arXiv subjects

Tianle Chen

Publications and source records attributed to Tianle Chen.

At least 19 recordsLinked to original sources

arXiv2026

Belief-Conditioned Multi-Agent Actor Critic for Decentralized LLM Collaboration

Recent work has explored optimizing LLM collaboration through Multi-Agent Reinforcement Learning (MARL), but many approaches still rely on predefined workflows, limiting efficiency and scalability. Decentralized LLM collaboration is challenging because agents execute independently with limited communication, share a joint objective, and must infer useful information from communication traces. In this paper, we propose BCMAAC, a Belief-Conditioned Multi-Agent Actor-Critic method that maintains beliefs over contextual information and uses them to facilitate training. Experiments on Minecraft game-playing, web searching and tool-using tasks show that BCMAAC improves task performance and collaborative behavior compared to naive actor-critic baselines. The improvements are particularly pronounced in multi-turn tasks with richer contextual information, where belief modeling provides a clear advantage.

cs.AI↗

arXiv2026

Beyond the Teacher: Understanding and Preserving Distillation Overshoot in Agentic On-Policy Distillation

On-policy distillation (OPD) trains agents on their own trajectories using supervision from a stronger teacher. We uncover a counterintuitive phenomenon in multi-turn agentic tasks: a smaller, initially weaker student can substantially outperform its teacher during training, only to regress as distillation continues. We term this phenomenon distillation overshoot. Through decision- and task-level analyses, we trace overshoot to two competing effects of OPD: repairing weak student decisions and eroding successful student choices that the teacher disfavors. Early in training, repair outpaces erosion, allowing the student to combine the teacher's corrections with its own strengths and surpass the teacher. As training continues, erosion overtakes repair, causing these gains to vanish. Motivated by this finding, we propose Success-Preserving On-Policy Distillation (SPOD), which selectively masks conflicting teacher supervision on successful student choices and introduces a soft probability floor to counteract their erosion, while retaining teacher guidance elsewhere. SPOD requires only binary episode success and model probabilities already available during OPD. Across two model families and three agentic benchmarks, SPOD consistently outperforms ordinary OPD and evaluated variants, improving final success by up to 15.2 percentage points and surpassing the teacher by up to 20.0 points. Moreover, SPOD's final checkpoints outperform even ordinary OPD's best intermediate checkpoints in all six settings, demonstrating that beyond-teacher gains can be not only preserved but further extended. Code: https://github.com/Jike338/SPOD

cs.LG↗

arXiv2026

DeLoop: Repeated Actions as Self-Supervision for Long-Horizon Agents

Language-model agents that run long tasks often redo work they have already done: they re-issue a tool call they made earlier and get back exactly the same observation, a step that adds no new information. We identify this failure as a loop. Loops appear natively in Qwen, Llama, and Gemini agents and become far more common once the agent's history is compressed to fit its context window. To reduce these loops, we propose DeLoop, a self-supervised training method. DeLoop builds on a simple observation: the agent's own rollout can provide the supervision needed to learn from its loops. We use a looping action as a negative example and the first later non-looping action in the same rollout as a positive example. DeLoop learns from these pairs through direct preference optimization and does not require task rewards, gold answers, or LLM judges. On CodeGym under compression, DeLoop reduces the share of Qwen3-8B tasks containing loops from 57.3% to 6.1%. It also uses about one third as many steps per task, with similar task success. On Qwen3-32B, DeLoop reduces loops and steps while raising task success from 37.7% to 44.8%.

cs.AI↗

arXiv2026

DTDD: Divergence-Triggered Dynamic Distillation for Reliable On-Policy Supervision

On-policy distillation (OPD) provides dense token-level supervision on student rollouts, but this supervision can become unreliable when the student and teacher diverge, either because of a large initial policy gap or compounding errors in long-horizon reasoning and tool use. We introduce Divergence-Triggered Dynamic Distillation (DTDD), an adaptive OPD framework that uses segment-level student-teacher divergence to trigger teacher takeover during training. When divergence triggers an intervention during reasoning, the student prefix continues to receive standard OPD supervision. The teacher then generates a recovery suffix, which the student learns through behavior cloning, using cross-entropy on the teacher-generated tokens. We evaluate DTDD in three regimes: long-horizon reasoning, distillation from a base initialization without offline supervised fine-tuning (SFT), and multi-turn agents. We also compare alternative trigger designs and teacher-intervention rates. DTDD reduces the post-warmup peak gradient norm by $8.6\times$ and, from a base initialization, reaches $21.3\%$ held-out math accuracy in $2.6\times$ fewer optimizer updates than OPD; both methods finish near $22\%$. It improves best-observed performance by $4.8$ and $4.1$ percentage points on ALFWorld with LoRA and agentic code, respectively. Ablations show that segment-averaged triggers outperform token-spike triggers. Behavior cloning is also more effective than the tested OPD surrogate for training teacher-generated recoveries. Across 23 experimental arms, takeover tends to help when the teacher substantially outperforms the OPD-trained student, provided that the teacher can recover from student-visited states, the student can learn those recoveries, and teacher intervention is neither excessive nor insufficient.

cs.LG↗

arXiv2026

A Systematic Study of Cross-Modal Typographic Attacks on Audio-Visual Reasoning

As audio-visual multi-modal large language models (MLLMs) are increasingly deployed in safety-critical applications, understanding their cross-modal vulnerabilities is essential. To this end, we introduce Multi-Modal Typography, a systematic study examining how perturbations added to speech and on-screen text impact audio-visual reasoning of closed- and open-source models. For instance, on the WorldSense benchmark, the attack success rate (ASR) against Gemini 3.7 Flash increased from 9.53\% to 43.41\%, highlighting its vulnerability. We analyze the cross-modal effects of unimodal and multimodal attacks and evaluate how specific properties of the attacks impact the overall multi-modal reasoning. By applying fine-tuning as a mitigation, we lower the ASR of Qwen2.5-Omni-7B on WorldSense from $60.6\%$ to $52.0\%$. Our findings across multi-modal reasoning tasks, content moderation, and offline robot task selection benchmarks establish multi-modal typography as a critical and underexplored attack strategy in multi-modal reasoning. Code and data will be made publicly available.

cs.CV↗

arXiv2026

AIM: Adaptive Interaction Modeling Networks for Real-to-Sim Soft-Body Simulation

Deformable-object manipulation is essential for robotic tasks such as folding laundry and handling food, where robots must control shape changes as well as object motion. Predictive soft-body simulation supports these tasks by anticipating deformation under external interactions. However, spatial neighborhoods can misrepresent deformation dependencies, introducing local errors that accumulate over successive predictions. Models fitted to individual scenes must also accommodate changes in object geometry and manipulation conditions. In this work, we propose AIM, an Adaptive Interaction Modeling framework that treats real-to-sim soft-body simulation as a local-global interaction modeling problem. AIM uses motion history and geometry to adapt particle relations over current spatial neighbors and retained connections, while geometry-conditioned global communication coordinates object-wide responses. A unified kinematic control-point interface represents different manipulation configurations, and multi-step autoregressive supervision trains the model on its own predicted trajectories. Experiments on PhysTwin and PGND demonstrate improved motion accuracy and visual fidelity, with a 20.0% reduction in future-prediction tracking error relative to PhysTwin and a 22.8% reduction in mean long-horizon particle error across six object categories relative to PGND. The framework further supports transfer across actions, object instances, and scenes, including zero-shot transfer from robot interactions to human manipulation without target-domain dynamics fitting.

cs.RO↗

arXiv2026

Improving LLM Collaboration via Multi-Agent Preference Learning

Several works have explored multi-agent reinforcement learning (MARL) in LLM collaboration. However, constructing reliable rewards is difficult in practice, as complete and accurate metrics are often unavailable and hard to aggregate. Preference learning provides an alternative by learning from comparative human or AI feedback. Yet, its extension to multi-agent systems remains underexplored. To address this gap, we formulate preference-based multi-agent systems (MAS) from decentralized and centralized collaboration perspectives. We also introduce a general multi-agent preference learning framework (MAPL) to solve these problems. MAPL allows iterative updates by comparing the current solution with decentralized or centralized solutions generated by various agents. We instantiate MAPL using MARL from human feedback (MARLHF) with a learned reward model and multi-agent direct preference optimization (MADPO). Experiments on collaborative writing, coding, tool use, and travel planning show that MAPL can improve collaboration quality and efficiency while approaching the performance of MARL with fixed, well-defined rewards. Within MAPL, MARLHF generally outperforms MADPO on most tasks but remains sensitive to data coverage, agent and comparator models, and the underlying MARL algorithms.

cs.AI↗

arXiv2026

Multiclass Semantic Segmentation of Wildland Fire Images Using Context-Aware Centralized Copy-Paste Data Augmentation

Producing accurate annotations for deep learning based image segmentation is both costly and labor intensive. This challenge is especially evident in wildland fire applications, where accurately labeled datasets are scarce due to the difficulty of collecting and annotating dynamic fire scenes. To address this problem, our previous work introduced the Centralized Copy-Paste Data Augmentation (CCPDA) method for semantic segmentation of wildland fire imagery, which generates artificial training samples by randomly pasting fire clusters from source images onto target images. However, random placement can produce contextually unrealistic scenes, such as fire burning on asphalt. In this paper, we present a context-aware strategy designed specifically to improve data quality and realism in small multiclass wildland fire datasets, ensuring that augmented samples remain contextually meaningful. The proposed method restricts fire placement to semantically valid target regions and selects the location whose Ash-Vegetation composition most closely matches the source context. This approach preserves existing fire regions in the target image, prevents unrealistic placements, and maintains contextual accuracy by generating images that resemble real wildland fire scenes. We evaluate the Context-Aware CCPDA strategy through numerical analysis and comparisons with other augmentation methods by a weighted sum-based multi-objective optimization (MOO) approach. The results confirm that the context-aware data augmentation strategy leads to improved segmentation performance and contextual realism, outperforming other augmentation procedures.

cs.CV↗

arXiv2026

EgoAsk: Egocentric Teaching of Personalized Object Knowledge for Household Robots

Unlike users, who know their own belongings and routines, household robots cannot easily acquire such personalized object knowledge automatically and depend on users to teach them. User-initiated teaching requires users to arrange dedicated teaching sessions and decide what to teach, even when they are unsure what the robot needs to learn. We introduce EgoAsk, a smart-glasses-based system that proactively embeds personalized object teaching into everyday activities. EgoAsk shares the user's first-person view with the robot, identifies gaps in personalized object knowledge, and analyzes ongoing activity to ask context-relevant questions that support future household assistance. To examine how teaching initiative and question timing affect users' teaching experiences, we conducted a within-subjects study with 18 participants and found lower reported knowledge-gap monitoring burden with robot-initiated questioning and less need for context reconstruction with EgoAsk. These findings characterize teaching burdens and timing preferences, offering design implications for egocentric robot-teaching systems.

cs.HC↗

arXiv2026

Fewer Tokens, Smaller Cache: Reward-Coordinated Efficient Reasoning

Large Reasoning Models (LRMs) excel on complex tasks through long chain-of-thought (CoT) reasoning, but their lengthy intermediate steps cause severe overthinking that inflates inference cost. KV-cache compression is a common solution, yet existing reasoning-oriented methods apply a uniform policy across the trajectory and judge compression only by what it removes from the cache. Two observations point the other way. First, a reasoning state's tolerance to context loss varies along the trajectory, and process reward tracks it: deleting tokens at high-reward steps preserves accuracy far better than deleting the same budget at random. Second, compression is not free on the generation side, since a smaller cache leads the model to generate more tokens, partly canceling the saving. Together these motivate coordinating both sides under a single process reward. We propose ReCo (Reward-Coordinated Compression), a step-wise framework in which a lightweight process-reward estimator scores each completed step and drives three components: (1) reward-adaptive KV-cache compression that shrinks the retained cache harder at high-reward steps and less at low-reward ones, (2) a reward-banded penalty on reflection tokens that curbs redundant generation, and (3) confidence-based early stopping that triggers when the reasoning is reliable. Across three reasoning models and six benchmarks, ReCo reduces generated tokens by 37%-65% and end-to-end latency by 2.08x-2.35x over Full CoT, all while largely preserving accuracy.

cs.AI↗

arXiv2026

FieryGS: In-the-Wild Fire Synthesis with Physics-Integrated Gaussian Splatting

We consider the problem of synthesizing photorealistic, physically plausible combustion effects in in-the-wild 3D scenes. Traditional CFD and graphics pipelines can produce realistic fire effects but rely on handcrafted geometry, expert-tuned parameters, and labor-intensive workflows, limiting their scalability to the real world. Recent scene modeling advances like 3D Gaussian Splatting (3DGS) enable high-fidelity real-world scene reconstruction, yet lack physical grounding for combustion. To bridge this gap, we propose FieryGS, a physically-based framework that integrates physically-accurate and user-controllable combustion simulation and rendering within the 3DGS pipeline, enabling realistic fire synthesis for real scenes. Our approach tightly couples three key modules: (1) multimodal large-language-model-based physical material reasoning, (2) efficient volumetric combustion simulation, and (3) a unified renderer for fire and 3DGS. By unifying reconstruction, physical reasoning, simulation, and rendering, FieryGS removes manual tuning and automatically generates realistic, controllable fire dynamics consistent with scene geometry and materials. Our framework supports complex combustion phenomena -- including flame propagation, smoke dispersion, and surface carbonization -- with precise user control over fire intensity, airflow, ignition location and other combustion parameters. Evaluated on diverse indoor and outdoor scenes, FieryGS outperforms all comparative baselines in visual realism, physical fidelity, and controllability. Project page can be found at https://pku-vcl-geometry.github.io/FieryGS/.

cs.GR↗

arXiv2026

S1-Omni: A Unified Multimodal Reasoning Model for Scientific Understanding, Prediction, and Generation

We present S1-Omni, a unified multimodal reasoning model for scientific understanding, prediction, and generation. AI for Science (AI4S) has advanced significantly through domain-specific models, tool-augmented LLMs, and scientific language models. However, model capabilities remain highly fragmented, limiting the joint modeling of heterogeneous data, scientific laws, and expert knowledge. S1-Omni addresses this gap by consolidating these capabilities into a single, coherent scientific reasoning model. The architecture of S1-Omni is built upon three core components: unified representation of scientific data, natural-world knowledge alignment, and decoding for domain-specific tasks. First, S1-Omni maps natural-language instructions and scientific objects, including CIF, SMILES, protein sequences, spectra, and scientific images, into a shared representation space. Second, it incorporates scientific laws and expert knowledge into data construction and training, enabling the model to reason from scientific evidence. Third, it performs task-specific decoding to support a broad range of applications, including property prediction, spectrum-to-molecular generation, protein site and structure prediction, and scientific image generation and editing. S1-Omni is trained on S1-Omni-Corpus, which covers 200 scientific tasks and contains millions of reasoning samples, and is evaluated on over 60 scientific benchmarks. It outperforms GPT-5.5 and Gemini-3.1-Pro on most benchmarks and matches or surpasses domain-specific models on several benchmarks. Overall, S1-Omni provides a practical path toward unified scientific modeling.

cs.AI↗

arXiv2026

Imprinting topology on thermal light

Topological structuring of light inevitably leverages on optical coherence to ensure that the imparted spatial phases are preserved, requiring highly coherent sources or coherence engineering embedded in the design. Now we show that thermal light can be spatially engineered to carry optical topologies in the form of Skyrmions. Such topologies are immune to time averaged decoherence, a fact we leverage on in reverse to create metasurface mediated incoherent topologies from a thermal source. The pristine nature of our measured Skyrmions validates the approach, while simulations reveal how coherence management in the metasurface design would further enhance the functionality. Remarkably, the generation stage inherits robustness from the topology, remaining immune to material and fabrication defects. Our work reports the first topologies from purely thermal light, opening a path to exploiting topology in ubiquitous everyday light sources.

physics.optics↗

arXiv2026

Learning Decentralized LLM Collaboration with Multi-Agent Actor Critic

Recent work has explored optimizing LLM collaboration through Multi-Agent Reinforcement Learning (MARL). However, most MARL fine-tuning approaches rely on predefined execution protocols, which often require centralized execution. Decentralized LLM collaboration is more appealing in practice, as agents can run inference in parallel with flexible deployments. Also, current approaches use Monte Carlo methods for fine-tuning, which suffer from high variance and thus require more samples to train effectively. Actor-critic methods are prevalent in MARL for dealing with these issues; thus, we developed Multi-Agent Actor-Critic (MAAC) methods to optimize decentralized LLM collaboration. In this paper, we analyze when and why these MAAC methods are beneficial. We propose 2 MAAC approaches, \textbf{CoLLM-CC} with a \textbf{C}entralized \textbf{C}ritic and \textbf{CoLLM-DC} with \textbf{D}ecentralized \textbf{C}ritics. Our experiments across writing, coding, and game-playing domains show that Monte Carlo methods and CoLLM-DC can achieve performance comparable to CoLLM-CC in short-horizon and dense-reward settings. However, they both underperform CoLLM-CC on long-horizon or sparse-reward tasks, where Monte Carlo methods require substantially more samples and CoLLM-DC struggles to converge.

cs.AI↗

arXiv2026

Centralized Copy-Paste: Enhanced Data Augmentation Strategy for Wildland Fire Semantic Segmentation

Collecting and annotating images for the purpose of training segmentation models is often cost prohibitive. In the domain of wildland fire science, this challenge is further compounded by the scarcity of reliable public datasets with labeled ground truth. This paper presents the Centralized Copy-Paste Data Augmentation (CCPDA) method, for the purpose of assisting with the training of deep-learning multiclass segmentation models, with special focus on improving segmentation outcomes for the fire-class. CCPDA has three main steps: (i) identify fire clusters in the source image, (ii) apply a centralization technique to focus on the core of the fire area, and (iii) paste the refined fire clusters onto a target image. This method increases dataset diversity while preserving the essential characteristics of the fire class. The effectiveness of this augmentation technique is demonstrated via numerical analysis and comparison against various other augmentation methods using a weighted sum-based multi-objective optimization approach. This approach helps elevate segmentation performance metrics specific to the fire class, which carries significantly more operational significance than other classes (fuel, ash, or background). Numerical performance assessment validates the efficacy of the presented CCPDA method in alleviating the difficulties associated with small, manually labeled training datasets. It also illustrates that CCPDA outperforms other augmentation strategies in the application scenario considered, particularly in improving fire-class segmentation performance.

cs.CV↗

arXiv2026

VLM-UQBench: A Benchmark for Modality-Specific and Cross-Modality Uncertainties in Vision Language Models

Uncertainty quantification (UQ) is vital for ensuring that vision-language models (VLMs) behave safely and reliably. A central challenge is to localize uncertainty to its source, determining whether it arises from the image, the text, or misalignment between the two. We introduce VLM-UQBench, a benchmark for modality-specific and cross-modal data uncertainty in VLMs, It consists of 600 real-world samples drawn from the VizWiz dataset, curated into clean, image-, text-, and cross-modal uncertainty subsets, and a scalable perturbation pipeline with 8 visual, 5 textual, and 3 cross-modal perturbations. We further propose two simple metrics that quantify the sensitivity of UQ scores to these perturbations and their correlation with hallucinations, and use them to evaluate a range of UQ methods across four VLMs and three datasets. Empirically, we find that: (i) existing UQ methods exhibit strong modality-specific specialization and substantial dependence on the underlying VLM, (ii) modality-specific uncertainty frequently co-occurs with hallucinations while current UQ scores provide only weak and inconsistent risk signals, and (iii) although UQ methods can rival reasoning-based chain-of-thought baselines on overt, group-level ambiguity, they largely fail to detect the subtle, instance-level ambiguity introduced by our perturbation pipeline. These results highlight a significant gap between current UQ practices and the fine-grained, modality-aware uncertainty required for reliable VLM deployment.

cs.CV↗

arXiv2025

LLM Collaboration With Multi-Agent Reinforcement Learning

A large amount of work has been done in Multi-Agent Systems (MAS) for modeling and solving problems with multiple interacting agents. However, most LLMs are pretrained independently and not specifically optimized for coordination. Existing LLM fine-tuning frameworks rely on individual rewards, which require complex reward designs for each agent to encourage collaboration. To address these challenges, we model LLM collaboration as a cooperative Multi-Agent Reinforcement Learning (MARL) problem. We develop a multi-agent, multi-turn algorithm, Multi-Agent Group Relative Policy Optimization (MAGRPO), to solve it, building on current RL approaches for LLMs as well as MARL techniques. Our experiments on LLM writing and coding collaboration demonstrate that fine-tuning MAS with MAGRPO enables agents to generate high-quality responses efficiently through effective cooperation. Our approach opens the door to using other MARL methods for LLMs and highlights the associated challenges. Our code is available at https://github.com/OpenMLRL/CoMLRL.

cs.AI↗

arXiv2025

Some Modalities are More Equal Than Others: Decoding and Architecting Multimodal Integration in MLLMs

Despite remarkable advancements in Multimodal Large Language Models (MLLMs), a fundamental question remains: are MLLMs robust to contradicting modalities? To rigorously study this, we introduce MMA-Bench comprising videos and tasks that probe a model's reliance on specific modalities. Using black-box and white-box interpretability techniques, we provide a critical analysis of the brittleness of both open- and closed-sourced MLLMs. We show that current MLLMs struggle under misaligned audio-visual pairs and simple misleading text, thereby lacking robust multi-modal reasoning. Building on these findings, we propose a modality alignment tuning strategy to teach the model when to prioritize, leverage, or ignore specific modality cues. Through extensive experiments and analysis, we show that our alignment tuning yields demonstrably stronger multimodal grounding. This work provides both interpretability tools and a clear path toward developing MLLMs with intrinsically reliable cross-modal reasoning. Code and dataset will be publicly available.

cs.CV↗