​Join us for an exclusive event tailored for developers to explore open-source LLM models and their synergy with MongoDB. Dive into in-depth sessions led by experts, uncovering LLM applications from NLP to content generation. Discover MongoDB’s pivotal role as a database platform seamlessly integrated with LLM development. Through workshops and demonstrations, learn how MongoDB enhances data storage and retrieval for scalable LLM solutions. Network with peers and industry leaders, whether you’re a seasoned developer or new to LLM and MongoDB. Don’t miss this opportunity to expand your horizons and shape the future of software development.

​Sponsors: MongoDB, Boldcap, Petavue

​Presented by: Petavue & Xylem AI

​Community Partner: AI Brains

​Agenda

​Session 1: Open LLM and Evaluations (09:30 AM to 11:00 AM)

​Session speakers: Pranav and Jeyaraj

• ​Introduction to Language Models

  • ​An overview of the evolution and impact of large language models in the industry

• ​Evaluating NL-SQL

  • ​Techniques and Benchmarks: A comprehensive report.

  • ​Discuss the benchmarking process, focusing on the Bird dataset, including accuracy, latency, model size, and cost metrics.

  • ​Insights into overall learning.

​Break (11:00 AM to 11:30 AM)

​Session 2: Workshop - RAG with MongoDB(11:30 AM to 01:30 PM)

​Session speakers: Jeyaraj

• ​Understanding RAG

  • ​An exploration of what RAG is and its applications.

  • ​Discussion on the effectiveness of RAG: scenarios where it excels and its limitations.

• ​Exploring RAG Techniques

  • ​Dive into various strategies such as HyDE, Re-Ranking, and Self-RAG.

• ​Interactive RAG Workshop

  • ​Participants will be divided into groups to engage in hands-on practice with different RAG techniques using tools like Google Colab, sample datasets, and access to a free MongoDB vector cluster.

  • ​Review of the RAG evaluation report from Petavue.

​Lunch Break (01:30 PM to 02:30 PM)

​Session 3: Workshop - Fine-Tuning Strategies(02:30 PM to 04:30 PM)

​Session Speaker: Pranav

• ​The Essentials of Fine-Tuning

  • ​Why and when fine-tuning is necessary.

  • ​Overview of various techniques, including LoRA and PEFT.

• ​Fine-Tuning Demonstration

  • ​Live demonstration of fine-tuning Mistral 7B model using Colab Pro.

• ​Practical Fine-Tuning Session

  • ​A guided, practical session on fine-tuning with Mistral 7B for NL-to-SQL use case

• ​Data Collection for Fine-Tuning

  • ​Tips on collecting effective training data, including the use of synthetic data generation and logs from GPT/Claude.

​Networking Session (04:30 PM to 05:00 PM)

​An opportunity for attendees to network, discuss the day's learnings and explore potential collaborations.