Digital health chatbots powered by Large Language Models (LLMs) have the potential to significantly improve personal health management for chronic conditions by providing accessible and on-demand health coaching and question-answering. However, these chatbots risk providing unverified and inaccurate information because LLMs generate responses based on patterns learned from diverse internet data. Retrieval Augmented Generation (RAG) can help mitigate hallucinations and inaccuracies in LLM responses by grounding it on reliable content. However, efficiently and accurately retrieving most relevant set of content for real-time user questions remains a challenge. In this work, we introduce Query-Based Retrieval Augmented Generation (QB-RAG), a novel approach that pre-computes a database of potential queries from a content base using LLMs. For an incoming patient question, QB-RAG efficiently matches it against this pre-generated query database using vector search, improving alignment between user questions and the content. We establish a theoretical foundation for QB-RAG and provide a comparative analysis of existing retrieval enhancement techniques for RAG systems. Finally, our empirical evaluation demonstrates that QB-RAG significantly improves the accuracy of healthcare question answering, paving the way for robust and trustworthy LLM applications in digital health.

The deployment of social robots in real-world scenarios is increasing, supporting humans in various contexts. However, they still struggle to grasp social dynamics, especially in public spaces, sometimes resulting in violations of social norms, such as interrupting human conversations. This behavior, originating from a limited processing of social norms, might be perceived as robot-centered. Understanding social dynamics, particularly in group-robot-interactions (GRI), underscores the need for further research and development in human-robot-interaction (HRI). Enhancing the interaction abilities of social robots, especially in GRIs, can improve their effectiveness in real-world applications on a micro-level, as group interactions lead to increased motivation and comfort. In this study, we assessed the influence of the interaction condition (dyadic vs. triadic) on the perceived extraversion (ext.) of social robots in public spaces. The research involved 40 HRIs, including 24 dyadic (i.e., one human and one robot) interactions and 16 triadic interactions, which involve at least three entities, including the robot.

Large language models (LLMs) hold great promise in summarizing medical evidence. Most recent studies focus on the application of proprietary LLMs. Using proprietary LLMs introduces multiple risk factors, including a lack of transparency and vendor dependency. While open-source LLMs allow better transparency and customization, their performance falls short compared to proprietary ones. In this study, we investigated to what extent fine-tuning open-source LLMs can further improve their performance in summarizing medical evidence. Utilizing a benchmark dataset, MedReview, consisting of 8,161 pairs of systematic reviews and summaries, we fine-tuned three broadly-used, open-sourced LLMs, namely PRIMERA, LongT5, and Llama-2. Overall, the fine-tuned LLMs obtained an increase of 9.89 in ROUGE-L (95% confidence interval: 8.94-10.81), 13.21 in METEOR score (95% confidence interval: 12.05-14.37), and 15.82 in CHRF score (95% confidence interval: 13.89-16.44). The performance of fine-tuned LongT5 is close to GPT-3.5 with zero-shot settings. Furthermore, smaller fine-tuned models sometimes even demonstrated superior performance compared to larger zero-shot models. The above trends of improvement were also manifested in both human and GPT4-simulated evaluations. Our results can be applied to guide model selection for tasks demanding particular domain knowledge, such as medical evidence summarization.

Building efficient inference framework has gained increasing interests for research community. Early-exit models, a variant of LLMs, improves the inference efficiency of LLMs by skipping rest layers and directly generate output tokens when they are confident enough. However, there is no work of LLM inference framework that takes early-exit models into consideration. This is non-trivial as prior art on LLM inference cannot be directly applied to early-exit models. In this work, we solves two key challenges in building efficient inference framework for early-exit models: (1) batch inference at iteration-level granularity; and (2) KV cache management. For the former, we propose to process the batch until all sequences surpass the early-exit confidence threshold. For the latter, we propose to fill the KV cache of rest layers before the iteration terminates. Our evaluation shows that, compared with the original vLLM operating at full layers, our solution achieves up to 1.25x speed up.

Large language models (LLMs) with billions of parameters excel at predicting the next token in a sequence. Recent work computes non-vacuous compression-based generalization bounds for LLMs, but these bounds are vacuous for large models at the billion-parameter scale. Moreover, these bounds are obtained through restrictive compression techniques, bounding compressed models that generate low-quality text. Additionally, the tightness of these existing bounds depends on the number of IID documents in a training set rather than the much larger number of non-IID constituent tokens, leaving untapped potential for tighter bounds. In this work, we instead use properties of martingales to derive generalization bounds that benefit from the vast number of tokens in LLM training sets. Since a dataset contains far more tokens than documents, our generalization bounds not only tolerate but actually benefit from far less restrictive compression schemes. With Monarch matrices, Kronecker factorizations, and post-training quantization, we achieve non-vacuous generalization bounds for LLMs as large as LLaMA2-70B. Unlike previous approaches, our work achieves the first non-vacuous bounds for models that are deployed in practice and generate high-quality text.

--Cyber-physical systems have become an essential part of the modern healthcare industry. The healthcare cyber-physical systems (HCPS) combine physical and cyber components to improve the healthcare industry. While HCPS has many advantages, it also has some drawbacks, such as a lengthy data entry process, a lack of real-time processing, and limited real-time patient visualization. T o overcome these issues, this paper represents an innovative approach to integrating large language model (LLM) to enhance the efficiency of the healthcare system. By incorporating LLM at various layers, HCPS can leverage advanced AI capabilities to improve patient outcomes, advance data processing, and enhance decision-making. I NTRODUCTION Healthcare cyber-physical systems (HCPS) embody the convergence of cybernetics, software, and physical components, offering profound advancements in medical care [1].

--Integrating Generative AI (GenAI) into educational contexts presents a transformative potential for enhancing learning experiences. This paper introduces CourseGPT, a generative AI tool designed to support instructors and enhance the educational experiences of undergraduate students. Built on open-source Large Language Models (LLMs) from Mistral AI, CourseGPT offers continuous instructor support and regular updates to course materials, enriching the learning environment. By utilizing course-specific content, such as slide decks and supplementary readings and references, CourseGPT provides precise, dynamically generated responses to student inquiries. Unlike generic AI models, CourseGPT allows instructors to manage and control the responses, thus extending the course scope without overwhelming details. The paper demonstrates the application of CourseGPT using the CPR E 431: Basics of Information System Security course as a pilot. This course, with its large enrollment and diverse curriculum, serves as an ideal testbed for CourseGPT . The tool aims to enhance the learning experience, accelerate feedback processes, and streamline administrative tasks. The study evaluates CourseGPT's impact on student outcomes, focusing on correctness scores, context recall, and faithfulness of responses. Results indicate that the Mixtral-8x7b model, with a higher parameter count, outperforms smaller models, achieving an 88.0% correctness score and a 66.6% faithfulness score. Additionally, feedback from former students and teaching assistants on CourseGPT's accuracy, helpfulness, and overall performance was collected. The outcomes revealed that a significant majority found CourseGPT to be highly accurate and beneficial in addressing their queries, with many praising its ability to provide timely and relevant information.

Representations from deep neural networks (DNNs) have proven remarkably predictive of neural activity involved in both visual and linguistic processing. Despite these successes, most studies to date concern unimodal DNNs, encoding either visual or textual input but not both. Yet, there is growing evidence that human meaning representations integrate linguistic and sensory-motor information. Here we investigate whether the integration of multimodal information operated by current vision-and-language DNN models (VLMs) leads to representations that are more aligned with human brain activity than those obtained by language-only and vision-only DNNs. We focus on fMRI responses recorded while participants read concept words in the context of either a full sentence or an accompanying picture. Our results reveal that VLM representations correlate more strongly than language- and vision-only DNNs with activations in brain areas functionally related to language processing. A comparison between different types of visuo-linguistic architectures shows that recent generative VLMs tend to be less brain-aligned than previous architectures with lower performance on downstream applications. Moreover, through an additional analysis comparing brain vs. behavioural alignment across multiple VLMs, we show that -- with one remarkable exception -- representations that strongly align with behavioural judgments do not correlate highly with brain responses. This indicates that brain similarity does not go hand in hand with behavioural similarity, and vice versa.

We present factgenie: a framework for annotating and visualizing word spans in textual model outputs. Annotations can capture various span-based phenomena such as semantic inaccuracies or irrelevant text. With factgenie, the annotations can be collected both from human crowdworkers and large language models. Our framework consists of a web interface for data visualization and gathering text annotations, powered by an easily extensible codebase.

Visual-Language Alignment (VLA) has gained a lot of attention since CLIP's groundbreaking work. Although CLIP performs well, the typical direct latent feature alignment lacks clarity in its representation and similarity scores. On the other hand, lexical representation, a vector whose element represents the similarity between the sample and a word from the vocabulary, is a natural sparse representation and interpretable, providing exact matches for individual words. However, lexical representations is difficult to learn due to no ground-truth supervision and false-discovery issues, and thus requires complex design to train effectively. In this paper, we introduce LexVLA, a more interpretable VLA framework by learning a unified lexical representation for both modalities without complex design. We use DINOv2 as our visual model for its local-inclined features and Llama 2, a generative language model, to leverage its in-context lexical prediction ability. To avoid the false discovery, we propose an overuse penalty to refrain the lexical representation from falsely frequently activating meaningless words. We demonstrate that these two pre-trained uni-modal models can be well-aligned by fine-tuning on modest multi-modal dataset and avoid intricate training configurations. On cross-modal retrieval benchmarks, LexVLA, trained on the CC-12M multi-modal dataset, outperforms baselines fine-tuned on larger datasets (e.g., YFCC15M) and those trained from scratch on even bigger datasets (e.g., 1.1B data, including CC-12M). We conduct extensive experiments to analyze LexVLA.