This material may not be published, broadcast, rewritten, or redistributed. Quotes displayed in real-time or delayed by at least 15 minutes. Market data provided by Factset . Powered and implemented by FactSet Digital Solutions . Mutual Fund and ETF data provided by Refinitiv Lipper .

Attention mechanisms are central to the success of large language models (LLMs), enabling them to capture intricate token dependencies and implicitly assign importance to each token. Recent studies have revealed the sink token, which receives disproportionately high attention despite their limited semantic role. In this paper, we first expand the relationship between the sink token and other tokens, moving beyond attention to explore their similarity in hidden states, considering the layer depth. We observe that as the layers get deeper, the cosine similarity between the normalized hidden states of the sink token and those of other tokens increases, and that the normalized hidden states of the sink token exhibit negligible changes. These imply that other tokens consistently are directed toward the sink token throughout the layers. Next, we propose a dynamic token selection method, called OrthoRank, using these findings to select important tokens. Specifically, in a certain layer, we define token importance by the speed at which the token moves toward the sink token. This is converted into orthogonality with the sink token, meaning that tokens that are more orthogonal to the sink token are assigned greater importance. Finally, through extensive experiments, we demonstrated that our method results in lower perplexity and higher zero-shot accuracy compared to layer pruning methods at the same sparsity ratio with comparable throughput, while also achieving superior performance on LongBench.

In arbitrary-order language models, it is an open question how to sample tokens in parallel from the correct joint distribution. With discrete diffusion models, the more tokens they generate in parallel, the less their predicted distributions adhere to the originally learned data distribution, as they rely on a conditional independence assumption that only works with infinitesimally small timesteps. We find that a different class of models, any-subset autoregressive models (AS-ARMs), holds the solution. As implied by the name, AS-ARMs can generate tokens in any order, and in parallel. Moreover, AS-ARMs support parallelized joint probability density estimation, allowing them to correct their own parallel-generated token distributions, via our Any-Subset Speculative Decoding (ASSD) algorithm. ASSD provably enables generation of tokens from the correct joint distribution, with the number of neural network calls upper bounded by the number of tokens predicted. We empirically verify that ASSD speeds up language generation, without sacrificing quality. Furthermore, we provide a mathematically justified scheme for training AS-ARMs for generation, and show that AS-ARMs achieve state-of-the-art performance among sub-200M parameter models on infilling benchmark tasks, and nearly match the performance of models 50X larger on code generation. Our theoretical and empirical results indicate that the once-forgotten AS-ARMs are a promising direction of language modeling.

Childhood vaccination is a cornerstone of public health, yet disparities in vaccination coverage persist across England. These disparities are shaped by complex interactions among various factors, including geographic, demographic, socioeconomic, and cultural (GDSC) factors. Previous studies mostly rely on cross-sectional data and traditional statistical approaches that assess individual or limited sets of variables in isolation. Such methods may fall short in capturing the dynamic and multivariate nature of vaccine uptake. In this paper, we conducted a longitudinal machine learning analysis of childhood vaccination coverage across 150 districts in England from 2021 to 2024. Using vaccination data from NHS records, we applied hierarchical clustering to group districts by vaccination coverage into low- and high-coverage clusters. A CatBoost classifier was then trained to predict districts' vaccination clusters using their GDSC data. Finally, the SHapley Additive exPlanations (SHAP) method was used to interpret the predictors' importance. The classifier achieved high accuracies of 92.1, 90.6, and 86.3 in predicting districts' vaccination clusters for the years 2021-2022, 2022-2023, and 2023-2024, respectively. SHAP revealed that geographic, cultural, and demographic variables, particularly rurality, English language proficiency, the percentage of foreign-born residents, and ethnic composition, were the most influential predictors of vaccination coverage, whereas socioeconomic variables, such as deprivation and employment, consistently showed lower importance, especially in 2023-2024. Surprisingly, rural districts were significantly more likely to have higher vaccination rates. Additionally, districts with lower vaccination coverage had higher populations whose first language was not English, who were born outside the UK, or who were from ethnic minority groups.

We introduce a new dynamic vocabulary for language models. It can involve arbitrary text spans during generation. These text spans act as basic generation bricks, akin to tokens in the traditional static vocabularies. We show that, the ability to generate multi-tokens atomically improve both generation quality and efficiency (compared to the standard language model, the MAUVE metric is increased by 25%, the latency is decreased by 20%). The dynamic vocabulary can be deployed in a plug-and-play way, thus is attractive for various downstream applications. For example, we demonstrate that dynamic vocabulary can be applied to different domains in a training-free manner. It also helps to generate reliable citations in question answering tasks (substantially enhancing citation results without compromising answer accuracy).

Data stream forecasts are essential inputs for decision making at digital platforms. Machine learning algorithms are appealing candidates to produce such forecasts. Yet, digital platforms require a large-scale forecast framework that can flexibly respond to sudden performance drops. Re-training ML algorithms at the same speed as new data batches enter is usually computationally too costly. On the other hand, infrequent re-training requires specifying the re-training frequency and typically comes with a severe cost of forecast deterioration. To ensure accurate and stable forecasts, we propose a simple data-driven monitoring procedure to answer the question when the ML algorithm should be re-trained. Instead of investigating instability of the data streams, we test if the incoming streaming forecast loss batch differs from a well-defined reference batch. Using a novel dataset constituting 15-min frequency data streams from an on-demand logistics platform operating in London, we apply the monitoring procedure to popular ML algorithms including random forest, XGBoost and lasso. We show that monitor-based re-training produces accurate forecasts compared to viable benchmarks while preserving computational feasibility. Moreover, the choice of monitoring procedure is more important than the choice of ML algorithm, thereby permitting practitioners to combine the proposed monitoring procedure with one's favorite forecasting algorithm.

Article comprehension is an important challenge in natural language processing with many applications such as article generation or image-to-article retrieval. Prior work typically encodes all tokens in articles uniformly using pretrained language models. However, in many applications, such as understanding news stories, these articles are based on real-world events and may reference many named entities that are difficult to accurately recognize and predict by language models. To address this challenge, we propose an ENtity-aware article GeneratIoN and rEtrieval (ENGINE) framework, to explicitly incorporate named entities into language models. ENGINE has two main components: a named-entity extraction module to extract named entities from both metadata and embedded images associated with articles, and an entity-aware mechanism that enhances the model's ability to recognize and predict entity names. We conducted experiments on three public datasets: GoodNews, VisualNews, and WikiText, where our results demonstrate that our model can boost both article generation and article retrieval performance, with a 4-5 perplexity improvement in article generation and a 3-4% boost in recall@1 in article retrieval. We release our implementation at https://github.com/Zhongping-Zhang/ENGINE .

Two brothers whose grandmother was left unaided for hours after a fall, have developed AI technology to help local authorities reach hundreds of shielding, isolated or digitally excluded residents. Monty and Hector Alexander are working with Hammersmith and Fulham council to pilot their automated voice call system that phones households every fortnight to ask whether they need help during the Covid-19 pandemic. At the ages of 24 and 26, the brothers said they decided to harness their tech expertise to find solutions to social issues rather than joining a big corporate machine or tech giant. The brothers, who live in White City, founded the start-up Yokeru while Monty was still studying mechanical engineering at Imperial College London last year in an attempt to "improve communication between vulnerable people and caregivers". Much of their motivations were personal after their grandmother was left stranded outside her residential home for eight hours after falling over, even with the home's 24/7 care.

Babylon, a UK start-up, plans to "put an accessible and affordable health service in the hands of every person on earth" by putting artificial intelligence (AI) tools to work. Currently, the company has operations in the UK and Rwanda and hopes to expand to the Middle East, the United States, and China. The company's strategy is to combine the power of AI with the medical expertise of humans to deliver unparalleled access to healthcare. The Amazing Ways Babylon Health Is Using Artificial Intelligence To Make Healthcare Universally ... [ ] Accessible Babylon's engineers, doctors, and scientists developed an AI system that can receive data about the symptoms someone is suffering from, compare the information to a database of known conditions and illnesses to find possible matches, and then identify a course of action and related risk factors. People can use the "Ask Babylon" feature to inquire about their medical concerns to get an initial understanding of what they might be dealing with, but this service is not intended to replace the expertise of a doctor or be used in a medical emergency.

Boris Johnson's most senior aide, Dominic Cummings, is facing conflict of interest accusations over a consultancy role he undertook for a government-endorsed healthcare startup that is in position to receive a share of a new £250m flagship public fund. Cummings advised Babylon Health, a controversial artificial intelligence (AI) firm working within the NHS, on its communications strategy and its senior recruitment, an investigation by the Guardian and the Bureau of Investigative Journalism can reveal. A GP app developed by the company was later backed publicly on multiple occasions by the health secretary, Matt Hancock. The former Vote Leave campaign director's formal role with Babylon concluded in July last year but he continued to advise the firm about recruitment until September 2018 – the same month Hancock visited the firm and told staff he wanted the NHS to help the company expand. In August this year, shortly after Boris Johnson entered No 10 with Cummings as his top adviser, Downing Street and the Department of Health announced a £250m fund to boost the use of AI in the NHS by using automated systems for diagnoses or data analysis.