Retrieval augmented generation (RAG) has become the standard in long-context question answering (QA) systems. However, typical implementations of RAG rely on a rather naive retrieval mechanism, in which texts whose embeddings are most similar to that of the query are deemed most relevant. This has consequences in subjective QA tasks, where the most relevant text may not directly contain the answer. In this work, we propose a novel extension to RAG systems, which we call Retrieval from AI Derived Documents (RAIDD). RAIDD leverages the full power of the LLM in the retrieval process by deriving inferred features, such as summaries and example questions, from the documents at ingest. We demonstrate that this approach significantly improves the performance of RAG systems on long-context QA tasks.

In this work, we present HiReview, a novel framework for hierarchical taxonomy-driven automatic literature review generation. With the exponential growth of academic documents, manual literature reviews have become increasingly labor-intensive and time-consuming, while traditional summarization models struggle to generate comprehensive document reviews effectively. Large language models (LLMs), with their powerful text processing capabilities, offer a potential solution; however, research on incorporating LLMs for automatic document generation remains limited. To address key challenges in large-scale automatic literature review generation (LRG), we propose a two-stage taxonomy-then-generation approach that combines graph-based hierarchical clustering with retrieval-augmented LLMs. First, we retrieve the most relevant sub-community within the citation network, then generate a hierarchical taxonomy tree by clustering papers based on both textual content and citation relationships. In the second stage, an LLM generates coherent and contextually accurate summaries for clusters or topics at each hierarchical level, ensuring comprehensive coverage and logical organization of the literature. Extensive experiments demonstrate that HiReview significantly outperforms state-of-the-art methods, achieving superior hierarchical organization, content relevance, and factual accuracy in automatic literature review generation tasks.

Large Language Models (LLMs) have achieved remarkable breakthroughs. However, the huge number of parameters in LLMs require significant amount of memory storage in inference, which prevents their practical deployment in many applications. To reduce memory storage of LLMs, singular value decomposition (SVD) provides a promising solution to approximate weight matrices for compressing LLMs. In this paper, we take a step further to explore parameter sharing across different layers with SVD to achieve more effective compression for LLMs. Specifically, weight matrices in different layers are decomposed and represented as a linear combination of a set of shared basis vectors and unique coefficients. The types of weight matrices and the layer selection for basis sharing are examined when compressing LLMs to maintain the performance. Comprehensive experiments demonstrate that Basis Sharing outperforms state-of-the-art SVD-based compression approaches and parameter sharing techniques, especially under large compression ratios. Large Language Models (LLMs) have revolutionized natural language processing by enabling machines to understand human language more accurately.

This paper proposes a simple yet effective jailbreak attack named FlipAttack against black-box LLMs. First, from the autoregressive nature, we reveal that LLMs tend to understand the text from left to right and find that they struggle to comprehend the text when noise is added to the left side. Motivated by these insights, we propose to disguise the harmful prompt by constructing left-side noise merely based on the prompt itself, then generalize this idea to 4 flipping modes. Second, we verify the strong ability of LLMs to perform the text-flipping task, and then develop 4 variants to guide LLMs to denoise, understand, and execute harmful behaviors accurately. These designs keep FlipAttack universal, stealthy, and simple, allowing it to jailbreak black-box LLMs within only 1 query. Experiments on 8 LLMs demonstrate the superiority of FlipAttack. Remarkably, it achieves $\sim$98\% attack success rate on GPT-4o, and $\sim$98\% bypass rate against 5 guardrail models on average. The codes are available at GitHub\footnote{https://github.com/yueliu1999/FlipAttack}.

Large language models (LLMs) have demonstrated remarkable capabilities, but they also pose risks related to the generation of toxic or harmful content. This work introduces Precision Knowledge Editing (PKE), an advanced technique that builds upon existing knowledge editing methods to more effectively identify and modify toxic parameter regions within LLMs. By leveraging neuron weight tracking and activation pathway tracing, PKE achieves finer granularity in toxic content management compared to previous methods like Detoxifying Instance Neuron Modification (DINM). Our experiments demonstrate that PKE significantly reduces the attack success rate (ASR) across various models, including Llama2-7b and Llama-3-8b-instruct, while maintaining overall model performance. Additionally, we also compared the performance of some closed-source models (gpt-4-0613 and Claude 3 Sonnet) in our experiments, and found that models adjusted using our method far outperformed the closed-source models in terms of safety. This research contributes to the ongoing efforts to make LLMs safer and more reliable for real-world applications.

Despite the increasing effectiveness of language models, their reasoning capabilities remain underdeveloped. In particular, causal reasoning through counterfactual question answering is lacking. This work aims to bridge this gap. We first derive novel metrics that balance accuracy in factual and counterfactual questions, capturing a more complete view of the reasoning abilities of language models than traditional factual-only based metrics. Second, we propose several fine-tuning approaches that aim to elicit better reasoning mechanisms, in the sense of the proposed metrics. Finally, we evaluate the performance of the fine-tuned language models in a variety of realistic scenarios. In particular, we investigate to what extent our fine-tuning approaches systemically achieve better generalization with respect to the base models in several problems that require, among others, inductive and deductive reasoning capabilities.

Efficient patient-doctor interaction is among the key factors for a successful disease diagnosis. During the conversation, the doctor could query complementary diagnostic information, such as the patient's symptoms, previous surgery, and other related information that goes beyond medical evidence data (test results) to enhance disease diagnosis. However, this procedure is usually time-consuming and less-efficient, which can be potentially optimized through computer-assisted systems. As such, we propose a diagnostic dialogue system to automate the patient information collection procedure. By exploiting medical history and conversation logic, our conversation agents, particularly the doctor agent, can pose multi-round clinical queries to effectively collect the most relevant disease diagnostic information. Moreover, benefiting from our two-stage recommendation structure, carefully designed ranking criteria, and interactive patient agent, our model is able to overcome the under-exploration and non-flexible challenges in dialogue generation. Our experimental results on a real-world medical conversation dataset show that our model can generate clinical queries that mimic the conversation style of real doctors, with efficient fluency, professionalism, and safety, while effectively collecting relevant disease diagnostic information.

Agentic AIs $-$ AIs that are capable and permitted to undertake complex actions with little supervision $-$ mark a new frontier in AI capabilities and raise new questions about how to safely create and align such systems with users, developers, and society. Because agents' actions are influenced by their attitudes toward risk, one key aspect of alignment concerns the risk profiles of agentic AIs. Risk alignment will matter for user satisfaction and trust, but it will also have important ramifications for society more broadly, especially as agentic AIs become more autonomous and are allowed to control key aspects of our lives. AIs with reckless attitudes toward risk (either because they are calibrated to reckless human users or are poorly designed) may pose significant threats. They might also open 'responsibility gaps' in which there is no agent who can be held accountable for harmful actions. What risk attitudes should guide an agentic AI's decision-making? How might we design AI systems that are calibrated to the risk attitudes of their users? What guardrails, if any, should be placed on the range of permissible risk attitudes? What are the ethical considerations involved when designing systems that make risky decisions on behalf of others? We present three papers that bear on key normative and technical aspects of these questions.

De-identification is important in protecting patients' privacy for healthcare text analytics. The MASK framework is one of the best on the de-identification shared task organised by n2c2/i2b2 challenges. This work enhances the MASK framework by integrating ClinicalBERT, a deep learning model specifically fine-tuned on clinical texts, alongside traditional de-identification methods like dictionary lookup and rule-based approaches. The system effectively identifies and either redacts or replaces sensitive identifiable entities within clinical documents, while also allowing users to customise the masked documents according to their specific needs. The integration of ClinicalBERT significantly improves the performance of entity recognition, achieving 0.9732 F1-score, especially for common entities such as names, dates, and locations. A risk assessment feature has also been developed, which analyses the uniqueness of context within documents to classify them into risk levels, guiding further de-identification efforts. While the system demonstrates strong overall performance, this work highlights areas for future improvement, including handling more complex entity occurrences and enhancing the system's adaptability to different clinical settings.

Knowledge tracing (KT) is a popular approach for modeling students' learning progress over time, which can enable more personalized and adaptive learning. However, existing KT approaches face two major limitations: (1) they rely heavily on expert-defined knowledge concepts (KCs) in questions, which is timeconsuming and prone to errors; and (2) KT methods tend to overlook the semantics of both questions and the given KCs. In this work, we address these challenges and present KCQRL, a framework for automated knowledge concept annotation and question representation learning that can improve the effectiveness of any existing KT model. First, we propose an automated KC annotation process using large language models (LLMs), which generates question solutions and then annotates KCs in each solution step of the questions. Second, we introduce a contrastive learning approach to generate semantically rich embeddings for questions and solution steps, aligning them with their associated KCs via a tailored false negative elimination approach. These embeddings can be readily integrated into existing KT models, replacing their randomly initialized embeddings. We demonstrate the effectiveness of KCQRL across 15 KT models on two large real-world Math learning datasets, where we achieve consistent performance improvements. The recent years have witnessed a surge in online learning platforms (Adedoyin & Soykan, 2023; Gros & García-Peñalvo, 2023), where students learn new knowledge concepts, which are then tested through exercises. Needless to say, personalization is crucial for effective learning: it allows that new knowledge concepts are carefully tailored to the current knowledge state of the student, which is more effective than one-size-fits-all approaches to learning (Cui & Sachan, 2023; Xu et al., 2024). However, such personalization requires that the knowledge of students is continuously assessed, which highlights the need for knowledge tracing (KT). In KT, one models the temporal dynamics of students' learning processes (Corbett & Anderson, 1994) in terms of a core set of skills, which are called knowledge concepts (KCs). KT models are typically time-series models that receive the past interactions of the learner as input (e.g., her previous exercises) in order to predict response of the learner to the next exercise. Yet, existing KT models have two main limitations that hinder their applicability in practice (see Figure 1). They require a comprehensive mapping between KCs and questions, which is typically done through manual annotations by experts. However, such KC annotation is both time-intensive and prone to errors (Clark, 2014; Bier et al., 2019). KT models overlook the semantics of both questions and KCs.