In this talk, we’ll discuss the Reinforcement Fine Tuning (RFT) platform, allowing you to train OpenAI models to reason better on your specific tasks to arrive at better answers. We’ll go over why RFT matters in the new era of Agents, how our customers have succeeded with RFT in their verticals, and how you can achieve success too. We’ll also cover the more theoretical aspects of reinforcement learning and why OpenAI models work uniquely well with RFT, as well as the more practical aspects of how to define effective evals, environments, and graders so that you can deploy better performing models with confidence.
Speaker
Wenjie Zi
Machine Learning @OpenAI, Community Leader Building TAPNET, Previously @Grammarly, 10+ Years of Industrial Experience in Artificial Intelligence Applications
Wenjie Zi holds a Master’s degree in Computer Science from the University of Toronto, specializing in Natural Language Processing (NLP). Currently, she serves as a Member of Technical Staff at OpenAI, bringing over ten years of industrial experience in artificial intelligence applications. Wenjie has successfully implemented and deployed various projects, including Retrieval Augmented Generation (RAG), recommendation systems, semantic parsing (natural language to SQL), and quantitative trading.
Her research has been published in leading conferences and workshops such as ACL, NeurIPS, and KDD. Additionally, Wenjie is a course instructor for the certificate program at the University of Toronto, where she teaches deep learning-related subjects.
In her spare time, Wenjie actively participates in the Canadian AI community, serving as a committee member of MLOps World and as the lead of the Women in AI Canada Sponsorship team. She co-founded the Toronto AI Practitioners Network (TAPNET) in early 2024 and has organized multiple meetups with over 100 participants, aiming to connect technology practitioners across North America.
Speaker
Speaker
Will Hang
