MLGdańsk #137 – Overview of Reinforcement Learning from Human Feedback

We would like to invite you to 135. MLGdańsk meeting on Monday, 6th March 2023 at 18.00 CET

The meeting will be held online on jitsi::

Our guests speaker is:
Nikita Pavlichenko

The topic of the meeting is:
“Overview of Reinforcement Learning from Human Feedback

Nikita Pavlichenko is a Research Scientist at Toloka, where he works on connecting deep learning models with the crowdsourcing and human evaluation of generative models.

Recent progress on Large Language Models introduced a new machine learning technique: reinforcement learning from human feedback (RLHF). RLHF powers models such as InstructGPT and ChatGPT.

In this talk, we are going to provide an introduction to why this technique is necessary, the basics of reinforcement learning, why RL needs human feedback, and how to solve practical tasks with RLHF.

The meeting is open to all interested – feel free to join!