论文检索

Deep Learning · Large Language Models

Mahesh Ramesh, Kaousheik Jayakumar, Aswinkumar Ramkumar, Pavan Thodima, Aniket Rege, Emmanouil-Vasileios Vlatakis-Gkaragkounis

Cooperative reasoning under incomplete information remains complex for both humans and multi-agent AI, requiring agents to transcend individual logic in favor of recursive Theory-of-Mind (ToM) and strategic coordination. To investigate these challenges, we conduct a large-scale evaluation of 17 state-of-the-art LLMs (4B–600B+) on Hanabi card game across 2–5 players. To examine their limitations, we analyze the impact of context engineering and scaffold robustness, ranging from minimal prompts (Watson setting) to Bayesian-motivated scaffolding (Sherlock setting) and multi-turn working memory (Mycroft setting). Our findings reveal that: (1) top-performing models can autonomously track game states via internal working memory, although not reliably, and (2) cross-play performance scales smoothly with model capability. However, even the best models (scoring ≈ 15/25) trail specialist human experts (> 20/25). We introduce and release two novel datasets: HanabiLogs (1,520 annotated trajectories) and HanabiRewards (560 games with dense move-level utilities). By fine-tuning a 4B open-weight model (Qwen3-Instruct) on our datasets, we achieve performance gains of up to 156%, bringing performance to within 3 points of a strong proprietary reasoning model (o4-mini) and surpassing the best non-reasoning model (GPT-4.1) by 52%. Crucially, our HanabiRewards RL-finetuned model further generalizes beyond Hanabi, improving performance on a cooperative group-guessing benchmark by 11%, temporal reasoning on EventQA by 6.4%, instruction-following on IFBench-800K by 1.7 Pass@10, and matching AIME 2025 mathematical reasoning Pass@10. Code and datasets are available at {redacted for double blind}.

Social Aspects · Safety

Shuhui Zhu, Yue Lin, Shriya Kaistha, Wenhao Li, Baoxiang Wang, Hongyuan Zha, Gillian Hadfield, Pascal Poupart

Indirect reciprocity, which means helping those who help others, is difficult to sustain among decentralized, self-interested LLM agents without reliable reputation systems. We introduce Agentic Linguistic Gossip Network (ALIGN), an automated framework where agents strategically share open-ended gossip using hierarchical tones to evaluate trustworthiness and coordinate social norms. We demonstrate that ALIGN consistently improves indirect reciprocity and resists malicious entrants by identifying and ostracizing defectors without changing intrinsic incentives. Notably, we find that stronger reasoning capabilities in LLMs lead to more incentive-aligned cooperation, whereas chat models often over-cooperate even when strategically suboptimal. These results suggest that leveraging LLM reasoning through decentralized gossip is a promising path for maintaining social welfare in agentic ecosystems.

Deep Learning · Large Language Models

Ruishu Zhu, Zhihao Huang, Jiacheng Sun, Ping Luo, Hongyuan Zhang, Xuelong Li

Motivated by discrete diffusion's success in language-vision modeling, we explore its potential for multi-view generation, a task dominated by continuous approaches. We introduce \textbf{ViewMask-1-to-3}, formulating multi-view synthesis as a discrete sequence modeling problem where each viewpoint is represented as visual tokens from MAGVIT-v2. Through \textbf{masked token prediction}, our approach \textbf{enables progressive multi-view generation via iterative token unmasking}, unifying language and vision in a shared token space. Importantly, simple random masking combined with self-attention naturally encourages cross-view consistency without specialized architectures or 3D geometric priors. Our method outperforms the baseline on the GSO and 3D-FUTURE benchmarks, ranking first on average across standard image metrics and improving IoU by 10.6\% on 3D-FUTURE. This validates discrete diffusion as a promising candidate for multi-view generation.

Deep Learning · Large Language Models

Zhiwei Zhang, Yudi Lin, Linlin Wu, Fali Wang, Yi Xin, Xiaomin Li, Minhua Lin, Xianfeng Tang, Qi He, Suhang Wang

Diffusion language models (DLMs) have recently emerged as an alternative to autoregressive approaches, enabling parallel sequence generation and flexible token generation orders. Machine unlearning plays a critical role in mitigating legal and ethical risks by removing the influence of specific training examples from trained models. While unlearning has been extensively studied for autoregressive language models, its applicability to DLMs remains unexplored. The architectural differences of DLMs raise new challenges for effective and robust unlearning that are not addressed by existing methods. In this paper, we present the first comprehensive study of unlearning for DLMs. Through systematic empirical analysis, we show that unlearning performance in DLMs is highly sensitive to generation hyperparameters, highlighting the need for evaluation across diverse generation settings. We further observe that DLMs tend to reproduce unlearned information when target inputs are embedded within informative contexts, due to their ability to incorporate both prefix and suffix conditioning, which increases vulnerability to elicitation attacks and weakens the robustness of existing unlearning methods. To design a robust unlearning method, we propose an adversarial reinforcement learning framework, where a context generator adversarially produces informative contexts to elicit unlearned knowledge, while the DLM is optimized to suppress undesired recall. We further introduce novel components to address credit assignment and stability issues in this adversarial learning setup. Extensive experiments demonstrate that our method significantly improves unlearning effectiveness while preserving model utility. Our code is available at: https://anonymous.4open.science/r/dllm_unlearning-771D/

Deep Learning · Large Language Models

Taoyuze Lv, Alexander Chen, Fengyu Xie, Chu Wu, Jeffrey Meng, Dongzhan Zhou, Yingheng Wang, Bram Hoex, Zhicheng Zhong, Tong Xie

Large language models (LLMs) have shown promising potential in materials science, enabling tasks ranging from knowledge retrieval to property prediction. Existing materials science benchmarks mainly focus on perceptual or knowledge-based tasks, largely ignoring the structure modelling tasks, a core challenge in real scientific workflows. In practice, constructing and manipulating atomic structures is one of the most creative and least automated steps in materials research. In this work, we introduce AtomWorld, a benchmark designed to evaluate the abilities of LLMs on structure modifications. The benchmark includes ten fundamental actions under four widely used modelling categories, enabling verifiable evaluation metrics. We find that Gemini 2.5 Pro generally performs the best. While the success rate decreases markedly with increasing modelling complexity, with particularly low success rates (below 12\% for rotation) for operations involving complex spatial relations. Our results suggest that contemporary LLMs are better suited as copilots for materials structure modelling rather than fully unsupervised autonomous scientific agents. Beyond evaluation, AtomWorld also serves as a testbed and playground for developing future structure-aware models, including reinforcement learning and agentic approaches.

Deep Learning · Large Language Models

Zizhe Chen, Jiqian Dong, Yizhou Tian, Garry YANG, Yongqiang Chen, Zhitang Chen, James Cheng

Reinforcement Learning (RL) refines large language models (LLMs) by directly optimizing model behavior with reward signals. Although accurate state value estimation is essential for stable training in classical RL settings, it remains an understudied challenge in LLM post-training. In this work, we demonstrate that accurate value estimation can stabilize and improve post-training. First, we construct State Value Estimation Benchmark (SVEB) and show that critics of standard approaches like PPO simply degenerate toward a coarse group-average baseline. To overcome this, we propose two techniques. One is a heuristic method *Numca*, which uses numbers in responses as state representation to calculate state value. Another is a general hidden-state-based framework *Hista*, which utilize the semantic information in hidden states to group disjoint responses. Experiments show that when equipped with these improved estimates, training gains better performance consistently with different RL algorithms.

Deep Learning · Large Language Models

Mihir Prabhudesai, Aryan Satpathy, Yangmin Li, Zheyang Qin, Nikash Bhardwaj, Amir Zadeh, Chuan Li, Katerina Fragkiadaki, Deepak Pathak

We have witnessed remarkable advances in LLM reasoning capabilities with the advent of DeepSeek-R1. However, much of this progress has been fueled by the abundance of internet question–answer (QA) pairs—a major bottleneck going forward, since such data is limited in scale and concentrated mainly in domains like mathematics. In contrast, other sciences such as physics lack sufficient large-scale QA datasets to effectively train reasoning-capable models. In this work, we show that physics simulators can serve as a powerful alternative source of supervision for training LLMs for physical reasoning. We generate random scenes in physics engines, create synthetic question–answer pairs from simulated interactions, and train LLMs using reinforcement learning on this synthetic data. Our models exhibit zero-shot sim-to-real transfer to real-world physics benchmarks: for example, training solely on synthetic simulated data improves performance on IPhO (International Physics Olympiad) problems by 5–10 percentage points across different model sizes. These results demonstrate that physics simulators can act as scalable data generators, enabling LLMs to acquire deep physical reasoning skills beyond the limitations of internet-scale QA data.

Deep Learning · Large Language Models

Yubing Ren, Ping Guo, Yanan Cao

Large generative models raise growing concerns about provenance, misinformation, and impersonation. Digital watermarking offers a principled solution, yet extending it to natural language remains challenging due to text discreteness and sensitivity to semantic perturbations. Existing text watermarking methods either operate at the token level requiring white-box access and remaining fragile to paraphrasing, or at the sentence level, which supports black-box deployment but suffers from low Watermark Success Rate (WSR). We show that low WSR in sentence-level watermarking primarily stems from low injection success probability caused by a mismatch between posterior embedding distributions and semantic accept regions. Based on this insight, we propose \textbf{X-Guard}, a geometry-aware sentence-level watermarking framework that improves injection success by jointly optimizing embedding distributions and semantic space partitioning. X-Guard learns a more isotropic embedding space and introduces \textbf{A$^2$PQ}, a centroid-aligned partitioning scheme that approximately equalizes probability mass across regions. Extensive experiments across multiple models, languages, and attack settings demonstrate that X-Guard consistently improves robustness while preserving text fluency and practical deployability.

Deep Learning · Large Language Models

Jiayu Bai, Danchen Yu, Zhenyu Liao, TianQi Hou, Feng Zhou, Robert Qiu, Zenan Ling

Kronecker adapters have emerged as a promising approach for fine-tuning large-scale models, enabling high-rank updates through tunable component structures. However, existing work largely treats the component structure as a fixed or heuristic design choice, leaving the dimensions and number of Kronecker components underexplored. In this paper, we identify component structure as a key factor governing the capacity of Kronecker adapters. We perform a fine-grained analysis of both the dimensions and number of Kronecker components. In particular, we show that the alignment between Kronecker adapters and full fine-tuning depends on component configurations. Guided by these insights, we propose Component Designed Kronecker Adapters (CDKA). We further provide parameter-budget–aware configuration guidelines and a tailored training stabilization strategy for practical deployment. Experiments across various natural language processing tasks demonstrate the effectiveness of CDKA.

Social Aspects · Accountability, Transparency, and Interpretability

Peng Cui, Jiahao Zhang, Lijie Hu

While Contrastive Learning (CL) has revolutionized self-supervised representation learning, its latent representations remain highly entangled and opaque, limiting their interpretability in safety-critical applications. We identify that a fundamental cause of this entanglement is the reliance on deterministic similarity measures, which treat all feature dimensions equally. In compositional scenes, this creates an Optimization Conflict: common background features, such as, "blue sky", are encouraged to align in positive pairs but simultaneously repelled in negative pairs, causing gradient oscillations that hinder precise semantic disentanglement. To address this, we propose **BayesNCL** (Bayesian Gated Non-Negative Contrastive Learning). Unlike standard approaches, BayesNCL introduces a probabilistic gating mechanism that dynamically filters out task-irrelevant, high-frequency common features while selectively retaining discriminative semantics. By formalizing feature selection as a variational inference problem with a sparse Bernoulli prior, our method effectively resolves the optimization conflict. Empirical experimental results on benchmark datasets demonstrate that BayesNCL achieves a remarkable 142.1\% improvement in semantic consistency compared to state-of-the-art baselines, yielding highly interpretable representations without compromising downstream task performance.

Applications · Computer Vision

Nils Morbitzer, Jonathan Evers, Artem Savkin, Thomas Stauner, Nassir Navab, Federico Tombari, Stefano Gasperini

Forecasting the evolution of dynamic environments is crucial for autonomous agents. While generative world models have recently achieved high photorealism in 2D video synthesis, by mixing within the image plane ego-motion and environmental dynamics, they exhibit physical inconsistencies, such as morphing or vanishing objects, especially over long time horizons. In this paper, we propose FR3D, a world model that predicts a persistent 3D latent representation for future dynamic 3D reconstruction. Unlike prior works that treat the world as a sequence of image-based features, FR3D explicitly decouples the 3D evolution of the scene from the agent's trajectory, treating the inferred ego-motion as a latent proxy for action. This disentanglement resolves the ambiguities between self-motion and world-motion, ensuring geometric consistency into the future. Furthermore, we introduce a teacher-student distillation strategy that leverages the spatial "common sense" of off-the-shelf foundation models, leading to robust zero-shot generalization. Extensive experiments demonstrate FR3D's strong performance for future dynamic 3D reconstruction from monocular observations across multiple datasets, even 2 seconds into the future.

Deep Learning · Large Language Models

Dulhan Jayalath, Shashwat Goel, Thomas Foster, Parag Jain, Suchin Gururangan, Cheng Zhang, Anirudh Goyal, Alan Schelten

Where do learning signals come from when there is no ground truth in post-training? We show that inference compute itself can serve as supervision. By generating parallel rollouts and converting them into reference estimates, models can learn without human labels—critically, even in non-verifiable domains like healthcare guidance where no programmatic checker exists. We call this framework *Compute as Teacher (CaT)* and it turns inference-time compute from parallel rollouts into supervision for RL training. The framework has two components: (1) reference estimation which aggregates rollouts into a pseudo-reference answer, and (2) reward derivation which converts that pseudo-reference into RL rewards. For (1), we explore a simple method we call *synthesis*, but the framework admits any aggregator. For (2), we introduce self-proposed rubrics for non-verifiable domains. These are binary, auditable criteria generated from the pseudo-reference and scored by an LLM judge. On HealthBench, models trained with CaT match or exceed inference-time aggregation quality while using 9× less test-time compute. Here, CaT also competes with learning from expert physician annotations, yielding up to +30% relative improvement over the initial policy. The framework extends naturally to verifiable rewards, matching the best existing baselines on MATH-500 in test-time RL and demonstrating 'drop-in' versatility across both types of domains.

Deep Learning · Large Language Models

Mengfei Zhang, Yu Mi, Leijing Zhou

Autoregressive Large Language Models (LLMs) often fail in complex reasoning because early-stage errors remain uncorrectable in subsequent steps—a limitation fundamentally rooted in the inherent irreversibility of the Transformer architecture. In this paper, we propose HEdit, a lightweight reasoning enhancement paradigm that equips models with a "hindsight-like" capability for dynamic error correction during generation. Our core insight involves deconstructing reasoning failures into two pivotal stages: latent representational biases emerging at logical anchors, and the subsequent eruption of explicit cognitive dissonance at trigger points. Based on these observations, the HEdit framework detects internal inconsistency signals at trigger points in real-time, actively backtracks to critical anchors, and utilizes a lightweight trainable editor to precisely refine their Key-Value (KV) caches. This mechanism effectively breaks the unidirectional constraints of autoregressive inference. Empirical results demonstrate that HEdit significantly enhances the performance of various models on mathematical reasoning tasks—with average accuracy improvements ranging from 2.2\% to 10.8\%—while maintaining extremely low overhead (add parameters $<0.5\%$). HEdit provides a dynamic, pluggable and lightweight solution, making it particularly beneficial for users in low-resource environments. Our code can be found at anonymous github: https://anonymous.4open.science/r/ac3d2-51CF/

Deep Learning · Large Language Models

Paulius Rauba, Dominykas Seputis, Patrikas Vanagas, Mihaela van der Schaar

Least privilege is a core security principle: grant each request only the minimum access needed to achieve its goal. Deployed language models almost never follow it, instead being exposed through a single API endpoint that serves all users and requests. This gap exists not because least privilege would be unhelpful—deployments would benefit greatly from reducing unnecessary capability exposure. The real obstacle is definitional and mechanistic: what does "access" mean inside a language model, and how can we enforce it without retraining or deploying multiple models? We take inspiration from least privilege in computer systems and define a class of models called *least-privilege language models*, where privilege is *reachable internal computation* during the forward pass. In this view, lowering privilege literally shrinks the model's accessible function class (as opposed to denying access via learned policies). We formalize deployment-time control as a monitor--allocator--enforcer stack, separating (i) request-time signals, (ii) a decision rule that allocates privilege, and (iii) an inference-time mechanism that selects privilege. We then propose *Nested Least-Privilege Networks*, a shape-preserving, rank-indexed intervention that provides a smooth, reversible control knob. We show that this knob yields policy-usable privilege--utility frontiers and enables selective suppression of targeted capabilities with limited collateral degradation across various policies. Most importantly, we see this as a defense of a completely new deployment paradigm which challenges the premise that we can only have output-level control of language models.

Deep Learning · Graph Neural Networks

Ali Azizpour, Madeline Navarro, Santiago Segarra

We propose an adaptive node feature selection approach for graph neural networks (GNNs) that identifies and removes unnecessary features during training. The ability to measure how features contribute to model output is key for interpreting decisions and reducing dimensionality by eliminating unhelpful variables. However, graph-structured data introduces complex dependencies that may be unsuited to classical feature importance metrics. Inspired by this, we present a data-, model-, and task-agnostic method that determines relevant features during training based on changes in validation performance upon permuting feature values. We theoretically motivate our approach by characterizing how the relationships between node data and graph structure influences GNN performance. Empirically, we show that (i) our highly general approach rivals the performance of tailored feature selection approaches that exploit prior assumptions; (ii) we return meaningful feature importance scores well before the GNN is fully trained; and (iii) our scores demonstrably extract relevant properties that inform feature importance for various graph learning settings.

Deep Learning · Large Language Models

Joshua Ong, Yu Zhao, Mihaela C. Stoian, Wenda Li, Shay Cohen, Eleonora Giunchiglia

While plan-and-infill decoding in Masked Diffusion Models (MDMs) shows promise for mathematical and code reasoning, performance remains highly sensitive to slot infilling order, often yielding substantial output variance. We introduce DiffuSearch, a framework that formulates slot selection as decision making and optimises infilling orders through Monte Carlo Tree Search (MCTS). DiffuSearch uses look-ahead simulations to evaluate partial completions before commitment, systematically exploring the combinatorial space of generation orders. Experiments show an average improvement of 3.2% over autoregressive baselines and 8.0% over baseline plan-and-infill, with notable gains of 19.5% on MBPP and 4.9% on MATH500. Our analysis reveals that while DiffuSearch predominantly follows sequential ordering, incorporating non-sequential generation is essential for maximising performance. We observe that larger exploration constants, rather than increased simulations, are necessary to overcome model confidence biases and discover effective orderings. These findings establish MCTS-based planning as an effective approach for enhancing generation quality in MDMs.

Hongze Mi, Yibo Feng, Wenjie Lu, Song Cao, Jinyuan Li, Yanming Li, Xuelin Zhang, Haotian Luo, Songyang Peng, He Cui 等

Multimodal Large Language Model (MLLM) agents facilitate Graphical User Interface (GUI) automation but struggle with long-horizon, cross-application tasks due to limited context windows. While memory systems provide a viable solution, existing paradigms struggle to adapt to dynamic GUI environments, suffering from a granularity mismatch between high-level intent and low-level execution, and context pollution where the static accumulation of outdated experiences drives agents into hallucination. To address these bottlenecks, we propose the Darwinian Memory System (DMS), a self-evolving architecture that constructs memory as a dynamic ecosystem governed by the law of "survival of the fittest." DMS decomposes complex trajectories into independent, reusable units for compositional flexibility, and implements Utility-driven Natural Selection to track survival value, actively pruning suboptimal paths and inhibiting high-risk plans. This evolutionary pressure compels the agent to derive superior strategies. Extensive experiments on real-world multi-app benchmarks validate that DMS boosts general-purpose MLLMs without training costs or architectural overhead, achieving average gains of 18.0% in success rate and 33.9% in execution stability, while reducing task latency, establishing it as an effective self-evolving memory system for GUI tasks. The implementation is available at https://anonymous.4open.science/r/DMS-C48C.

Deep Learning · Large Language Models

Shengrui Li, Fei zhao, Kaiyan Zhao, Jieying Ye, Haifeng Liu, Fangcheng Shi, Zheyong Xie, Yao Hu, Shaosheng Cao

Determining an effective data mixture is a key factor in Large Language Model (LLM) pre-training, where models must balance general competence with proficiency on hard tasks such as math and code. However, identifying an optimal mixture remains an open challenge, as existing approaches either rely on unreliable tiny-scale proxy experiments or require prohibitively expensive large-scale exploration. To address this, we propose Decouple Searching from Training Mix (DeMix), a novel framework that leverages model merging to predict optimal data ratios. Instead of training proxy models for every sampled mixture, DeMix trains component models on candidate datasets at scale and derives data mixture proxies via weighted model merging. This paradigm decouples search from training costs, enabling evaluation of unlimited sampled mixtures without extra training burden and thus facilitating better mixture discovery through more search trials. Extensive experiments demonstrate that DeMix breaks the trade-off between sufficiency, accuracy and efficiency, obtaining the optimal mixture with higher benchmark performance at lower search cost. Additionally, we release the DeMix Corpora, a comprehensive 22T-token dataset comprising high-quality pre-training data with validated mixtures to facilitate open research.

Satwik Bhattamishra, Kulin Shah, Michael Hahn, Varun Kanade

We study the problem of learning Transformer-based sequence models with black-box access to their outputs. In this setting, a learner may adaptively query the oracle with any sequence of vectors and observe the output of the target function. We begin with studying the learnability of the simplest formulation, that is, learning a single-head attention-based regressor with queries. We show that for a model with width $d$, there is an elementary algorithm to learn the parameters of single-head attention with $O(d^2)$ queries. Further, we show that if there exists an algorithm to learn ReLU feedforward networks (FFNs), then the single-head algorithm can be easily adapted to learn one-layer Transformers with single-head attention. Next, we show that, in the common regime where the head dimension $r \ll d$, single-head attention-based models can be learned with $O(rd)$ queries via compressed sensing arguments. We also study robustness to noisy oracle access, proving that under mild norm and margin conditions, the parameters can be estimated to $\varepsilon$ accuracy with a polynomial number of queries even when outputs are only provided up to additive tolerance. Finally, we consider the learnability of multi-head attention and show that they are not identifiable from queries, and hence, learnability in the same sense is not feasible without additional assumptions. We discuss potential approaches to learn multi-head attention-based models under certain structural assumptions.

Deep Learning · Generative Models and Autoencoders

Hongzhou Zhu, Min Zhao, Guande He, Hang Su, Chongxuan Li, Jun Zhu

To achieve real-time video generation, current approaches distill pretrained bidirectional video diffusion models into few-step autoregressive (AR) models. This process involves an *architectural gap*, as it converts full attention into causal attention. In this paper, we demonstrate that existing methods fail to bridge this gap theoretically, leading to suboptimal performance. Specifically, these methods employ ODE distillation to initialize the AR student, where a key requirement is *injectivity*. We figure out that for an AR student, *frame-level injectivity* must hold: each noisy frame must map to a unique clean frame under the PF-ODE of the *AR teacher*. We theoretically prove that existing methods, which distill an AR student from a bidirectional teacher, violate this frame-level injectivity. Consequently, the student fails to recover the teacher's flow map and instead learns a conditional expectation, resulting in subpar performance. To address this issue, we propose *Causal Forcing*, which employs an AR teacher for ODE initialization, thereby effectively bridging the architectural gap. Empirical results show that our method outperforms all baselines across all metrics, surpassing the SOTA Self-Forcing by 19.3\% in Dynamic Degree, 8.7\% in VisionReward, and 16.7\% in Instruction Following.