AI300
  1. Ben Burtenshaw / Hugging Face Researcher
    1411.
    I'm kinda worried about the dev (software) scene and its relationship to AI/ the basics of ML.›
    LLM score 65 · 4 months ago
  2. Shane Legg / Chief AGI Scientist at DeepMind
    1412.
    To whom do the fruits of advanced technological innovation belong?›
    LLM score 65 · 4 months ago
  3. Sayak Paul / Hugging Face Researcher
    1413.
    If you're attending ICML'26 do drop by our poster on leveraging video priors for image editing!›
    LLM score 15 · 4 months ago
  4. Shuchao Bi / Meta Researcher
    1414.
    It has been wonderful watching the team cook together.›
    LLM score 25 · 4 months ago
  5. Cameron Wolfe / Researcher at Netflix
    1415.
    Agentic RL requires new algorithm modifications.›
    LLM score 78 · 4 months ago
  6. Jerry Tworek / ex OpenAI VP of RL
    1416.
    First step to replacing transformer is appreciating it deeply for how far it was able to carry us
    LLM score 35 · 4 months ago
  7. Andy Jones / Anthropic Research Engineer
    1417.
    my feed's j-space reactions are * 50% 'the authors claim too little about AI consciousness'›
    LLM score 25 · 4 months ago
  8. Mehtaab Sawhney / OpenAI for Science
    1418.
    A unique experience in mathematics is being told that one has carried the "wrong perspective" on one's own work.›
    LLM score 65 · 4 months ago
  9. Ben Burtenshaw / Hugging Face Researcher
    1419.
    BREAKING - Pre-training Attacks! AI labs have been attacking the whole internet, github, public libraries, and training models on them for years!›
    LLM score 25 · 4 months ago
  10. Thomas Wolf / Hugging Face Cofounder
    1420.
    i like this idea of fine-tuning LLMs for efficient reasoning, especially when the intervention remains as non-invasive as possible and the resulting model behaves very similarly to the original checkpoint›
    LLM score 72 · 4 months ago
  11. Lilian Weng / Thinking Machines Cofounder
    1421.
    new post on harness engineering for AI self-improvement: https://t.co/XNCycrAZbM›
    LLM score 72 · 4 months ago
  12. Devansh Pandey / Standard Intelligence Cofounder
    1422.
    replicates with arbitrary prompts in binary afaict :P https://t.co/MAAiSfkZJC https://t.co/W1ZFvty1g4
    LLM score 15 · 4 months ago
  13. Victoria Lin / Thinking Machines Researcher
    1423.
    If you're excited about efficient parallel reasoning 🧠⚡ in LLMs, don't miss the ThreadWeaver paper presentation at #ICML2026.›
    LLM score 35 · 4 months ago
  14. Victoria Lin / Thinking Machines Researcher
    1424.
    If you're excited about efficient parallel reasoning 🧠⚡ in LLMs, don't miss the ThreadWeaver paper presentation at #ICML2026.›
    LLM score 35 · 4 months ago
  15. John Carmack
    1425.
    Memory cost and capacity are significant issues for AI accelerators.›
    LLM score 75 · 4 months ago
  16. Merve Noyan / Hugging Face ML Engineer
    1426.
    OpenClaw + local models ❤️ filter OpenClaw-able models and copy commands on how to serve them conveniently 🦞 https://t.co/BwVeeHXygr
    LLM score 65 · 4 months ago
  17. Sholto Douglas / Researcher at Anthropic
    1427.
    A good time to reflect on https://t.co/xb3Y3CRiGT.›
    LLM score 35 · 4 months ago
  18. Emmanuel Ameisen / Anthropic Interpretability Researcher
    1428.
    We locate a small subset of the representations inside a language model (the J-space), and find that it resembles an internal monologue.›
    LLM score 78 · 4 months ago
  19. Devansh Pandey / Standard Intelligence Cofounder
    1429.
    just asked "what is 100x100" in binary to fable 5, which hit safety classifiers and redirected to opus 4.8, which hit safety classfiers and redirected to sonnet 4.6, which hit safety classifiers and redirected to haiku 4.5, which finally answered 10,000 :P
    LLM score 25 · 4 months ago
  20. Behnam Neyshabur / Anthropic Researcher
    1430.
    At ICML and curious about @mirendil? We have been heads-down building, but from the Mirendil team, my amazing PhD advisor Nati Srebro is there — and he's a great person to chat with about our mission.›
    LLM score 15 · 4 months ago
  21. Sayak Paul / Hugging Face Researcher
    1431.
    Over the past few months, we have worked relentlessly towards making 🤗 Kernels more secure, reliable, and agent-optimized.›
    LLM score 35 · 4 months ago
  22. Thomas Wolf / Hugging Face Cofounder
    1432.
    Fable weekend project: agent collaboration, but make it a tiny civilization 🌇🗺️🏦🏭›
    LLM score 65 · 4 months ago
  23. Cameron Wolfe / Researcher at Netflix
    1433.
    Really interesting work. I find it especially interesting that evaluating w.r.t.›
    LLM score 72 · 4 months ago
  24. Ben Burtenshaw / Hugging Face Researcher
    1434.
    I'm not sold on all the taste chat.›
    LLM score 65 · 4 months ago
  25. Andy Jones / Anthropic Research Engineer
    1435.
    i suspect humans overrate the importance of continual learning because humans don't come with a make-me-generally-smarter lever.›
    LLM score 65 · 4 months ago
  26. Andrew White / Edison Scientific Cofounder
    1436.
    I like taking quotes from ~100 year-old political movements and replacing the political movement's name with "AGI" to see how well they fit.›
    LLM score 15 · 4 months ago
  27. Ronak Malde / DeepMind Researcher
    1437.
    This is a crazy finding by @waterloo_intern , distilling Fable reasoning traces into a 4b elicits mode collapse into one fundamental truth ›
    LLM score 65 · 4 months ago
  28. Ben Burtenshaw / Hugging Face Researcher
    1438.
    live stream next tuesday [7/7] on distillation in post training.›
    LLM score 65 · 4 months ago
  29. Thomas Wolf / Hugging Face Cofounder
    1439.
    this is literally documented in the published Fable 5 System Card https://t.co/hd6pnjBnn9 https://t.co/zGQKA3dfkH
    LLM score 35 · 4 months ago
  30. Merve Noyan / Hugging Face ML Engineer
    1440.
    this was in DeepSeek? https://t.co/pfRVkXvh4k
    LLM score 25 · 4 months ago
More