LLMOps Day: Bringing Practitioners, Researchers, and Students, Together
On April 26th 2024 in Polytechnique Montreal
aaa
SEMLA presents: LLMOps Day
The event will cover a variety of topics related to training, deploying, and using LLMs as part of building applications. We will focus on the challenges faced by application developers and data scientists. We will delve into the specific of use cases and the current and future infrastructure needs.
As we have done in our prior edition, we will step back and try to grasp the concepts involved in the emerging practices. What challenges are seen as an old problem within a new context? What makes prior solutions not work? What brand new concepts and techniques will be born as a result?
Last but not least, we will hold a panel discussing the future of LLM models in enterprise applications. Our goal is to understand future trends, e.g., small specialized models vs. large more general models, need for fine-tuning, how to building trust and reliability in solutions.
Stay tuned!
Speakers
Dayi Lin
Staff Researcher - Huawei Centre for Software Excellence
Essam Mansour
Assistant Professor - Concordia University
Heng Li
Assistant Professor - Polytechnique Montreal
Ian Arawjo
Assistant Professor - Université de Montréal
Jenn Gamble
Founding Head of Data Science - Distyl AI
Josh McGrath
Founding Member of Technical Staff - DatologyAI
Orlando E Marquez
NLP Staff Applied Research Scientist - ServiceNow
Sarath Chandar
Assistant Professor - Polytechnique de Montréal
Organizers
Amine Mhedhbi
Assistant Professor - Polytechnique Montreal
Foutse Khomh
Full Professor - Polytechnique Montreal
Houssem Ben Braiek
Lead ML Eng. - Sycodal
Maxime Lamothe
Assistant Professor - Polytechnique Montreal
Schedule
| Time | Topic | Event |
|---|---|---|
| 9:00 - 9:15 | MLOps Day Opening | A Welcome Introduction |
| 9:15 - 10:15 | Keynote | Implementing LLMs in Enterprise Systems: Practical Strategies and Future Directions – Jenn Gamble |
| 10:15 - 10:30 | Break | Morning Coffee |
| 1st Theme –> | LLM Fundamentals & Evaluations | |
| 10:30 - 11:00 | LLM Limitations | Limitations of Large Language Models – Sarath Chandar |
| 11:00 - 11:30 | LLM Evaluations | Lessons from ChainForge, a Visual Tool for Prompt Engineering and LLM Hypothesis Testing – Ian Arawjo |
| 11:30 - 12:00 | Data Monitoring | Quality In, Quality Out: Ensuring the Quality of Data Inputs for Intelligent Systems – Heng Li |
| 12:00 - 14:00 | Break | Lunch & Networking |
| 2nd Theme –> | System Building and Deployments | |
| 14:00 - 14:30 | Workflow Building | Shipping Enterprise GenAI: From Research to Production – Orlando E Marquez |
| 14:30 - 15:00 | Data Management | ChatGPT versus Traditional Question Answering for Knowledge Graphs – Essam Mansour |
| 15:00 - 15:30 | Software Engineering | Building enterprise-grade Foundation Model applications – Dayi Lin |
| 15:30 - 15:45 | Break | Evening Coffee |
| 3rd Theme –> | LLM Training & Future of LLMs in the Enterprise | |
| 15:45 - 16:15 | Dataset Curation | Data curation at the scale of pretraining – John Mcgrath |
| 16:15 - 17:00 | LLMs - Agents - Infra | Panel on future of LLM usage within enterprise Panelists: Jenn Gamble, Josh McGrath, Orlando Marquez, Dayi Lin, and Ahmed E. Hassan |
Location
Polytechnique Montréal - Lecture Room (M-1010)
- 2500 Chem. de Polytechnique, Montréal, QC H3T 1J4
- 5 minutes walk from Université-de-Montréal Metro station
