AI300
  1. Ben Burtenshaw / Hugging Face Researcher
    121.
    We are so back! Muse Spark Open is a beast for agentoc stuff, computer use and multimodal reasoning.›
    LLM score 25 · 11 days ago
  2. Merve Noyan / Hugging Face ML Engineer
    122.
    llama serve -hf ggml-org/Muse-Spark-Open-816B-A42B-GGUF:Q2_K›
    LLM score 35 · 11 days ago
  3. Thomas Wolf / Hugging Face Cofounder
    123.
    we’re so back https://t.co/rc64gc3vmU
    LLM score 25 · 11 days ago
  4. Boris Cherny / Creator of Claude Code
    124.
    Over the last few years, we have made significant progress on prompt injection for Claude models.›
    LLM score 65 · 11 days ago
  5. Shuchao Bi / Meta Researcher
    125.
    It is a really good open-weights model https://t.co/0HA2N59nXW
    LLM score 35 · 11 days ago
  6. Leandro von Werra / Hugging Face Head of Research
    126.
    Gave Claude a real pen and a paper with a simple task: participate in an art contest.›
    LLM score 35 · 11 days ago
  7. Leandro von Werra / Hugging Face Head of Research
    127.
    Gave Claude a real pen and a paper with a simple task: participate in an art contest.›
    LLM score 15 · 11 days ago
  8. Sayak Paul / Hugging Face Researcher
    128.
    It's time. It's fucking time! Flux.3 Image Klein is out and we bring you full support for it in Diffusers.›
    LLM score 35 · 11 days ago
  9. Igor Babuschkin / Cofounder of xAI
    129.
    Nemotron 3.5 Super VL is now available on the River API for inference, finetuning and RL.›
    LLM score 65 · 11 days ago
  10. Jason Weston / Meta Research Scientist
    130.
    Also ... did @gu_yuling make a: (A) chart crime ❌›
    LLM score 15 · 11 days ago
  11. Merve Noyan / Hugging Face ML Engineer
    131.
    this is huge https://t.co/SKo7F6ojdK
    LLM score 15 · 11 days ago
  12. Thomas Wolf / Hugging Face Cofounder
    132.
    As always, great to read METR’s thoughtful and in-depth work on this topic https://t.co/skfAB5eyWt
    LLM score 25 · 11 days ago
  13. Lewis Tunstall / Hugging Face Researcher
    133.
    Async RL works best with large batch sizes, but then the fwd/bwd pass becomes a major bottleneck, especially with agentic training where rollouts are >100k tokens long 😢›
    LLM score 78 · 11 days ago
  14. Eric Jang / ex VP of AI at 1X Robotics
    134.
    Jungle basketball inches ever closer https://t.co/JdcBTz9wio
    LLM score 25 · 11 days ago
  15. Yingru Li / xAI Researcher
    135.
    To be precise, this is a property of the estimator, not of RL, and sparsity isn’t necessarily bad.›
    LLM score 75 · 11 days ago
  16. Horace He / Thinking Machines Founding Engineer
    136.
    As @johnschulman2 has put it, there are many ways we can "spend compute to save our customers compute".›
    LLM score 65 · 11 days ago
  17. Yingru Li / xAI Researcher
    137.
    Interesting result: under on-policy training, Fisher information gives a task-agnostic reason most parameters receive only tiny expected gradients.›
    LLM score 78 · 11 days ago
  18. Tri Dao / Chief Scientist at Together
    138.
    TPUs have different bottlenecks from GPUs: few cores so you don’t need yo think about parallelism as much, large VMEM so easier to fuse kernels, but vector units have lower throughput so anything non matmul require much more care https://t.co/4ZdhowD46c
    LLM score 72 · 11 days ago
  19. Sholto Douglas / Researcher at Anthropic
    139.
    preparing to watch pint-eval get absolutely smashed, they're going to have to put it on log scale https://t.co/JCkkIsLM44
    LLM score 15 · 11 days ago
  20. Lewis Tunstall / Hugging Face Researcher
    140.
    My favourite part about @typesafeai is that they are one of the rare (only?) AI companies with a sense of humour https://t.co/3mSTMiTRzg
    LLM score 15 · 11 days ago
  21. Eric Jang / ex VP of AI at 1X Robotics
    141.
    Was at a business-suit wearing investor event in NYC where people were discussing "what are the signs of overinvestment?" ›
    LLM score 65 · 11 days ago
  22. Jason Weston / Meta Research Scientist
    142.
    Muse Wonder is out, with strong performance.›
    LLM score 35 · 11 days ago
  23. Dan Fu / VP of Kernels at Together
    143.
    Great work from @KumbongHermann, check it out! https://t.co/e8xyEUAS4O
    LLM score 25 · 12 days ago
  24. Lewis Tunstall / Hugging Face Researcher
    144.
    Great thread on why theoretical physics conjectures / theories are a harder nut to crack than in mathematics https://t.co/t4JXFto3kL
    LLM score 15 · 12 days ago
  25. Thomas Wolf / Hugging Face Cofounder
    145.
    wat. nanoGPT, but it trains inside a tweet. GPU-poor final form›
    LLM score 25 · 12 days ago
  26. Zixuan Li / Lead Z.ai
    146.
    GLM just helped uncover 23 Windows vulnerabilities, all acknowledged by Microsoft MSRC.›
    LLM score 65 · 12 days ago
  27. Zixuan Li / Lead Z.ai
    147.
    Thrilled to welcome Yuki Chen (also known as 古河/Guhe) to https://t.co/GTCIA6vbgh!›
    LLM score 25 · 12 days ago
  28. Merve Noyan / Hugging Face ML Engineer
    148.
    join us next week to learn about speculative decoding! https://t.co/6vhiob19TO
    LLM score 35 · 12 days ago
  29. Lewis Tunstall / Hugging Face Researcher
    149.
    Here's a quirk when evaluating closed models on tasks that trigger refusals (cyber, bio etc):›
    LLM score 78 · 12 days ago
  30. Eric Jang / ex VP of AI at 1X Robotics
    150.
    Revenge of the GPU poor roboticists (MIT)
    LLM score 15 · 12 days ago
More