Machine Learning Meet-up - Fine-tuning LLMs for open-form relation classification

We are delighted to announce the first Machine Learning Meet Up of 2024 in conjunction with Quantexa, Join us on Monday 29th January at Portal (formerly Tangent) Trinity’s ideas workspace (Directions here: where two speakers from Quantexa will discuss the company’s knowledge graph component for working with very large networks and graphs and another speaker will provide an insight into their NLP pipeline.

Quantexa is a decision intelligence platform for the banking, insurance, and government sectors. In March 2023 Quantexa announced a Series E funding round, becoming the first UK-based “Unicorn” of 2023.
In early 2023, Quantexa acquired Aylien, an Irish startup, adding Aylien’s News Intelligence Platform to the Quantexa product portfolio and bringing the full Aylien team into Quantexa. As part of the acquisition, a new Centre of Excellence in Natural Language Processing was created in Dublin, led by Dr. Chris Hokamp.

The first talk of the evening is titled “Fine-tuning LLMs for open-form relation classification” and will be led by speaker, Shashank Mangla. – A high-level presentation of our pipeline for fine-tuning our own cloud-gapped LLMs for a specific usecase around open-form relation classification. We will also discuss our journey through building various prototypes using generative models for this usecasse.

The second talk is titled “Practical Graph ML for real-world Knowledge Graphs” and will be led by speaker, Aaron Zolnai-Lucas – A look into how open source tools like NetworkX, GraphBLAS & spark can be used to process knowledge graphs at scale in ML pipelines using tricks from linear algebra. We also discuss types of network features, representations and models that yield the best results at scale, including how Graph Neural Networks (GNNs) can be practically deployed.

The final speaker of the evening is by Dr Harshvardhan J. Pandit, an Assistant Professor at the School of Computing in Dublin City University and the ADAPT Centre. His talk is titled, “AI Act: What’s changing in the way we do AI?” His research interests are focused on the application of semantics towards solving real-world challenges associated with privacy, legal and regulatory compliance, and consent. Dr Pandit will be looking at the AI Act what it means and what ‘developers’ need to think about next in terms of regulation and upcoming changes.

The evening will finish off with drinks and pizza. It’s a great opportunity to meet and talk with a range of interesting and informed individuals.