As software systems evolve, developers face a dual challenge: maintaining correctness by fixing bugs and continuously adapting functionality to meet new user demands. Traditional software engineering processes rely heavily on human developers to interpret requirements, fix errors, and ensure correctness against specifications. With the advancement of Large Language Models (LLMs) adept at code generation, the opportunity arises to shift portions of these responsibilities onto machine-driven processes. However, simply prompting an LLM to solve a complex programming task--be it eliminating a subtle bug or implementing a new feature--often falls short. Complex codebases exceed the model's context window, specification details are not always fully captured in a single prompt, and correctness requires iterative refinement guided by tests, analysis, and verification. This paper proposes a holistic, iterative framework that enables an LLM to evolve a codebase from an initial, potentially buggy state to one that satisfies not only pre-existing correctness criteria but also newly introduced feature demands. Key contributions include: 1. Unified Framework for Bug-to-Demand Resolution: We present a method by which the LLM iteratively refines code, starting from an imperfect state (with known or unknown bugs) and incrementally adjusting the codebase to meet a set of evolving functional and nonfunctional requirements introduced over time.

In this work, we propose the first motion transfer approach in diffusion transformer through Mixture of Score Guidance (MSG), a theoretically-grounded framework for motion transfer in diffusion models. Our key theoretical contribution lies in reformulating conditional score to decompose motion score and content score in diffusion models. By formulating motion transfer as a mixture of potential energies, MSG naturally preserves scene composition and enables creative scene transformations while maintaining the integrity of transferred motion patterns. This novel sampling operates directly on pre-trained video diffusion models without additional training or fine-tuning. Through extensive experiments, MSG demonstrates successful handling of diverse scenarios including single object, multiple objects, and cross-object motion transfer as well as complex camera motion transfer. Additionally, we introduce MotionBench, the first motion transfer dataset consisting of 200 source videos and 1000 transferred motions, covering single/multi-object transfers, and complex camera motions.

Despite impressive advancements in video understanding, most efforts remain limited to coarse-grained or visual-only video tasks. However, real-world videos encompass omni-modal information (vision, audio, and speech) with a series of events forming a cohesive storyline. The lack of multi-modal video data with fine-grained event annotations and the high cost of manual labeling are major obstacles to comprehensive omni-modality video perception. To address this gap, we propose an automatic pipeline consisting of high-quality multi-modal video filtering, semantically coherent omni-modal event boundary detection, and cross-modal correlation-aware event captioning. In this way, we present LongVALE, the first-ever Vision-Audio-Language Event understanding benchmark comprising 105K omni-modal events with precise temporal boundaries and detailed relation-aware captions within 8.4K high-quality long videos. Further, we build a baseline that leverages LongVALE to enable video large language models (LLMs) for omni-modality fine-grained temporal video understanding for the first time. Extensive experiments demonstrate the effectiveness and great potential of LongVALE in advancing comprehensive multi-modal video understanding.

Blockchain smart contracts have catalyzed the development of decentralized applications across various domains, including decentralized finance. However, due to constraints in computational resources and the prevalence of data silos, current smart contracts face significant challenges in fully leveraging the powerful capabilities of Large Language Models (LLMs) for tasks such as intelligent analysis and reasoning. To address this gap, this paper proposes and implements a universal framework for integrating LLMs with blockchain data, {\sysname}, effectively overcoming the interoperability barriers between blockchain and LLMs. By combining semantic relatedness with truth discovery methods, we introduce an innovative data aggregation approach, {\funcname}, which significantly enhances the accuracy and trustworthiness of data generated by LLMs. To validate the framework's effectiveness, we construct a dataset consisting of three types of questions, capturing Q\&A interactions between 10 oracle nodes and 5 LLM models. Experimental results demonstrate that, even with 40\% malicious nodes, the proposed solution improves data accuracy by an average of 17.74\% compared to the optimal baseline. This research not only provides an innovative solution for the intelligent enhancement of smart contracts but also highlights the potential for deep integration between LLMs and blockchain technology, paving the way for more intelligent and complex applications of smart contracts in the future.

TOBU is a novel mobile application that captures and retrieves `personal memories' (pictures/videos together with stories and context around those moments) in a user-engaging AI-guided conversational approach. Our initial prototype showed that existing retrieval techniques such as retrieval-augmented generation (RAG) systems fall short due to their limitations in understanding memory relationships, causing low recall, hallucination, and unsatisfactory user experience. We design TOBUGraph, a novel graph-based retrieval approach. During capturing, TOBUGraph leverages large language models (LLMs) to automatically create a dynamic knowledge graph of memories, establishing context and relationships of those memories. During retrieval, TOBUGraph combines LLMs with the memory graph to achieve comprehensive recall through graph traversal. Our evaluation using real user data demonstrates that TOBUGraph outperforms multiple RAG implementations in both precision and recall, significantly improving user experience through improved retrieval accuracy and reduced hallucination.

Hate speech online remains an understudied issue for marginalized communities, and has seen rising relevance, especially in the Global South, which includes developing societies with increasing internet penetration. In this paper, we aim to provide marginalized communities living in societies where the dominant language is low-resource with a privacy-preserving tool to protect themselves from hate speech on the internet by filtering offensive content in their native languages. Our contribution in this paper is twofold: 1) we release REACT (REsponsive hate speech datasets Across ConTexts), a collection of high-quality, culture-specific hate speech detection datasets comprising seven distinct target groups in eight low-resource languages, curated by experienced data collectors; 2) we propose a solution to few-shot hate speech detection utilizing federated learning (FL), a privacy-preserving and collaborative learning approach, to continuously improve a central model that exhibits robustness when tackling different target groups and languages. By keeping the training local to the users' devices, we ensure the privacy of the users' data while benefitting from the efficiency of federated learning. Furthermore, we personalize client models to target-specific training data and evaluate their performance. Our results indicate the effectiveness of FL across different target groups, whereas the benefits of personalization on few-shot learning are not clear.

LLM vendors expend substantial effort to secure their models and make them unhelpful to adversaries like cybercriminals (Touvron et al. 2023, Section 4.3) (OpenAI et al. 2024, Section 3) (OpenAI 2024a). However, LLMs have been repeatedly "jailbroken" out of these constraints (Chao et al. 2024; Mazeika et al. 2024; Souly et al. 2024). No robust LLM security measures are known. Classic jailbreaks encode LLM prompts to bypass model safeguards. They tend to be unstable, add a token overhead, and reduce model performance (Chao et al. 2024; Mazeika et al. 2024; Souly et al. 2024).

Protecting patient data privacy is a critical concern when deploying machine learning algorithms in healthcare. Differential privacy (DP) is a common method for preserving privacy in such settings and, in this work, we examine two key trade-offs in applying DP to the NLP task of medical coding (ICD classification). Regarding the privacy-utility trade-off, we observe a significant performance drop in the privacy preserving models, with more than a 40% reduction in micro F1 scores on the top 50 labels in the MIMIC-III dataset. From the perspective of the privacy-fairness trade-off, we also observe an increase of over 3% in the recall gap between male and female patients in the DP models. Further understanding these trade-offs will help towards the challenges of real-world deployment.

Surgical interventions, particularly in neurology, represent complex and high-stakes scenarios that impose substantial cognitive burdens on surgical teams. Although deliberate education and practice can enhance cognitive capabilities, surgical training opportunities remain limited due to patient safety concerns. To address these cognitive challenges in surgical training and operation, we propose SurgBox, an agent-driven sandbox framework to systematically enhance the cognitive capabilities of surgeons in immersive surgical simulations. Specifically, our SurgBox leverages large language models (LLMs) with tailored Retrieval-Augmented Generation (RAG) to authentically replicate various surgical roles, enabling realistic training environments for deliberate practice. In particular, we devise Surgery Copilot, an AI-driven assistant to actively coordinate the surgical information stream and support clinical decision-making, thereby diminishing the cognitive workload of surgical teams during surgery. By incorporating a novel Long-Short Memory mechanism, our Surgery Copilot can effectively balance immediate procedural assistance with comprehensive surgical knowledge. Extensive experiments using real neurosurgical procedure records validate our SurgBox framework in both enhancing surgical cognitive capabilities and supporting clinical decision-making. By providing an integrated solution for training and operational support to address cognitive challenges, our SurgBox framework advances surgical education and practice, potentially transforming surgical outcomes and healthcare quality. The code is available at https://github.com/franciszchen/SurgBox.

Augmenting LLMs with context leads to improved performance across many applications. Despite much research on Retrieval Augmented Generation (RAG) systems, an open question is whether errors arise because LLMs fail to utilize the context from retrieval or the context itself is insufficient to answer the query. To shed light on this, we develop a new notion of sufficient context, along with a way to classify instances that have enough information to answer the query. We then use sufficient context to analyze several models and datasets. By stratifying errors based on context sufficiency, we find that proprietary LLMs (Gemini, GPT, Claude) excel at answering queries when the context is sufficient, but often output incorrect answers instead of abstaining when the context is not. We further categorize cases when the context is useful, and improves accuracy, even though it does not fully answer the query and the model errs without the context. Building on our findings, we explore ways to reduce hallucinations in RAG systems, including a new selective generation method that leverages sufficient context information for guided abstention. Our method improves the fraction of correct answers among times where the model responds by 2-10% for Gemini, GPT, and Gemma. Providing Large Language Models (LLMs) with additional context, such as in Retrieval Augmented Generation (RAG) systems, has led to major improvements in LLM factuality and verifiability when adapting to new domains (Lewis et al., 2020). In the case of open-domain question answering, a retrieval model provides context at inference time in the form of snippets or long-form text (Zhu et al., 2021). Then, the model synthesizes the query along with this added context to generate the answer. The ideal outcome is for the LLM to output the correct answer if the provided context contains enough information to answer the question when combined with the model's parametric knowledge. Otherwise, the model should abstain from answering and/or ask for more information. One core challenge in achieving this ideal outcome is building models that can use the provided context only when it helps answer the question correctly. Several works have investigated this issue by evaluating models in the presence of irrelevant information in the context (discussed in Section 2). However, "relevant information" can range from directly containing the answer to simply being topically related Work done during an internship at Google. Work done during an internship at Google. Question: Who is Lya L. married to?