hackertube.io

John Schulman - Reinforcement Learning from Human Feedback: Progress and Challenges

UC Berkeley EECS · talk · Watch on YouTube

Machine learning & deep learningLLMs & generative AI

What it is

UC Berkeley EECS colloquium by OpenAI's John Schulman on reinforcement learning from human feedback. He covers why RL helps with truthfulness, hallucination, and the limits of behavior cloning.

Why Hacker News recommends it

Recommended in LLM threads as an authoritative explanation from a ChatGPT architect of why models hallucinate and why imitation is not enough.

  • 2 people recommended it
  • 2 different months
  • 0 strong recommendations
  • first 2023, last 2023

What HN commenters said

  1. “Sure, here you go”

    Schulman on truthfulness, uncertainty, and why models hallucinate.

    mike_hearn, 2023-04 · Recommendation · in “It's not only you, there's an explosion of ChatGPT on HN”

  2. “There's a very good talk by John Schulman, chief architect of ChatGPT”

    Schulman explains RL's role in reducing hallucination.

    KKKKkkkk1, 2023-05 · Recommendation · in “Don Knuth Plays with ChatGPT”

Excerpts from Hacker News comments, linked to the originals. Summaries and labels were written by a language model from those comments.

Topics: rlhf, chatgpt, hallucination, behavior cloning, openai

Also recommended by the same people