Powerful artificial intelligence based on large language models can be computationally split to run on several smartphones. That could enable people to use AIs locally without relying on a cloud service's data centres – and without needing to share sensitive queries or personal information with a tech company. "Our key motivation was privacy," says Sangeetha Abdu Jyothi at the University of California, Irvine.

Google is creating a new, more sophisticated Android AI assistant called Pixie set to arrive with its Pixel 9 phone, according to a report from The Information. Based on the company's new Gemini large language model (LLM), it'll be able to perform "complex and multimodal tasks" like giving you directions to the nearest store to buy a product you photographed on your smartphone. The assistant will be exclusive to Google's Pixel devices and use data from Google products like Gmail and Maps. That would help it "evolve into a far more personalized version of the Google Assistant," the report states. It appears to be a separate product from Google's Assistant with Bard showed off at Made By Google in October.

Large-scale language-image pre-trained models (e.g., CLIP) have shown superior performances on many cross-modal retrieval tasks. However, the problem of transferring the knowledge learned from such models to video-based person re-identification (ReID) has barely been explored. In addition, there is a lack of decent text descriptions in current ReID benchmarks. To address these issues, in this work, we propose a novel one-stage text-free CLIP-based learning framework named TF-CLIP for video-based person ReID. More specifically, we extract the identity-specific sequence feature as the CLIP-Memory to replace the text feature. Meanwhile, we design a Sequence-Specific Prompt (SSP) module to update the CLIP-Memory online. To capture temporal information, we further propose a Temporal Memory Diffusion (TMD) module, which consists of two key components: Temporal Memory Construction (TMC) and Memory Diffusion (MD). Technically, TMC allows the frame-level memories in a sequence to communicate with each other, and to extract temporal information based on the relations within the sequence. MD further diffuses the temporal memories to each token in the original features to obtain more robust sequence features. Extensive experiments demonstrate that our proposed method shows much better results than other state-of-the-art methods on MARS, LS-VID and iLIDS-VID. The code is available at https://github.com/AsuradaYuci/TF-CLIP.

Recently, several works have explored if and how robotic coaches can promote and maintain mental well-being in different settings. However, findings from these studies revealed that these robotic coaches are not ready to be used and deployed in real-world settings due to several limitations that span from technological challenges to coaching success. To overcome these challenges, this paper presents VITA, a novel multi-modal LLM-based system that allows robotic coaches to autonomously adapt to the coachee's multi-modal behaviours (facial valence and speech duration) and deliver coaching exercises in order to promote mental well-being in adults. We identified five objectives that correspond to the challenges in the recent literature, and we show how the VITA system addresses these via experimental validations that include one in-lab pilot study (N=4) that enabled us to test different robotic coach configurations (pre-scripted, generic, and adaptive models) and inform its design for using it in the real world, and one real-world study (N=17) conducted in a workplace over 4 weeks. Our results show that: (i) coachees perceived the VITA adaptive and generic configurations more positively than the pre-scripted one, and they felt understood and heard by the adaptive robotic coach, (ii) the VITA adaptive robotic coach kept learning successfully by personalising to each coachee over time and did not detect any interaction ruptures during the coaching, (iii) coachees had significant mental well-being improvements via the VITA-based robotic coach practice. The code for the VITA system is openly available via: https://github.com/Cambridge-AFAR/VITA-system.

Large Language Models (LLMs) have demonstrated significant success across various domains. However, their application in complex decision-making tasks frequently necessitates intricate prompt engineering or fine-tuning, leading to challenges in unseen downstream tasks and heavy demands on computational resources. Meanwhile, Reinforcement Learning (RL) has been recognized as effective in decision-making problems but struggles in environments with sparse rewards, such as open-world games. To overcome these challenges, we introduce AdaRefiner, a novel framework designed to enhance the synergy between LLMs and RL feedback. The key component of AdaRefiner is a lightweight Adapter Language Model (LM), which automatically refines task comprehension based on feedback from RL agents. This method mitigates the need for intricate prompt engineering and intensive LLM fine-tuning while maintaining the LLMs' generalization abilities and enhancing their decision-making capabilities in downstream tasks. Empirical evaluations of AdaRefiner on 22 diverse tasks within the open-world game Crafter have demonstrated its superior effectiveness, especially in guiding agents towards higher-level and common-sense skills. Our work makes contributions to the automatic self-refinement of LLMs with RL feedback, offering a more adaptable and efficient solution for complex decision-making problems.

This paper introduces PowerInfer, a high-speed Large Language Model (LLM) inference engine on a personal computer (PC) equipped with a single consumer-grade GPU. The key underlying the design of PowerInfer is exploiting the high locality inherent in LLM inference, characterized by a power-law distribution in neuron activation. This distribution indicates that a small subset of neurons, termed hot neurons, are consistently activated across inputs, while the majority, cold neurons, vary based on specific inputs. PowerInfer exploits such an insight to design a GPU-CPU hybrid inference engine: hot-activated neurons are preloaded onto the GPU for fast access, while cold-activated neurons are computed on the CPU, thus significantly reducing GPU memory demands and CPU-GPU data transfers. PowerInfer further integrates adaptive predictors and neuron-aware sparse operators, optimizing the efficiency of neuron activation and computational sparsity. Evaluation shows that PowerInfer attains an average token generation rate of 13.20 tokens/s, with a peak of 29.08 tokens/s, across various LLMs (including OPT-175B) on a single NVIDIA RTX 4090 GPU, only 18% lower than that achieved by a top-tier server-grade A100 GPU. This significantly outperforms llama.cpp by up to 11.69x while retaining model accuracy.

In the era of modern healthcare, swiftly generating medical question summaries is crucial for informed and timely patient care. Despite the increasing complexity and volume of medical data, existing studies have focused solely on text-based summarization, neglecting the integration of visual information. Recognizing the untapped potential of combining textual queries with visual representations of medical conditions, we introduce the Multimodal Medical Question Summarization (MMQS) Dataset. This dataset, a major contribution to our work, pairs medical queries with visual aids, facilitating a richer and more nuanced understanding of patient needs. We also propose a framework, utilizing the power of Contrastive Language Image Pretraining(CLIP) and Large Language Models(LLMs), consisting of four modules that identify medical disorders, generate relevant context, filter medical concepts, and craft visually aware summaries. Our comprehensive framework harnesses the power of CLIP, a multimodal foundation model, and various general-purpose LLMs, comprising four main modules: the medical disorder identification module, the relevant context generation module, the context filtration module for distilling relevant medical concepts and knowledge, and finally, a general-purpose LLM to generate visually aware medical question summaries. Leveraging our MMQS dataset, we showcase how visual cues from images enhance the generation of medically nuanced summaries. This multimodal approach not only enhances the decision-making process in healthcare but also fosters a more nuanced understanding of patient queries, laying the groundwork for future research in personalized and responsive medical care

Current approaches to evaluating large language models (LLMs) with pre-existing Knowledge Graphs (KG) mostly ignore the structure of the KG and make arbitrary choices of which part of the graph to evaluate. In this paper, we introduce KGLens, a method to evaluate LLMs by generating natural language questions from a KG in a structure aware manner so that we can characterize its performance on a more aggregated level. KGLens uses a parameterized KG, where each edge is augmented with a beta distribution that guides how to sample edges from the KG for QA testing. As the evaluation proceeds, different edges of the parameterized KG are sampled and assessed appropriately, converging to a more global picture of the performance of the LLMs on the KG as a whole. In our experiments, we construct three domain-specific KGs for knowledge assessment, comprising over 19,000 edges, 700 relations, and 21,000 entities. The results demonstrate that KGLens can not only assess overall performance but also provide topic, temporal, and relation analyses of LLMs. This showcases the adaptability and customizability of KGLens, emphasizing its ability to focus the evaluation based on specific criteria.

Voice plays an important role in our lives by facilitating communication, conveying emotions, and indicating health. Therefore, tracking vocal interactions can provide valuable insight into many aspects of our lives. This paper presents our ongoing efforts to design a new vocal tracking system we call VoCopilot. VoCopilot is an end-to-end system centered around an energy-efficient acoustic hardware and firmware combined with advanced machine learning models. As a result, VoCopilot is able to continuously track conversations, record them, transcribe them, and then extract useful insights from them. By utilizing large language models, VoCopilot ensures the user can extract useful insights from recorded interactions without having to learn complex machine learning techniques. In order to protect the privacy of end users, VoCopilot uses a novel wake-up mechanism that only records conversations of end users. Additionally, all the rest of pipeline can be run on a commodity computer (Mac Mini M2). In this work, we show the effectiveness of VoCopilot in real-world environment for two use cases.

The success of large language models has shifted the evaluation paradigms in natural language processing (NLP). The community's interest has drifted towards comparing NLP models across many tasks, domains, and datasets, often at an extreme scale. This imposes new engineering challenges: efforts in constructing datasets and models have been fragmented, and their formats and interfaces are incompatible. As a result, it often takes extensive (re)implementation efforts to make fair and controlled comparisons at scale. Catwalk aims to address these issues. Catwalk provides a unified interface to a broad range of existing NLP datasets and models, ranging from both canonical supervised training and fine-tuning, to more modern paradigms like in-context learning. Its carefully-designed abstractions allow for easy extensions to many others. Catwalk substantially lowers the barriers to conducting controlled experiments at scale. For example, we finetuned and evaluated over 64 models on over 86 datasets with a single command, without writing any code. Maintained by the AllenNLP team at the Allen Institute for Artificial Intelligence (AI2), Catwalk is an ongoing open-source effort: https://github.com/allenai/catwalk.