开启时会模糊预览图，关闭后正常显示
For instance, you don't need RL on something you have verifiably true data for that doesn't use any trajectory to get to the outcome for. aka single turn, no reasoning math problems for instance
Cofounder and Head of Post Training @NousResearch, prev @StabilityAI Github: https://t.co/LZwHTUFwPq HuggingFace: https://t.co/sN2FFU8PVE
A brain is a machine for dealing with combinatorial explosions.
Professor of computer science at UW and author of '2040' and 'The Master Algorithm'. Into machine learning, AI, and anything that makes me curious.
I kind of want to write a blog post on some of the constraints, limitations, and things of that nature on RL with LLMs
For instance, you don't need RL on something you have verifiably true data for that doesn't use any trajectory to get to the outcome for. aka single turn, no reasoning math problems for instance
RT @mddanishyusuf: 📤 Maillayer v2.0 is live on @ProductHunt A self-hosted Mailchimp alternative that you buy once, host yourself, and use…
"The Micro Startups Guy" ❯ https://t.co/hwZ0eO0l5D ❯ https://t.co/RkKck3vdIO ❯ https://t.co/PyEJHvxCRn ❯ https://t.co/5hDIulx6OL Sold @nocodeapi for 6 figures
RT @BoredElonMusk: We WILL cancel daylight savings time someday.
FOLLOWS YOU. Artificial Intelligence, Cognitive Architectures, Computation. The goal is integrity, not conformity. https://t.co/rFUNzdYXuK
Socialists don’t understand compounding.
Former Quant Investor, now building @lumera (formerly called Pastel Network) | My Open Source Projects: https://t.co/9qbOCDlaqM