Cheese to Coffee: Exploring LLMs and Hierarchical Small World Diagrams

Kickstart your day on the 16th April with an early morning 'before work' shot of technology-focussed talk and chat where we will be hosting Daisy Everard where we will be asking "How far between Cheese and Coffee?" and discussing how LLM's use Heirarchical Navigable Small World Diagrams.

This will be a 'hybrid event' - both in-person at Fothergill House Nottingham, and available to join online too. Those who mark themselves as attending will be sent a link to join the webinar closer to the time if you're unable to join us for breakfast personally. You are welcome to join us at Fothergill House from 7am. The links will also be made available via social media posts so make sure you follow us on LinkedIn and X to keep up to date.

About the talk
Hierarchical Navigable Small Worlds (HNSWs) address the challenge of efficient approximate nearest neighbour searches in high-dimensional data by leveraging a multi-layered graph structure that combines long-range and local connections, enabling fast, scalable, and accurate searches. This is particularly critical for large language models (LLMs), which use HNSWs to efficiently retrieve semantically similar embeddings from vast high-dimensional spaces, ensuring rapid and accurate processing in tasks like text generation, context retrieval, and language understanding for applications such as conversational AI and recommendation systems.

About the speaker
Daisy is a Data Engineer who has been working on an AI driven analytics platform with an interest in the tech and mathematics behind AI systems.