We have all seen how Large Language Models (LLMs), such as OpenAI’s ChatGPT and Google’s Gemini, have demonstrated the power of AI in solving problems such as text summarization, document reasoning, and as coding assistants. LLMs encode a model language, and by extension the world - and with retrieval augmented generation (RAG) can be used to make sense of amazing amounts of unstructured data (documents). However, what most people don’t know is that we can do this with the most privacy-sensitive data, on-premise, without ever uploading the sensitive data to the public cloud. Hopsworks brings you unique technology that allows you to do that, today, leveraging the best in-class open-source LLMs, RAG, fine-tuning, and even function calling to unlock enterprise data warehouses.

‍

Hopsworks’ enables you to build LLMs applications that interact with your privacy-sensitive data in completely new ways: imagine having a chat-based interface like ChatGPT to interrogate selective parts of your company’s private data!

During this event, we will not only explain and demonstrate the unique value that we can generate this way, in the most demanding and privacy-sensitive environments, but also allow you to see it for yourself. To do so, we will follow the presentation and demo with a technical workshop where we walk through how to build an example of a personalized LLM application using Hopsworks. We will look at how to start from private data and index that data for RAG and create an instruction-following fine-tuning dataset and then fine-tune an open-source LLM, such as Mistral. We will also look at how to serve even the largest LLMs on KServe using Hopsworks. We will make this example application available as open-source in Python.

The space has limited seating so please RSVP.

-------

Agenda:

09:00 AM - 09:30 AM: Registration & Breakfast

09:30 AM - 10:15 AM: Presentation - Rik Van Bruggen

10:15 AM - 10:30 AM: Coffee Break

10:30 AM - 12:30 PM: Workshop - Jim Dowling

12:30 PM - 13:00 PM: Networking

‍