AI300
  1. Jonathan Ross / TPU Creator
    31.
    Why is GPU+LPU so much more efficient? Consider building a transportation network for the United States and your options are 18 wheelers or delivery vans.›
    LLM score 85 · 4 days ago
  2. Jeff Dean / Chief Scientist at DeepMind
    32.
    The video of my conversation with Bill Dally at GTC last week is up.›
    LLM score 92 · 5 days ago
  3. Leo Gao / OpenAI Alignment Researcher
    33.
    new post: Corporations seem evil because we anthropomorphize them›
    LLM score 85 · 5 days ago
  4. John Schulman
    34.
    Great work by Chroma training a search agent with SoTA efficiency.›
    LLM score 92 · 5 days ago
  5. Chi Jin / OpenAI Researcher
    35.
    Thrilled to share our Goedel-series project on code! We can now automatically generate not only verifiable mathematical proofs, but also verifiable coding/programs.›
    LLM score 92 · 5 days ago
  6. Leandro von Werra / Hugging Face Head of Research
    36.
    There is almost unanimous agreement in the thread that data is the driver of models getting stronger, not architecture.›
    LLM score 92 · 5 days ago
  7. Jonathan Ross / TPU Creator
    37.
    Someone told me yesterday they don't trust AI for research because it gets the current President wrong.›
    LLM score 82 · 5 days ago
  8. Fei-Fei Li
    38.
    AI can generate 100 million splats, but it’s one creator’s imagination that made this uniquely beautiful world possible!🤩 https://t.co/3yIemVLLPH
    LLM score 82 · 5 days ago
  9. Jeremy Bernstein / Thinking Machines Researcher
    39.
    I found this to be a wonderful paper. I made a related observation last year: the hparam scalings for learning rate and batch size in our signSGD paper (from 2018!) have been found to be compute optimal in recent LLM scaling studies›
    LLM score 92 · 5 days ago
  10. Jonas Geiping / AI Researcher in Tübingen
    40.
    Can we automate incremental research in a narrow AI security subfield?›
    LLM score 92 · 5 days ago
  11. Andrej Karpathy / AI researcher
    41.
    When I built menugen ~1 year ago, I observed that the hardest part by far was not the code itself, it was the plethora of services you have to assemble like IKEA furniture to make it real, the DevOps: services, payments, auth, database, security, domain names, etc...›
    LLM score 92 · 5 days ago
  12. Jeff Dean / Chief Scientist at DeepMind
    42.
    📢 Another exciting step forward today with the launch of Gemini 3.1 Flash Live.›
    LLM score 25 · 5 days ago
  13. Merve Noyan / Hugging Face ML Engineer
    43.
    Cohere just topped Open ASR Leaderboard with a 2B model 👑›
    LLM score 75 · 5 days ago
  14. Noam Shazeer
    44.
    Gemini 3.1 Flash Live is now available, and it’s built for production-ready reliability.›
    LLM score 12 · 5 days ago
  15. Ben Burtenshaw / Hugging Face Researcher
    45.
    OpenEnv is ready for prime time.›
    LLM score 92 · 5 days ago
  16. Vahid Kazemi / ex MTS at xAI
    46.
    I used to think being behind in the AI race would hurt Apple, but so far it's been the opposite.›
    LLM score 85 · 6 days ago
  17. Kushal Thaman / Researcher at Flapping Airplanes
    47.
    Druids is made by extremely sharp folks, and feels like a new programming layer since coding agents first came out.›
    LLM score 92 · 6 days ago
  18. Behnam Neyshabur / Anthropic Researcher
    48.
    We have been heads down but wanted to share a bit about what we are doing 🧵 https://t.co/ICZlLwZ3Fq
    LLM score 75 · 6 days ago
  19. Andrej Karpathy / AI researcher
    49.
    One common issue with personalization in all LLMs is how distracting memory seems to be for the models.›
    LLM score 92 · 6 days ago
  20. Merve Noyan / Hugging Face ML Engineer
    50.
    does anyone know of a good multimodal tool calling dataset to fine-tune models on?›
    LLM score 92 · 6 days ago
  21. Ben Burtenshaw / Hugging Face Researcher
    51.
    Give coding agent a sharable workspace with persistent storage.›
    LLM score 92 · 6 days ago
  22. Merve Noyan / Hugging Face ML Engineer
    52.
    fav papers from 3DV #2 SAIL-Recon does reconstruction for inputs above 100 images (video)›
    LLM score 92 · 6 days ago
  23. Sander Dieleman / DeepMind Research Scientist
    53.
    This looks like it will be really good! Everything you've ever wanted to know about generalisation in diffusion models.›
    LLM score 75 · 6 days ago
  24. Leandro von Werra / Hugging Face Head of Research
    54.
    Which LLM would be better: - today's best architecture trained on 2023's best data›
    LLM score 82 · 6 days ago
  25. Thomas Wolf / Hugging Face Cofounder
    55.
    What are the best current techniques to have autoresearch behave better than (slightly improved) random search?›
    LLM score 92 · 6 days ago
  26. Sayak Paul / Hugging Face Researcher
    56.
    Introducing the first discrete diffusion pipeline for text in Diffusers -- LLaDA2 by @TheInclusionAI 🔥›
    LLM score 75 · 6 days ago
  27. Demis Hassabis / CEO of DeepMind
    57.
    Excited to partner with Agile Robots! Looking forward to seeing our models being deployed through Agile Robots incredible platform to help solve some of the most complex industrial challenges https://t.co/gsx6eGWwUI
    LLM score 75 · 6 days ago
  28. Naman Jain / Researcher at Cursor
    58.
    Check out the tech report detailing our continued pre-training and RL setup behind Composer2! Also sharing some example CursorBench problems by popular demand https://t.co/Ki9dDLcFX7 https://t.co/8FX9UfwnMx
    LLM score 85 · 7 days ago
  29. Dan Fu / VP of Kernels at Together
    59.
    As the COLM deadline creeps up - check out @NeelGuha's blog on how to write ML research papers! https://t.co/S3tyJspX2k
    LLM score 92 · 7 days ago
  30. Niklas Muennighoff / AI Researcher at Stanford
    60.
    One gem from Composer paper is that RL improved both pass@k & pass@1.›
    LLM score 92 · 7 days ago
More