doc-R1 outperforms previous baselines by 10.4%. Furthermore, SPO demonstrates superior performance over GRPO, boosting results by 5.0% with Qwen3-8B and 6.1% with Qwen3-4B. These results highlight the effectiveness of our integrated framework and novel training algorithm in advancing the state-of-the-art for complex, long-document visual question answering.

cs.CL↗

On blow up NLS with a multiplicative noise

It is of significant interest to understand whether a noise will speed up or prevent blow up. Under certain nondegenerate conditions, \cite{dD2005Blowup} proved a multiplicative noise will speed up blow up of NLS, in the sense that, blow up can happen in any short time with positive probability. We prove that such probability is indeed quite small, and provide a large deviation type upper bound.

math.AP↗

CL-bench: A Benchmark for Context Learning

Current language models (LMs) excel at reasoning over prompts using pre-trained knowledge. However, real-world tasks are far more complex and context-dependent: models must learn from task-specific context and leverage new knowledge beyond what is learned during pre-training to reason and resolve tasks. We term this capability context learning, a crucial ability that humans naturally possess but has been largely overlooked. To this end, we introduce CL-bench, a real-world benchmark consisting of 500 complex contexts, 1,899 tasks, and 31,607 verification rubrics, all crafted by experienced domain experts. Each task is designed such that the new content required to resolve it is contained within the corresponding context. Resolving tasks in CL-bench requires models to learn from the context, ranging from new domain-specific knowledge, rule systems, and complex procedures to laws derived from empirical data, all of which are absent from pre-training. This goes far beyond long-context tasks that primarily test retrieval or reading comprehension, and in-context learning tasks, where models learn simple task patterns via instructions and demonstrations. Our evaluations of ten frontier LMs find that models solve only 17.2% of tasks on average. Even the best-performing model, GPT-5.1, solves only 23.7%, revealing that LMs have yet to achieve effective context learning, which poses a critical bottleneck for tackling real-world, complex context-dependent tasks. CL-bench represents a step towards building LMs with this fundamental capability, making them more intelligent and advancing their deployment in real-world scenarios.

cs.CL↗