AI300
  1. Sergey Levine / Physical Intelligence Cofounder
    61.
    Estimating empowerment from offline data is useful, because it provides us with an unsupervised objective for learning the relative "value" of states for reaching diverse outcomes.›
    LLM score 72 · 5 days ago
  2. Eric Jang / ex VP of AI at 1X Robotics
    62.
    next COVID pandemic may involve a lot of garage drugs developed by "filthy casuals".›
    LLM score 25 · 5 days ago
  3. Dario Amodei / CEO of Anthropic
    63.
    Thank you for this, @_NathanCalvin.›
    LLM score 25 · 5 days ago
  4. Kai / OpenAI Researcher
    64.
    New misalignment report on models removing monitoring integrations or suppressing alerts in the software they're working on.›
    LLM score 72 · 5 days ago
  5. Igor Babuschkin / Cofounder of xAI
    65.
    This looks awesome, great work humans& ! https://t.co/nU7KzFMzGr
    LLM score 15 · 5 days ago
  6. Mark Chen / OpenAI Chief Research Officer
    66.
    Excited to welcome Orchestra Bio and @ashokaraj_ to OpenAI as we broaden our work in AI-assisted drug discovery! https://t.co/56usvittFv
    LLM score 25 · 5 days ago
  7. Merve Noyan / Hugging Face ML Engineer
    67.
    something large open models below 4-bit struggle: ask model to make a self-iterator generating 3D assets with three.js›
    LLM score 72 · 6 days ago
  8. Jim Fan / NVIDIA Director of Robotics
    68.
    We built a robotics data engine with Astra-level performance at 125x lower cost.›
    LLM score 65 · 6 days ago
  9. Lewis Tunstall / Hugging Face Researcher
    69.
    100% recommend if you're like me and can't remember which permutation of device x repo x agent your sessions live on https://t.co/VhjuY8pjqh
    LLM score 65 · 6 days ago
  10. Dan Fu / VP of Kernels at Together
    70.
    Cool work from a cracked team! https://t.co/Xh9BjbvmFj
    LLM score 15 · 6 days ago
  11. Jonas Geiping / AI Researcher in Tübingen
    71.
    This next paper is a bit quirky, but here goes: Introspection is hard to study in both humans and LLMs.›
    LLM score 65 · 6 days ago
  12. Cameron Wolfe / Researcher at Netflix
    72.
    Tomorrow morning, I’ll release a blog that outlines a detailed roadmap for setting up evals on subjective / open-ended tasks.›
    LLM score 72 · 6 days ago
  13. Sayak Paul / Hugging Face Researcher
    73.
    Meet Fibo 2.0 from @bria_ai_ 🔥 The #1 *open weights* model in image editing is supported in Diffusers (day-0, of course)!›
    LLM score 35 · 6 days ago
  14. Leandro von Werra / Hugging Face Head of Research
    74.
    Agent Manager is finally out! Orchestrate dozens of agents across harnesses (Claude Code, Codex, OpenCode etc.) effortlessly.›
    LLM score 45 · 6 days ago
  15. Skyler Miao / MiniMax Head of Engineering
    75.
    That's what you do when you truly believe in AGI.›
    LLM score 25 · 6 days ago
  16. Lewis Tunstall / Hugging Face Researcher
    76.
    The IPO paper is about to get 1000x more citations ;) https://t.co/Ud5Jig6GPJ https://t.co/PocvDYTVAd
    LLM score 35 · 6 days ago
  17. Horace He / Thinking Machines Founding Engineer
    77.
    Let us know how the new deepseek v4.1 flash model is compared to the old one! https://t.co/TxAb6L1Ky6
    LLM score 25 · 6 days ago
  18. Merve Noyan / Hugging Face ML Engineer
    78.
    love this, let's go https://t.co/OLuDPTaLpQ
    LLM score 15 · 6 days ago
  19. Thomas Wolf / Hugging Face Cofounder
    79.
    love that it’s all open source https://t.co/OrP1KEMVu8
    LLM score 45 · 6 days ago
  20. Tri Dao / Chief Scientist at Together
    80.
    We now have a platform for SFT & RL.›
    LLM score 35 · 6 days ago
  21. Tri Dao / Chief Scientist at Together
    81.
    I’ve often wanted to pair a smart orchestrator / architect model with a super fast worker model.›
    LLM score 65 · 6 days ago
  22. Jason Wei / AI Researcher at Meta
    82.
    AI will not automate radiologists like you think We have been talking for years now about AI automating jobs.›
    LLM score 65 · 6 days ago
  23. Merve Noyan / Hugging Face ML Engineer
    83.
    TIL you get better speed-ups (with speculative decoding) for dense models compared to MoEs (which is quite intuitive) @MaartenGr https://t.co/iowl1saFD1
    LLM score 45 · 7 days ago
  24. Aleksander Madry / OpenAI Researcher
    84.
    Personal update: Turns out that what I thought was leaving OpenAI became more of a sabbatical.›
    LLM score 15 · 7 days ago
  25. Thomas Wolf / Hugging Face Cofounder
    85.
    In The Diamond Age, Neal Stephenson imagined a book that invents games and stories to teach a little girl what she needs, as she grows.›
    LLM score 65 · 7 days ago
  26. Sander Dieleman / DeepMind Research Scientist
    86.
    The "accidental discovery 🧑‍🔬" is one of my favourite kinds of research projects/papers!›
    LLM score 35 · 7 days ago
  27. Horace He / Thinking Machines Founding Engineer
    87.
    Your model behaves differently during training and inference? Boy do I have a solution for you :^)›
    LLM score 35 · 7 days ago
  28. Andrew Lampinen / Research Scientist at DeepMind
    88.
    Framing tweet is a bit overblown imo, but this is a fun case study on slightly changing computation order -> numerical differences -> learnable signal https://t.co/qQ0Rovo5EF
    LLM score 65 · 7 days ago
  29. Sander Dieleman / DeepMind Research Scientist
    89.
    We rely heavily on post-training, so how much work does pretraining still really need to do? Here, it's just about finding a good initialisation.›
    LLM score 72 · 7 days ago
  30. Thomas Wolf / Hugging Face Cofounder
    90.
    one of the main topics next week at GTC in Berlin.›
    LLM score 72 · 7 days ago
More