Analytical placers rely on differentiable objective functions to guide placement, typically combining intermediate surrogate metrics such as half-perimeter wirelength (HPWL) and cell-density penalties. However, these placement-stage surrogates remain misaligned with downstream routed and timing quality. Prior work reduces this gap with human-designed terms or learned black-box surrogates, but the former requires expert retuning and the latter is difficult to explain, debug, or deploy in analytical placement flows. CoEvoP&R addresses these limitations with a large language model (LLM)-based framework that automatically evolves analytical placement objectives. At each generation, the prompt combines the restricted objective interface, baseline context, and archived prior candidates with routing-related feedback from placement, timing proxy, and routing tools. The LLM proposes readable differentiable objectives, which are embedded and validated in DREAMPlace, evaluated through a timing proxy and an actual router, and stored with their feedback to guide later generations. Across eight ChiP-Bench Nangate45 designs and three seeds, CoEvoP&R reduces post-route routed wirelength and congestion by 16.9% and 36.7%, with gains of 0.70 ns in worst negative slack and a 912 ns reduction in total negative slack magnitude over native DREAMPlace. Across eight ICCAD 2015 Superblue designs, it reduces post-route routed wirelength and congestion by 5.4% and 23.2%. Code is available at https://github.com/FCHXWH823/CoEvoP-R.git.
展开完整摘要收起摘要↓
Analytical placers rely on differentiable objective functions to guide placement, typically combining intermediate surrogate metrics such as half-perimeter wirelength (HPWL) and cell-density penalties. However, these placement-stage surrogates remain misaligned with downstream routed and timing quality. Prior work reduces this gap with human-designed terms or learned black-box surrogates, but the former requires expert retuning and the latter is difficult to explain, debug, or deploy in analytical placement flows. CoEvoP&R addresses these limitations with a large language model (LLM)-based framework that automatically evolves analytical placement objectives. At each generation, the prompt combines the restricted objective interface, baseline context, and archived prior candidates with routing-related feedback from placement, timing proxy, and routing tools. The LLM proposes readable differentiable objectives, which are embedded and validated in DREAMPlace, evaluated through a timing proxy and an actual router, and stored with their feedback to guide later generations. Across eight ChiP-Bench Nangate45 designs and three seeds, CoEvoP&R reduces post-route routed wirelength and congestion by 16.9% and 36.7%, with gains of 0.70 ns in worst negative slack and a 912 ns reduction in total negative slack magnitude over native DREAMPlace. Across eight ICCAD 2015 Superblue designs, it reduces post-route routed wirelength and congestion by 5.4% and 23.2%. Code is available at https://github.com/FCHXWH823/CoEvoP-R.git.
作者Cheng Yan, Guangyang Ye, Wuyang Zhang, Fan Xu, Zhijun Fan, Xiang Xia, Yanyong Zhang
Test-time scaling empowers Large Reasoning Models (LRMs) to tackle complex tasks via extensive Chain-of-Thought (CoT). However, this often induces the "overthinking" paradox, where redundant reasoning increases computational overhead without guaranteeing accuracy. Existing test-time efficiency optimization methods primarily fall into two categories: information-theoretic approaches, which are prone to "deceptive convergence" where low uncertainty masks hallucinations, and latent representation analyses, which are often post-hoc, lacking the real-time sensitivity for dynamic reasoning. To bridge this gap, we first posit the Phase-Momentum Alignment Hypothesis, asserting that reasoning correctness hinges on the temporal synchronization between geometric momentum and uncertainty resolution. We then theoretically formulate the Cognitive-Energy Model to characterize these dynamics through two orthogonal dimensions: Geometric Cognitive Effort, quantified by latent velocity and tortuosity, and Entropic Cognitive Uncertainty. To operationalize this, we introduce PUMA (Phase-Uncertainty Momentum Alignment), a training-free framework employing a tiered diagnostic architecture. By coupling lightweight phase monitoring with event-triggered geometric analysis, PUMA effectively distinguishes active exploration from passive stagnation, enabling precise interventions through adaptive truncation or corrective measures. Extensive experiments on LRMs spanning 1.5B to 32B demonstrate that PUMA consistently outperforms state-of-the-art baselines across diverse benchmarks, achieving a superior accuracy-efficiency trade-off and robust cross-domain generalization.
展开完整摘要收起摘要↓
Test-time scaling empowers Large Reasoning Models (LRMs) to tackle complex tasks via extensive Chain-of-Thought (CoT). However, this often induces the "overthinking" paradox, where redundant reasoning increases computational overhead without guaranteeing accuracy. Existing test-time efficiency optimization methods primarily fall into two categories: information-theoretic approaches, which are prone to "deceptive convergence" where low uncertainty masks hallucinations, and latent representation analyses, which are often post-hoc, lacking the real-time sensitivity for dynamic reasoning. To bridge this gap, we first posit the Phase-Momentum Alignment Hypothesis, asserting that reasoning correctness hinges on the temporal synchronization between geometric momentum and uncertainty resolution. We then theoretically formulate the Cognitive-Energy Model to characterize these dynamics through two orthogonal dimensions: Geometric Cognitive Effort, quantified by latent velocity and tortuosity, and Entropic Cognitive Uncertainty. To operationalize this, we introduce PUMA (Phase-Uncertainty Momentum Alignment), a training-free framework employing a tiered diagnostic architecture. By coupling lightweight phase monitoring with event-triggered geometric analysis, PUMA effectively distinguishes active exploration from passive stagnation, enabling precise interventions through adaptive truncation or corrective measures. Extensive experiments on LRMs spanning 1.5B to 32B demonstrate that PUMA consistently outperforms state-of-the-art baselines across diverse benchmarks, achieving a superior accuracy-efficiency trade-off and robust cross-domain generalization.
Value-aware AI systems require explicit computational representations of human values (groundings) and their aggregation into value systems in order to align their decisions with ours. As such representations are difficult to elicit, value learning seeks to infer them by observing human behaviour. This work addresses the lack of grounded value learning methods in generative AI: existing approaches typically replicate human preferences without awareness of the multidimensional structure of value alignment, or lack principled value system elicitation methods. To address these gaps, we adapt a previously validated value system learning method to the generative AI setting, which, based on pairwise prompt-response preference data, simultaneously learns: i) an implementation of a grounding for a set of values given by a multi-objective reward model, and ii) a value system representation in the form of a weighted linear scalarization of the previous grounding model. To ensure that the learned value systems are based on coherent value representations, our algorithm dynamically prioritizes the grounding learning process. We evaluate the method against baselines and a contemporary method on prompt-response preference datasets. Results show competitive performance and minimal trade-offs against the baselines, while improving explainability.
展开完整摘要收起摘要↓
Value-aware AI systems require explicit computational representations of human values (groundings) and their aggregation into value systems in order to align their decisions with ours. As such representations are difficult to elicit, value learning seeks to infer them by observing human behaviour. This work addresses the lack of grounded value learning methods in generative AI: existing approaches typically replicate human preferences without awareness of the multidimensional structure of value alignment, or lack principled value system elicitation methods. To address these gaps, we adapt a previously validated value system learning method to the generative AI setting, which, based on pairwise prompt-response preference data, simultaneously learns: i) an implementation of a grounding for a set of values given by a multi-objective reward model, and ii) a value system representation in the form of a weighted linear scalarization of the previous grounding model. To ensure that the learned value systems are based on coherent value representations, our algorithm dynamically prioritizes the grounding learning process. We evaluate the method against baselines and a contemporary method on prompt-response preference datasets. Results show competitive performance and minimal trade-offs against the baselines, while improving explainability.
We introduce the Deep Second-Order Stochastic Residual Method (D2SRM) for high-dimensional, Hessian-dependent fully nonlinear parabolic PDEs. A single scalar space--time network generates derivative-consistent approximations of the solution, gradient, and Hessian, which are trained jointly through second-order Brownian one-step residuals and terminal value and gradient penalties. For globally Lipschitz equations with identity diffusion and sufficiently weak Hessian coupling, we establish well-posedness in a Brownian occupation space and develop a population-level convergence theory. Under additional regularity, an a posteriori estimate bounds the squared full-jet occupation error of any admissible candidate by the time step and its population objective. For approximate population minimizers, the error bound separates time discretization, neural approximation, and population suboptimality; when the latter two terms are $O(h)$, the full-jet occupation norm is $O(h^{1/2})$. Experiments on a 100-dimensional manufactured benchmark compare terminal treatments, probe Hessian couplings inside and outside the proved small-gain range, and show decreasing errors as the time step decreases. The code is available at https://github.com/ZZHPKU/D2SRM.
展开完整摘要收起摘要↓
We introduce the Deep Second-Order Stochastic Residual Method (D2SRM) for high-dimensional, Hessian-dependent fully nonlinear parabolic PDEs. A single scalar space--time network generates derivative-consistent approximations of the solution, gradient, and Hessian, which are trained jointly through second-order Brownian one-step residuals and terminal value and gradient penalties. For globally Lipschitz equations with identity diffusion and sufficiently weak Hessian coupling, we establish well-posedness in a Brownian occupation space and develop a population-level convergence theory. Under additional regularity, an a posteriori estimate bounds the squared full-jet occupation error of any admissible candidate by the time step and its population objective. For approximate population minimizers, the error bound separates time discretization, neural approximation, and population suboptimality; when the latter two terms are $O(h)$, the full-jet occupation norm is $O(h^{1/2})$. Experiments on a 100-dimensional manufactured benchmark compare terminal treatments, probe Hessian couplings inside and outside the proved small-gain range, and show decreasing errors as the time step decreases. The code is available at https://github.com/ZZHPKU/D2SRM.
作者Wendi Guo, Søren Byg Vilsen, Daniel Ioan Stroe, Yaqi Li, Yicun Huang, Ashima Verma, Daniel Brandell
Supercharging of lithium-ion batteries (LiBs) requires robust health monitoring to ensure durability, safety, and user confidence, particularly for emerging vehicle-to-grid applications with bidirectional energy flows. Yet battery management remains largely disconnected from the material and structural origins of aging, limiting both interpretable health assessment and informed battery design. Here we propose a physics-informed learning framework with virtual sensing that infers hard-to-measure design parameters, including solid-state diffusion coefficient, electrode thickness, ion concentration, and particle size, directly from standard battery management system (BMS) measurements. Across diverse fast-charging strategies and driving profiles, embedding a digital-twin-derived particle-cracking mechanism as a soft constraint reduces trajectory and lifetime prediction errors by 6-8 times relative to state-of-the-art machine learning baselines using only 2% early-life observations. We further show that accurate degradation extrapolation does not require fully resolved governing equations; validated partial mechanisms, jointly refined with limited data, provide sufficient guidance. Virtual sensing transforms standard charging signals into latent design variables without additional sensors, bridging observable battery behavior and underlying aging processes while reducing capacity loss error by up to 39%, end-of-life (EOL) error by 17%, and prediction variability by up to 54%, enabling real-time exploration of new battery configurations. More broadly, the proposed framework establishes a practical feedback loop between deployment and development, demonstrating how real-world operation can continuously inform upstream design decisions across complex multiphysics systems.
展开完整摘要收起摘要↓
Supercharging of lithium-ion batteries (LiBs) requires robust health monitoring to ensure durability, safety, and user confidence, particularly for emerging vehicle-to-grid applications with bidirectional energy flows. Yet battery management remains largely disconnected from the material and structural origins of aging, limiting both interpretable health assessment and informed battery design. Here we propose a physics-informed learning framework with virtual sensing that infers hard-to-measure design parameters, including solid-state diffusion coefficient, electrode thickness, ion concentration, and particle size, directly from standard battery management system (BMS) measurements. Across diverse fast-charging strategies and driving profiles, embedding a digital-twin-derived particle-cracking mechanism as a soft constraint reduces trajectory and lifetime prediction errors by 6-8 times relative to state-of-the-art machine learning baselines using only 2% early-life observations. We further show that accurate degradation extrapolation does not require fully resolved governing equations; validated partial mechanisms, jointly refined with limited data, provide sufficient guidance. Virtual sensing transforms standard charging signals into latent design variables without additional sensors, bridging observable battery behavior and underlying aging processes while reducing capacity loss error by up to 39%, end-of-life (EOL) error by 17%, and prediction variability by up to 54%, enabling real-time exploration of new battery configurations. More broadly, the proposed framework establishes a practical feedback loop between deployment and development, demonstrating how real-world operation can continuously inform upstream design decisions across complex multiphysics systems.
Animal re-identification (Re-ID) relies on fine-grained identity cues that can be disrupted by blur, noise, compression, and other visual degradations. Existing robustness strategies based on degradation-augmented training or pixel-level restoration improve robustness indirectly, but do not explicitly repair shifts in the identity retrieval space. We study corruption-robust animal Re-ID as input-conditioned feature-space repair and introduce DARA, a lightweight retrofit for compact Re-ID models. DARA freezes the fine-tuned backbone and learns routed low-rank residual experts to adapt degraded-input embeddings without corruption-type annotations. To stabilize this adaptive repair, original-to-corrupted distillation uses an original-image teacher to preserve individual embeddings and retrieval relations. Experiments on ATRW, FriesianCattle2017, MPDD, and SeaStarReID2023 show that DARA improves corrupted-query retrieval over standard and augmentation-based fine-tuning, generalizes to unseen corruptions and cross-domain evaluation, and recovers 77.0% of the corrupted-query mAP gap to full corrupted fine-tuning while adding only 0.49% parameters and 0.05% FLOPs.
展开完整摘要收起摘要↓
Animal re-identification (Re-ID) relies on fine-grained identity cues that can be disrupted by blur, noise, compression, and other visual degradations. Existing robustness strategies based on degradation-augmented training or pixel-level restoration improve robustness indirectly, but do not explicitly repair shifts in the identity retrieval space. We study corruption-robust animal Re-ID as input-conditioned feature-space repair and introduce DARA, a lightweight retrofit for compact Re-ID models. DARA freezes the fine-tuned backbone and learns routed low-rank residual experts to adapt degraded-input embeddings without corruption-type annotations. To stabilize this adaptive repair, original-to-corrupted distillation uses an original-image teacher to preserve individual embeddings and retrieval relations. Experiments on ATRW, FriesianCattle2017, MPDD, and SeaStarReID2023 show that DARA improves corrupted-query retrieval over standard and augmentation-based fine-tuning, generalizes to unseen corruptions and cross-domain evaluation, and recovers 77.0% of the corrupted-query mAP gap to full corrupted fine-tuning while adding only 0.49% parameters and 0.05% FLOPs.
作者Haolin Ren, Ziyang Huang, Chenhao Yuan, Jun Zhao, Kang Liu
Diffusion large language models (dLLMs) are a promising alternative to autoregressive generation. However, reasoning-oriented post-training for dLLMs remains challenging. Supervised fine-tuning (SFT) for dLLMs requires dense but often off-policy masked states, while reinforcement learning (RL) relies on sparse rewards or value modeling. This paper proposes trace-based on-policy distillation (TOPD), a teacher-supervised framework that transfers reasoning ability to a target dLLM without reward estimation. The key idea is to supervise a dLLM on its own denoising trajectory, focusing on the trace-aligned token decisions that form the final response. Specifically, TOPD samples on-policy diffusion trajectories from the target dLLM, obtains teacher token distributions from a teacher model on the corresponding partially denoised states, and updates the target dLLM with a token-level Reverse Kullback-Leibler (Reverse-KL) objective. This design preserves dense teacher supervision while aligning training with the model's own denoising states. On mathematical reasoning benchmarks, TOPD enables SDAR-4B-Chat to match the MATH500 accuracy of its RL-trained counterpart TraDo-4B-Instruct, with gains of +5.7 under static evaluation and +4.5 under dynamic evaluation. Compared with the RL-trained counterpart, TOPD achieves this with 4$\times$ fewer rollout rounds, corresponding to an estimated 96.0$\times$ to-accuracy model-compute speedup.
展开完整摘要收起摘要↓
Diffusion large language models (dLLMs) are a promising alternative to autoregressive generation. However, reasoning-oriented post-training for dLLMs remains challenging. Supervised fine-tuning (SFT) for dLLMs requires dense but often off-policy masked states, while reinforcement learning (RL) relies on sparse rewards or value modeling. This paper proposes trace-based on-policy distillation (TOPD), a teacher-supervised framework that transfers reasoning ability to a target dLLM without reward estimation. The key idea is to supervise a dLLM on its own denoising trajectory, focusing on the trace-aligned token decisions that form the final response. Specifically, TOPD samples on-policy diffusion trajectories from the target dLLM, obtains teacher token distributions from a teacher model on the corresponding partially denoised states, and updates the target dLLM with a token-level Reverse Kullback-Leibler (Reverse-KL) objective. This design preserves dense teacher supervision while aligning training with the model's own denoising states. On mathematical reasoning benchmarks, TOPD enables SDAR-4B-Chat to match the MATH500 accuracy of its RL-trained counterpart TraDo-4B-Instruct, with gains of +5.7 under static evaluation and +4.5 under dynamic evaluation. Compared with the RL-trained counterpart, TOPD achieves this with 4$\times$ fewer rollout rounds, corresponding to an estimated 96.0$\times$ to-accuracy model-compute speedup.
On-policy knowledge distillation transfers reasoning from large teachers to compact students, but existing approaches suffer three compounding failure modes: (i) cold-start collapse, where a fresh student assigns near-zero mass to teacher-preferred tokens; (ii) state-agnostic divergence scheduling, where time-only forward/reverse-KL interpolation ignores the student's coverage state; and (iii) binary reward sparsity, where pass/fail signals discard information from partially correct traces. We present CADENCE, a unified framework with a targeted fix for each. Its DRIFT mechanism schedules a per-token convex mixture of forward-KL and reverse-KL surrogate objectives on student-sampled trajectories (per-token surrogates, not sequence-level KL gradient estimators). Six components extend it: (A) COVA, a coverage-adaptive $β$ schedule accelerating the forward-to-reverse transition; (B) FTB, a forking-token boost concentrating gradient at high-entropy positions via a globally-normalized entropy reference; (C) CCD, a dense reward adding numerical-proximity partial credit for incorrect-but-close traces; (D) LAP, brevity-preferential correct-rollout reinforcement; (E) EMR, an entropy-matching calibration regularizer; (F) BSD, a bootstrapped self-distillation phase. On GSM8K and MATH-500 (corrected 512-token protocol, 5 seeds, reported std), CADENCE distills a 0.5B student from a 1.5B teacher to 69.8 $\pm$ 0.5% GSM8K pass@1 (from 48.7% pretrained; 63.2% of the teacher gap closed) and to 72.1 $\pm$ 0.4% with a 3B teacher (76.2% closed), beating the strongest matched-compute label-using baseline (DRIFT+binary reward) by +4.4 $\pm$ 0.7 points. All experiments run on a single Apple Mac Studio (M-series, 64GB unified memory), showing principled distillation reaches strong reasoning quality without datacenter-scale hardware.
展开完整摘要收起摘要↓
On-policy knowledge distillation transfers reasoning from large teachers to compact students, but existing approaches suffer three compounding failure modes: (i) cold-start collapse, where a fresh student assigns near-zero mass to teacher-preferred tokens; (ii) state-agnostic divergence scheduling, where time-only forward/reverse-KL interpolation ignores the student's coverage state; and (iii) binary reward sparsity, where pass/fail signals discard information from partially correct traces. We present CADENCE, a unified framework with a targeted fix for each. Its DRIFT mechanism schedules a per-token convex mixture of forward-KL and reverse-KL surrogate objectives on student-sampled trajectories (per-token surrogates, not sequence-level KL gradient estimators). Six components extend it: (A) COVA, a coverage-adaptive $β$ schedule accelerating the forward-to-reverse transition; (B) FTB, a forking-token boost concentrating gradient at high-entropy positions via a globally-normalized entropy reference; (C) CCD, a dense reward adding numerical-proximity partial credit for incorrect-but-close traces; (D) LAP, brevity-preferential correct-rollout reinforcement; (E) EMR, an entropy-matching calibration regularizer; (F) BSD, a bootstrapped self-distillation phase. On GSM8K and MATH-500 (corrected 512-token protocol, 5 seeds, reported std), CADENCE distills a 0.5B student from a 1.5B teacher to 69.8 $\pm$ 0.5% GSM8K pass@1 (from 48.7% pretrained; 63.2% of the teacher gap closed) and to 72.1 $\pm$ 0.4% with a 3B teacher (76.2% closed), beating the strongest matched-compute label-using baseline (DRIFT+binary reward) by +4.4 $\pm$ 0.7 points. All experiments run on a single Apple Mac Studio (M-series, 64GB unified memory), showing principled distillation reaches strong reasoning quality without datacenter-scale hardware.
作者Sijing Wu, Yunhao Li, Huiyu Duan, Yucheng Zhu, Xiongkuo Min, Patrick Le Callet, Guangtao Zhai
AI-generated human-centric videos play a crucial role in a wide range of modern applications. However, they often suffer from quality issues and semantic mismatches, underscoring the importance of effective quality assessment for such videos. To this end, we extend our previous dataset HVEval with pairwise preference annotations, resulting in HVEval+, the largest holistic quality assessment dataset for AI-generated human-centric videos, which comprises 1k prompts based on a comprehensive taxonomy, 20k videos generated by 24 text-to-video (T2V) models, and extensive human annotations, including 60k mean opinion scores (MOSs) and 60k preference pairs across 3 dimensions (i.e., spatial quality, temporal quality, and text-video correspondence), as well as 20k category-specific question-answer (Q&A) pairs. Along with the HVEval+ dataset, we further propose MoE-Rater, a Mixture-of-Experts (MoE)-inspired and multimodal large language model (MLLM)-based all-in-one method that supports multi-dimensional quality rating, multi-dimensional pairwise comparison, and category-specific question answering within a single model. Specifically, we introduce Mixture of Projector Experts (MoPE) and Mixture of LoRA Experts (MoLE), together with a three-stage training strategy consisting of task-aware pre-training, task-specific adaptation, and adaptive routing optimization, to effectively unify multiple tasks, resulting in superior performance on both HVEval+ and Human-AGVQA datasets. Extensive experiments and comprehensive analysis demonstrate the significant potential of the HVEval+ dataset and the MoE-Rater method in advancing AI-generated video quality assessment and further facilitating the evaluation and optimization of T2V models.
展开完整摘要收起摘要↓
AI-generated human-centric videos play a crucial role in a wide range of modern applications. However, they often suffer from quality issues and semantic mismatches, underscoring the importance of effective quality assessment for such videos. To this end, we extend our previous dataset HVEval with pairwise preference annotations, resulting in HVEval+, the largest holistic quality assessment dataset for AI-generated human-centric videos, which comprises 1k prompts based on a comprehensive taxonomy, 20k videos generated by 24 text-to-video (T2V) models, and extensive human annotations, including 60k mean opinion scores (MOSs) and 60k preference pairs across 3 dimensions (i.e., spatial quality, temporal quality, and text-video correspondence), as well as 20k category-specific question-answer (Q&A) pairs. Along with the HVEval+ dataset, we further propose MoE-Rater, a Mixture-of-Experts (MoE)-inspired and multimodal large language model (MLLM)-based all-in-one method that supports multi-dimensional quality rating, multi-dimensional pairwise comparison, and category-specific question answering within a single model. Specifically, we introduce Mixture of Projector Experts (MoPE) and Mixture of LoRA Experts (MoLE), together with a three-stage training strategy consisting of task-aware pre-training, task-specific adaptation, and adaptive routing optimization, to effectively unify multiple tasks, resulting in superior performance on both HVEval+ and Human-AGVQA datasets. Extensive experiments and comprehensive analysis demonstrate the significant potential of the HVEval+ dataset and the MoE-Rater method in advancing AI-generated video quality assessment and further facilitating the evaluation and optimization of T2V models.
作者Cheng-Yao Hong, Yifan Wang, Yuewei Lin, Chenyu You
Attention-based models often develop attention sinks, where a small number of tokens repeatedly attract attention and accumulate unusually large activations. In vision transformers, these outliers are closely related to registers, which have been diagnostically linked to global, low-frequency image structure. Existing work has largely studied registers through interpretability analyses and linear probes, leaving open whether they can be operationalized as plug-and-play signals for generation without retraining. We revisit this question in tokenized image generation. Using OpenCLIP and DINOv2 on ImageNet, we find that test-time register features exhibit stronger low-frequency concentration than both [CLS] readouts and patch-mean features, and show a consistent (albeit moderate) correlation with pixel-space DCT low-frequency energy. Motivated by these diagnostics, we introduce RegToken, a training-free procedure that converts register structure into a small set of global prior tokens by (i) NFN-based layer localization, (ii) TokenRank-guided subspace extraction, and (iii) a projection-and-conservation update on the register subspace. Inserted into a frozen compact 1D token generation pipeline, RegToken improves ImageNet generation and alignment metrics (e.g., FID-5k 20.5 to 20.1, SigLIP 3.6 to 3.9) without modifying pretrained weights, and accelerates test-time optimization (Steps@$τ$ 74 to 52). Overall, our results suggest that structures often viewed as attention artifacts can be repurposed as lightweight global priors for tokenized generation.
展开完整摘要收起摘要↓
Attention-based models often develop attention sinks, where a small number of tokens repeatedly attract attention and accumulate unusually large activations. In vision transformers, these outliers are closely related to registers, which have been diagnostically linked to global, low-frequency image structure. Existing work has largely studied registers through interpretability analyses and linear probes, leaving open whether they can be operationalized as plug-and-play signals for generation without retraining. We revisit this question in tokenized image generation. Using OpenCLIP and DINOv2 on ImageNet, we find that test-time register features exhibit stronger low-frequency concentration than both [CLS] readouts and patch-mean features, and show a consistent (albeit moderate) correlation with pixel-space DCT low-frequency energy. Motivated by these diagnostics, we introduce RegToken, a training-free procedure that converts register structure into a small set of global prior tokens by (i) NFN-based layer localization, (ii) TokenRank-guided subspace extraction, and (iii) a projection-and-conservation update on the register subspace. Inserted into a frozen compact 1D token generation pipeline, RegToken improves ImageNet generation and alignment metrics (e.g., FID-5k 20.5 to 20.1, SigLIP 3.6 to 3.9) without modifying pretrained weights, and accelerates test-time optimization (Steps@$τ$ 74 to 52). Overall, our results suggest that structures often viewed as attention artifacts can be repurposed as lightweight global priors for tokenized generation.
作者Han Wang, Zijun Wang, Shuoshuo Xue, Rui Cao, Fengjiao Cheng, Xiaodan Liang, Roy Ka-Wei Lee
Action-conditioned world models are a key component of embodied AI, serving as scalable policy evaluators that reduce reliance on expensive real-world rollouts. To accurately capture diverse action-induced dynamics, such models should satisfy three key objectives-Physical Plausibility (P), Action Adherence (A), and Visual Fidelity (V), collectively referred to as PAV-while remaining robust to both in-distribution (ID) expert demonstrations and out-of-distribution (OOD) actions. However, existing methods primarily rely on ID action-video pairs and pixel-level reconstruction losses, which do not explicitly optimize PAV objectives and generalize poorly beyond expert data. To address this, we propose PAVXploreRL, a reinforcement learning framework built on a pretrained latent world model that explicitly optimizes PAV objectives through reward-driven training. To improve action generalization, our method jointly leverages ID trajectories and noise-driven OOD action exploration, without paired video supervision. Experiments show that PAVXploreRL consistently outperforms pretrained baselines, achieving a 5.6% average gain across benchmarks and producing higher-quality PAV properties. As a policy evaluator, it also yields more reliable performance estimates and reduces the overestimation bias of prior expert-only world models such as Ctrl-World. Code: https://github.com/Social-AI-Studio/PAVXploreRL
展开完整摘要收起摘要↓
Action-conditioned world models are a key component of embodied AI, serving as scalable policy evaluators that reduce reliance on expensive real-world rollouts. To accurately capture diverse action-induced dynamics, such models should satisfy three key objectives-Physical Plausibility (P), Action Adherence (A), and Visual Fidelity (V), collectively referred to as PAV-while remaining robust to both in-distribution (ID) expert demonstrations and out-of-distribution (OOD) actions. However, existing methods primarily rely on ID action-video pairs and pixel-level reconstruction losses, which do not explicitly optimize PAV objectives and generalize poorly beyond expert data. To address this, we propose PAVXploreRL, a reinforcement learning framework built on a pretrained latent world model that explicitly optimizes PAV objectives through reward-driven training. To improve action generalization, our method jointly leverages ID trajectories and noise-driven OOD action exploration, without paired video supervision. Experiments show that PAVXploreRL consistently outperforms pretrained baselines, achieving a 5.6% average gain across benchmarks and producing higher-quality PAV properties. As a policy evaluator, it also yields more reliable performance estimates and reduces the overestimation bias of prior expert-only world models such as Ctrl-World. Code: https://github.com/Social-AI-Studio/PAVXploreRL
Unified multimodal models (UMMs) have recently demonstrated powerful instruction-based image editing capabilities, but they also raise serious concerns about unauthorized manipulation of personal portraits. Existing adversarial protection methods are mainly designed for either visual understanding or image generation models and often become ineffective when transferred to UMMs, which process an image through multiple complementary visual pathways. In this work, we first conduct a feature-level analysis of unified image editing. We observe that the ViT-based understanding branch and the VAE-based generation branch exhibit non-trivial structural agreement for the same input image. Although perturbing an individual branch can reduce this agreement and induce intermediate hidden-state deviations, such effects are asymmetric and gradually attenuated during multimodal fusion and generation. These observations reveal that single-branch feature distortion is insufficient for consistently disrupting unified image editing. Motivated by this finding, we propose CCS, a unified adversarial protection framework that jointly drives the ViT and VAE representations away from their clean counterparts while explicitly disrupting their cross-branch compatibility through linear CKA. By simultaneously removing stable information from both visual pathways and creating incompatible visual contexts, CCS prevents the UMM from recovering reliable identity information during editing. Extensive experiments demonstrate that CCS consistently outperforms existing protection methods in suppressing identity-preserving edits.
展开完整摘要收起摘要↓
Unified multimodal models (UMMs) have recently demonstrated powerful instruction-based image editing capabilities, but they also raise serious concerns about unauthorized manipulation of personal portraits. Existing adversarial protection methods are mainly designed for either visual understanding or image generation models and often become ineffective when transferred to UMMs, which process an image through multiple complementary visual pathways. In this work, we first conduct a feature-level analysis of unified image editing. We observe that the ViT-based understanding branch and the VAE-based generation branch exhibit non-trivial structural agreement for the same input image. Although perturbing an individual branch can reduce this agreement and induce intermediate hidden-state deviations, such effects are asymmetric and gradually attenuated during multimodal fusion and generation. These observations reveal that single-branch feature distortion is insufficient for consistently disrupting unified image editing. Motivated by this finding, we propose CCS, a unified adversarial protection framework that jointly drives the ViT and VAE representations away from their clean counterparts while explicitly disrupting their cross-branch compatibility through linear CKA. By simultaneously removing stable information from both visual pathways and creating incompatible visual contexts, CCS prevents the UMM from recovering reliable identity information during editing. Extensive experiments demonstrate that CCS consistently outperforms existing protection methods in suppressing identity-preserving edits.
Text-to-video (T2V) generation models have achieved strong visual realism, but improving physical plausibility can come at the cost of semantic consistency with the input text. This tension arises because physical preference is typically determined by comparing dynamics between two videos, without accounting for whether either video faithfully depicts the scene specified by the prompt, making physical-semantic conflict a systematic tendency under this supervision paradigm. We formulate this challenge as a constrained preference optimization problem and propose Physical and Semantic Direct Preference Optimization (PSDPO), which modulates each preference pair's contribution based on the agreement between its physical and semantic signals. A gradient-level analysis shows that PSDPO bounds the semantic drift from conflicting pairs to a controllable residual, and further motivates a staged optimization protocol that provably reduces cumulative drift. The resulting method operates entirely within the standard DPO framework, requiring no auxiliary models or additional loss terms. Experiments show that PSDPO improves physical plausibility by up to $2\times$ over the baseline on VideoPhy-2, while maintaining strong semantic consistency on VBench, achieving a more reliable balance than existing preference-based methods.
展开完整摘要收起摘要↓
Text-to-video (T2V) generation models have achieved strong visual realism, but improving physical plausibility can come at the cost of semantic consistency with the input text. This tension arises because physical preference is typically determined by comparing dynamics between two videos, without accounting for whether either video faithfully depicts the scene specified by the prompt, making physical-semantic conflict a systematic tendency under this supervision paradigm. We formulate this challenge as a constrained preference optimization problem and propose Physical and Semantic Direct Preference Optimization (PSDPO), which modulates each preference pair's contribution based on the agreement between its physical and semantic signals. A gradient-level analysis shows that PSDPO bounds the semantic drift from conflicting pairs to a controllable residual, and further motivates a staged optimization protocol that provably reduces cumulative drift. The resulting method operates entirely within the standard DPO framework, requiring no auxiliary models or additional loss terms. Experiments show that PSDPO improves physical plausibility by up to $2\times$ over the baseline on VideoPhy-2, while maintaining strong semantic consistency on VBench, achieving a more reliable balance than existing preference-based methods.
Conditional generative modeling remains a challenging problem in semi-supervised settings where labeled data is scarce but unlabeled samples are abundant. To effectively leverage structural information embedded within the unlabeled dataset and compensate for sparse conditioning signals, we propose a semi-supervised framework combining conditional stochastic interpolation with low-dimensional latent representations. RepG decomposes generation into two stages: label-dependent latent sampling and high-dimensional reconstruction. This isolates the supervised learning of conditional dependencies to a low-dimensional space, requiring few labels while utilizing the abundant unlabeled data purely for reconstruction. Theoretically, we establish an error decomposition showing that the Kullback-Leibler divergence of RepG comprises stage-wise estimation errors and a structural bias quantified by conditional mutual information. For deep neural network estimators, we derive non-asymptotic convergence rates proving that RepG significantly improves sample complexity. By confining the supervised estimation burden to the low intrinsic dimension of the latent representation, RepG achieves a strictly faster convergence rate. Complemented by a minimax lower bound, our theoretical results demonstrate that this method effectively mitigates the curse of dimensionality inherent in direct ambient-space generative modeling.
展开完整摘要收起摘要↓
Conditional generative modeling remains a challenging problem in semi-supervised settings where labeled data is scarce but unlabeled samples are abundant. To effectively leverage structural information embedded within the unlabeled dataset and compensate for sparse conditioning signals, we propose a semi-supervised framework combining conditional stochastic interpolation with low-dimensional latent representations. RepG decomposes generation into two stages: label-dependent latent sampling and high-dimensional reconstruction. This isolates the supervised learning of conditional dependencies to a low-dimensional space, requiring few labels while utilizing the abundant unlabeled data purely for reconstruction. Theoretically, we establish an error decomposition showing that the Kullback-Leibler divergence of RepG comprises stage-wise estimation errors and a structural bias quantified by conditional mutual information. For deep neural network estimators, we derive non-asymptotic convergence rates proving that RepG significantly improves sample complexity. By confining the supervised estimation burden to the low intrinsic dimension of the latent representation, RepG achieves a strictly faster convergence rate. Complemented by a minimax lower bound, our theoretical results demonstrate that this method effectively mitigates the curse of dimensionality inherent in direct ambient-space generative modeling.
Conditional diffusion models have become a powerful and flexible framework for learning complex conditional distributions from labeled data. In practice, however, acquiring high-quality labels is costly and time-consuming, leaving large volumes of unlabeled data unused. To address this, we introduce label-augmented conditional diffusion (LACD), a simple and effective approach that incorporates unlabeled examples by assigning them a designated trivial label and performing joint denoising score matching over the augmented dataset. We provide sufficient conditions guaranteeing population-level identifiability of the target conditional distribution under this scheme. Moreover, we establish rigorous statistical guarantees: when sufficiently many unlabeled samples are available, the sampling distribution produced by LACD converges strictly faster than the purely supervised estimator in total variation distance, and at least as fast in Wasserstein-1 distance. Extensive experiments on synthetic, image, and tabular benchmarks corroborate our theory and show substantial gains in sample efficiency and generative performance compared with the purely supervised estimator.
展开完整摘要收起摘要↓
Conditional diffusion models have become a powerful and flexible framework for learning complex conditional distributions from labeled data. In practice, however, acquiring high-quality labels is costly and time-consuming, leaving large volumes of unlabeled data unused. To address this, we introduce label-augmented conditional diffusion (LACD), a simple and effective approach that incorporates unlabeled examples by assigning them a designated trivial label and performing joint denoising score matching over the augmented dataset. We provide sufficient conditions guaranteeing population-level identifiability of the target conditional distribution under this scheme. Moreover, we establish rigorous statistical guarantees: when sufficiently many unlabeled samples are available, the sampling distribution produced by LACD converges strictly faster than the purely supervised estimator in total variation distance, and at least as fast in Wasserstein-1 distance. Extensive experiments on synthetic, image, and tabular benchmarks corroborate our theory and show substantial gains in sample efficiency and generative performance compared with the purely supervised estimator.
Autonomous rendezvous and proximity operations (RPO) in adversarial orbital environments require guidance architectures balancing target pursuit, safety preservation, and real-time adaptability under dynamically evolving interaction conditions. Although learning-based approaches show promise, their application to safety-critical orbital robotics remains limited by concerns regarding interpretability, robustness, and constraint awareness. This work presents an adaptive Model Predictive Control (MPC) framework for autonomous spacecraft RPO in multi-agent adversarial scenarios. The proposed architecture combines a constrained receding-horizon MPC formulation with a data-driven supervisory tuning layer that adjusts controller parameters from offline closed-loop evaluation and online interaction geometry. Relative motion follows Clohessy-Wiltshire (CW) dynamics, enabling computationally efficient finite-horizon prediction and real-time quadratic optimization. The MPC formulation incorporates actuator limits, predictive keep-out-zone constraints, slack-variable feasibility handling, and optional Control Barrier Function (CBF) safety filtering. Rather than generating thrust commands directly, the adaptive layer modifies interpretable MPC parameters, including tracking weights, safety penalties, minimum-separation objectives, and keep-out-zone objectives. The framework was evaluated in the official Kerbal Space Program Differential Game (KSPDG) Capture-the-Satellite environment through Monte Carlo simulations. Results demonstrate improved closed-loop robustness, adaptive maneuvering behavior, and rendezvous performance compared with fixed-parameter MPC while preserving safety-aware operation and real-time feasibility, providing a modular, interpretable foundation for adaptive spacecraft RPO.
展开完整摘要收起摘要↓
Autonomous rendezvous and proximity operations (RPO) in adversarial orbital environments require guidance architectures balancing target pursuit, safety preservation, and real-time adaptability under dynamically evolving interaction conditions. Although learning-based approaches show promise, their application to safety-critical orbital robotics remains limited by concerns regarding interpretability, robustness, and constraint awareness. This work presents an adaptive Model Predictive Control (MPC) framework for autonomous spacecraft RPO in multi-agent adversarial scenarios. The proposed architecture combines a constrained receding-horizon MPC formulation with a data-driven supervisory tuning layer that adjusts controller parameters from offline closed-loop evaluation and online interaction geometry. Relative motion follows Clohessy-Wiltshire (CW) dynamics, enabling computationally efficient finite-horizon prediction and real-time quadratic optimization. The MPC formulation incorporates actuator limits, predictive keep-out-zone constraints, slack-variable feasibility handling, and optional Control Barrier Function (CBF) safety filtering. Rather than generating thrust commands directly, the adaptive layer modifies interpretable MPC parameters, including tracking weights, safety penalties, minimum-separation objectives, and keep-out-zone objectives. The framework was evaluated in the official Kerbal Space Program Differential Game (KSPDG) Capture-the-Satellite environment through Monte Carlo simulations. Results demonstrate improved closed-loop robustness, adaptive maneuvering behavior, and rendezvous performance compared with fixed-parameter MPC while preserving safety-aware operation and real-time feasibility, providing a modular, interpretable foundation for adaptive spacecraft RPO.
Social navigation requires the robot to reason and respond in complex real-world environments. While recent works attempt to incorporate human-level intelligence into robot planning using large Vision-Language Models (VLMs), end-to-end frameworks often create an unpredictable black-box, and existing instruction-following methods are not designed for full autonomy. To bridge this gap, we present G2-Nav, a novel framework that grounds abstract social reasoning and guards safe real-world deployment. Instead of asking the VLM for direct planning decisions, G2-Nav translates its semantic reasoning into a vision-language costmap with reliability and interpretability. The VLM evaluates traversable regions and social agents from open-set perception, mapping social context into the costmap. To improve real-world robustness, the VLM performs semantic verification on upstream tracking, and we introduce a high-frequency safety check to guard against system latency prior to trajectory generation. We demonstrate through real-world experiments that G2-Nav delivers safe, efficient, and socially compliant autonomous navigation in unstructured environments. Code will be made publicly available.
展开完整摘要收起摘要↓
Social navigation requires the robot to reason and respond in complex real-world environments. While recent works attempt to incorporate human-level intelligence into robot planning using large Vision-Language Models (VLMs), end-to-end frameworks often create an unpredictable black-box, and existing instruction-following methods are not designed for full autonomy. To bridge this gap, we present G2-Nav, a novel framework that grounds abstract social reasoning and guards safe real-world deployment. Instead of asking the VLM for direct planning decisions, G2-Nav translates its semantic reasoning into a vision-language costmap with reliability and interpretability. The VLM evaluates traversable regions and social agents from open-set perception, mapping social context into the costmap. To improve real-world robustness, the VLM performs semantic verification on upstream tracking, and we introduce a high-frequency safety check to guard against system latency prior to trajectory generation. We demonstrate through real-world experiments that G2-Nav delivers safe, efficient, and socially compliant autonomous navigation in unstructured environments. Code will be made publicly available.
Scientists at European XFEL conduct experiments that generate very large and complex datasets. The subsequent data analysis is challenging as scientists must combine their domain expertise with facility- and software-specific knowledge scattered across documentation, tools, and support channels. To address this problem, we designed and evaluated an agentic AI system tailored to the scientists' needs and integrated with the high-performance computing environment of European XFEL. Using a design science research approach, we conducted a rapid literature review, a systematic evaluation of 16 AI tools, multiple interviews, a focus group, and a user study with experts at European XFEL to develop and evaluate two prototypes. Our study identifies key knowledge challenges in scientific data analysis, derives requirements for an AI agent that supports knowledge retrieval and source code generation, and proposes design recommendations for a specialized system adaptable to the evolving AI tool landscape. These findings provide guidance for developing maintainable AI support in highly specialized scientific environments.
展开完整摘要收起摘要↓
Scientists at European XFEL conduct experiments that generate very large and complex datasets. The subsequent data analysis is challenging as scientists must combine their domain expertise with facility- and software-specific knowledge scattered across documentation, tools, and support channels. To address this problem, we designed and evaluated an agentic AI system tailored to the scientists' needs and integrated with the high-performance computing environment of European XFEL. Using a design science research approach, we conducted a rapid literature review, a systematic evaluation of 16 AI tools, multiple interviews, a focus group, and a user study with experts at European XFEL to develop and evaluate two prototypes. Our study identifies key knowledge challenges in scientific data analysis, derives requirements for an AI agent that supports knowledge retrieval and source code generation, and proposes design recommendations for a specialized system adaptable to the evolving AI tool landscape. These findings provide guidance for developing maintainable AI support in highly specialized scientific environments.
作者Roshan Klein-Seetharaman, Daniel Wang, Andrew Xu
Existing tool-use benchmarks report a single success rate for complex, multistep tasks. Inspired by ideas from cognitive science, we distinguish tool use from tool discovery and decompose the latter into curiosity (the model's ability to discover the parts needed to build the tool), recognition (the model's ability to discover the process of creating the tool), and efficiency (the model's use of the tool after creation). We show that this framework can be applied to existing discovery tasks, such as Voyager. In addition, we provide evidence that recognition inversely scales with model size, and we introduce and analyze a class of combinatorial games that demonstrates this. We further observe inverse scaling in a separate environment designed to emulate real-world tasks.
展开完整摘要收起摘要↓
Existing tool-use benchmarks report a single success rate for complex, multistep tasks. Inspired by ideas from cognitive science, we distinguish tool use from tool discovery and decompose the latter into curiosity (the model's ability to discover the parts needed to build the tool), recognition (the model's ability to discover the process of creating the tool), and efficiency (the model's use of the tool after creation). We show that this framework can be applied to existing discovery tasks, such as Voyager. In addition, we provide evidence that recognition inversely scales with model size, and we introduce and analyze a class of combinatorial games that demonstrates this. We further observe inverse scaling in a separate environment designed to emulate real-world tasks.
作者Baochen Fu, Wenzhi Deng, Baihao Jin, Yang Li, Zihan Nie, Kailin Jiang, Yuntao Du, Weiye Song
Optical coherence tomography (OCT) imaging is essential for the diagnosis and treatment of retinal diseases. Although multimodal large language models (MLLMs) have demonstrated considerable potential in medical image analysis, existing benchmarks largely reduce OCT understanding to coarse-grained disease classification or isolated visual question answering, leaving the complete cognitive process from visual perception to clinical reasoning insufficiently evaluated. To address this limitation, we introduce OCT-Bench, a comprehensive benchmark dedicated to OCT image understanding. OCT-Bench comprises 10,076 high-quality multiple-choice questions constructed from 4,137 OCT images across seven public datasets. Following the real-world clinical interpretation workflow, we establish a hierarchical capability taxonomy consisting of 20 fine-grained tasks across three dimensions: Perception, Cognition, and Reasoning. These tasks cover a broad range of capabilities, including imaging attributes, retinal anatomy, lesion characteristics, spatial relationships, disease assessment, therapeutic decision-making, and prognostic management. We systematically evaluate 20 representative MLLMs, including proprietary models, open-source general-purpose models, and medical-domain models. Experimental results demonstrate that current models remain substantially short of reliable OCT understanding. Moreover, neither medical-domain adaptation nor increased model scale consistently improves performance across capability levels. OCT-Bench enables comprehensive and fine-grained evaluation of MLLMs, providing a foundation for identifying capability bottlenecks and advancing clinically grounded OCT understanding.
展开完整摘要收起摘要↓
Optical coherence tomography (OCT) imaging is essential for the diagnosis and treatment of retinal diseases. Although multimodal large language models (MLLMs) have demonstrated considerable potential in medical image analysis, existing benchmarks largely reduce OCT understanding to coarse-grained disease classification or isolated visual question answering, leaving the complete cognitive process from visual perception to clinical reasoning insufficiently evaluated. To address this limitation, we introduce OCT-Bench, a comprehensive benchmark dedicated to OCT image understanding. OCT-Bench comprises 10,076 high-quality multiple-choice questions constructed from 4,137 OCT images across seven public datasets. Following the real-world clinical interpretation workflow, we establish a hierarchical capability taxonomy consisting of 20 fine-grained tasks across three dimensions: Perception, Cognition, and Reasoning. These tasks cover a broad range of capabilities, including imaging attributes, retinal anatomy, lesion characteristics, spatial relationships, disease assessment, therapeutic decision-making, and prognostic management. We systematically evaluate 20 representative MLLMs, including proprietary models, open-source general-purpose models, and medical-domain models. Experimental results demonstrate that current models remain substantially short of reliable OCT understanding. Moreover, neither medical-domain adaptation nor increased model scale consistently improves performance across capability levels. OCT-Bench enables comprehensive and fine-grained evaluation of MLLMs, providing a foundation for identifying capability bottlenecks and advancing clinically grounded OCT understanding.
作者Mohammad Arvan, Amber E. Osterholt, Bailee Rue, Yuvaneswaren R. Sureshbabu, Krishna R. Patel, Rebecca T. Feinstein, Bethany C. Bray, Niranjan S. Karnik
Introduction. Clinical and Translational Science Award (CTSA) programs must document their scholars' research impact, but assembling each scholar's record by hand takes staff an estimated 15 hours and does not scale to a full cohort. An artificial intelligence (AI) agent could serve as a tool to gather scholar data across platforms and disciplines. Methods. We built a human-in-the-loop AI agent that assembles a dossier of sourced evidence for each scholar and drafts one-sentence Translational Science Benefits Model (TSBM) impact summaries for staff review. We evaluated it in the impact-reporting workflow of one CTSA hub across 10 career-development (KL2/K12) scholars. Two evaluation staff independently coded all 507 findings as accept, edit, or reject; the primary measure was the unanimous usable rate, defined as the share both accepted or edited. Results. Both reviewers accepted or edited 81.7% of the agent's findings. Reviewers each spent a median of 14 minutes per scholar, replacing an estimated 15 hours of manual assembly. Inter-rater agreement was moderate (Cohen's kappa 0.43 on the usable-versus-reject decision). A profile discovery study found the agent's recall close to human search. The agent's impact evidence spanned all four TSBM domains, and about a third of the reviewed findings fell in non-scholarly categories that routine processes tend to miss. Reviewers rated synthesis accuracy 4.5 and usefulness 4.8 on a 5-point scale. Conclusions. A human-in-the-loop AI agent can serve as the first-pass author of a scholar's impact record, shifting staff from collecting and writing to reviewing, and making cohort-scale impact reporting feasible.
展开完整摘要收起摘要↓
Introduction. Clinical and Translational Science Award (CTSA) programs must document their scholars' research impact, but assembling each scholar's record by hand takes staff an estimated 15 hours and does not scale to a full cohort. An artificial intelligence (AI) agent could serve as a tool to gather scholar data across platforms and disciplines. Methods. We built a human-in-the-loop AI agent that assembles a dossier of sourced evidence for each scholar and drafts one-sentence Translational Science Benefits Model (TSBM) impact summaries for staff review. We evaluated it in the impact-reporting workflow of one CTSA hub across 10 career-development (KL2/K12) scholars. Two evaluation staff independently coded all 507 findings as accept, edit, or reject; the primary measure was the unanimous usable rate, defined as the share both accepted or edited. Results. Both reviewers accepted or edited 81.7% of the agent's findings. Reviewers each spent a median of 14 minutes per scholar, replacing an estimated 15 hours of manual assembly. Inter-rater agreement was moderate (Cohen's kappa 0.43 on the usable-versus-reject decision). A profile discovery study found the agent's recall close to human search. The agent's impact evidence spanned all four TSBM domains, and about a third of the reviewed findings fell in non-scholarly categories that routine processes tend to miss. Reviewers rated synthesis accuracy 4.5 and usefulness 4.8 on a 5-point scale. Conclusions. A human-in-the-loop AI agent can serve as the first-pass author of a scholar's impact record, shifting staff from collecting and writing to reviewing, and making cohort-scale impact reporting feasible.
作者Bo Tang, Yang Zhang, Guomian Zhuang, Wenqiang Wei, Gaoyang Zheng, Lindong Xie, Yanchao Tan, Feiyu Xiong, Qingyu Yang, Edward Chung, Zhiyu li
Existing memory systems for long-horizon LLM agents often retrieve prior traces as passive context rather than converting them into executable capabilities. In this paper, we propose MSCE, a training-free Memory--Skill Co-Evolution framework that organizes agent experience into grounded step traces, reusable procedural policies, and declarative environmental cognition. MSCE crystallizes evidence-backed L2 policies with positive estimated gain into callable skills that retain evidence links, applicability boundaries, decision guidance, verification rules, and reliability estimates. It further introduces reflection-weighted value backfilling, which propagates sparse terminal feedback through dense local self-reflections to produce evidence-calibrated trace values for governing memory and skill evolution. Experiments on EvoAgentBench and LoCoMo demonstrate that MSCE significantly outperforms state-of-the-art skill-augmented and memory-driven agent baselines, exhibiting strong cross-domain transferability and lifelong-evolution capabilities.
展开完整摘要收起摘要↓
Existing memory systems for long-horizon LLM agents often retrieve prior traces as passive context rather than converting them into executable capabilities. In this paper, we propose MSCE, a training-free Memory--Skill Co-Evolution framework that organizes agent experience into grounded step traces, reusable procedural policies, and declarative environmental cognition. MSCE crystallizes evidence-backed L2 policies with positive estimated gain into callable skills that retain evidence links, applicability boundaries, decision guidance, verification rules, and reliability estimates. It further introduces reflection-weighted value backfilling, which propagates sparse terminal feedback through dense local self-reflections to produce evidence-calibrated trace values for governing memory and skill evolution. Experiments on EvoAgentBench and LoCoMo demonstrate that MSCE significantly outperforms state-of-the-art skill-augmented and memory-driven agent baselines, exhibiting strong cross-domain transferability and lifelong-evolution capabilities.
Long-horizon AI agents are becoming increasingly capable, yet their interaction with users remains surprisingly thin. In most workflows, users give an initial instruction, receive only selective textual updates, and lose a clear sense of what the agent is doing or when to step in. This leaves a missing part in the current agent ecosystem: an always-on Jarvis-style mediator that keeps the agent continuously reachable to the user. Such a mediator should support real-time spoken interaction with the user, answer questions without interrupting the worker, proactively report progress or confusion, and inject user guidance back into the agent's execution when useful. In this work, we introduce JarvisBench, a benchmark for measuring the dual value of mediation in long-horizon agent workflows. JarvisBench contains two complementary tracks: an agent-collaboration track that measures whether mediation improves downstream task completion, and a user-interaction track that measures whether mediation makes ongoing execution more understandable, responsive, and accessible to users. We instantiate the benchmark with a modular reference Jarvis prototype and evaluate it on 34 text-only WildClaw tasks executed in OpenClaw. Preliminary results with GPT-5.5, Claude Opus 4.7, Gemini-based, and GPT-based worker agents suggest that Jarvis-style mediation can provide trace-grounded responses to user questions and improve task performance when sparse user guidance is injected at appropriate moments. The results also show that effectiveness depends strongly on the mediator's LLM brain, highlighting both the promise of this missing middle layer and the need for broader community effort. Demo page https://cchen1436.github.io/jarvis
展开完整摘要收起摘要↓
Long-horizon AI agents are becoming increasingly capable, yet their interaction with users remains surprisingly thin. In most workflows, users give an initial instruction, receive only selective textual updates, and lose a clear sense of what the agent is doing or when to step in. This leaves a missing part in the current agent ecosystem: an always-on Jarvis-style mediator that keeps the agent continuously reachable to the user. Such a mediator should support real-time spoken interaction with the user, answer questions without interrupting the worker, proactively report progress or confusion, and inject user guidance back into the agent's execution when useful. In this work, we introduce JarvisBench, a benchmark for measuring the dual value of mediation in long-horizon agent workflows. JarvisBench contains two complementary tracks: an agent-collaboration track that measures whether mediation improves downstream task completion, and a user-interaction track that measures whether mediation makes ongoing execution more understandable, responsive, and accessible to users. We instantiate the benchmark with a modular reference Jarvis prototype and evaluate it on 34 text-only WildClaw tasks executed in OpenClaw. Preliminary results with GPT-5.5, Claude Opus 4.7, Gemini-based, and GPT-based worker agents suggest that Jarvis-style mediation can provide trace-grounded responses to user questions and improve task performance when sparse user guidance is injected at appropriate moments. The results also show that effectiveness depends strongly on the mediator's LLM brain, highlighting both the promise of this missing middle layer and the need for broader community effort. Demo page https://cchen1436.github.io/jarvis
作者Ran Wei, Le Zhu, Haochi Wang, Ruizhe Yang, Jiapeng Guan, Siyuan Ji, Yuchen Hu, Zhe Jiang, Xiangyang Ji
Software complexity is a long-standing challenge for system engineers. Model-Driven Engineering (MDE) addresses it by treating models as first-class artefacts, but a typical MDE process spans many tools and produces heterogeneous models of different system aspects, making traceability, maintenance, and change management difficult. We propose RADIANT, an engineering methodology that combines MDE with Multi-Agent Large Language Models (LLMs) for complete model-based system development, with a focus on safety-critical systems. From a carefully specified requirement model, RADIANT automatically generates heterogeneous models across engineering phases — a concept model, a domain-specific modelling language, a conforming system model, and a behaviour model — together with executable, element-level traceability links, on top of which it provides exact, automated change-impact analysis. Generated behaviour models are translated into CSP and formally verified (e.g.\ for deadlock freedom and convergence) with a counterexample-driven repair loop. Evaluating RADIANT across three LLMs, we find that the multi-agent decomposition reliably improves the syntactic validity of the generated formal artefacts over a single-agent baseline — and their executability where the model's code generation permits — while gains in semantic accuracy are model-dependent. A six-participant study shows an order-of-magnitude ($10$--$15\times$) reduction in development time, and the unmodified pipeline transfers to a second domain.
展开完整摘要收起摘要↓
Software complexity is a long-standing challenge for system engineers. Model-Driven Engineering (MDE) addresses it by treating models as first-class artefacts, but a typical MDE process spans many tools and produces heterogeneous models of different system aspects, making traceability, maintenance, and change management difficult. We propose RADIANT, an engineering methodology that combines MDE with Multi-Agent Large Language Models (LLMs) for complete model-based system development, with a focus on safety-critical systems. From a carefully specified requirement model, RADIANT automatically generates heterogeneous models across engineering phases — a concept model, a domain-specific modelling language, a conforming system model, and a behaviour model — together with executable, element-level traceability links, on top of which it provides exact, automated change-impact analysis. Generated behaviour models are translated into CSP and formally verified (e.g.\ for deadlock freedom and convergence) with a counterexample-driven repair loop. Evaluating RADIANT across three LLMs, we find that the multi-agent decomposition reliably improves the syntactic validity of the generated formal artefacts over a single-agent baseline — and their executability where the model's code generation permits — while gains in semantic accuracy are model-dependent. A six-participant study shows an order-of-magnitude ($10$--$15\times$) reduction in development time, and the unmodified pipeline transfers to a second domain.
Large language models (LLMs) are increasingly used to automate data-processing workflows, yet coding agents typically produce scripts that are not automatically materialized as persistent, editable platform artifacts. We call this disconnect the NL2Pipeline gap. To bridge it, we introduce \textsc{DataFlow-Harness}, a platform that guides an LLM agent to construct platform-native directed acyclic graphs (DAGs) through typed, incremental mutations rather than free-form scripts. The platform combines \textsc{DataFlow-Skills} for procedural guidance, a Model Context Protocol (MCP) layer that exposes the live operator registry and current pipeline state, and \textsc{DataFlow-WebUI}, which synchronizes conversational authoring with a visual DAG editor. On a 12-task data-engineering benchmark, \textsc{DataFlow-Harness} achieves a 93.3% observed end-to-end pass rate. Relative to Vanilla Claude Code, it reduces measured monetary cost by 72.5% and generation latency by 49.9%; its observed pass rate is within 0.9 percentage points of the Context-Aware Claude Code baseline while its cost is 42.8% lower. Per-task analysis indicates that Skills are most useful when construction depends on implicit procedural knowledge. These results show that live platform grounding can produce persistent, editable workflow artifacts with an observed reliability close to script-generation baselines and with lower measured construction cost and latency.
展开完整摘要收起摘要↓
Large language models (LLMs) are increasingly used to automate data-processing workflows, yet coding agents typically produce scripts that are not automatically materialized as persistent, editable platform artifacts. We call this disconnect the NL2Pipeline gap. To bridge it, we introduce \textsc{DataFlow-Harness}, a platform that guides an LLM agent to construct platform-native directed acyclic graphs (DAGs) through typed, incremental mutations rather than free-form scripts. The platform combines \textsc{DataFlow-Skills} for procedural guidance, a Model Context Protocol (MCP) layer that exposes the live operator registry and current pipeline state, and \textsc{DataFlow-WebUI}, which synchronizes conversational authoring with a visual DAG editor. On a 12-task data-engineering benchmark, \textsc{DataFlow-Harness} achieves a 93.3% observed end-to-end pass rate. Relative to Vanilla Claude Code, it reduces measured monetary cost by 72.5% and generation latency by 49.9%; its observed pass rate is within 0.9 percentage points of the Context-Aware Claude Code baseline while its cost is 42.8% lower. Per-task analysis indicates that Skills are most useful when construction depends on implicit procedural knowledge. These results show that live platform grounding can produce persistent, editable workflow artifacts with an observed reliability close to script-generation baselines and with lower measured construction cost and latency.
Deploying LLM agents typically requires a compact test-time student, even if a stronger teacher is available during training. We study knowledge brewing: distilling a teacher's interactive experience into a persistent external memory for the student. Crucially, this requires no weight updates, expert demonstrations, ground-truth labels, or test-time teacher access. This setting poses two challenges: environments provide only sparse, binary feedback, and teacher-authored notes must be inherently tailored to be concretely executable by a substantially weaker student. To address these hurdles, we propose AgentBrew, comprising two coupled components. First, a failure-triggered teacher--Ralph Loop mitigates sparse feedback by transforming student failures into environment-validated notes. Second, student-aware synthesis calibrates teacher knowledge to the weak executor's operational granularity, yielding model-specific, actionable guidance. Extensive evaluations and comprehensive ablations across coding, math, and tool-use tasks demonstrate that this asymmetric, training-free brewing paradigm produces highly capable yet deployable LLM agents.
展开完整摘要收起摘要↓
Deploying LLM agents typically requires a compact test-time student, even if a stronger teacher is available during training. We study knowledge brewing: distilling a teacher's interactive experience into a persistent external memory for the student. Crucially, this requires no weight updates, expert demonstrations, ground-truth labels, or test-time teacher access. This setting poses two challenges: environments provide only sparse, binary feedback, and teacher-authored notes must be inherently tailored to be concretely executable by a substantially weaker student. To address these hurdles, we propose AgentBrew, comprising two coupled components. First, a failure-triggered teacher--Ralph Loop mitigates sparse feedback by transforming student failures into environment-validated notes. Second, student-aware synthesis calibrates teacher knowledge to the weak executor's operational granularity, yielding model-specific, actionable guidance. Extensive evaluations and comprehensive ablations across coding, math, and tool-use tasks demonstrate that this asymmetric, training-free brewing paradigm produces highly capable yet deployable LLM agents.
We describe DS@GT's submission to the eRisk 2026 Task 1 challenge on conversational depression screening, in which systems interview LLM personas that simulate individuals with varying depression profiles and produce a Beck Depression Inventory II (BDI-II) score plus four key symptoms per persona, without directly asking sensitive mental health questions. Our pipeline evolved through three stages: a monolithic single-model prototype to start off, a baseline multi-agent architecture that separates conversational interviewing from BDI-II scoring under a coordinating orchestration layer, and a final hybrid configuration that replaces the paid GPT-5-nano interviewer with the open-source Gemma 27B. To offset the model's weaker reasoning and instruction-following, the hybrid adds three algorithmic components: a precomputed dialogue tree that standardizes interview openers and follow-ups, a reliability-weighted consensus aggregation inspired by the Weaver framework, and a cluster-based imputation step for unprobed symptoms. We submitted three fully automated runs across all 20 personas, with Run 1 from the paid baseline and Runs 2 and 3 from the hybrid. Hybrid Run 3 achieved an ADODL of 0.9063, ranking 3rd among all complete-submission runs and placing DS@GT 2nd among the 21 teams overall, while outperforming our paid baseline Run 1 (0.8841) at roughly one-quarter of the per-persona API cost. These results support our central hypothesis that with sufficient algorithmic supervision, a weaker open-source model can compete with a stronger proprietary model in the conversational interviewer role. Our source code is available at https://github.com/dsgt-arc/erisk-task1-2026.
展开完整摘要收起摘要↓
We describe DS@GT's submission to the eRisk 2026 Task 1 challenge on conversational depression screening, in which systems interview LLM personas that simulate individuals with varying depression profiles and produce a Beck Depression Inventory II (BDI-II) score plus four key symptoms per persona, without directly asking sensitive mental health questions. Our pipeline evolved through three stages: a monolithic single-model prototype to start off, a baseline multi-agent architecture that separates conversational interviewing from BDI-II scoring under a coordinating orchestration layer, and a final hybrid configuration that replaces the paid GPT-5-nano interviewer with the open-source Gemma 27B. To offset the model's weaker reasoning and instruction-following, the hybrid adds three algorithmic components: a precomputed dialogue tree that standardizes interview openers and follow-ups, a reliability-weighted consensus aggregation inspired by the Weaver framework, and a cluster-based imputation step for unprobed symptoms. We submitted three fully automated runs across all 20 personas, with Run 1 from the paid baseline and Runs 2 and 3 from the hybrid. Hybrid Run 3 achieved an ADODL of 0.9063, ranking 3rd among all complete-submission runs and placing DS@GT 2nd among the 21 teams overall, while outperforming our paid baseline Run 1 (0.8841) at roughly one-quarter of the per-persona API cost. These results support our central hypothesis that with sufficient algorithmic supervision, a weaker open-source model can compete with a stronger proprietary model in the conversational interviewer role. Our source code is available at https://github.com/dsgt-arc/erisk-task1-2026.
Safe and socially compliant navigation in open human-robot environments requires robots to reason about heterogeneous participants with different dynamics, autonomy levels, and social roles. Existing trajectory prediction and planning methods often rely on homogeneous interaction assumptions or enforce only geometric collision constraints, making it difficult to jointly model asymmetric interactions, coupled prediction-planning, and soft social norms. This paper proposes SAGE, a socially-aware generative engine for heterogeneous multi-agent navigation. SAGE represents robots and surrounding entities as a directed heterogeneous graph and employs a Heterogeneous Graph Transformer (HGT) to encode type-specific asymmetric interactions. Conditioned on the resulting context, a diffusion-based generative module jointly models future entity trajectories and robot trajectory plans. During inference, a training-free safety-social energy guidance mechanism refines sampled robot trajectories using differentiable collision, kinematic, task-progress, and role-conditioned social-compliance terms. Extensive experiments on real-world (ETH/UCY and SDD) and synthetic datasets verify the effectiveness of SAGE in improving safety and social compliance while maintaining task performance. The proposed guidance mechanism consistently reduces collision and social-violation rates, scales to teams of up to 20 robots, and enables explicit control of the safety-accuracy-task trade-off without retraining. These findings demonstrate the potential of SAGE as a scalable framework for socially-aware multi-agent navigation in complex environments.
展开完整摘要收起摘要↓
Safe and socially compliant navigation in open human-robot environments requires robots to reason about heterogeneous participants with different dynamics, autonomy levels, and social roles. Existing trajectory prediction and planning methods often rely on homogeneous interaction assumptions or enforce only geometric collision constraints, making it difficult to jointly model asymmetric interactions, coupled prediction-planning, and soft social norms. This paper proposes SAGE, a socially-aware generative engine for heterogeneous multi-agent navigation. SAGE represents robots and surrounding entities as a directed heterogeneous graph and employs a Heterogeneous Graph Transformer (HGT) to encode type-specific asymmetric interactions. Conditioned on the resulting context, a diffusion-based generative module jointly models future entity trajectories and robot trajectory plans. During inference, a training-free safety-social energy guidance mechanism refines sampled robot trajectories using differentiable collision, kinematic, task-progress, and role-conditioned social-compliance terms. Extensive experiments on real-world (ETH/UCY and SDD) and synthetic datasets verify the effectiveness of SAGE in improving safety and social compliance while maintaining task performance. The proposed guidance mechanism consistently reduces collision and social-violation rates, scales to teams of up to 20 robots, and enables explicit control of the safety-accuracy-task trade-off without retraining. These findings demonstrate the potential of SAGE as a scalable framework for socially-aware multi-agent navigation in complex environments.
作者Nguyen Viet Tuan Kiet, Bui Dinh Pham, Duong Quoc Chinh, Dao Van Tung, Tran Cong Dao, Huynh Thi Thanh Binh
Multi-agent planning becomes substantially harder when agents must improve specialized decision-making skills while keeping their internal implementations private. This regime arises when agents are developed independently, expose different interfaces and capabilities, and must nevertheless coordinate without sharing executable policies. Prior research has largely assumed centralized optimization, shared policy access, or common skill representations, making it poorly suited to privacy-constrained cooperation. We introduce RELIC, a framework for learning interpretable and composable skills via revealed principles. Each agent refines its own programmatic skill through private LLM-guided search, while a trusted orchestrator evaluates proposed updates solely through team-level performance. Successful behaviors are not broadcast as code; instead, they are abstracted into portable principles that other agents can instantiate within their own interfaces and recombine with local strategies. This separates coordination from implementation sharing, enabling cross-agent transfer under heterogeneous skill signatures. RELIC thus introduces a new paradigm for privacy-preserving skill learning and coordination in multi-agent planning.
展开完整摘要收起摘要↓
Multi-agent planning becomes substantially harder when agents must improve specialized decision-making skills while keeping their internal implementations private. This regime arises when agents are developed independently, expose different interfaces and capabilities, and must nevertheless coordinate without sharing executable policies. Prior research has largely assumed centralized optimization, shared policy access, or common skill representations, making it poorly suited to privacy-constrained cooperation. We introduce RELIC, a framework for learning interpretable and composable skills via revealed principles. Each agent refines its own programmatic skill through private LLM-guided search, while a trusted orchestrator evaluates proposed updates solely through team-level performance. Successful behaviors are not broadcast as code; instead, they are abstracted into portable principles that other agents can instantiate within their own interfaces and recombine with local strategies. This separates coordination from implementation sharing, enabling cross-agent transfer under heterogeneous skill signatures. RELIC thus introduces a new paradigm for privacy-preserving skill learning and coordination in multi-agent planning.
Large language models and LLM-based agents are widely used as personal chat assistants, enterprise copilots, and autonomous workflow agents. In all these applications, memory (the ability to retain, access, and reason over information accumulated over long contexts and multiple interactions) plays a crucial role in determining the reliability of any agent. We introduce RECON (Reasoning over Extended Contexts with Obfuscated Narratives), a benchmark for evaluating compositional reasoning over long contexts. RECON spans 24 case files across three domains (criminal, medical, and financial), each ranging from 50k to 100k tokens, and tests agents on six memory intensive tasks: reconstructing multi-hop evidence chains, propagating cascading invalidations, resolving source conflicts, counterfactual reasoning, satisfying temporal constraints, and temporal fact retrieval. Recent memory benchmarks evaluate whether agents can retrieve scattered facts or detect if a fact has changed whereas RECON evaluates what happens after the change, whether agents can trace which downstream conclusions are affected, which survive through independent support, and how alternative timelines would have unfolded. Our evaluation reveals substantial limitations across current architectures: even the strongest non-Oracle system reaches only 22.4% Accuracy, with retrieval and reasoning each surfacing as challenges.
展开完整摘要收起摘要↓
Large language models and LLM-based agents are widely used as personal chat assistants, enterprise copilots, and autonomous workflow agents. In all these applications, memory (the ability to retain, access, and reason over information accumulated over long contexts and multiple interactions) plays a crucial role in determining the reliability of any agent. We introduce RECON (Reasoning over Extended Contexts with Obfuscated Narratives), a benchmark for evaluating compositional reasoning over long contexts. RECON spans 24 case files across three domains (criminal, medical, and financial), each ranging from 50k to 100k tokens, and tests agents on six memory intensive tasks: reconstructing multi-hop evidence chains, propagating cascading invalidations, resolving source conflicts, counterfactual reasoning, satisfying temporal constraints, and temporal fact retrieval. Recent memory benchmarks evaluate whether agents can retrieve scattered facts or detect if a fact has changed whereas RECON evaluates what happens after the change, whether agents can trace which downstream conclusions are affected, which survive through independent support, and how alternative timelines would have unfolded. Our evaluation reveals substantial limitations across current architectures: even the strongest non-Oracle system reaches only 22.4% Accuracy, with retrieval and reasoning each surfacing as challenges.