I'll be available at NeurIPS on Friday to meet up with anyone who would like to talk about OpenAI, deep learning, or the future of the field. Particularly looking to talk with people whose perspective is different from mine. Ping if you'd like to meet up — gdb@openai.com!
3
6
2
89
We're releasing "Dota 2 with Large Scale Deep Reinforcement Learning", a scientific paper analyzing our findings from our 3-year Dota project: openai.com/projects/five/
One highlight — we trained a new agent, Rerun, which has a 98% win rate vs the version that beat @OGEsports.
Dec 13, 2019 · 5:43 PM UTC
2

