LogisticsVLN: Vision-Language Navigation For Low-Altitude Terminal Delivery Based on Agentic UA Vs Xinyuan Zhang, Y onglin Tian, Fei Lin, Y ue Liu, Jing Ma, Korn elia S ara Szatm ary, Fei-Y ue Wang Abstract --The growing demand for intelligent logistics, particularly fine-grained terminal delivery, underscores the need for autonomous UA V (Unmanned Aerial V ehicle)-based delivery systems. However, most existing last-mile delivery studies rely on ground robots, while current UA V-based Vision-Language Navigation (VLN) tasks primarily focus on coarse-grained, long-range goals, making them unsuitable for precise terminal delivery. T o bridge this gap, we propose LogisticsVLN, a scalable aerial delivery system built on multimodal large language models (MLLMs) for autonomous terminal delivery. LogisticsVLN integrates lightweight Large Language Models (LLMs) and Visual-Language Models (VLMs) in a modular pipeline for request understanding, floor localization, object detection, and action-decision making. T o support research and evaluation in this new setting, we construct the Vision-Language Delivery (VLD) dataset within the CARLA simulator . In addition, we conduct subtask-level evaluations of each module of our system, offering valuable insights for improving the robustness and real-world deployment of foundation model-based vision-language delivery systems. I NTRODUCTION Driven by the rapid growth of e-commerce and urbanization, logistics has become an increasingly critical component of modern society [1]. In particular, there is a growing demand for stable, efficient, and user-centric terminal delivery, This work is partly supported by the Science and Technology Development Fund, Macao SAR (File no. Xinyuan Zhang is with the School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049, China (e-mail: zhangxinyuan23@mails.ucas.ac.cn).

Membership Inference Attacks (MIAs) have recently been employed to determine whether a specific text was part of the pre-training data of Large Language Models (LLMs). However, existing methods often misinfer non-members as members, leading to a high false positive rate, or depend on additional reference models for probability calibration, which limits their practicality. To overcome these challenges, we introduce a novel framework called Automatic Calibration Membership Inference Attack (ACMIA), which utilizes a tunable temperature to calibrate output probabilities effectively. This approach is inspired by our theoretical insights into maximum likelihood estimation during the pre-training of LLMs. We introduce ACMIA in three configurations designed to accommodate different levels of model access and increase the probability gap between members and non-members, improving the reliability and robustness of membership inference. Extensive experiments on various open-source LLMs demonstrate that our proposed attack is highly effective, robust, and generalizable, surpassing state-of-the-art baselines across three widely used benchmarks. Our code is available at: Github. 1 Introduction Large Language Models (LLMs), pre-trained on massive text corpora, have shown impressive human-level language understanding, reasoning, and decision-making capabilities [4, 28, 1, 23]. However, their tendency to memorize training data also introduces significant ethical and security concerns [14, 31, 2, 21, 22].

Comprehending long videos remains a significant challenge for Large Multi-modal Models (LMMs). Current LMMs struggle to process even minutes to hours videos due to their lack of explicit memory and retrieval mechanisms. To address this limitation, we propose RA VU(R etrieval A ugmented V ideo U nderstanding), a novel framework for video understanding enhanced by retrieval with compositional reasoning over a spatiotemporal graph. We construct a graph representation of the video, capturing both spatial and temporal relationships between entities. This graph serves as a long-term memory, allowing us to track objects and their actions across time. To answer complex queries, we decompose the queries into a sequence of reasoning steps and execute these steps on the graph, retrieving relevant key information. Our approach enables more accurate understanding of long videos, particularly for queries that require multi-hop reasoning and tracking objects across frames. Our approach demonstrate superior performances with limited retrieved frames (5-10) compared with other SOT A methods and baselines on two major video QA datasets, NExT -QA and EgoSchema. 1 Introduction Understanding videos inherently requires the ability to memorize multi-modal information and retrieve it according to a given task. Recent advancements in Large Multi-modal Models (LMMs) have shown promise in tackling this challenge [ Song et al., 2024a; He et al., 2024; Wang et al., 2024a ] .

In recent years, the compression of large language models (LLMs) has emerged as a key problem in facilitating LLM deployment on resource-limited devices, reducing compute costs, and mitigating the environmental footprint due to large-scale AI infrastructure. Here, we establish the foundations of LLM quantization from a rate-distortion theory perspective and propose a quantization technique based on simple rate-distortion optimization. Our technique scales to models containing hundreds of billions of weight parameters and offers users the flexibility to compress models, post-training, to a model size or accuracy specified by the user.

Large language models (LLMs) have shown promise in synthetic tabular data generation, yet existing methods struggle to preserve complex feature dependencies, particularly among categorical variables. This work introduces a probability-driven prompting approach that leverages LLMs to estimate conditional distributions, enabling more accurate and scalable data synthesis. The results highlight the potential of prompting probability distributions to enhance the statistical fidelity of LLM-generated tabular data.

Uncontrollable autonomous replication of language model agents poses a critical safety risk. To better understand this risk, we introduce RepliBench, a suite of evaluations designed to measure autonomous replication capabilities. RepliBench is derived from a decomposition of these capabilities covering four core domains: obtaining resources, exfiltrating model weights, replicating onto compute, and persisting on this compute for long periods. We create 20 novel task families consisting of 86 individual tasks. We benchmark 5 frontier models, and find they do not currently pose a credible threat of self-replication, but succeed on many components and are improving rapidly. Models can deploy instances from cloud compute providers, write self-propagating programs, and exfiltrate model weights under simple security setups, but struggle to pass KYC checks or set up robust and persistent agent deployments. Overall the best model we evaluated (Claude 3.7 Sonnet) has a >50% pass@10 score on 15/20 task families, and a >50% pass@10 score for 9/20 families on the hardest variants. These findings suggest autonomous replication capability could soon emerge with improvements in these remaining areas or with human assistance.

Vision-language models (VLMs) allow to embed texts and images in a shared representation space. However, it has been shown that these models are subject to a modality gap phenomenon meaning there exists a clear separation between the embeddings from one modality and another in the embedding space. While this misalignment is detrimental for downstream tasks such as multimodal retrieval, multimodal clustering or zero-shot classification, etc. no generic and practical methods have so far been proposed to assess it precisely and even reduce it. We therefore propose novel measures and effective techniques (spectral- and optimal transport-based methods) to achieve this goal. Extensive experiments conducted on several image-text datasets and models demonstrate their effectiveness and beneficial effects on downstream tasks. Our code is available at the URL provided in the paper's abstract.

Our work contributes to the fast-growing literature on the use of Large Language Models (LLMs) to perform graph-related tasks. In particular, we focus on usage scenarios that rely on the visual modality, feeding the model with a drawing of the graph under analysis. We investigate how the model's performance is affected by the chosen layout paradigm, the aesthetics of the drawing, and the prompting technique used for the queries. We formulate three corresponding research questions and present the results of a thorough experimental analysis. Our findings reveal that choosing the right layout paradigm and optimizing the readability of the input drawing from a human perspective can significantly improve the performance of the model on the given task. Moreover, selecting the most effective prompting technique is a challenging yet crucial task for achieving optimal performance.

Large language models (LLMs) have evolved from simple chatbots into autonomous agents capable of performing complex tasks such as editing production code, orchestrating workflows, and taking higher-stakes actions based on untrusted inputs like webpages and emails. These capabilities introduce new security risks that existing security measures, such as model fine-tuning or chatbot-focused guardrails, do not fully address. Given the higher stakes and the absence of deterministic solutions to mitigate these risks, there is a critical need for a real-time guardrail monitor to serve as a final layer of defense, and support system level, use case specific safety policy definition and enforcement. We introduce LlamaFirewall, an open-source security focused guardrail framework designed to serve as a final layer of defense against security risks associated with AI Agents. Our framework mitigates risks such as prompt injection, agent misalignment, and insecure code risks through three powerful guardrails: PromptGuard 2, a universal jailbreak detector that demonstrates clear state of the art performance; Agent Alignment Checks, a chain-of-thought auditor that inspects agent reasoning for prompt injection and goal misalignment, which, while still experimental, shows stronger efficacy at preventing indirect injections in general scenarios than previously proposed approaches; and CodeShield, an online static analysis engine that is both fast and extensible, aimed at preventing the generation of insecure or dangerous code by coding agents. Additionally, we include easy-to-use customizable scanners that make it possible for any developer who can write a regular expression or an LLM prompt to quickly update an agent's security guardrails.

In this paper, we introduce OSUniverse: a benchmark of complex, multimodal desktop-oriented tasks for advanced GUI-navigation AI agents that focuses on ease of use, extensibility, comprehensive coverage of test cases, and automated validation. We divide the tasks in increasing levels of complexity, from basic precision clicking to multistep, multiapplication tests requiring dexterity, precision, and clear thinking from the agent. In version one of the benchmark, presented here, we have calibrated the complexity of the benchmark test cases to ensure that the SOTA (State of the Art) agents (at the time of publication) do not achieve results higher than 50%, while the average white collar worker can perform all these tasks with perfect accuracy. The benchmark can be scored manually, but we also introduce an automated validation mechanism that has an average error rate less than 2%. Therefore, this benchmark presents solid ground for fully automated measuring of progress, capabilities and the effectiveness of GUI-navigation AI agents over the short and medium-term horizon. The source code of the benchmark is available at https://github.com/agentsea/osuniverse.