London Large Language Models

In this inaugural session, we shall discuss important subjects pertaining to Large Language Models and Vector Databases, as well as the innovative applications constructed with these advanced technologies.

Agenda:

Speaker: Nikolay Donets (Revolut)

• Title: Large Language Models in Regulated Industries: Opportunities and Challenges
• Description. In regulated industries, we’ll discuss how large language models can be used and the opportunities they bring. These models can make a big difference in areas such as finance. However, there are challenges to using them, such as keeping information private, ensuring security, avoiding bias and complying with regulations. By sharing practical examples and lessons learned, this talk aims to help everyone understand how large language models can be used effectively, while addressing the specific difficulties faced by regulated industries.

Speaker: Trung Huynh (Actable AI)

• Title: Practical aspects of using Large Language Models
• Description:

1. The latest techniques for fine-tuning such as QLoRA and Unlimiformer for long-context
2. Improving Contextual Retrieval
   Strategies for incorporating more information into the model's context to enhance retrieval capabilities.
3. Instruction Compliance and Output Formatting
   Methods for achieving better compliance from LLMs and directing them to produce outputs in a specific format.
4. Generating Citations in Q&A Sessions
   Techniques for prompting LLMs to produce accurate citations in a question-and-answer setting.

Speaker: Siddharth Sharma
Title: How to easily build personalised chat experiences?
Description:

• Hallucinations, how to avoid them?
• Context/Memory - what is it?
• What are the different types of memory?
• Showcase PeachDB: Infra for building chatbots that can truly do back-and-forth and learn user preferences over time

Speaker: Ivan Liagushkin
Title: User first: what LLM applications need to success
Description:
- What it takes to build an AI writing assistant today
- How the architecture of such a system evolves over time
- How to handle challenges that LLMs bring
- How good old software engineering can help in AI-based projects
- How to succeed in this very competitive field
------------------------------------------------------------------------------

To facilitate a seamless check-in process on the day of the event, the venue's Information Security team requires some information from all attendees. This standard procedure ensures the security and safety of all participants, and your cooperation is greatly appreciated.

Please message me directly with the following details:

Full Name
Email Address
LinkedIn URL (if applicable)

Due to some messaging restrictions on Meetup, you can send your details directly to my email address, [fedor.shabashev@gmail.com](mailto:fedor.shabashev@gmail.com), or via WhatsApp/Telegram at +44 7565655455.

Your information will be used solely for this security check. If there happen to be any issues with your entry, we will inform you promptly.

Thank you for your understanding and cooperation. We look forward to seeing you at the meetup!

Best regards,
Fedor Shabashev