Large Language Models (LLMs) have shown remarkable performance in various natural language tasks, but they often struggle with planning problems that require structured reasoning. To address this limitation, the conversion of planning problems into the Planning Domain Definition Language (PDDL) has been proposed as a potential solution, enabling the use of automated planners. However, generating accurate PDDL files typically demands human inputs or correction, which can be time-consuming and costly. In this paper, we propose a novel approach that leverages LLMs and environment feedback to automatically generate PDDL domain and problem description files without the need for human intervention. Our method introduces an iterative refinement process that generates multiple problem PDDL candidates and progressively refines the domain PDDL based on feedback obtained from interacting with the environment. To guide the refinement process, we develop an Exploration Walk (EW) metric, which provides rich feedback signals for LLMs to update the PDDL file. We evaluate our approach on $10$ PDDL environments. We achieve an average task solve rate of 66\% compared to a 29\% solve rate by GPT-4's intrinsic planning with chain-of-thought prompting. Our work enables the automated modeling of planning environments using LLMs and environment feedback, eliminating the need for human intervention in the PDDL translation process and paving the way for more reliable LLM agents in challenging problems.

LLM-powered coding agents, which operate in iterative loops (turns) to solve software engineering tasks, are becoming increasingly powerful. However, their practical deployment is hindered by significant and unpredictable costs. This challenge arises from a combination of factors: quadratically growing token counts with each turn, the high price of models, the large number of turns required for real-world tasks, and the tendency of agents to take inefficient or unnecessary actions. While existing research focuses on optimizing individual turns, the strategic control of the total number of turns remains an underexplored area for managing agent performance and cost. To address this gap, we conduct a comprehensive empirical study on SWE-bench using three state-of-the-art models and evaluate the impact of three distinct turn-control strategies: an unrestricted baseline, a fixed-turn limit with reminders, and a novel dynamic-turn strategy that grants extensions on-demand. Our findings first reveal a fundamental trade-off in the unrestricted setting, where no single model excels across performance, cost, and turn efficiency. We then show that a fixed-turn limit, specifically at the 75th percentile of the baseline, serves as a "sweet spot", substantially reducing costs (by 24%-68%) with minimal impact on solve rates. Most significantly, the dynamic-turn strategy consistently outperforms fixed-limit approaches, achieving comparable or better solve rates while further reducing costs by an additional 12%-24% by intelligently allocating resources only to tasks that need them. This work provides the first systematic analysis of turn-control strategies, offering simple yet effective guidelines for developers to balance cost and efficacy. We demonstrate that dynamic resource allocation is a superior, easy-to-implement approach for deploying powerful yet economically viable coding agents.

Abstract-- Multipatterning is an essential decomposition strategy in electronic design automation (EDA) that overcomes lithographic limitations when printing dense circuit layouts. Although heuristic-based backtracking and SA T solvers can address these challenges, they often struggle to simultaneously handle both complex constraints and secondary objectives. In this study, we present a hybrid workflow that casts multipatterning as a variant of a constrained graph coloring problem with the primary objective of minimizing feature violations and a secondary objective of balancing the number of features on each mask. Our pipeline integrates two main components: (1) A GNN-based agent, trained in an unsupervised manner to generate initial color predictions, which are refined by (2) refinement strategies (a GNN-based heuristic and simulated annealing) that together enhance solution quality and balance. Experimental evaluation in both proprietary data sets and publicly available open source layouts demonstrate complete conflict-free decomposition and consistent color balancing. The proposed framework provides a reproducible, data-efficient and deployable baseline for scalable layout decomposition in EDA workflows. As semiconductor technology progresses, the demand for higher circuit densities continues to surpass the limits of conventional lithographic techniques. The ongoing reduction in feature size introduces increasingly complex manufacturing constraints, making it difficult to accurately print intricate patterns on a single mask without defects. To address these challenges, modern electronic design automation (EDA) tools and fabrication processes rely on multipatterning, which is a layout decomposition technique that ensures manufacturability while preserving design integrity. In modern integrated circuit (IC) design, Design Rule Checking (DRC) is a critical step that ensures that the physical layout complies with a set of rules derived from the manufacturing constraints. These rules include the requirements on spacing, width, enclosure, and other geometric and connectivity constraints.

Large language models (LLMs) have great potential to enhance productivity in many disciplines, such as software engineering. However, it is unclear to what extent they can assist in the design process of electronic circuits. This paper focuses on the application of LLMs to switched-mode power supply (SMPS) design for printed circuit boards (PCBs). We present multiple LLM-based workflows that combine reasoning, retrieval-augmented generation (RAG), and a custom toolkit that enables the LLM to interact with SPICE simulations to estimate the impact of circuit modifications. Two benchmark experiments are presented to analyze the performance of LLM-based assistants for different design tasks, including parameter tuning, topology adaption and optimization of SMPS circuits. Experiment results show that SPICE simulation feedback and current LLM advancements, such as reasoning, significantly increase the solve rate on 269 manually created benchmark tasks from 15% to 91%. Furthermore, our analysis reveals that most parameter tuning design tasks can be solved, while limits remain for certain topology adaption tasks. Our experiments offer insights for improving current concepts, for example by adapting text-based circuit representations

Large Language Model (LLM)-based agents solve complex tasks through iterative reasoning, exploration, and tool-use, a process that can result in long, expensive context histories. While state-of-the-art Software Engineering (SE) agents like OpenHands or Cursor use LLM-based summarization to tackle this issue, it is unclear whether the increased complexity offers tangible performance benefits compared to simply omitting older observations. We present a systematic comparison of these approaches within SWE-agent on SWE-bench Verified across five diverse model configurations. Moreover, we show initial evidence of our findings generalizing to the OpenHands agent scaffold. We find that a simple environment observation masking strategy halves cost relative to the raw agent while matching, and sometimes slightly exceeding, the solve rate of LLM summarization. Additionally, we introduce a novel hybrid approach that further reduces costs by 7% and 11% compared to just observation masking or LLM summarization, respectively. Our findings raise concerns regarding the trend towards pure LLM summarization and provide initial evidence of untapped cost reductions by pushing the efficiency-effectiveness frontier. We release code and data for reproducibility.

Large language models (LLMs) can transform education, but their optimization for direct question-answering often undermines effective pedagogy which requires strategically withholding answers. To mitigate this, we propose an online reinforcement learning (RL)-based alignment framework that can quickly adapt LLMs into effective tutors using simulated student-tutor interactions by emphasizing pedagogical quality and guided problem-solving over simply giving away answers. We use our method to train a 7B parameter tutor model without human annotations which reaches similar performance to larger proprietary models like LearnLM. We introduce a controllable reward weighting to balance pedagogical support and student solving accuracy, allowing us to trace the Pareto frontier between these two objectives. Our models better preserve reasoning capabilities than single-turn SFT baselines and can optionally enhance interpretability through thinking tags that expose the model's instructional planning.

Among the hardest tasks for humans are those found in competitive programming where problems require sophisticated algorithmic thinking, puzzle solving, and the creation of effective code. As a domain to assess language models (LMs), it has not received enough attention, though. This study presents the ICPC benchmark, which consists of 254 international collegiate programming contest (ICPC) tasks. Each problem includes official analysis, reference code, and sample, high-quality unit, and hidden tests. We are able to develop and evaluate a variety of LM inference techniques for competitive programming with these resources. With zero-shot chain-of-thought prompting, we find that o1 only achieves a 19.1\% pass@1 solve rate. With our best inference technique, which combines multi-turn self-judge with reflection and retrieval over episodic information, raises this to 42.2\%. Furthermore, we conduct a new human-in-the-loop investigation to gain a deeper understanding of the remaining difficulties. Surprisingly, we discover that o1 can solve 17 out of 18 problems that were previously unsolvable by any model or technique with just a few specific instructions. A footstep toward LMs with grounded, imaginative, and algorithmic thinking is provided by our quantitative findings and qualitative research. We open-source our code and data at https://github.com/kraritt/zolve.

Retrosynthesis planning remains a central challenge in molecular discovery due to the vast and complex chemical reaction space. While traditional template-based methods offer tractability, they suffer from poor scalability and limited generalization, and template-free generative approaches risk generating invalid reactions. In this work, we propose TempRe, a generative framework that reformulates template-based approaches as sequence generation, enabling scalable, flexible, and chemically plausible retrosynthesis. We evaluated TempRe across single-step and multi-step retrosynthesis tasks, demonstrating its superiority over both template classification and SMILES-based generation methods. On the PaRoutes multi-step benchmark, TempRe achieves strong top-k route accuracy. Furthermore, we extend TempRe to direct multi-step synthesis route generation, providing a lightweight and efficient alternative to conventional single-step and search-based approaches. These results highlight the potential of template generative modeling as a powerful paradigm in computer-aided synthesis planning.

Aim: With the advent of LLMs, sophisticated agentic program repair has become viable at large organizations with large codebases. In this work, we develop an Engineering Agent that fixes the source code based on test failures at scale across diverse software offerings internally. Method: Using Llama as the base, we employ the ReAct harness to develop an agent. We start with a test failure that was triaged by a rule-based test failure bot. We then set up an agentic harness and allow the agent to reason and run a set of 15 actions from reading a file to generating a patch. We provide feedback to the agent through static analysis and test failures so it can refine its solution. We leverage an LLM-as-a-Judge to ensure that the patch conforms to the standards followed by a human review to land fixes. Benchmark Findings: We curated offline benchmarks for our patch generator, the Engineering Agent loop, and the LLM-as-a-Judge. In offline evaluations we found that a specialized 70B model is highly competitive with the much larger but vanilla Llama-405B. In an ablation study, we found that the ReAct harness (neural model) benefited from the symbolic information from static analysis tools and test execution traces. A model that strikes a balance between the solve rate and error rate vs the cost and latency has a benchmark solve rate of 42.3% using an average 11.8 feedback iterations. Production Findings: In a three month period, 80% of the generated fixes were reviewed, of which 31.5% were landed (25.5% of the total number of generated fixes). Feedback from Engineers: We used open coding to extract qualitative themes from engineers' feedback. We saw positive feedback in the form of quick approvals, gratitude, and surprise. We also found mixed feedback when the Engineering Agent's solution was partially correct and it served as a good starting point.

Large reasoning models (LRMs) achieve higher performance on challenging reasoning tasks by generating more tokens at inference time, but this verbosity often wastes computation on easy problems. Existing solutions, including supervised finetuning on shorter traces, user-controlled budgets, or RL with uniform penalties, either require data curation, manual configuration, or treat all problems alike regardless of difficulty. We introduce Adaptive Length Penalty (ALP), a reinforcement learning objective tailoring generation length to per-prompt solve rate. During training, ALP monitors each prompt's online solve rate through multiple rollouts and adds a differentiable penalty whose magnitude scales inversely with that rate, so confident (easy) prompts incur a high cost for extra tokens while hard prompts remain unhindered. Posttraining DeepScaleR-1.5B with ALP cuts average token usage by 50\% without significantly dropping performance. Relative to fixed-budget and uniform penalty baselines, ALP redistributes its reduced budget more intelligently by cutting compute on easy prompts and reallocating saved tokens to difficult ones, delivering higher accuracy on the hardest problems with higher cost.