Abstract
We propose an epoch-based reinforcement learning algorithm for infinite-horizon average-cost Markov decision processes (MDPs) that leverages a partial order over a policy class. In this structure, $\pi' \leq \pi$ if data collected under $\pi$ can be used to estimate the performance of $\pi'$, enabling counterfactual inference without additional environment interaction. Leveraging this partial order, we show that our algorithm achieves a regret bound of $O(\sqrt{w \log(|\Theta|) T})$, where $w$ is the width of the partial order. Notably, the bound is independent of the state and action space sizes. We illustrate the applicability of these partial orders in many domains in operations research, including inventory control and queuing systems. For each, we apply our framework to that problem, yielding new theoretical guarantees and strong empirical results without imposing extra assumptions such as convexity in the inventory model or specialized arrival-rate structure in the queuing model.

Abstract
We present a simple yet theoretically motivated improvement to Supervised Fine-Tuning (SFT) for the Large Language Model (LLM), addressing its limited generalization compared to reinforcement learning (RL). Through mathematical analysis, we reveal that standard SFT gradients implicitly encode a problematic reward structure that may severely restrict the generalization capabilities of model. To rectify this, we propose Dynamic Fine-Tuning (DFT), stabilizing gradient updates for each token by dynamically rescaling the objective function with the probability of this token. Remarkably, this single-line code change significantly outperforms standard SFT across multiple challenging benchmarks and base models, demonstrating greatly improved generalization. Additionally, our approach shows competitive results in offline RL settings, offering an effective yet simpler alternative. This work bridges theoretical insight and practical solutions, substantially advancing SFT performance. The code will be available at https://github.com/yongliang-wu/DFT.

Abstract
The term 'agent' in artificial intelligence has long carried multiple interpretations across different subfields. Recent developments in AI capabilities, particularly in large language model systems, have amplified this ambiguity, creating significant challenges in research communication, system evaluation and reproducibility, and policy development. This paper argues that the term 'agent' requires redefinition. Drawing from historical analysis and contemporary usage patterns, we propose a framework that defines clear minimum requirements for a system to be considered an agent while characterizing systems along a multidimensional spectrum of environmental interaction, learning and adaptation, autonomy, goal complexity, and temporal coherence. This approach provides precise vocabulary for system description while preserving the term's historically multifaceted nature. After examining potential counterarguments and implementation challenges, we provide specific recommendations for moving forward as a field, including suggestions for terminology standardization and framework adoption. The proposed approach offers practical tools for improving research clarity and reproducibility while supporting more effective policy development.

Abstract
Translating natural language problem descriptions into formal constraint models remains a fundamental challenge in constraint programming, requiring deep expertise in both the problem domain and modeling frameworks. Previous approaches to automating this translation have employed fixed workflows with predetermined modeling steps, failing on a significant number of benchmark problems. We present a new approach using a pure agentic strategy without any fixed pipeline. We developed a general-purpose Python coding agent based on the ReAct (Reason and Act) principle, utilizing a persistent IPython kernel for stateful code execution and iterative development. Rather than embedding constraint programming logic into the agent architecture, domain-specific expertise is injected solely through a carefully crafted project prompt. The agent combines this prompt-encoded knowledge with access to file operations and code execution tools, enabling it to test hypotheses, debug failures, and verify solutions dynamically. Implemented in just a few hundred lines of code, this architecture successfully solves all 101 problems of the CP-Bench constraint programming benchmark set. The results suggest that constraint modeling tasks require the combination of general coding tools and domain expertise encoded in prompts, rather than specialized agent architectures or predefined workflows.

Abstract
Issue localization, the process of identifying code locations that need modification to resolve software issues, is a critical yet challenging task in software development. The semantic gap between natural language issue descriptions and faulty code requires complex multi-hop reasoning through code dependencies. Existing LLM-based agents attempt to address this by integrating repository retrieval tools. However, this transforms issue localization into a demanding task we call Repo Deep Search, which requires the LLM to effectively utilize various repository retrieval tools throughout a multi-step reasoning and navigation process. To tackle this challenge, we present ToolTrain, a two-stage tool-integrated training framework combining rejection-sampled supervised fine-tuning and tool-integrated reinforcement learning to enhance LLMs' ability to use retrieval tools for issue localization. Experimental results show that ToolTrain-trained models achieve state-of-the-art performance, with our 32B model even surpassing Claude-3.7 on function-level localization. The results also show that improved localization performance translates to better end-to-end issue resolution performance. This further demonstrates that training for issue localization is a viable and effective strategy for improving automated software development.

Abstract
Graph Neural Networks (GNNs) have emerged as a powerful, data-driven approach for Quantum Error Correction (QEC) decoding, capable of learning complex noise characteristics directly from syndrome data. However, the robustness of these decoders against subtle, adversarial perturbations remains a critical open question. This work introduces a novel framework to systematically probe the vulnerabilities of a GNN decoder using a reinforcement learning (RL) agent. The RL agent is trained as an adversary with the goal of finding minimal syndrome modifications that cause the decoder to misclassify. We apply this framework to a Graph Attention Network (GAT) decoder trained on experimental surface code data from Google Quantum AI. Our results show that the RL agent can successfully identify specific, critical vulnerabilities, achieving a high attack success rate with a minimal number of bit flips. Furthermore, we demonstrate that the decoder's robustness can be significantly enhanced through adversarial training, where the model is retrained on the adversarial examples generated by the RL agent. This iterative process of automated vulnerability discovery and targeted retraining presents a promising methodology for developing more reliable and robust neural network decoders for fault-tolerant quantum computing.