Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Spinning Up in Deep RL

spinningup.openai.com

Post reply on HN
Print view View original

11–20 of 53 posts

Re: Spinning Up in Deep RL

#11

Post by slavik81 »

Findeton wrote: post #5
colordrops wrote: post #3

Earlier quoted context omitted.

I assume this comment is generated? The link is a standard Sphinx doc.

And now we're going to need a turing test captcha.

I'd be more worried about bots gaming the voting. I'm perfectly happy to share this website with intelligent machines if they make insightful comments.

Whether odomojuli's post was written by a human, robot or dog is rather immaterial. It's the content of the post that makes it good or bad. It can be evaluated without knowing the author.

Re: Spinning Up in Deep RL

#12

Post by svalorzen »

If you are ever interested in the topic of RL, but wish to start learning the concepts on simpler algorithms and keep the "deep" part for later, I maintain a library that has most of the same design goals:

https://github.com/Svalorzen/AI-Toolbox

Each algorithm is extensively commented, self-contained (aside from general utilities), and the interfaces are as similar as I could make them be. One of my goals is specifically to help people try out simple algorithms so they can inspect and understand what is happening, before trying out more powerful but less transparent algorithms.

I'd be happy to receive feedback on accessibility, presentation, docs or even more algorithms that you'd like to see implemented (or even general questions on how things work).

Re: Spinning Up in Deep RL

#13

Post by janhenr »

I enormously appreciate the resources OpenAI provides to start out in DRL such as this one. However, OpenAI has (purposely?) left out the brittleness of their algorithms to parameter choice and code-level optimizations [1] in the past. As a researcher myself, I would be more than surprised to hear that OpenAI did not explore this behaviour themselves. Instead, my guess would be that these "inconveniences" would do harm to the Marketing of OpenAI and its algos. Such deeds are far more harmful to proper understanding of DRL and applications than a nice UI is beneficial imo.

[1]https://gradientscience.org/policy_gradients_pt1/

Re: Spinning Up in Deep RL

#14

Post by odyssey7 »

slavik81 wrote: post #11
Findeton wrote: post #5

Earlier quoted context omitted.

And now we're going to need a turing test captcha.

I'd be more worried about bots gaming the voting. I'm perfectly happy to share this website with intelligent machines if they make insightful comments. Whether odomojuli's post was written by a human, robot or dog is rather immaterial. It's the content of the post that makes it good or bad. It can be evaluated without knowing the author.

Perhaps this will keep us on our toes. I tend to find something more believable if it's presented in nice prose (the AI's forte) -- which is really not how I should be evaluating information.

Re: Spinning Up in Deep RL

#16

Post by BiasRegularizer »

4k05 wrote: post #9
ajfjrbfbf wrote: post #6

Earlier quoted context omitted.

> feel comfortable saying that the biggest obstacle in progress for AI is UI Yep.

Do you have a source for this? Not nitpicking, i actually need it for a project...

I believe he's referring to the absurdity of this statement, especially in relation to SpinningUp (it doesn't have any UI as it's a RL library, and the docs site are generated using standard Sphinx doc generator)

Re: Spinning Up in Deep RL

#17

Post by 317070 »

Davidzheng wrote: post #8
colordrops wrote: post #3

Earlier quoted context omitted.

I assume this comment is generated? The link is a standard Sphinx doc.

If that comment is generated, I will quit my current job and work full time on AI. I don't believe it.

All the blogs posted by e.g. this user [0] were generated by GPT-3. [1] Some of those reached the top of HN.

That comment indeed looks a lot like it is generated. It has correlated a bunch of words, but it did not understand that the link between UI and AI is tenuous. It is probably one of the few comments where it is so glaringly obvious. There are likely a lot more comments around which are generated but which went unnoticed.

This comment is not generated, as the links below are dated after the GPT-3 dataset was scraped.

[0] https://news.ycombinator.com/submitted?id=adolos

[1] https://adolos.substack.com/p/what-i-would-do-with-gpt-3-if-...

Re: Spinning Up in Deep RL

#18

Post by GnarlyWhale »

Plug for the RL specialization out of the University of Alberta, hosted on coursera: https://www.coursera.org/specializations/reinforcement-learn... All courses in the specialization are free to audit.

For those unaware, the university of Alberta is Rich Sutton's home institution, and he approves of and promotes the course.

Re: Spinning Up in Deep RL

#19

Post by lazzlazzlazz »

Davidzheng wrote: post #8
colordrops wrote: post #3

Earlier quoted context omitted.

I assume this comment is generated? The link is a standard Sphinx doc.

If that comment is generated, I will quit my current job and work full time on AI. I don't believe it.

It's definitely generated. But excellent nonetheless.

Re: Spinning Up in Deep RL

#20

Post by yamrzou »

colordrops wrote: post #3
_5659 wrote: post #2

"Pray, who is the candidate's tailor?" -Hilbert Who is responsible for OpenAI's UI/UX design? It is immaculate and should be the standard for the community. I'm always dazzled by the impeccable standards of OpenAI with regards to tone, presentation, accessibility. The documentation is both familiar but distinct, an impressive achievement! I have my own personal qualms on OpenAI's ethics and virtues but am nevertheles…

I assume this comment is generated? The link is a standard Sphinx doc.

An AI having their own personal qualms on OpenAI's ethics and virtues? I doubt it. It would be hilarious if this was generated.
Post reply on HN