04-18, 10:20–10:55 (US/Eastern), Auditorium 5
Reasoning models like OpenAI's o3 and DeepSeek's R1 herald a new paradigm that leverages test-time compute to solve tasks requiring reasoning. These models represent a departure from traditional LLMs, upending long-held assumptions about them. In this session, we will discuss the different dimensions along which test-time compute can be expended and scaled. We will showcase best practices for prompting reasoning models as well as how to direct test-time compute towards achieving desired results. Finally, we will demonstrate how to train our own reasoning models specific to our domain or use case.
The objectives of this session are to:
1. Highlight differences between mainstream LLMs and reasoning models
2. Understand test-time compute and the different dimensions along which they can be scaled.
2. Demonstrate experimental results with reasoning models from DeepSeek and OpenAI
3. Learn how to prompt reasoning models effectively.
4. Showcase how to leverage test time compute at the application level to achieve good results.
Previous knowledge expected
Suhas Pai is a NLP researcher and co-founder/CTO at Hudson Labs, a Toronto based Y-combinator backed startup. He is the author of the book 'Designing Large Language Model Applications', published by O'Reilly Media. He has contributed to the development of several open-source LLMs, including being the co-lead of the Privacy working group at BigScience, as part of the BLOOM LLM project. Suhas is active in the ML community, being Chair of the TMLS (Toronto Machine Learning Summit) conference since 2021. He is also a frequent speaker at AI conferences worldwide, and hosts regular seminars discussing the latest research in the field of NLP.