President & Co-Founder @OpenAI
Joined July 2010
- Tweets 4,815
- Following 65
- Followers 497,251
- Likes 1,386
And our MuseNet concert is live with over 1k viewers! This concert is an experiment (we haven't done any curation / filtering on these songs), so will be very curious to see what people think.
5
9
2
46
Another cool composition from the community:
This work from @mcleavey of @OpenAI has me floored.
Train GPT-2 on a large dataset, and you get an amazing composer that lets anyone create music.
Here is a riff on Poker Face created in 2 minutes.
instaud.io/3B78
Notice the coherence of the composition!
8
Pretty cool MuseNet rendition of "Let It Go": nitter.vloup.ch/PYEEDM/status/11… (composed using the MuseNet co-composer we released today: openai.com/blog/musenet/#try)
2
7
Trained on MIDI! From the post: "MuseNet was not explicitly programmed with our understanding of music, but instead discovered patterns of harmony, rhythm, and style by learning to predict the next token in hundreds of thousands of MIDI files"
1
5
24
Some pretty great sounding AI-generated music, from an unsupervised neural network called MuseNet: nitter.vloup.ch/OpenAI/status/11…
All generated using the same technology as GPT-2.
Come listen to MuseNet play a concert today on livestream from 12p-3p PT: twitch.tv/openai
Introducing MuseNet, a neural network which discovered how to generate music using many different instruments and styles.
Listen & interact: openai.com/blog/musenet/
MuseNet will play an experimental concert today from 12–3pmPT on livestream: twitch.tv/openai
5
10
2
67
Great work to @rewonfc @scottgray76 @AlecRad @ilyasut! Spare Transformer has become a core piece of OpenAI infrastructure, and we'll have some exciting results to share using it in upcoming months.
Releasing some work today with @scottgray76 @AlecRad and @ilyasut. Contains some simple adaptations for Transformers that extend them to long sequences.
1
1
42
Unsupervised learning is starting to work, largely due to the Transformer model (which is what powers e.g. GPT-2). Before: could learn on sequence lengths of a few thousand. Now: can easily learn on tens of thousands.
Releasing the Sparse Transformer, a network which sets records at predicting what comes next in a sequence — whether text, images, or sound. Improvements to neural 'attention' let it extract patterns from sequences 30x longer than possible previously: openai.com/blog/sparse-trans…
3
51
1
246
Welcome :)!
Today I joined @openai as a full-time researcher focusing on AI interpretability following at five month trial. I'm so grateful to be here. Our team is hiring, and if you're interested in helping make AI go well, please DM me (even if you feel unqualified)
1
3
31
Super excited to welcome @mcleavey to the OpenAI team full-time!
First day full time @OpenAI! It's been quite a journey from my 1st @deeplearningai_ class on neural nets, to the wonderful @fastdotai diversity fellowship, to OpenAI's own scholars/fellows program. Thankful for all the support & looking forward to lots more to learn and create.
2
26
There's also an article in Russian about their strategy: cybersport.ru/dota-2/intervi…
Watched many of their games live :) — was very exciting to observe! Here's a big Reddit thread where people collected all weaknesses of Five: old.reddit.com/r/DotA2/comments/…
1
Final OpenAI Five stats:
- Competitive: 7,215 wins, 42 losses (99.4% winrate, 15,019 total players)
- Cooperative: 35,466 games (18,689 total players)
gg all!
venturebeat.com/2019/04/22/o…
5
70
5
312