Meetup #23 - WiMLDS Poznań

SAVE THE DATE! Our next Meetup is coming in June, 18th. 🚀

This time we will be hosted by Fandom whose Poznań office is located on Baraniaka 6 Street.

We will have two amazing presentations followed by another awesome networking with pizza 🍕. Can't wait to meet you all! 😊

Neon Zhang (Fandom)
ML and GenAI in the Entertainment World - Start from Embeddings
Neon will present her work on the latest Data & Games Hackathon Project at Fandom, an entertainment site that hosts wikis about TV, Movies, and games for fans.
The project aimed to solve the need to go through all the reviews when interested in only a certain aspect of a game e.g. characters. In the project, bite-sized keywords were produced to evaluate games from different perspectives based on user reviews and capture their overall sentiment of those aspects using Embeddings.
The presentation will give a peek at how simple Embeddings can offer significant business value for entertainment sites, and showcase the technologies behind the project (including UAE-Large-V1 and FAISS) as well as the product demo. In addition, it will also introduce the applications of Machine Learning and GenAI in Fandom to highlight their impact on the entertainment world.

Maria Knorps (Tweag/Modus Create)
Evaluating Retrieval in RAGs
Maria will present the work of the GenAI team at Tweag, the innovation arm of Modus Create, where the team developed an evaluation framework for Retrieval-Augmented Generation (RAG) systems. RAG systems provide an easy and low-cost way to extend the knowledge of Large Language Models (LLMs), but measuring their performance is challenging.
The presentation will review existing evaluation frameworks, ranging from those based on the traditional ML approach using ground truth datasets, including Tweag's framework, to those that use LLMs to compute evaluation metrics. It will also delve into the practical implementation of Tweag's chatbot over two distinct document datasets, providing insights on chunking, embedding, and comparing open-source and commercial LLMs.

Agenda:
18:30 Welcome from WiMLDS (10min)
18:40 Evaluating Retrieval in RAGs - Maria Knorps (Tweag/Modus Create) (25min)
19:05 Q&A (10min)
19:15 ML and GenAI in the Entertainment World - Start from Embeddings - Neon Zhang (Fandom) (25min)
19:25 Q&A (10min)
19:35 Closing remarks (5min)
19:40 Networking