JudgeFlow: Agentic Workflow Optimization via Block Judge

Optimizing LLM-based agentic workflows is challenging for scaling AI capabilities. Current methods rely on coarse, end-to-end evaluation signals and lack fine-grained signals on where to refine, often resulting in inefficient or low-impact modifications. To address these limitations, we propose {\our{}}, an Evaluation-Judge-Optimization-Update pipeline. We incorporate reusable, configurable logic blocks into agentic workflows to capture fundamental forms of logic. On top of this abstraction, we design a dedicated Judge module that inspects execution traces -- particularly failed runs -- and assigns rank-based responsibility scores to problematic blocks. These fine-grained diagnostic signals are then leveraged by an LLM-based optimizer, which focuses modifications on the most problematic block in the workflow. Our approach improves sample efficiency, enhances interpretability through block-level diagnostics, and provides a scalable foundation for automating increasingly complex agentic workflows. We evaluate {\our{}} on mathematical reasoning and code generation benchmarks, where {\our{}} achieves superior performance and efficiency compared to existing methods. The source code is publicly available at https://github.com/ma-zihan/JudgeFlow.

Priority-Aware Multi-Robot Coverage Path Planning

Multi-robot systems are widely used for coverage tasks that require efficient coordination across large environments. In Multi-Robot Coverage Path Planning (MCPP), the objective is typically to minimize the makespan by generating non-overlapping paths for full-area coverage. However, most existing methods assume uniform importance across regions, limiting their effectiveness in scenarios where some zones require faster attention. We introduce the Priority-Aware MCPP (PA-MCPP) problem, where a subset of the environment is designated as prioritized zones with associated weights. The goal is to minimize, in lexicographic order, the total priority-weighted latency of zone coverage and the overall makespan. To address this, we propose a scalable two-phase framework combining (1) greedy zone assignment with local search, spanning-tree-based path planning, and (2) Steiner-tree-guided residual coverage. Experiments across diverse scenarios demonstrate that our method significantly reduces priority-weighted latency compared to standard MCPP baselines, while maintaining competitive makespan. Sensitivity analyses further show that the method scales well with the number of robots and that zone coverage behavior can be effectively controlled by adjusting priority weights.

Adaptive Replay Buffer for Offline-to-Online Reinforcement Learning

Offline-to-Online Reinforcement Learning (O2O RL) faces a critical dilemma in balancing the use of a fixed offline dataset with newly collected online experiences. Standard methods, often relying on a fixed data-mixing ratio, struggle to manage the trade-off between early learning stability and asymptotic performance. To overcome this, we introduce the Adaptive Replay Buffer (ARB), a novel approach that dynamically prioritizes data sampling based on a lightweight metric we call 'on-policyness'. Unlike prior methods that rely on complex learning procedures or fixed ratios, ARB is designed to be learning-free and simple to implement, seamlessly integrating into existing O2O RL algorithms. It assesses how closely collected trajectories align with the current policy's behavior and assigns a proportional sampling weight to each transition within that trajectory. This strategy effectively leverages offline data for initial stability while progressively focusing learning on the most relevant, high-rewarding online experiences. Our extensive experiments on D4RL benchmarks demonstrate that ARB consistently mitigates early performance degradation and significantly improves the final performance of various O2O RL algorithms, highlighting the importance of an adaptive, behavior-aware replay buffer design.

VRPAgent: LLM-Driven Discovery of Heuristic Operators for Vehicle Routing Problems

Designing high-performing heuristics for vehicle routing problems (VRPs) is a complex task that requires both intuition and deep domain knowledge. Large language model (LLM)-based code generation has recently shown promise across many domains, but it still falls short of producing heuristics that rival those crafted by human experts. In this paper, we propose VRPAgent, a framework that integrates LLM-generated components into a metaheuristic and refines them through a novel genetic search. By using the LLM to generate problem-specific operators, embedded within a generic metaheuristic framework, VRPAgent keeps tasks manageable, guarantees correctness, and still enables the discovery of novel and powerful strategies. Across multiple problems, including the capacitated VRP, the VRP with time windows, and the prize-collecting VRP, our method discovers heuristic operators that outperform handcrafted methods and recent learning-based approaches while requiring only a single CPU core. To our knowledge, \VRPAgent is the first LLM-based paradigm to advance the state-of-the-art in VRPs, highlighting a promising future for automated heuristics discovery.

Diffusion Alignment as Variational Expectation-Maximization

Diffusion alignment aims to optimize diffusion models for the downstream objective. While existing methods based on reinforcement learning or direct backpropagation achieve considerable success in maximizing rewards, they often suffer from reward over-optimization and mode collapse. We introduce Diffusion Alignment as Variational Expectation-Maximization (DAV), a framework that formulates diffusion alignment as an iterative process alternating between two complementary phases: the E-step and the M-step. In the E-step, we employ test-time search to generate diverse and reward-aligned samples. In the M-step, we refine the diffusion model using samples discovered by the E-step. We demonstrate that DAV can optimize reward while preserving diversity for both continuous and discrete tasks: text-to-image synthesis and DNA sequence design.

Active Attacks: Red-teaming LLMs via Adaptive Environments

We address the challenge of generating diverse attack prompts for large language models (LLMs) that elicit harmful behaviors (e.g., insults, sexual content) and are used for safety fine-tuning. Rather than relying on manual prompt engineering, attacker LLMs can be trained with reinforcement learning (RL) to automatically generate such prompts using only a toxicity classifier as a reward. However, capturing a wide range of harmful behaviors is a significant challenge that requires explicit diversity objectives. Existing diversity-seeking RL methods often collapse to limited modes: once high-reward prompts are found, exploration of new regions is discouraged. Inspired by the active learning paradigm that encourages adaptive exploration, we introduce \textit{Active Attacks}, a novel RL-based red-teaming algorithm that adapts its attacks as the victim evolves. By periodically safety fine-tuning the victim LLM with collected attack prompts, rewards in exploited regions diminish, which forces the attacker to seek unexplored vulnerabilities. This process naturally induces an easy-to-hard exploration curriculum, where the attacker progresses beyond easy modes toward increasingly difficult ones. As a result, Active Attacks uncovers a wide range of local attack modes step by step, and their combination achieves wide coverage of the multi-mode distribution. Active Attacks, a simple plug-and-play module that seamlessly integrates into existing RL objectives, unexpectedly outperformed prior RL-based methods -- including GFlowNets, PPO, and REINFORCE -- by improving cross-attack success rates against GFlowNets, the previous state-of-the-art, from 0.07% to 31.28% (a relative gain greater than $400\ \times$) with only a 6% increase in computation. Our code is publicly available \href{https://github.com/dbsxodud-11/active_attacks}{here}.

TrajEvo: Trajectory Prediction Heuristics Design via LLM-driven Evolution

Trajectory prediction is a critical task in modeling human behavior, especially in safety-critical domains such as social robotics and autonomous vehicle navigation. Traditional heuristics based on handcrafted rules often lack accuracy and generalizability. Although deep learning approaches offer improved performance, they typically suffer from high computational cost, limited explainability, and, importantly, poor generalization to out-of-distribution (OOD) scenarios. In this paper, we introduce TrajEvo, a framework that leverages Large Language Models (LLMs) to automatically design trajectory prediction heuristics. TrajEvo employs an evolutionary algorithm to generate and refine prediction heuristics from past trajectory data. We propose two key innovations: Cross-Generation Elite Sampling to encourage population diversity, and a Statistics Feedback Loop that enables the LLM to analyze and improve alternative predictions. Our evaluations demonstrate that TrajEvo outperforms existing heuristic methods across multiple real-world datasets, and notably surpasses both heuristic and deep learning methods in generalizing to an unseen OOD real-world dataset. TrajEvo marks a promising step toward the automated design of fast, explainable, and generalizable trajectory prediction heuristics. We release our source code to facilitate future research at https://github.com/ai4co/trajevo.

USPR: Learning a Unified Solver for Profiled Routing

The Profiled Vehicle Routing Problem (PVRP) extends the classical VRP by incorporating vehicle-client-specific preferences and constraints, reflecting real-world requirements such as zone restrictions and service-level preferences. While recent reinforcement learning (RL) solvers have shown promise, they require retraining for each new profile distribution, suffer from poor representation ability, and struggle to generalize to out-of-distribution instances. In this paper, we address these limitations by introducing USPR (Unified Solver for Profiled Routing), a novel framework that natively handles arbitrary profile types. USPR introduces three key innovations: (i) Profile Embeddings (PE) to encode any combination of profile types; (ii) Multi-Head Profiled Attention (MHPA), an attention mechanism that models rich interactions between vehicles and clients; (iii) Profile-aware Score Reshaping (PSR), which dynamically adjusts decoder logits using profile scores to improve generalization. Empirical results on diverse PVRP benchmarks demonstrate that USPR achieves state-of-the-art results among learning-based methods while offering significant gains in flexibility and computational efficiency. We make our source code publicly available to foster future research at https://github.com/ai4co/uspr.

TrajEvo: Designing Trajectory Prediction Heuristics via LLM-driven Evolution

Trajectory prediction is a crucial task in modeling human behavior, especially in fields as social robotics and autonomous vehicle navigation. Traditional heuristics based on handcrafted rules often lack accuracy, while recently proposed deep learning approaches suffer from computational cost, lack of explainability, and generalization issues that limit their practical adoption. In this paper, we introduce TrajEvo, a framework that leverages Large Language Models (LLMs) to automatically design trajectory prediction heuristics. TrajEvo employs an evolutionary algorithm to generate and refine prediction heuristics from past trajectory data. We introduce a Cross-Generation Elite Sampling to promote population diversity and a Statistics Feedback Loop allowing the LLM to analyze alternative predictions. Our evaluations show TrajEvo outperforms previous heuristic methods on the ETH-UCY datasets, and remarkably outperforms both heuristics and deep learning methods when generalizing to the unseen SDD dataset. TrajEvo represents a first step toward automated design of fast, explainable, and generalizable trajectory prediction heuristics. We make our source code publicly available to foster future research at https://github.com/ai4co/trajevo.

The Iterative Chainlet Partitioning Algorithm for the Traveling Salesman Problem with Drone and Neural Acceleration

This study introduces the Iterative Chainlet Partitioning (ICP) algorithm and its neural acceleration for solving the Traveling Salesman Problem with Drone (TSP-D). The proposed ICP algorithm decomposes a TSP-D solution into smaller segments called chainlets, each optimized individually by a dynamic programming subroutine. The chainlet with the highest improvement is updated and the procedure is repeated until no further improvement is possible. The number of subroutine calls is bounded linearly in problem size for the first iteration and remains constant in subsequent iterations, ensuring algorithmic scalability. Empirical results show that ICP outperforms existing algorithms in both solution quality and computational time. Tested over 1,059 benchmark instances, ICP yields an average improvement of 2.75% in solution quality over the previous state-of-the-art algorithm while reducing computational time by 79.8%. The procedure is deterministic, ensuring reliability without requiring multiple runs. The subroutine is the computational bottleneck in the already efficient ICP algorithm. To reduce the necessity of subroutine calls, we integrate a graph neural network (GNN) to predict incremental improvements. We demonstrate that the resulting Neuro ICP (NICP) achieves substantial acceleration while maintaining solution quality. Compared to ICP, NICP reduces the total computational time by 49.7%, while the objective function value increase is limited to 0.12%. The framework's adaptability to various operational constraints makes it a valuable foundation for developing efficient algorithms for truck-drone synchronized routing problems.