arXiv subjects

Nan Jiang

Publications and source records attributed to Nan Jiang.

At least 109 records · Page 6Linked to original sources

arXiv2025

Show Me Why It's Correct: Saving 1/3 of Debugging Time in Program Repair with Interactive Runtime Comparison

Automated Program Repair (APR) holds the promise of alleviating the burden of debugging and fixing software bugs. Despite this, developers still need to manually inspect each patch to confirm its correctness, which is tedious and time-consuming. This challenge is exacerbated in the presence of plausible patches, which accidentally pass test cases but may not correctly fix the bug. To address this challenge, we propose an interactive approach called iFix to facilitate patch understanding and comparison based on their runtime difference. iFix performs static analysis to identify runtime variables related to the buggy statement and captures their runtime values during execution for each patch. These values are then aligned across different patch candidates, allowing users to compare and contrast their runtime behavior. To evaluate iFix, we conducted a within-subjects user study with 28 participants. Compared with manual inspection and a state-of-the-art interactive patch filtering technique, iFix reduced participants' task completion time by 36% and 33% while also improving their confidence by 50% and 20%, respectively. Besides, quantitative experiments demonstrate that iFix improves the ranking of correct patches by at least 39% compared with other patch ranking methods and is generalizable to different APR tools.

cs.SE↗

arXiv2025

Self-rewarding correction for mathematical reasoning

We study self-rewarding reasoning large language models (LLMs), which can simultaneously generate step-by-step reasoning and evaluate the correctness of their outputs during the inference time-without external feedback. This integrated approach allows a single model to independently guide its reasoning process, offering computational advantages for model deployment. We particularly focus on the representative task of self-correction, where models autonomously detect errors in their responses, revise outputs, and decide when to terminate iterative refinement loops. To enable this, we propose a two-staged algorithmic framework for constructing self-rewarding reasoning models using only self-generated data. In the first stage, we employ sequential rejection sampling to synthesize long chain-of-thought trajectories that incorporate both self-rewarding and self-correction mechanisms. Fine-tuning models on these curated data allows them to learn the patterns of self-rewarding and self-correction. In the second stage, we further enhance the models' ability to assess response accuracy and refine outputs through reinforcement learning with rule-based signals. Experiments with Llama-3 and Qwen-2.5 demonstrate that our approach surpasses intrinsic self-correction capabilities and achieves performance comparable to systems that rely on external reward models.

cs.AI↗

arXiv2025

Improving LLM General Preference Alignment via Optimistic Online Mirror Descent

Reinforcement learning from human feedback (RLHF) has demonstrated remarkable effectiveness in aligning large language models (LLMs) with human preferences. Many existing alignment approaches rely on the Bradley-Terry (BT) model assumption, which assumes the existence of a ground-truth reward for each prompt-response pair. However, this assumption can be overly restrictive when modeling complex human preferences. In this paper, we drop the BT model assumption and study LLM alignment under general preferences, formulated as a two-player game. Drawing on theoretical insights from learning in games, we integrate optimistic online mirror descent into our alignment framework to approximate the Nash policy. Theoretically, we demonstrate that our approach achieves an $O(T^{-1})$ bound on the duality gap, improving upon the previous $O(T^{-1/2})$ result. More importantly, we implement our method and show through experiments that it outperforms state-of-the-art RLHF algorithms across multiple representative benchmarks.

cs.LG↗

arXiv2025

GameArena: Evaluating LLM Reasoning through Live Computer Games

Evaluating the reasoning abilities of large language models (LLMs) is challenging. Existing benchmarks often depend on static datasets, which are vulnerable to data contamination and may get saturated over time, or on binary live human feedback that conflates reasoning with other abilities. As the most prominent dynamic benchmark, Chatbot Arena evaluates open-ended questions in real-world settings, but lacks the granularity in assessing specific reasoning capabilities. We introduce GameArena, a dynamic benchmark designed to evaluate LLM reasoning capabilities through interactive gameplay with humans. GameArena consists of three games designed to test specific reasoning capabilities (e.g., deductive and inductive reasoning), while keeping participants entertained and engaged. We analyze the gaming data retrospectively to uncover the underlying reasoning processes of LLMs and measure their fine-grained reasoning capabilities. We collect over 2000 game sessions and provide detailed assessments of various reasoning capabilities for five state-of-the-art LLMs. Our user study with 100 participants suggests that GameArena improves user engagement compared to Chatbot Arena. For the first time, GameArena enables the collection of step-by-step LLM reasoning data in the wild.

cs.AI↗

arXiv2025

SELP: Generating Safe and Efficient Task Plans for Robot Agents with Large Language Models

Despite significant advancements in large language models (LLMs) that enhance robot agents' understanding and execution of natural language (NL) commands, ensuring the agents adhere to user-specified constraints remains challenging, particularly for complex commands and long-horizon tasks. To address this challenge, we present three key insights, equivalence voting, constrained decoding, and domain-specific fine-tuning, which significantly enhance LLM planners' capability in handling complex tasks. Equivalence voting ensures consistency by generating and sampling multiple Linear Temporal Logic (LTL) formulas from NL commands, grouping equivalent LTL formulas, and selecting the majority group of formulas as the final LTL formula. Constrained decoding then uses the generated LTL formula to enforce the autoregressive inference of plans, ensuring the generated plans conform to the LTL. Domain-specific fine-tuning customizes LLMs to produce safe and efficient plans within specific task domains. Our approach, Safe Efficient LLM Planner (SELP), combines these insights to create LLM planners to generate plans adhering to user commands with high confidence. We demonstrate the effectiveness and generalizability of SELP across different robot agents and tasks, including drone navigation and robot manipulation. For drone navigation tasks, SELP outperforms state-of-the-art planners by 10.8% in safety rate (i.e., finishing tasks conforming to NL commands) and by 19.8% in plan efficiency. For robot manipulation tasks, SELP achieves 20.4% improvement in safety rate. Our datasets for evaluating NL-to-LTL and robot task planning will be released in github.com/lt-asset/selp.

cs.RO↗

arXiv2025

LeDex: Training LLMs to Better Self-Debug and Explain Code

In the domain of code generation, self-debugging is crucial. It allows LLMs to refine their generated code based on execution feedback. This is particularly important because generating correct solutions in one attempt proves challenging for complex tasks. Prior works on self-debugging mostly focus on prompting methods by providing LLMs with few-shot examples, which work poorly on small open-sourced LLMs. In this work, we propose LeDex, a training framework that significantly improves the self-debugging capability of LLMs. Intuitively, we observe that a chain of explanations on the wrong code followed by code refinement helps LLMs better analyze the wrong code and do refinement. We thus propose an automated pipeline to collect a high-quality dataset for code explanation and refinement by generating a number of explanations and refinement trajectories from the LLM itself or a larger teacher model and filtering via execution verification. We perform supervised fine-tuning (SFT) and further reinforcement learning (RL) on both success and failure trajectories with a novel reward design considering code explanation and refinement quality. SFT improves the pass@1 by up to 15.92% and pass@10 by 9.30% over four benchmarks. RL training brings additional up to 3.54% improvement on pass@1 and 2.55% improvement on pass@10. The trained LLMs show iterative refinement ability and can keep refining code continuously. Lastly, our human evaluation shows that the LLMs trained with our framework generate more useful code explanations and help developers better understand bugs in source code.

cs.CL↗

arXiv2025

LATTE: Improving Latex Recognition for Tables and Formulae with Iterative Refinement

Portable Document Format (PDF) files are dominantly used for storing and disseminating scientific research, legal documents, and tax information. LaTeX is a popular application for creating PDF documents. Despite its advantages, LaTeX is not WYSWYG -- what you see is what you get, i.e., the LaTeX source and rendered PDF images look drastically different, especially for formulae and tables. This gap makes it hard to modify or export LaTeX sources for formulae and tables from PDF images, and existing work is still limited. First, prior work generates LaTeX sources in a single iteration and struggles with complex LaTeX formulae. Second, existing work mainly recognizes and extracts LaTeX sources for formulae; and is incapable or ineffective for tables. This paper proposes LATTE, the first iterative refinement framework for LaTeX recognition. Specifically, we propose delta-view as feedback, which compares and pinpoints the differences between a pair of rendered images of the extracted LaTeX source and the expected correct image. Such delta-view feedback enables our fault localization model to localize the faulty parts of the incorrect recognition more accurately and enables our LaTeX refinement model to repair the incorrect extraction more accurately. LATTE improves the LaTeX source extraction accuracy of both LaTeX formulae and tables, outperforming existing techniques as well as GPT-4V by at least 7.03% of exact match, with a success refinement rate of 46.08% (formula) and 25.51% (table).

cs.CV↗

arXiv2025

How Effective are Large Language Models in Generating Software Specifications?

Software specifications are essential for many Software Engineering (SE) tasks such as bug detection and test generation. Many existing approaches are proposed to extract the specifications defined in natural language form (e.g., comments) into formal machine readable form (e.g., first order logic). However, existing approaches suffer from limited generalizability and require manual efforts. The recent emergence of Large Language Models (LLMs), which have been successfully applied to numerous SE tasks, offers a promising avenue for automating this process. In this paper, we conduct the first empirical study to evaluate the capabilities of LLMs for generating software specifications from software comments or documentation. We evaluate LLMs performance with Few Shot Learning (FSL) and compare the performance of 13 state of the art LLMs with traditional approaches on three public datasets. In addition, we conduct a comparative diagnosis of the failure cases from both LLMs and traditional methods, identifying their unique strengths and weaknesses. Our study offers valuable insights for future research to improve specification generation.

cs.SE↗

arXiv2025

Active Symbolic Discovery of Ordinary Differential Equations via Phase Portrait Sketching

The symbolic discovery of Ordinary Differential Equations (ODEs) from trajectory data plays a pivotal role in AI-driven scientific discovery. Existing symbolic methods predominantly rely on fixed, pre-collected training datasets, which often result in suboptimal performance, as demonstrated in our case study in Figure 1. Drawing inspiration from active learning, we investigate strategies to query informative trajectory data that can enhance the evaluation of predicted ODEs. However, the butterfly effect in dynamical systems reveals that small variations in initial conditions can lead to drastically different trajectories, necessitating the storage of vast quantities of trajectory data using conventional active learning. To address this, we introduce Active Symbolic Discovery of Ordinary Differential Equations via Phase Portrait Sketching (APPS). Instead of directly selecting individual initial conditions, our APPS first identifies an informative region within the phase space and then samples a batch of initial conditions from this region. Compared to traditional active learning methods, APPS mitigates the gap of maintaining a large amount of data. Extensive experiments demonstrate that APPS consistently discovers more accurate ODE expressions than baseline methods using passively collected datasets.

cs.LG↗

arXiv2025

3rd Workshop on Maritime Computer Vision (MaCVi) 2025: Challenge Results

The 3rd Workshop on Maritime Computer Vision (MaCVi) 2025 addresses maritime computer vision for Unmanned Surface Vehicles (USV) and underwater. This report offers a comprehensive overview of the findings from the challenges. We provide both statistical and qualitative analyses, evaluating trends from over 700 submissions. All datasets, evaluation code, and the leaderboard are available to the public at https://macvi.org/workshop/macvi25.

cs.CV↗

arXiv2025

Regularized MIP Model for Integrating Energy Storage Systems and its Application for Solving a Trilevel Interdiction Problem

Incorporating energy storage systems (ESS) into power systems has been studied in many recent works, where binary variables are often introduced to model the complementary nature of battery charging and discharging. A conventional approach for these ESS optimization problems is to relax binary variables and convert the problem into a linear program. However, such linear programming relaxation models can yield unrealistic fractional solutions, such as simultaneous charging and discharging. In this paper, we develop a regularized Mixed-Integer Programming (MIP) model for the ESS optimal power flow (OPF) problem. We prove that under mild conditions, the proposed regularized model admits a zero integrality gap with its linear programming relaxation; hence, it can be solved efficiently. By studying the properties of the regularized MIP model, we show that its optimal solution is also near-optimal to the original ESS OPF problem, thereby providing a valid and tight upper bound for the ESS OPF problem. The use of the regularized MIP model allows us to solve a trilevel min-max-min network contingency problem which is otherwise intractable to solve.

math.OC↗

arXiv2024

Class-RAG: Real-Time Content Moderation with Retrieval Augmented Generation

Robust content moderation classifiers are essential for the safety of Generative AI systems. In this task, differences between safe and unsafe inputs are often extremely subtle, making it difficult for classifiers (and indeed, even humans) to properly distinguish violating vs. benign samples without context or explanation. Scaling risk discovery and mitigation through continuous model fine-tuning is also slow, challenging and costly, preventing developers from being able to respond quickly and effectively to emergent harms. We propose a Classification approach employing Retrieval-Augmented Generation (Class-RAG). Class-RAG extends the capability of its base LLM through access to a retrieval library which can be dynamically updated to enable semantic hotfixing for immediate, flexible risk mitigation. Compared to model fine-tuning, Class-RAG demonstrates flexibility and transparency in decision-making, outperforms on classification and is more robust against adversarial attack, as evidenced by empirical studies. Our findings also suggest that Class-RAG performance scales with retrieval library size, indicating that increasing the library size is a viable and low-cost approach to improve content moderation.

cs.AI↗

arXiv2024

Interfacial Perpendicular Magnetic Anisotropy of Ultrathin Fe(001) Film Grown on CoO(001) Surface

Exploring novel systems with perpendicular magnetic anisotropy (PMA) is vital for advancing memory devices. In this study, we report an intriguing PMA system involving an ultrathin Fe layer on an antiferromagnetic (AFM) CoO(001) surface. The measured perpendicular anisotropy field is inversely proportional to the Fe thickness, indicating an interfacial origin of PMA. Temperature-dependent measurements reveal that the antiferromagnetism of CoO has a negligible effect on the PMA. By leveraging the magneto-optical Kerr effect and birefringence effect, we achieved concurrent visualization of ferromagnetic (FM) and AFM domains. A pronounced coupling effect between these domains was observed near the spin reorientation transition, contrasting sharply with areas of stronger PMA that exhibited weak coupling. This research not only establishes a new FM/AFM bilayer PMA system but also significantly advances the understanding of FM/AFM interfacial interactions.

cond-mat.mtrl-sci↗

arXiv2024

Symbol Preference Aware Generative Models for Recovering Variable Names from Stripped Binary

Decompilation aims to recover the source code form of a binary executable. It has many security applications, such as malware analysis, vulnerability detection, and code hardening. A prominent challenge in decompilation is to recover variable names. We propose a novel technique that leverages the strengths of generative models while mitigating model biases. We build a prototype, GenNm, from pre-trained generative models CodeGemma-2B, CodeLlama-7B, and CodeLlama-34B. We finetune GenNm on decompiled functions and teach models to leverage contextual information. GenNm includes names from callers and callees while querying a function, providing rich contextual information within the model's input token limitation. We mitigate model biases by aligning the output distribution of models with symbol preferences of developers. Our results show that GenNm improves the state-of-the-art name recovery precision by 5.6-11.4 percentage points on two commonly used datasets and improves the state-of-the-art by 32% (from 17.3% to 22.8%) in the most challenging setup where ground-truth variable names are not seen in the training dataset.

cs.SE↗

arXiv2024

Commit0: Library Generation from Scratch

With the goal of benchmarking generative systems beyond expert software development ability, we introduce Commit0, a benchmark that challenges AI agents to write libraries from scratch. Agents are provided with a specification document outlining the library's API as well as a suite of interactive unit tests, with the goal of producing an implementation of this API accordingly. The implementation is validated through running these unit tests. As a benchmark, Commit0 is designed to move beyond static one-shot code generation towards agents that must process long-form natural language specifications, adapt to multi-stage feedback, and generate code with complex dependencies. Commit0 also offers an interactive environment where models receive static analysis and execution feedback on the code they generate. Our experiments demonstrate that while current agents can pass some unit tests, none can yet fully reproduce full libraries. Results also show that interactive feedback is quite useful for models to generate code that passes more unit tests, validating the benchmarks that facilitate its use.

cs.SE↗

arXiv2024

Improved Analytic Modeling of Neutron Star Interiors

Studies of neutron stars are extremely timely given the recent detection of gravitational waves from a binary neutron star merger GW170817, and an International Space Station payload NICER currently in operation that aims to determine radii of neutron stars to a precision better than 5%. In many cases, neutron star solutions are constructed numerically due to the complexity of the field equations with realistic equations of state. However, in order to relate observables like the neutron star mass and radius to interior quantities like central density and pressure, it would be useful to provide an accurate, analytic modeling of a neutron star interior. One such solution for static and isolated neutron stars is the Tolman VII solution characterized only by two parameters (e.g. mass and radius), though its agreement with numerical solutions is not perfect. We here introduce an improved analytic model based on the Tolman VII solution by introducing an additional parameter to make the analytic density profile agree better with the numerically obtained one. This additional parameter can be fitted in terms of the stellar mass, radius and central density in an equation-of-state-insensitive way. In most cases, we find that the new model more accurately describes realistic profiles than the original Tolman VII solution by a factor of 2-5. Our results are first-step calculations towards constructing analytic interior solutions for more realistic neutron stars under rotation or tidal deformation.

gr-qc↗

arXiv2024

Online Iterative Reinforcement Learning from Human Feedback with General Preference Model

We investigate Reinforcement Learning from Human Feedback (RLHF) in the context of a general preference oracle. In particular, we do not assume the existence of a reward function and an oracle preference signal drawn from the Bradley-Terry model as most of the prior works do. We consider a standard mathematical formulation, the reverse-KL regularized minimax game between two LLMs for RLHF under general preference oracle. The learning objective of this formulation is to find a policy so that it is consistently preferred by the KL-regularized preference oracle over any competing LLMs. We show that this framework is strictly more general than the reward-based one, and propose sample-efficient algorithms for both the offline learning from a pre-collected preference dataset and online learning where we can query the preference oracle along the way of training. Empirical studies verify the effectiveness of the proposed framework.

cs.LG↗

arXiv2024

RLHF Workflow: From Reward Modeling to Online RLHF

We present the workflow of Online Iterative Reinforcement Learning from Human Feedback (RLHF) in this technical report, which is widely reported to outperform its offline counterpart by a large margin in the recent large language model (LLM) literature. However, existing open-source RLHF projects are still largely confined to the offline learning setting. In this technical report, we aim to fill in this gap and provide a detailed recipe that is easy to reproduce for online iterative RLHF. In particular, since online human feedback is usually infeasible for open-source communities with limited resources, we start by constructing preference models using a diverse set of open-source datasets and use the constructed proxy preference model to approximate human feedback. Then, we discuss the theoretical insights and algorithmic principles behind online iterative RLHF, followed by a detailed practical implementation. Our trained LLM achieves impressive performance on LLM chatbot benchmarks, including AlpacaEval-2, Arena-Hard, and MT-Bench, as well as other academic benchmarks such as HumanEval and TruthfulQA. We have shown that supervised fine-tuning (SFT) and iterative RLHF can obtain state-of-the-art performance with fully open-source datasets. Further, we have made our models, curated datasets, and comprehensive step-by-step code guidebooks publicly available. Please refer to https://github.com/RLHFlow/RLHF-Reward-Modeling and https://github.com/RLHFlow/Online-RLHF for more detailed information.

cs.LG↗