arXiv subjects

Xiaomo Liu

Publications and source records attributed to Xiaomo Liu.

At least 19 recordsLinked to original sources

arXiv2026

WorldMemArena: Evaluating Multimodal Agent Memory Through Action-World Interaction

Multimodal large language models are increasingly deployed as long-horizon agents, where memory must do more than recall: it must track an evolving world, revise what has gone stale, and surface the right evidence at decision time. Existing benchmarks measure recall over static dialogue, collapse memory into a single end-of-task accuracy, and reduce visual observations to captions, leaving us unable to localize failures to writing, maintenance, retrieval, or use. The rise of agent harnesses that author their own memory sharpens this gap, since we have no principled way to compare hand-designed pipelines with self-managing alternatives. To close these gaps, we formulate multimodal agent memory as an Action-World Interaction Loop with an observable four-stage lifecycle, and instantiate it in WorldMemArena: 400 multi-session multimodal tasks spanning Lifelong Evolution (evolving personal and task states) and Agentic Execution (memory from real observations, actions, and feedback), annotated with gold memory points, updates, distractors, and evidence chains for stage-level diagnosis. This enables the first head-to-head comparison of long-context, manually designed (RAG and external memory systems), and harness-based memory agents. Results show that: (1) better memory writing and storage do not guarantee better performance; (2) multimodal memory still struggles to fully use visual evidence; (3) systems are unstable across domains and degrade on realistic agentic trajectories; and (4) harness memory is more flexible but remains costly and less reliable.

cs.CV↗

arXiv2026

Deep FinResearch Bench: Evaluating AI's Ability to Conduct Professional Financial Investment Research

We introduce Deep FinResearch Bench, a practical and comprehensive evaluation framework for deep research (DR) agents in financial investment research. The benchmark assesses three dimensions of report quality: qualitative rigor, quantitative forecasting and valuation accuracy, and claim credibility and verifiability. Particularly, we define corresponding qualitative and quantitative evaluation metrics and implement an automated scoring procedure to enable scalable assessment. Applying the benchmark to financial reports from frontier DR agents and comparing them with reports authored by financial professionals, we find that AI-generated reports still fall short across these dimensions. These findings underscore the need for domain-specialized DR agents tailored to finance, and we hope the work establishes a foundation for standardized benchmarking of DR agents in financial research.

cs.AI↗

arXiv2026

Detecting Non-Membership in LLM Training Data via Rank Correlations

As large language models (LLMs) are trained on increasingly vast and opaque text corpora, determining which data contributed to training has become essential for copyright enforcement, compliance auditing, and user trust. While prior work focuses on detecting whether a dataset was used in training (membership inference), the complementary problem -- verifying that a dataset was not used -- has received little attention. We address this gap by introducing PRISM, a test that detects dataset-level non-membership using only grey-box access to model logits. Our key insight is that two models that have not seen a dataset exhibit higher rank correlation in their normalized token log probabilities than when one model has been trained on that data. Using this observation, we construct a correlation-based test that detects non-membership. Empirically, PRISM reliably rules out membership in training data across all datasets tested while avoiding false positives, thus offering a framework for verifying that specific datasets were excluded from LLM training.

cs.CL↗

arXiv2026

Distill and Align Decomposition for Enhanced Claim Verification

Complex claim verification requires decomposing sentences into verifiable subclaims, yet existing methods struggle to align decomposition quality with verification performance. We propose a reinforcement learning (RL) approach that jointly optimizes decomposition quality and verifier alignment using Group Relative Policy Optimization (GRPO). Our method integrates: (i) structured sequential reasoning; (ii) supervised finetuning on teacher-distilled exemplars; and (iii) a multi-objective reward balancing format compliance, verifier alignment, and decomposition quality. Across six evaluation settings, our trained 8B decomposer improves downstream verification performance to (71.75%) macro-F1, outperforming prompt-based approaches ((+1.99), (+6.24)) and existing RL methods ((+5.84)). Human evaluation confirms the high quality of the generated subclaims. Our framework enables smaller language models to achieve state-of-the-art claim verification by jointly optimising for verification accuracy and decomposition quality.

cs.AI↗

arXiv2026

ExStrucTiny: A Benchmark for Schema-Variable Structured Information Extraction from Document Images

Enterprise documents, such as forms and reports, embed critical information for downstream applications like data archiving, automated workflows, and analytics. Although generalist Vision Language Models (VLMs) perform well on established document understanding benchmarks, their ability to conduct holistic, fine-grained structured extraction across diverse document types and flexible schemas is not well studied. Existing Key Entity Extraction (KEE), Relation Extraction (RE), and Visual Question Answering (VQA) datasets are limited by narrow entity ontologies, simple queries, or homogeneous document types, often overlooking the need for adaptable and structured extraction. To address these gaps, we introduce ExStrucTiny, a new benchmark dataset for structured Information Extraction (IE) from document images, unifying aspects of KEE, RE, and VQA. Built through a novel pipeline combining manual and synthetic human-validated samples, ExStrucTiny covers more varied document types and extraction scenarios. We analyze open and closed VLMs on this benchmark, highlighting challenges such as schema adaptation, query under-specification, and answer localization. We hope our work provides a bedrock for improving generalist models for structured IE in documents.

cs.CL↗

arXiv2025

Perturb Your Data: Paraphrase-Guided Training Data Watermarking

Training data detection is critical for enforcing copyright and data licensing, as Large Language Models (LLM) are trained on massive text corpora scraped from the internet. We present SPECTRA, a watermarking approach that makes training data reliably detectable even when it comprises less than 0.001% of the training corpus. SPECTRA works by paraphrasing text using an LLM and assigning a score based on how likely each paraphrase is, according to a separate scoring model. A paraphrase is chosen so that its score closely matches that of the original text, to avoid introducing any distribution shifts. To test whether a suspect model has been trained on the watermarked data, we compare its token probabilities against those of the scoring model. We demonstrate that SPECTRA achieves a consistent p-value gap of over nine orders of magnitude when detecting data used for training versus data not used for training, which is greater than all baselines tested. SPECTRA equips data owners with a scalable, deploy-before-release watermark that survives even large-scale LLM training.

cs.CL↗

arXiv2025

ALMAS: an Autonomous LLM-based Multi-Agent Software Engineering Framework

Multi-agent Large Language Model (LLM) systems have been leading the way in applied LLM research across a number of fields. One notable area is software development, where researchers have advanced the automation of code implementation, code testing, code maintenance, inter alia, using LLM agents. However, software development is a multifaceted environment that extends beyond just code. As such, a successful LLM system must factor in multiple stages of the software development life-cycle (SDLC). In this paper, we propose a vision for ALMAS, an Autonomous LLM-based Multi-Agent Software Engineering framework, which follows the above SDLC philosophy such that it may work within an agile software development team to perform several tasks end-to-end. ALMAS aligns its agents with agile roles, and can be used in a modular fashion to seamlessly integrate with human developers and their development environment. We showcase the progress towards ALMAS through our published works and a use case demonstrating the framework, where ALMAS is able to seamlessly generate an application and add a new feature.

cs.SE↗

arXiv2025

LLM Agents for Automated Dependency Upgrades

As a codebase expands over time, its library dependencies can become outdated and require updates to maintain innovation and security. However, updating a library can introduce breaking changes in the code, necessitating significant developer time for maintenance. To address this, we introduce a framework of LLM agents to be used in combination with migration documentation to automatically recommend and apply code updates and ensure compatibility with new versions. Our solution can automatically localize updated library usages in live Java codebases and implement recommended fixes in a user-friendly manner. The system architecture consists of multiple key components: a Summary Agent, Control Agent, and Code Agent. To validate our approach, we apply the framework on an industrial use case by which we create three synthetic code repositories with major Upgrade changes and benchmark our approach against state-of-the-art methods. Results show that our approach not only performs upgrades using fewer tokens across all cases but also achieves a precision of 71.4%, highlighting its efficiency and effectiveness compared to state-of-the-art methods.

cs.SE↗

arXiv2025

The Oracle Has Spoken: A Multi-Aspect Evaluation of Dialogue in Pythia

Dialogue is one of the landmark abilities of large language models (LLMs). Despite its ubiquity, few studies actually distinguish specific ingredients underpinning dialogue behavior emerging during post-training. We employ a comprehensive suite of model-based metrics, each targeting a distinct fine-grained aspect of dialogue, motivated by linguistic theory. We evaluate how the performance of pre-trained Pythia models changes with respect to each of those dimensions, depending on model size and as a result of supervised fine-tuning on conversational datasets. We observe only a mild impact of raw model size on most metrics, whereas fine-tuning quickly saturates the scores for all but the smallest models tested. Somewhat contrary to our expectations, many metrics show very similar trends, especially if they are all rooted in the same evaluator model, which raises the question of their reliability in measuring a specific dimension. To that end, we conduct additional analyses of score distributions, metric correlations, and term frequencies in generated responses to help explain our observations.

cs.CL↗

arXiv2025

CoCoLex: Confidence-guided Copy-based Decoding for Grounded Legal Text Generation

Due to their ability to process long and complex contexts, LLMs can offer key benefits to the Legal domain, but their adoption has been hindered by their tendency to generate unfaithful, ungrounded, or hallucinatory outputs. While Retrieval-Augmented Generation offers a promising solution by grounding generations in external knowledge, it offers no guarantee that the provided context will be effectively integrated. To address this, context-aware decoding strategies have been proposed to amplify the influence of relevant context, but they usually do not explicitly enforce faithfulness to the context. In this work, we introduce Confidence-guided Copy-based Decoding for Legal Text Generation (CoCoLex)-a decoding strategy that dynamically interpolates the model produced vocabulary distribution with a distribution derived based on copying from the context. CoCoLex encourages direct copying based on the model's confidence, ensuring greater fidelity to the source. Experimental results on five legal benchmarks demonstrate that CoCoLex outperforms existing context-aware decoding methods, particularly in long-form generation tasks.

cs.CL↗

arXiv2025

Meta-RAG on Large Codebases Using Code Summarization

Large Language Model (LLM) systems have been at the forefront of applied Artificial Intelligence (AI) research in a multitude of domains. One such domain is software development, where researchers have pushed the automation of a number of code tasks through LLM agents. Software development is a complex ecosystem, that stretches far beyond code implementation and well into the realm of code maintenance. In this paper, we propose a multi-agent system to localize bugs in large pre-existing codebases using information retrieval and LLMs. Our system introduces a novel Retrieval Augmented Generation (RAG) approach, Meta-RAG, where we utilize summaries to condense codebases by an average of 79.8\%, into a compact, structured, natural language representation. We then use an LLM agent to determine which parts of the codebase are critical for bug resolution, i.e. bug localization. We demonstrate the usefulness of Meta-RAG through evaluation with the SWE-bench Lite dataset. Meta-RAG scores 84.67 % and 53.0 % for file-level and function-level correct localization rates, respectively, achieving state-of-the-art performance.

cs.SE↗

arXiv2024

"What is the value of {templates}?" Rethinking Document Information Extraction Datasets for LLMs

The rise of large language models (LLMs) for visually rich document understanding (VRDU) has kindled a need for prompt-response, document-based datasets. As annotating new datasets from scratch is labor-intensive, the existing literature has generated prompt-response datasets from available resources using simple templates. For the case of key information extraction (KIE), one of the most common VRDU tasks, past work has typically employed the template "What is the value for the {key}?". However, given the variety of questions encountered in the wild, simple and uniform templates are insufficient for creating robust models in research and industrial contexts. In this work, we present K2Q, a diverse collection of five datasets converted from KIE to a prompt-response format using a plethora of bespoke templates. The questions in K2Q can span multiple entities and be extractive or boolean. We empirically compare the performance of seven baseline generative models on K2Q with zero-shot prompting. We further compare three of these models when training on K2Q versus training on simpler templates to motivate the need of our work. We find that creating diverse and intricate KIE questions enhances the performance and robustness of VRDU models. We hope this work encourages future studies on data quality for generative model training.

cs.CL↗

arXiv2024

Fine-Tuning Language Models with Differential Privacy through Adaptive Noise Allocation

Language models are capable of memorizing detailed patterns and information, leading to a double-edged effect: they achieve impressive modeling performance on downstream tasks with the stored knowledge but also raise significant privacy concerns. Traditional differential privacy based training approaches offer robust safeguards by employing a uniform noise distribution across all parameters. However, this overlooks the distinct sensitivities and contributions of individual parameters in privacy protection and often results in suboptimal models. To address these limitations, we propose ANADP, a novel algorithm that adaptively allocates additive noise based on the importance of model parameters. We demonstrate that ANADP narrows the performance gap between regular fine-tuning and traditional DP fine-tuning on a series of datasets while maintaining the required privacy constraints.

cs.AI↗

arXiv2024

CodeMirage: Hallucinations in Code Generated by Large Language Models

Large Language Models (LLMs) have shown promising potentials in program generation and no-code automation. However, LLMs are prone to generate hallucinations, i.e., they generate text which sounds plausible but is incorrect. Although there has been a recent surge in research on LLM hallucinations for text generation, similar hallucination phenomenon can happen in code generation. Sometimes the generated code can have syntactical or logical errors as well as more advanced issues like security vulnerabilities, memory leaks, etc. Given the wide adaptation of LLMs to enhance efficiency in code generation and development in general, it becomes imperative to investigate hallucinations in code generation. To the best of our knowledge, this is the first attempt at studying hallucinations in the code generated by LLMs. We start by introducing the code hallucination definition and a comprehensive taxonomy of code hallucination types. We propose the first benchmark CodeMirage dataset for code hallucinations. The benchmark contains 1,137 GPT-3.5 generated hallucinated code snippets for Python programming problems from two base datasets - HumanEval and MBPP. We then propose the methodology for code hallucination detection and experiment with open source LLMs such as CodeLLaMA as well as OpenAI's GPT-3.5 and GPT-4 models using one-shot prompt. We find that GPT-4 performs the best on HumanEval dataset and gives comparable results to the fine-tuned CodeBERT baseline on MBPP dataset. Towards the end, we discuss various mitigation strategies for code hallucinations and conclude our work.

cs.SE↗

arXiv2024

BuDDIE: A Business Document Dataset for Multi-task Information Extraction

The field of visually rich document understanding (VRDU) aims to solve a multitude of well-researched NLP tasks in a multi-modal domain. Several datasets exist for research on specific tasks of VRDU such as document classification (DC), key entity extraction (KEE), entity linking, visual question answering (VQA), inter alia. These datasets cover documents like invoices and receipts with sparse annotations such that they support one or two co-related tasks (e.g., entity extraction and entity linking). Unfortunately, only focusing on a single specific of documents or task is not representative of how documents often need to be processed in the wild - where variety in style and requirements is expected. In this paper, we introduce BuDDIE (Business Document Dataset for Information Extraction), the first multi-task dataset of 1,665 real-world business documents that contains rich and dense annotations for DC, KEE, and VQA. Our dataset consists of publicly available business entity documents from US state government websites. The documents are structured and vary in their style and layout across states and types (e.g., forms, certificates, reports, etc.). We provide data variety and quality metrics for BuDDIE as well as a series of baselines for each task. Our baselines cover traditional textual, multi-modal, and large language model approaches to VRDU.

cs.CL↗

arXiv2024

Translating between SQL Dialects for Cloud Migration

Migrations of systems from on-site premises to the cloud has been a fundamental endeavor by many industrial institutions. A crucial component of such cloud migrations is the transition of databases to be hosted online. In this work, we consider the difficulties of this migration for SQL databases. While SQL is one of the prominent methods for storing database procedures, there are a plethora of different SQL dialects (e.g., MySQL, Postgres, etc.) which can complicate migrations when the on-premise SQL dialect differs to the dialect hosted on the cloud. Tools exist by common cloud provides such as AWS and Azure to aid in translating between dialects in order to mitigate the majority of the difficulties. However, these tools do not successfully translate $100\%$ of the code. Consequently, software engineers must manually convert the remainder of the untranslated database. For large organizations, this task quickly becomes intractable and so more innovative solutions are required. We consider this challenge a novel yet vital industrial research problem for any large corporation that is considering cloud migrations. Furthermore, we introduce potential avenues of research to tackle this challenge that have yielded promising preliminary results.

cs.DB↗

arXiv2024

Search-based Optimisation of LLM Learning Shots for Story Point Estimation

One of the ways Large Language Models (LLMs) are used to perform machine learning tasks is to provide them with a few examples before asking them to produce a prediction. This is a meta-learning process known as few-shot learning. In this paper, we use available Search-Based methods to optimise the number and combination of examples that can improve an LLM's estimation performance, when it is used to estimate story points for new agile tasks. Our preliminary results show that our SBSE technique improves the estimation performance of the LLM by 59.34% on average (in terms of mean absolute error of the estimation) over three datasets against a zero-shot setting.

cs.SE↗

arXiv2023

DocLLM: A layout-aware generative language model for multimodal document understanding

Enterprise documents such as forms, invoices, receipts, reports, contracts, and other similar records, often carry rich semantics at the intersection of textual and spatial modalities. The visual cues offered by their complex layouts play a crucial role in comprehending these documents effectively. In this paper, we present DocLLM, a lightweight extension to traditional large language models (LLMs) for reasoning over visual documents, taking into account both textual semantics and spatial layout. Our model differs from existing multimodal LLMs by avoiding expensive image encoders and focuses exclusively on bounding box information to incorporate the spatial layout structure. Specifically, the cross-alignment between text and spatial modalities is captured by decomposing the attention mechanism in classical transformers to a set of disentangled matrices. Furthermore, we devise a pre-training objective that learns to infill text segments. This approach allows us to address irregular layouts and heterogeneous content frequently encountered in visual documents. The pre-trained model is fine-tuned using a large-scale instruction dataset, covering four core document intelligence tasks. We demonstrate that our solution outperforms SotA LLMs on 14 out of 16 datasets across all tasks, and generalizes well to 4 out of 5 previously unseen datasets.

cs.CL↗