President & Co-Founder @OpenAI
Joined July 2010
- Tweets 4,815
- Following 65
- Followers 497,251
- Likes 1,386
Longest-ever episode of the web series Two Minute Papers — clocking in at 11 min 30 seconds, despite the name — on OpenAI Five's victory over the Dota 2 world champions: youtube.com/watch?v=tfb6aEUM…
3
10
1
86
Winter is coming again — someone fine-tuned GPT-2 to generate alternate Game of Thrones endings. buff.ly/30HutaE
7
68
5
220
Very impressed by the work done by our second class of OpenAI Scholars! Check out their projects here: openai.com/blog/openai-schol…
4
11
59
"Starting out in a new field is a bit like learning a new language. At first everything feels foreign and overwhelming, and it's important to press on through that feeling." — @mcleavey's advice on getting started in deep learning resonates with me!
deeplearning.ai/blog/deep-le…
2
21
3
123
Surprisingly mesmerizing: nitter.vloup.ch/SamBouiss/status…. Cool to see PPO solve a game that's simple enough to anticipate what move it should make at each step.
2
3
40
The @BBCPhilharmonic will soon be playing a piece composed in collaboration with MuseNet
Thrilled to work with the talented and creative @Robert_Laidlow setting up #MuseNet as a part of the composition process. Can't wait to hear the performance!
1
20
1
87
Congrats @gralston! YC is in great hands.
I'm delighted @gralston is taking over as President of YC! He is YC at its best, and I'm excited to see where he takes things. blog.ycombinator.com/geoff-r…
1
3
The program starts with teaching deep learning from the beginning. So usually better to apply for a full-time role if you're already doing deep learning! (We'll consider applications from people doing other forms of ML though.)
3
Our Fellows program trains people from other fields to make world-class machine learning contributions within 6 months. Kinda crazy that this is possible: nitter.vloup.ch/OpenAI/status/11…
Applications are open for the next class of Fellows. Acceptances on a rolling basis, so apply today!
In just 6 months, our Fellows went from beginner to producing world-class ML contributions. Their accomplishments: MuseNet, analysis of bias/variance of gradients, implementations of very large scaling transformers, and learning adaptive weight sparsity: openai.com/blog/openai-fello…
4
9
85
Great podcast from @mcleavey on MuseNet and AI ethics: towardsdatascience.com/opena…
2
13
63
Very encouraging to see others engaging with the ethics of their NLP releases:
This release was a great occasion to discuss our ethics and values: open-source, education & community. We wrote a blog post discussing it.
Tutorial: medium.com/huggingface/how-t…
Code: github.com/huggingface/trans…
Demo: convai.huggingface.co
Ethics & values: medium.com/huggingface/ethic…
2
20
Please do! Many people currently working at OpenAI applied multiple times before getting an offer. One advantage is we get to see your growth over time.
1
9
The Scholars program is really inspiring — always amazing to see what people coming from other fields can accomplish in 3 months of training. Come see what's been built:
On Tues 5/14, we'll be hosting Scholars Demo Day. See what's been accomplished in the past 3 months: from intrinsic motivation for robotic tasks to fine-tuning GPT-2 for question answering.
Join us in person! Request an invite: bit.ly/2vKlFlZ
9
51
Looks really cool. Great work!
Yesterday I launched Talk to Transformer, a site where you can try out @OpenAI's new text-generating language model on your own custom text. talktotransformer.com/ One example is below. @AlecRad @gdb
2
8
4
37
Nope! Team's been working super hard on this since GPT-2 launch, and we just reached the point where we felt confident enough to take a next step (with a minor internal push to get it done by ICLR: iclr.cc/).
2