arXiv:2605.17497v1 Announce Type: new
Abstract: GRPO-style RLVR trains reasoning models from multiple on-policy attempts per prompt, but typically uses these attempts only through terminal rewards. We show that a mixed group contains a richer process signal: a correct completion is a self-generated witness of how the current policy can solve the problem, while a wrong completion provides on-policy prefixes where the policy needs correction. We introduce \emph{Self-Supervised On-Policy Distillation} (SSOPD), which distills a teacher distribution conditioned on the shortest correct completion into prefixes of the longest wrong completion. This converts intra-group correct--wrong contrast into dense process supervision without external solution traces. A stopping-time view motivates the shortest-correct / longest-wrong rule as a finite-group approximation to editing persistent failures toward fast-success actions, and a prompt-level frontier weight concentrates the auxiliary loss where correct and wrong branches coexist. Across AIME 2024, AIME 2025, and HMMT 2025, SSOPD improves over GRPO in all nine model-benchmark settings. On Qwen3-8B, it reaches a macro Avg@12 of 65.6, outperforming GRPO by 1.6 points and the solution-conditioned OPSD baseline by 0.8 points. Code will be released at https://github.com/tzq1999/SSOPD.
Science Journals
arXiv:2605.14005v2 Announce Type: replace
Abstract: Speculative decoding has become a widely adopted technique for accelerating large language model (LLM) inference by drafting multiple candidate tokens and verifying them with a target model in parallel. Its efficiency, however, critically depends on the average accepted length $\tau$, i.e., how many draft tokens survive each verification step. In this work, we identify a new mechanism-level vulnerability in model-based speculative decoding: the drafter is trained to approximate the target model distribution, but this approximation is inevitably imperfect. Such a drafter-target mismatch creates a hidden attack surface where small perturbations can preserve the target model's visible behavior while substantially reducing draft-token acceptability. We propose Mistletoe, a stealthy acceleration-collapse attack against speculative decoding. Mistletoe directly targets the acceptance mechanism of speculative decoding. It jointly optimizes a degradation objective that decreases drafter-target agreement and a semantic-preservation objective that constrains the target model's output distribution. To resolve the conflict between these objectives, we introduce a null-space projection mechanism, where degradation gradients are projected away from the local semantic-preserving direction, suppressing draft acceptance while minimizing semantic drift. Experiments on various speculative decoding systems show that Mistletoe substantially reduces average accepted length $\tau$, collapses speedup, and lowers averaged token throughput, while preserving output quality and perplexity. Our work highlights that speculative decoding introduces a mechanism-level attack surface beyond existing output robustness, calling for more robust designs of LLM acceleration systems.
arXiv:2605.14271v2 Announce Type: replace
Abstract: LLM agents increasingly run inside execution harnesses that dispatch tools, allocate resources, and route messages between specialized components. However, a harness can return a correct, benign answer over a trajectory that accesses unauthorized resources or leaks context to the wrong agent. Output-level evaluation cannot see these failures, yet most safety benchmarks score only final outputs or terminal states, even though many violations occur mid-trajectory rather than at termination. The central question is whether the harness respects user intent, permission boundaries, and information-flow constraints throughout execution. To address this gap, we propose HarnessAudit, a framework that audits full execution trajectories across boundary compliance, execution fidelity, and system stability, with a focus on multi-agent harnesses where these risks are most pronounced. We further introduce HarnessAudit-Bench, a benchmark of 210 tasks across eight real-world domains, instantiated in both single-agent and multi-agent configurations with embedded safety constraints. Evaluating ten harness configurations across frontier models and three multi-agent frameworks, we find that: (i) task completion is misaligned with safe execution, and violations accumulate with trajectory length; (ii) safety risks vary across domains, task types, and agent roles; (iii) most violations concentrate in resource access and inter-agent information transfer; and (iv) multi-agent collaboration expands the safety risk surface, while harness design sets the upper bound of safe deployment.
arXiv:2605.18635v1 Announce Type: new
Abstract: Credit default prediction is a tabular learning problem with severe class imbalance, heterogeneous features, and tight latency budgets. Tabular Foundation Models (TFMs) approach this problem through in-context learning, which makes their predictions sensitive to how the context window is built. We benchmark four classical models and five TFMs on the Home Credit and Lending Club datasets, varying the context-construction strategy (seven options) and the context size (1K to 50K). On both datasets, the choice of context strategy explains more variance in AUC-ROC than the choice of TFM family: balanced and hybrid sampling add 3 to 4 AUC points over uniform sampling, and the gap exceeds the spread between TFMs. With a balanced context of 5K to 10K examples, the strongest TFMs reach the AUC of classical baselines trained on the full data, while also recovering meaningful default-class recall that default-threshold GBDTs do not. We frame this as evidence that context construction, rather than architecture choice, is the primary deployment lever for TFMs in imbalanced credit-risk settings.
arXiv:2605.18632v1 Announce Type: new
Abstract: Neural network checkpoints have quietly become a large-scale data resource: millions of trained weight vectors now exist, each encoding task-, domain-, and architecture-specific knowledge. This position paper argues that model checkpoints should be treated as a first-class data modality, and that generative modeling in weight space should be standardized as a core machine learning primitive. Recent advances demonstrate that neural weights can be synthesized on demand, often matching fine-tuning performance while reducing adaptation cost by orders of magnitude. We contend that these results reflect an underlying structural fact: high-performing models occupy low-dimensional, highly structured regions of weight space shaped by symmetry, flatness, modularity, and shared subspaces. Building on this view, we organize existing methods into a five-stage pipeline, survey applications where the approach is already practical, and clarify current limits: adapter-scale and conditional generation are advancing rapidly, while unrestricted frontier-scale checkpoint synthesis remains open. Our goal is to shift the community's default mindset from optimizing models per task to sampling models from learned weight distributions, accelerating toward an era in which AI systems routinely improve or create other AI systems.
arXiv:2605.16376v1 Announce Type: cross
Abstract: Kelvin is a lightweight learned pre-encoder that sits in front of an unmodified libx264 encoder. It applies content-adaptive pixel adjustments, bounded at +/-1/255 per channel, so that the encoder allocates bits where they matter most perceptually, while emitting a standard H.264 bitstream compatible with every existing decoder, player, and CDN. On the seven-sequence 1080p UVG benchmark, Kelvin v1.0 achieves a mean BD-VMAF of -27.62% (7 of 7 wins) and BD-VMAF-NEG of -5.18% (6 of 7 wins) relative to baseline libx264 at preset medium. On the 30-sequence MCL-JCV public set (28 unseen by training), the same checkpoint wins on 28 of 30 clips by BD-VMAF; with the two diagnosable failures removed the mean is -27.70% BD-VMAF and -5.37% BD-VMAF-NEG, consistent with UVG to within one percentage point. A central engineering challenge is the non-differentiability of H.264: we describe a hybrid codec proxy that combines a calibrated differentiable rate estimator (Spearman rho = 0.986 vs. real libx264 bits-per-pixel) with a U-Net distortion proxy trained on real encoder outputs. We publish full per-sequence rate-distortion data, a named failure-mode taxonomy on MCL-JCV (rate-floor violation, distribution shift, metric saturation), a five-baseline sanity panel (hqdn3d, unsharp, -tune psnr, -tune ssim, x265 medium), and honest positioning: x265 medium beats Kelvin on every metric on the same corpus. Kelvin is therefore designed for workloads where remaining on H.264 is a constraint rather than a choice.
arXiv:2605.18630v1 Announce Type: new
Abstract: Large Language Models (LLMs) are increasingly deployed as scientific AI as- sistants, and a growing body of benchmarks evaluates their capabilities across knowledge retrieval, reasoning, code generation, and tool use. These evaluations, however, typically assume the scientific problem is already well-posed, whereas practical scientific assistance often begins with an ill-posed user request that must be refined through dialogue before any computation, analysis, or experiment can be carried out reliably. We introduce SCICONVBENCH, a benchmark for multi- turn clarification in scientific task formulation across four computational science problem domains: fluid mechanics, solid mechanics, materials science, and par- tial differential equations (PDEs). SCICONVBENCH targets two complementary capabilities: eliciting missing information (disambiguation) and detecting and correcting erroneous requests containing internally contradictory information (in- consistency resolution). Our benchmark pairs a structured task ontology with a rubric-based evaluation framework, enabling systematic measurement of LLM per- formance across three dimensions: clarification behavior, conversational grounding, and final-specification fidelity. Current frontier models perform relatively well on inconsistency resolution, but even the best model resolves only 52.7% of the disambiguation cases in fluid mechanics. We further find that frontier LLMs fre- quently make silent assumptions and perform implicit specification repairs that are not grounded in the conversation with users. SCICONVBENCH establishes a foundation for evaluating the upstream conversational reasoning that a reliable computational science assistant requires. The code and data can be found at https://github.com/csml-rpi/SciConvBench.
Overcoming the Intrinsic Performance Limitations of MEMS IMU via Diffusion-Based Generative Learning
arXiv:2605.16391v1 Announce Type: cross
Abstract: Inertial measurement units (IMUs) are fundamental sensing components in multi-source integrated navigation systems, and their performance directly determines the accuracy and reliability of solutions. However, the precision of low-cost IMUs is inherently constrained by hardware limitations. Recently, generative artificial intelligence has demonstrated remarkable capability in modeling complex data distributions and reconstructing high-fidelity signals. Motivated by this, we propose a diffusion-based generative learning framework for synthesizing high-fidelity virtual IMU data from low-cost IMU measurements. Specifically, a conditional diffusion model based on a U-Net architecture is constructed, where high-grade IMU measurements are utilized as ground-truth priors and low-cost IMU measurements are employed as conditional inputs. The virtual IMU data generated by the model is used for subsequent navigation and localization tasks. Experimental results demonstrate that the generated virtual IMU data significantly outperform the original low-cost IMU measurements in both positioning and attitude estimation. Furthermore, we transfer the model to airborne mapping experiments, where the proposed method produces thinner and more consistent point clouds. Overall, the proposed framework breaks the performance limits of low-cost IMU and demonstrates the potential of diffusion-based generative learning for virtual high-grade IMU data.
arXiv:2603.20904v5 Announce Type: replace-cross
Abstract: Spectral gaps, Kramers escape rates, and position-dependent relaxation timescales are dynamical invariants encoded in the infinitesimal generator $\Lop$ of a stochastic flow. We show that weak projection of the governing It\^{o} SDE onto temporal test functions produces an endogeneity bias of order $O(T\,\dt^{3/2})$ that grows with the observation window and cannot be eliminated by additional data. Projecting instead onto spatial Gaussian kernels removes the bias exactly: $\mathcal{F}_{t_n}$-measurability and the tower property guarantee unbiased regression rows at every step. The resulting framework jointly identifies the drift $b(x)$ and diffusion $a(x)$ from a single sparse regression, producing an explicit symbolic enerator amenable to spectral analysis. Validation on three benchmark systems yields coefficient errors below 5%, stationary-density total-variation distances below 0.01, and autocorrelation functions that faithfully reproduce true relaxation timescales.
arXiv:2605.16685v1 Announce Type: cross
Abstract: We study mathematical programs with equilibrium constraints, in which a leader knows their own cost function, but lacks a model of the followers' response. Instead, the leader can only query this response at specific points. While this setting precludes the use of gradient-based methods, existing zeroth-order approaches treat the composed objective entirely as a black box, deploying zeroth-order tools across both the leader and follower. Such approaches are inefficient, as they discard information the leader already possesses about their own cost function. In this work we instead propose to deploy zeroth-order tools only where they are truly needed: to handle the unknown, non-smooth followers' response. Specifically, we first propose PZOS, an algorithm that combines exact partial gradients of the leader's cost with zeroth-order Jacobian estimates of the followers' response in a chain-rule-inspired manner, and establish that it achieves a strictly lower variance bound than the black-box baseline. Second, we introduce the partial Goldstein subdifferential, a stationarity notion tailored to this composite structure, and prove convergence of our algorithm to both standard and partial Goldstein stationary points. Finally, we validate our method on two application domains -- toll optimization in routing games and defense-attack investment in security games -- demonstrating consistent improvements over black-box baselines in convergence speed, objective value, and estimator variance, with robust performance even under few queries per iteration.
arXiv:2605.16689v1 Announce Type: cross
Abstract: AI-native 6G visions increasingly invoke wireless foundation models, large multimodal models, and wireless world models as the natural endpoint of AI-native networking, drawing an analogy to recent developments in large language models (LLMs). We argue that this analogy is structurally incomplete. The success of LLMs is based on a broad, reusable, and largely self-contained tokenized data substrate, whereas the wireless domain lacks an equivalent data foundation. Unlike text, code, or images, wireless data such as CSI tensors, IQ samples, or scheduler logs are not self-contained: their meaning is configuration-dependent, simulator-conditioned, task-disaggregated, and weakly grounded in operational feedback, all structural bottlenecks that undermine current pre- and post-training recipes. We therefore argue that monolithic models, including mixture-of-experts (MoE) and wireless world models, are not the most realistic near-term path toward deployable AI-native networks. Instead, emerging evidence points toward composable and agentic network architectures, where general reasoning models orchestrate specialized signal processing models, classical algorithms, digital twins, standards-aware retrieval, and safety checks through explicit programmable interfaces.
arXiv:2605.16817v1 Announce Type: cross
Abstract: Learned image compression has achieved competitive rate-distortion performance, but very-low-bitrate reconstruction remains difficult because the transmitted representation often cannot preserve fine textures and local structures. Perceptual and generative codecs address this problem by using learned reconstruction priors, and controllable codecs allow one model to cover different bitrate and reconstruction preferences. However, controllability alone does not resolve the decoder-side reconstruction-prior problem: under severe bit constraints, the decoder must infer missing details from limited transmitted information, while existing codebook-based controllable designs generally rely on single-codebook token-based priors. This paper proposes Adaptive Fused Prior Transfer for Controllable Generative Image Compression (AFP-GIC), a controllable codec that transfers an adaptive fused prior from a frozen pretrained AdaCode model. Encoder-side fused-prior features guide latent formation, while the decoder predicts a compatible fused prior from the compressed representation and selected control variables, enabling prior-guided reconstruction without transmitting the fused prior itself. A motivating analysis relates decoder-side fused-prior alignment to a reconstruction-error upper bound and shows that the fused-prior family contains single-codebook choices as special cases. Under the unified benchmark, AFP-GIC reduces decoder latency by 18.1% and the overall parameter count by 31.10 million (20.5%) relative to DC-VIC. Experiments on Kodak, CLIC2020, and DIV2K show competitive PSNR, with the clearest perceptual gains in NIQE scores and very-low-bitrate visual comparisons.
arXiv:2510.16046v3 Announce Type: replace
Abstract: We present CARDIO-Affect, a complex-systems theoretical framework for long-term emotional dynamics in bounded social groups, with explicit uncertainty quantification at every layer. Long-period naturalistic emotion in stable small groups exhibits hallmarks of complex systems -- multi-stable attractors, weak chaos, long-range memory, and sparse heterogeneous coupling -- invisible to conventional short-clip facial-emotion analysis. CARDIO-Affect treats individual emotion as a multi-stable nonlinear stochastic dynamical system and group emotion as a sparsely-coupled network with emergent macrostates, formalised through six propositions and four pillars: (i) statistical mechanics with neural-parameterised Hamiltonian SDE over asymmetric potentials; (ii) information geometry on a 45-dimensional Fisher-Rao manifold; (iii) topological data analysis for invariant trajectory signatures; (iv) HRV-inspired Emotional Variability Analytics (EVA) decomposing each person-day into multi-scale time/frequency/nonlinear measures. We validate on the first 30.1-month longitudinal in-the-wild facial-emotion corpus (companion: arXiv:2510.15221) by discovering three falsifiable paradoxes: Sparse-Contagion (R_0=0.36, density 2.7%, 8 BH-FDR edges), Asymmetric-Persistence (negative dwell 5.85x positive, 1.77D potential gap), and Crisis-Inversion (Shanghai 2022 lockdown naive d=-0.40 collapses to permutation-p=0.94 under BSTS + synthetic-control). On synthetic benchmarks, CARDIO-EBM v2 matches asymptotically optimal Granger on linear VAR data (Class A AUROC 0.984+/-0.012 vs Granger 0.997+/-0.001, 5 seeds) but fails on tanh-coupled nonlinear data (Class B AUROC 0.490 vs Granger 0.796), a documented limitation of the linear mask-self estimator. We release framework code and the full reproduction pipeline.
arXiv:2605.18069v1 Announce Type: cross
Abstract: This paper studies sampling error bounds for denoising diffusion probabilistic models (DDPMs) in the 2-Wasserstein distance. Our contributions are threefold. (i) Under general Lipschitz-type conditions on the score function and for a broad class of variance schedules, including the cosine schedule, we establish sharp upper bounds that are optimal in both the dimension and the number of steps, and recover several sharp error bounds previously obtained in the literature. (ii) We prove that the same Lipschitz-type conditions, which encompass those commonly imposed on the (learned) score, imply a logarithmic Sobolev inequality and hence a quadratic transportation cost inequality for the DDPM. As a consequence, in settings covered by existing work, an optimal Wasserstein bound, up to a logarithmic factor, follows from the recently obtained sharp error bound in the Kullback-Leibler divergence under geometric-type variance schedules. (iii) We show that for general log-concave target distributions, the optimal Wasserstein error bound remains attainable even without a quadratic transportation cost inequality for the target. Our analysis is based on viewing the DDPM sampler as a discretization of the F\"ollmer process rather than the conventional reverse Ornstein-Uhlenbeck process.
arXiv:2605.18607v1 Announce Type: new
Abstract: Progress in language model development is often driven by comparative decisions: which architecture to adopt, which pretraining corpus to use, or which training recipe to apply. Making these decisions well requires reliable performance forecasts, yet the two commonly used signals are fundamentally limited. Cross-entropy loss is poorly aligned with downstream capabilities, and direct downstream evaluation is expensive, sparse, and often uninformative at early training stages. Instead, we propose to construct proxy metrics by aggregating token-level statistics, such as entropy, top-k accuracy, and expert token rank, from a candidate model's next token distribution over expert-written solutions. Across three settings, our proxies consistently outperform loss- and compute-based baselines: 1) For cross-family model selection, they rank a heterogeneous population of reasoning models with mean Spearman Rho = 0.81 (vs. Rho = 0.36 for cross-entropy loss); 2) For pretraining data selection, they reliably rank 25 candidate corpora for a target model at roughly $10{,}000\times$ less compute than direct evaluation, pushing the Pareto frontier beyond existing methods; and 3) for training-time forecasting, they extrapolate downstream accuracy across an $18\times$ compute horizon with roughly half the error of existing alternatives. Together, these results suggest that expert trajectories are a broadly useful source of signal for assessing model capabilities, enabling reliable performance forecasting throughout the model development life cycle.
arXiv:2605.17198v1 Announce Type: cross
Abstract: To be useful for downstream applications, vision decoding models that are trained to reconstruct seen images from human brain activity must be able to generalize to internally generated visual representations, i.e., mental images. In an analysis of the recently released NSD-Imagery dataset, we demonstrated that while some modern vision decoders can perform quite well on mental image reconstruction, some fail, and that state-of-the-art (SOTA) performance on seen image reconstruction is no guarantee of SOTA performance on mental image reconstruction. Motivated by these findings, we developed MIRAGE, a method explicitly designed to train on vision datasets and cross-decode mental images from brain activity. MIRAGE employs a linear backbone and multi-modal text and image features as input to a diffusion model. Feature metrics and human raters establish MIRAGE as SOTA for mental image reconstruction on the NSD-Imagery benchmark. With ablation analysis we show that mental image reconstruction works best when decoders use image features with relatively few dimensions and include guidance from text-based and both high- and low-level image-based features. Our work indicates that--given the right architecture--existing large-scale datasets using external stimuli are viable training data for decoding mental images, and warrant optimism about the future success and utility of mental image reconstruction.
arXiv:2605.18708v1 Announce Type: cross
Abstract: We address a fundamental question: Can one determine whether a received signal is squeezed when each copy arrives with a different displacement/amplitude? We introduce an interaction Hamiltonian that converts quadrature squeezing into number squeezing. Using this conversion, we test whether the copies satisfy $g^{(2)}(0)<1$. The Hamiltonian itself does not create nonclassicality; it only transfers it from quadrature squeezing to number squeezing. This allows us to identify squeezing even when individual copies have random displacements.
arXiv:2605.14854v2 Announce Type: replace
Abstract: Human Mesh Recovery (HMR) is fundamentally ambiguous: under occlusion or weak depth cues, multiple 3D bodies can explain the same image evidence. This ambiguity is not uniform across the body, as torso pose and root structure are often relatively well constrained, whereas distal articulations such as the arms and legs are more uncertain. Building on this observation, we propose FactorizedHMR, a two-stage framework that treats these two regimes differently. A deterministic regression module first recovers a stable torso-root anchor, and a probabilistic flow-matching module then completes the remaining non-torso articulation. To make this completion reliable, we combine a composite target representation with geometry-aware supervision and feature-aware classifier-free guidance, preserving the torso-root anchor while improving single-reference recovery of ambiguity-prone articulation. We also introduce a synthetic data pipeline that provides the paired image-camera-motion supervision under diverse viewpoints. Across camera-space and world-space benchmarks, FactorizedHMR remains competitive with strong baselines, with the clearest gains in occlusion-heavy recovery and drift-sensitive world-space metrics.
arXiv:2605.14963v2 Announce Type: replace
Abstract: Stereo matching on top-bottom equirectangular images provides an effective framework for full-surround perception, as vertically aligned epipolar lines enable the use of advanced perspective stereo architectures that are largely driven by large-scale datasets and monocular priors. However, the performance of such adaptations is severely limited by the scarcity of omnidirectional stereo datasets and the degradation of perspective monocular priors under spherical distortions. To address these challenges, we propose H-OmniStereo, a zero-shot omnidirectional stereo matching framework. First, we construct high-quality synthetic dataset comprising over 2.8 million top-bottom equirectangular stereo pairs to scale up training. Second, we introduce an equirectangular monocular normal estimator, specifically operating in a heading-aligned coordinate system. Beyond providing distortion-robust and cross-view-consistent geometric priors for establishing reliable correspondences in stereo matching, this design boosts training efficiency and accommodates train-test FoV mismatches. Extensive experiments show that our approach achieves higher accuracy than existing methods on out-of-domain datasets and successfully generalizes to real-world consumer camera setups using a single model. The model and dataset will be released at https://github.com/JIANG-CX/H-OmniStereo.
arXiv:2406.13187v2 Announce Type: replace
Abstract: While long-tailed semi-supervised learning (LTSSL) has attracted growing attention in many real-world classification tasks, existing LTSSL algorithms typically assume that labeled and unlabeled data share nearly identical class distributions. When this assumption is violated, these methods can perform poorly because they rely on biased model-generated pseudo-labels. To address this issue, we propose a simple yet effective approach called DeCon for LTSSL with unknown unlabeled class distributions. Specifically, DeCon decouples learning into two specialized branches: a standard branch that focuses on head classes and a balanced branch that focuses on tail classes. During training, the two branches interact and gradually converge, allowing them to complement each other and ultimately achieve strong performance across all classes. Despite its simplicity, we show that DeCon achieves state-of-the-art performance on a variety of standard LTSSL benchmarks, e.g., an averaged 2.7\% absolute increase in test accuracy against existing algorithms when the class distributions of labeled and unlabeled data are mismatched. Even when the class distributions are identical, DeCon consistently outperforms many sophisticated LTSSL algorithms. Furthermore, we conduct extensive ablation analyses to tease apart the factors that are the most important to the success of DeCon. The source code is available at \url{https://github.com/Gank0078/DeCon}.
arXiv:2605.15177v2 Announce Type: replace
Abstract: Test-time compute scaling is a primary axis for improving LLM reasoning. Existing methods primarily scale depth by extending a single reasoning trace. Scaling breadth by sampling multiple candidates in parallel is straightforward, but introduces a selection bottleneck: choosing the best candidate without a ground-truth verifier, since pointwise LLM judging is noisy and biased. To address this, we introduce OpenDeepThink, a population-based test-time compute framework that selects via pairwise Bradley-Terry comparison. Each generation, the LLM judges random pairs of candidates and aggregates votes via Bradley-Terry into a global ranking; top-ranked candidates are preserved and the top three quarters are mutated using the natural-language critiques produced during comparison; the bottom quarter is discarded. OpenDeepThink raises Gemini 3.1 Pro's effective Codeforces Elo by +405 points in eight sequential LLM-call rounds (~27 minutes wall-clock). The pipeline transfers across weaker and stronger models without retuning, and on the multi-domain HLE benchmark, gains appear concentrated in objectively verifiable domains and reverse in subjective ones. We release CF-73, a curated set of 73 expert-rated Codeforces problems with International Grandmaster annotation and 99% local-evaluation agreement against the official verdict.
arXiv:2511.07288v2 Announce Type: replace
Abstract: Learning complex policies with Reinforcement Learning (RL) is often hindered by instability and slow convergence, a problem exacerbated by the difficulty of reward engineering. Imitation Learning (IL) from expert demonstrations bypasses this reliance on rewards. However, state-of-the-art IL methods, exemplified by Generative Adversarial Imitation Learning (GAIL)Ho et. al, suffer from severe sample inefficiency. This is a direct consequence of their foundational on-policy algorithms, such as TRPO Schulman et.al. In this work, we introduce an adversarial imitation learning algorithm that incorporates off-policy learning to improve sample efficiency. By combining an off-policy framework with auxiliary techniques specifically, in this case a double Q network based stabilization and value learning without reward function inference we demonstrate a reduction in the samples required to robustly match expert behavior.
arXiv:2605.18599v1 Announce Type: new
Abstract: Transformer-based models have advanced feedforward novel view synthesis (NVS). Current architectures such as GS-LRM and LVSM mix semantic information (e.g., RGB) and spatial information (e.g., Pl\"ucker rays) into a shared feature space. Since Pl\"ucker rays naturally carry lattice-like spatial structure, these designs can make the spatial bias interfere with appearance representation and degrade rendering fidelity. To this end, we propose to decouple the representation of feedforward NVS transformers into separate semantic and spatial tokens. The decoupled design keeps semantic and spatial information explicit in their branches while preserving cross-branch interaction through shared attention routing. Built on this design, we introduce optional categorized supervision and bidirectional modulation: the former provides branch-specific training signals, while the latter improves interaction between the two branches. Notably, the base decoupled design introduces virtually zero additional inference latency due to its architectural design. The proposed designs achieve consistent improvements, demonstrating effectiveness across decoder-only and encoder-decoder feedforward NVS models.
arXiv:2605.09826v2 Announce Type: replace
Abstract: Theory of Mind (ToM), the ability to track others epistemic state, makes humans efficient collaborators. AI agents need the same capacity in multi agent settings, yet existing benchmarks mostly test literal ToM by asking direct belief questions. The ability act optimally on implicit beliefs in embodied environments, called functional ToM, remains largely untested. We introduce EnactToM, an evolving benchmark of 300 embodied multi-agent tasks set in a 3D household with partial observability, private information, and constrained communication. Each task is formally verified for solvability and required epistemic depth, and new tasks are generated increase difficulty as models improve. On the hard split, all seven evaluated frontier models score 0.0% Pass^3 on functional task completion, while averaging 45.0% on literal belief probes. Manual analysis traces 93% of sampled failures to epistemic coordination breakdowns such as withheld information, ignored partner constraints, and misallocated messages, providing a concrete target for future work.
arXiv:2605.03409v2 Announce Type: replace
Abstract: We present Robust Agent Compensation (RAC), a log-based recovery paradigm (providing a safety net) implemented through an architectural extension that can be applied to most Agent frameworks to support reliable executions (avoiding unintended side effects). Users can choose to enable RAC without changing their current agent code (e.g., LangGraph agents). The proposed approach can be implemented in most existing agent frameworks via their existing extension points. We present an implementation based on LangChain, demonstrate its viability through the $\tau$-bench and REALM-Bench, and show that when solving complex problems, RAC is 1.5-8X or more better in both latency and token economy compared to state-of-the-art LLM-based recovery approaches.