| Tweetovi |
| Ilya Sutskever proslijedio/la je tweet | ||
|
Lilian Weng
@lilianweng
|
30. sij |
|
Humans learn from curriculum since birth. We can learn complicated math problems because we have accumulated enough prior knowledge. This could be true for training a ML/RL model as well. Let see how curriculum can help an RL agent learn: lilianweng.github.io/lil-log/2020/0…
|
||
|
||
| Ilya Sutskever proslijedio/la je tweet | ||
|
Wojciech Zaremba
@woj_zaremba
|
30. sij |
|
OpenAI + PyTorch = ❤
openai.com/blog/openai-py…
|
||
|
|
||
| Ilya Sutskever proslijedio/la je tweet | ||
|
PyTorch
@PyTorch
|
30. sij |
|
Welcome OpenAI to the PyTorch community! twitter.com/OpenAI/status/…
|
||
|
|
||
| Ilya Sutskever proslijedio/la je tweet | ||
|
OpenAI
@OpenAI
|
30. sij |
|
We're standardizing OpenAI's deep learning framework on PyTorch to increase our research productivity at scale on GPUs (and have just released a PyTorch version of Spinning Up in Deep RL): openai.com/blog/openai-py… pic.twitter.com/lgvqDdWDoB
|
||
|
|
||
|
Ilya Sutskever
@ilyasut
|
17. sij |
|
arxiv.org/abs/2001.04413
Cool theory paper presenting a problem that:
- can be efficiently learned by SGD with a DenseNet with x^2 nonlin,
- cannot be efficiently learned by any kernel method, including NTK.
|
||
|
|
||
| Ilya Sutskever proslijedio/la je tweet | ||
|
Elliot Turner
@eturner303
|
12. sij |
|
The funny thing about AIDungeon is due to all the GPT-2 pre-training on swaths of random web text, you can create an adventure game about literally anything. Here's one about installing OpenCV from source: pic.twitter.com/zJvtjBwcbq
|
||
|
|
||
| Ilya Sutskever proslijedio/la je tweet | ||
|
Greg Brockman
@gdb
|
9. sij |
|
Watched "The Squire — AI Written Short Film".
It's a live-action adaption of AI Dungeon gameplay. Super fun to see!
youtube.com/watch?v=Kx-2Py… pic.twitter.com/RZVLXMcYA8
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
6. sij |
|
Super excited to work with you, @jeffclune!!! twitter.com/jeffclune/stat…
|
||
|
|
||
| Ilya Sutskever proslijedio/la je tweet | ||
|
|
Greg Brockman
@gdb
|
30. pro |
|
I agree with this article that AI Dungeon 2 (the GPT-2 powered text-based game of infinite possibilities) is "one of the coolest video game experiments of 2019". You should definitely check it out: play.aidungeon.io. Great work @nickwalton00! twitter.com/verge/status/1…
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
3. sij |
|
Seconded! It'll be a truly great talk. twitter.com/gdb/status/121…
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
23. pro |
|
This is amazing twitter.com/ak92501/status…
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
15. pro |
|
an amazing game decisionproblem.com/paperclips/
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
14. pro |
|
Thanks Nando!
|
||
|
|
||
| Ilya Sutskever proslijedio/la je tweet | ||
|
Dota 2
@dota2updates
|
13. pro |
|
Dota 2 with Large Scale Deep Reinforcement Learning
cdn.openai.com/dota-2.pdf
Via @OpenAI
#Dota2 pic.twitter.com/yT0GLiTv56
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
13. pro |
|
The DotA paper is out!!! twitter.com/OpenAI/status/…
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
9. pro |
|
Cool short video on the history of AI: youtube.com/watch?v=IBe2o-…
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
6. pro |
|
Stop attacking junior researchers asking for feedback.
twitter.com/zacharylipton/…
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
5. pro |
|
If ML is applied to ML training, ML training should become faster and easier. Join us to help make it happen: twitter.com/woj_zaremba/st…
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
5. pro |
|
The double descent phenomenon occurs in practical deep learning settings. Interesting that it went unnoticed for so long! twitter.com/OpenAI/status/…
|
||
|
|
||
|
|
Ilya Sutskever
@ilyasut
|
3. pro |
|
Our contribution to increasing the entropy of RL environments: twitter.com/OpenAI/status/…
|
||
|
|
||