AI300
  1. Tri Dao / Chief Scientist at Together
    121.
    This is what we've been coking for the last 9 months: make MoEs training goes ~2x faster and ~2x less memory! Highlights:›
    LLM score 85 · about 1 month ago
  2. Noam Shazeer
    122.
    Gemini 3 Flash is live.›
    LLM score 70 · about 1 month ago
  3. Demis Hassabis / CEO of DeepMind
    123.
    For a fast model, Gemini 3 Flash offers incredible performance, allowing us to provide frontier intelligence to everyone globally.›
    LLM score 20 · about 1 month ago
  4. Noam Brown / OpenAI Research Scientist
    124.
    Our efforts at @OpenAI to advance scientific progress aren't just limited to math/physics/coding.›
    LLM score 20 · about 1 month ago
  5. Demis Hassabis / CEO of DeepMind
    125.
    Always enjoy discussing the big picture with @FryRsquared.›
    LLM score 20 · about 1 month ago
  6. Dan Fu / VP of Kernels at Together
    126.
    @maxencefrenette Good point! I think nuanced discussions of the task, and task-specific architectures are key for this.›
    LLM score 70 · about 2 months ago
  7. Tri Dao / Chief Scientist at Together
    127.
    Nvidia continues to put out some of the strongest and fastest open models.›
    LLM score 80 · about 2 months ago
  8. Dan Fu / VP of Kernels at Together
    128.
    @davieball @Tim_Dettmers Yes, exactly! One piece that didn’t make it into my post - some of the best innovations come resource constrained environments (eg DeepSeek).
    LLM score 70 · about 2 months ago
  9. Dan Fu / VP of Kernels at Together
    129.
    New blog post about paths to AGI and arguing why it’s too early to say AGI is resource limited.›
    LLM score 80 · about 2 months ago
  10. Dan Fu / VP of Kernels at Together
    130.
    My response to @Tim_Dettmers great post last week that we won't reach AGI because of resource limitations.›
    LLM score 65 · about 2 months ago
  11. Noam Brown / OpenAI Research Scientist
    131.
    @tszzl Yeah, the Claude 3 announcement from March 2024 still listed GSM8K as one of the benchmarks
    LLM score 70 · about 2 months ago
  12. Demis Hassabis / CEO of DeepMind
    132.
    First LLM contact from space 🛰️ using our highly efficient open source Gemma models! Huge congrats to @PhilipJohnston and the @Starcloud_Inc_ team! https://t.co/JFuh9Y8a1f
    LLM score 20 · about 2 months ago
  13. Noam Brown / OpenAI Research Scientist
    133.
    An important lesson that ARC-AGI has internalized, but not many others have, is that benchmark perf is a function of test-time compute.›
    LLM score 85 · about 2 months ago
  14. Noam Brown / OpenAI Research Scientist
    134.
    IMO GDPVal is the most important result from our @OpenAI GPT-5.2 launch.›
    LLM score 80 · about 2 months ago
  15. Noam Brown / OpenAI Research Scientist
    135.
    I'm also really happy that @OpenAI was willing to publish the original GDPVal results showing Claude ahead of ChatGPT.›
    LLM score 60 · about 2 months ago
  16. Demis Hassabis / CEO of DeepMind
    136.
    We’re also announcing a new partnership with @AISecurityInst that builds on two years working together & will focus on foundational safety and security research essential for realising AI’s potential to benefit humanity.›
    LLM score 45 · about 2 months ago
  17. Demis Hassabis / CEO of DeepMind
    137.
    I’ve always believed that AI will be the most transformational technology of our time, and partnerships like this are vital to turning that potential into real progress @SciTechgovuk.›
    LLM score 20 · about 2 months ago
  18. Demis Hassabis / CEO of DeepMind
    138.
    The UK is an amazing place for science & innovation.›
    LLM score 50 · about 2 months ago
  19. Andrej Karpathy / AI researcher
    139.
    Quick new post: Auto-grading decade-old Hacker News discussions with hindsight›
    LLM score 80 · about 2 months ago
  20. Tim Dettmers / Research Scientist at Ai2
    140.
    Many people think AI will continue improve towards AGI.›
    LLM score 80 · about 2 months ago
  21. Tim Dettmers / Research Scientist at Ai2
    141.
    My new blog post discusses the physical reality of computation and why this means we will not see AGI or any meaningful superintelligence: https://t.co/jsAKQ6T3gC
    LLM score 80 · about 2 months ago
  22. Andrej Karpathy / AI researcher
    142.
    In today's episode of programming horror... In the Python docs of random.seed() def, we're told›
    LLM score 40 · about 2 months ago
  23. Noam Brown / OpenAI Research Scientist
    143.
    From inception to release, the journal publication process can easily take over a year.›
    LLM score 70 · about 2 months ago
  24. Igor Babuschkin / Cofounder of xAI
    144.
    SGLang is the best inference framework for LLMs.›
    LLM score 20 · about 2 months ago
  25. Andrej Karpathy / AI researcher
    145.
    @ChrSzegedy I could certainly imagine that "nesting" the simulation might be too "effortful" for the model, compute or data density wise.›
    LLM score 60 · about 2 months ago
  26. Noam Brown / OpenAI Research Scientist
    146.
    @Thom_Wolf Putnam is more knowledge-based whereas IMO requires more creativity and more time per problem, so Putnam is easier for LLMs.
    LLM score 70 · about 2 months ago
  27. Andrej Karpathy / AI researcher
    147.
    @DimitrisPapail There is definitely work going into engineering the "you" simulation - the personality that gets all the rewards in verifiable problems, or all the upvotes from users/judge LLMs, or mimics the responses of SFT, and there is an emergent composite personality from that.›
    LLM score 30 · about 2 months ago
  28. Andrej Karpathy / AI researcher
    148.
    Don't think of LLMs as entities but as simulators.›
    LLM score 85 · about 2 months ago
  29. Demis Hassabis / CEO of DeepMind
    149.
    Gemini has always had exceptionally strong multimodal capabilities.›
    LLM score 15 · about 2 months ago
  30. Noam Brown / OpenAI Research Scientist
    150.
    @deredleritt3r Ah yeah that could have been worded better.›
    LLM score 75 · about 2 months ago
More