Accelerating the LLM Life Cycle on the Cloud

Experts from Lightning (creators of PyTorch Lightning, and core contributors to PyTorch) will give an in-depth, no-skip guide on developing and deploying LLMs at enterprise scale.

Join us in-person and virtually as we discuss how to better perform various challenging tasks involved in the LLM lifecycle. We will be discussing how to leverage PyTorch Lightning to train, tune and deploy models.

You will hear from William Falcon, Founder of PyTorch Lightning and CEO of Lightning AI, and Luca Antiga, CTO of Lightning AI.

AGENDA
6 – 7:30 p.m.

Part 1: Building agents with third-party LLMs
We’ll build agents using public APIs from Open AI, Mistral, etc.

Part 2: Build agents with private models
We’ll show how to build the same agents with private models running on your own infrastructure with your own data. We’ll fly through the full model development lifecycle from preparing data to deploying and optimizing a model API.

Step 1: Data preparation at scale

This step will show how to download, process and optimize a massive open-source dataset for training at scale. This is often an overlooked step that can improve model training speeds by at least 20x.

Step 2: Continued pretraining for LLMs

Next, we’ll continue pretraining a model on the dataset we created. You’ll learn how this is done on multi-node across 16 H100 GPUs with the latest tricks for multi-node training with fault-tolerance and more.

Step 3: Finetune an LLM

Once we’ve pretrained our model, we’ll finetune it to align it to answer questions in a way that is tailored to your industry (finance). In this example we’ll bias the model to sound more medical

Step 4: Deploy model API

We’ll deploy our model behind a high-performance API that auto-scales and can be plugged back into the agents.

Step 5: Profile high-performance deployment

We’ll benchmark the API and use our custom built profiling tools (in Studios) to highlight code bottlenecks that can be optimized to increase throughput, lower latency and maximize tokens/s speed.

Step 6: Pipeline

We’ll end with showing how this whole process can be automated with your favorite pipeline/workflow manager and the Studios SDK

Light bites and drink will be provided.

EVENT LOCATION: 84.51°, 433 W Van Buren St #610s, Chicago, IL 60607 "Old Post Office Building" & Virtual

Please note, our speakers will be virtual.

In-person Attendees will be added to the building visitor list. Please check in at the information desk at the main Van Buren lobby. After check-in, take the DS elevator bank to the 6th floor.

Microsoft Teams Info:
Join the meeting now
Meeting ID: 219 882 420 830
Passcode: mYP5sR

ABOUT 84.51° SCIENCE & TECHNOLOGY SOLUTIONS
We will share exciting science and technology use cases around various Engineering, Data Science, and other Technology topics occurring at 84.51°. Our goal is to network and collaborate with our local data science, engineering, and technology professionals. While anyone is welcome to attend, our topics are specific to Engineering and Data Science.

ABOUT 84.51°
84.51°is a retail data science, insights and media company. We help The Kroger Co., consumer packaged goods companies, agencies, publishers and affiliated partners create more personalized and valuable experiences for shoppers across the path to purchase. Powered by cutting edge science, we leverage 1st party retail data from over 62 million U.S. households to fuel a more customer-centric journey utilizing 84.51° Insights, 84.51° Loyalty Marketing and our retail media advertising solution, Kroger Precision Marketing.