@jsuarez

I build sane open-source RL tools. MIT PhD, creator of Neural MMO and founder of PufferAI. DM for business: non-LLM sim engineering, RL R&D, infra & support.

puffer.ai
Joined March 2019
  • Tweets10K
  • Following126
  • Followers31.6K
795Media
Releasing PufferLib 5.0: Train agents in under a second
47
156
30
1,840
172,101
The issue with AI is the people, not the tech. Devs have devoted their lives to a craft. Instead of focusing on how new tools can help them achieve the previously impossible, we have huge swaths of people who have never accomplished anything frothing at the mouth at the prospect of obsoleting people better than them. Directly spurred on by the execs of top "labs" marketing to the lowest of men.
Tim Sweeney
@TimSweeneyEpicOct 17
The games we've long wanted, which didn't exist because they'd cost several times more to make than they'd earn back, will soon become possible. As will the games developers wanted to build, but couldn't with the resources available. It'll be a new golden age of gaming.
35
71
3
898
31,398
As with usual on X, a heck of a lot of people seem mad about something I didn't write! @grok kindly explain the difference between what's actually in this post and how the comments have "read" it
2
13
1,143
My last post on this was serious. It would be awesome to send on release with a group of researchers. I know a disproportionate number of you are end-game OSRS players. Reply below!
Old School RuneScape
@OldSchoolRSOct 16
Only 4 Days until The Fractured Archive is out! 👀 Mods Ayiza and Kieren will be hosting a Watch Party live on Twitch so you can keep up with the progress as we watch to see who will be among the first to clear the raid! 📺 🔗 Check out the blog for more details: osrs.game/TFA-Need-To-Know
1
17
2,345
RLed the roadster with PufferLib
26
47
6
1,112
40,705
Reinforcement Learning with Joseph Suarez xitter.cf/i/broadcasts/1qxvveylv…
0:00

Joseph Suarez 🐡

Reinforcement Learning with Joseph Suarez

1
20
2,165
Joseph Suarez 🐡 retweeted
Ray
@raysan5Oct 15
Hey! **raylib 6.2** is out! 🔥 It was supposed to be a small patch version but it comes with +300 improvements from +100 new contributors! 🚀 Release notes: github.com/raysan5/raylib/re… Windows Installer: raysan5.itch.io/raylib
raylib 6.2 is out! +300 improvements!

ALT raylib 6.2 is out! +300 improvements!

21
58
8
753
13,240
PufferLib's RL Agents beat the Inferno and Colosseum, but how fast can their developers conquer The Fractured Archive? Raids with researchers starts on release, and you can join. Coordinate on our discord.gg/puffer.
2
2
44
4,580
Anyone following my work for a long time will know what a huge influence OSRS has had on NMMO and PufferLib, but this is mostly for fun. We may decide to RL some of the bosses (in sim, not live game) later if they are mechanically interesting enough.
1
14
1,377
Looking for reasonably experienced players for the first few hours (~inferno + quiver and a mega, or more exp less gear). Will open it up more casually depending on how hard it ends up being thereafter. All drops split.
10
780
Joseph Suarez 🐡 retweeted
Samuel Sokota@ssokotaSep 30
In our Nature paper, we introduce the first superhuman Stratego AI, which we built using general techniques that we developed for RL & test-time compute under imperfect information. 1/N
43
187
29
1,421
286,420
Joseph Suarez 🐡 retweeted
Finlay Sanders
@finlay_sandersSep 14

Beating AutoAscend’s NetHack Challenge Score with RL

Article · @finlay_sanders
4
15
6
112
25,824
If you're using raylib professionally, get your company to pitch in a couple hundred bucks a month. Or do it yourself.
Ray
@raysan5Sep 22
How is it possible to get that level of open source support while many other projects like raylib struggle to get enough finance to even pay a developer minimal salary? I don't know... 😞
1
2
89
8,379

You have SKILL_ISSUES.md

Article · @jsuarez
6
11
5
284
20,452

PufferLib 5.0's Streamlined RL Algorithm

Article · @jsuarez
3
15
141
9,650
Two tasks, ~30 minute and ~90 minute horizon respectively. Typically completed by the top 1-3% of players after 10's of hours of attempts. These results changed my understanding of what PufferLib is capable of doing!
valo
@ValtteriValoSep 14
i'd like to keep it low key and say this is no big deal, but in reality this has to be the coolest project i've worked on maybe ever, and there's a lot more to come!

Old School RuneScape in PufferLib 5.0

Article · @ValtteriValo
1
4
116
8,324
Note: we do NOT run any of this on the live game servers! It's all sim. OSRS is my favorite game and I hate seeing the econ + major account achievements ruined by botting
16
935
>Double RL SOTA w/ a very vanilla setup. Probably the first major point of comparison to existing lit for PufferLib! I consider this saturated without tweaking the enchant system. The current impl doesn't really fit an RL benchmark.
Daphne Cornelisse
@daphnesolvesSep 14

Lessons from reaching level 6 in Craftax through tabula rasa on-policy reinforcement learning

Article · @daphnesolves
2
54
6,183
Spencer took point on stable self-play this update. Robocode is an awesome RL benchmark. Super fast to sim, lots of interesting strategies, plenty of hard scripted opponents to beat!
Spencer Cheng
@spencchengSep 14

Selfplay in Games Without a Built In Curriculum

Article · @spenccheng
3
47
4,950
Joseph Suarez 🐡 retweeted
valo
@ValtteriValoSep 14
i'd like to keep it low key and say this is no big deal, but in reality this has to be the coolest project i've worked on maybe ever, and there's a lot more to come!

Old School RuneScape in PufferLib 5.0

Article · @ValtteriValo
10
10
3
81
15,172

Solving Breakout in Under a Second with PufferLib 5.0

Article · @jsuarez
4
6
1
110
5,095
Load more