• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
AI
Report

Trillium plans to open-source reinforcement-learning environments

a16z says Nathan Lambert wants to give away environments others would sell and show where AI models get stuck.

Nathan LambertNL
a16zA1
LishaLI
3 Sources, 1h ago, first seen 1h ago

TLDR

a16z says Nathan Lambert is launching Trillium as a nonprofit open AI lab and wants to give away reinforcement-learning environments others would sell. He also wants to show how his team uses AI to build AI, including where models get stuck. An interviewer says labs have reason to keep those environments private, but PhD students need them to study post-training.

Combined views

13K

3 Sources, first seen 1h ago

132 likes21 comments43 saves10 reposts

Combined views

13K

3 Sources, first seen 1h ago

132 likes21 comments43 saves10 reposts
Image: @a16z on X

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Featured Source

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

3 Sources

Lisha@lishali88Open souring more RL environments to demystify post training! @natolambert on why Trillium is releasing them: labs have every reason to keep them private, but they're exactly the infrastructure PhD students need to study.1h
Nathan Lambert@natolambertVery timely appearance on the a16z podcast with @lishali88, to try to understand how open models play into RSI dynamics, who captures value of open models, wrangling agent swarms, and why I'm building Trillium Labs.1h
a16z@a16zNathan Lambert of @trillium_labs says Chinese AI researchers' jaws drop to the floor when he tells them how much compute OpenAI has: "The Chinese labs are very technically excellent. They are more compute-starved." "Without a doubt, they have a comparable talent pool to draw on... To say that China is anything but equal on talent, you're gonna have a hard time because they just have so many people being trained at such a high level." "What they don't have is compute." "You talk to people there, and you're like, 'Oh, I heard an average OpenAI researcher has five thousand H100 equivalents that they can access at any time,' and like their jaw just falls to the floor. They're just like, 'I can't comprehend.'" @natolambert @lishali8856m
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI

    3 Sources

    Lisha@lishali88Open souring more RL environments to demystify post training! @natolambert on why Trillium is releasing them: labs have every reason to keep them private, but they're exactly the infrastructure PhD students need to study.1h
    Nathan Lambert@natolambertVery timely appearance on the a16z podcast with @lishali88, to try to understand how open models play into RSI dynamics, who captures value of open models, wrangling agent swarms, and why I'm building Trillium Labs.1h
    a16z@a16zNathan Lambert of @trillium_labs says Chinese AI researchers' jaws drop to the floor when he tells them how much compute OpenAI has: "The Chinese labs are very technically excellent. They are more compute-starved." "Without a doubt, they have a comparable talent pool to draw on... To say that China is anything but equal on talent, you're gonna have a hard time because they just have so many people being trained at such a high level." "What they don't have is compute." "You talk to people there, and you're like, 'Oh, I heard an average OpenAI researcher has five thousand H100 equivalents that they can access at any time,' and like their jaw just falls to the floor. They're just like, 'I can't comprehend.'" @natolambert @lishali8856m
    Today's Rank

    #12

    Today's Rank

    #12