i love the openai team so much
Ilya Kostrikov
119 posts
Ilya Kostrikov
@ikostrikov
Pre-Training Researcher @OpenAI, previously: RL/post-training @OpenAI, Postdoc at UC Berkeley @berkeley_ai, Researcher at Google Brain, PhD in CS @CILVRatNYU
- After spending 1.5 years as a postdoc at UC Berkeley, I have recently started working at @OpenAI, specifically in the ChatGPT team's RL subteam. I am thrilled to be a part of this team, and I can't wait to dive into the exciting projects and challenges!
- OpenAI is nothing without its people
- The era of post training has begun.Yall heard it from the man himself
- I’m excited to release my implementations of RL algorithms in Jax: github.com/ikostrikov/jax…. On my machine with Titan X (Pascal), it works more than twice faster (4x faster in some cases) than similar PyTorch implementations. Thanks to @nikishin_evg for helping with Jax!
- I'm excited to share my Jax implementation of SAC from pixels + image augmentations from DrQ (github.com/ikostrikov/jax…, see train_pixels.py). This Jax version is almost twice faster than our original implementation in PyTorch.
- Highly recommend! I had a blast working with Pavel at OpenAI, and NYU is a great place to do PhD.I am recruiting Ph.D. students for my new lab at @nyuniversity! Please apply, if you want to work with me on reasoning, reinforcement learning, understanding generalization and AI for science. Details on my website: izmailovpavel.github.io. Please spread the word!
- Excited to present our work with @ashvinair and @svlevine, Offline RL with Implicit Q-Learning (IQL), a simple method that achieves SOTA performance on D4RL arxiv.org/abs/2110.06169 and works 4x faster than prior SOTA github.com/ikostrikov/imp… Thread below
- Back to the top 🚀🚀🚀🚀🚀Exciting News from Chatbot Arena❤️🔥 Over the past week, the latest @OpenAI ChatGPT-4o (20241120) competed anonymously as "anonymous-chatbot", gathering 8,000+ community votes. The result? OpenAI reclaims the #1 spot, surpassing Gemini-Exp-1114 with an impressive 1361 score!
- When you work on one topic for almost a decade, you get a very particular set of skills.
- I deeply regret my participation in the board's actions. I never intended to harm OpenAI. I love everything we've built together and I will do everything I can to reunite the company.
