August : Agents, Experts and Extracting Structured Data

We return this month with an evening to cover a range of topics:

FOOD UPDATE : Google has told us that there WILL BE food at the event - and this time we'll be back to having pizza rather than their "in-house" catering!

Talks:

"LangExtract + New Gemma + Mystery Model" - Sam Witteveen

In this talk, Sam will look at a new library released by the Gemini team called LangExtract, which allows you to do traditional NLP tasks using a large language model and also at some of the recent releases from the Gemma family of models.

"Mixture of Experts Routing" - Florian Kowarsch

This talk will give a overview of MoE (Mixtures of Experts) especially how the routing part needs to be trained and then look at recent publications that explore different approaches without a centralized routing. Florian plans to conclude on a possible future with drop-in-replacement of small expert models.

"Agent Efficiency, Memory and Confidence" - Martin Andrews

In his talk, Martin will discuss the findings in recent research around how to efficiently balance the use of tokens to different tasks when creating an Agent system. He'll also cover some recent research on adding memory, and measuring on-going 'Confidence' in LLM outputs.

---

Talks will start at 7:15pm and end at around 8:45pm, at which point people normally come up to the front for a bit of a chat with each other, and the speakers.

As always, we're actively looking for more speakers - both '30 minutes long-form', and lightning talks. For the lightning talks, we welcome folks to come and talk about something cool they've done with keras, PyTorch, JAX and/or Deep Learning for 5-10mins (so, if you have slides, then #max=10). We believe that the key ingredient for the success of a Lightning Talk is simply the cool/interesting factor. It doesn't matter whether you're an expert or an enthusiastic beginner: Given the responses we have had to previous talks, we're sure there are lots of people who would love to hear what you've been playing with. If you're interested in talking, please just introduce yourself to Martin at one of the events.