AI300
  1. Demis Hassabis / CEO of DeepMind
    121.
    Always enjoy discussing the big picture with @FryRsquared. We talked about the frontiers of computability, the nature of the mind, and why I’m optimistic that AI will help us...
    LLM score 20 · about 1 month ago
  2. Dan Fu / VP of Kernels at Together
    122.
    @maxencefrenette Good point! I think nuanced discussions of the task, and task-specific architectures are key for this. For one example - video models usually do full...
    LLM score 70 · about 1 month ago
  3. Tri Dao / Chief Scientist at Together
    123.
    Nvidia continues to put out some of the strongest and fastest open models. Pretraining and post training data are released as well, something very few orgs have done...
    LLM score 80 · about 1 month ago
  4. Dan Fu / VP of Kernels at Together
    124.
    @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 1 month ago
  5. Dan Fu / VP of Kernels at Together
    125.
    New blog post about paths to AGI and arguing why it’s too early to say AGI is resource limited. Three main points: * Today’s AI models are extremely underutilized - tons of...
    LLM score 80 · about 2 months ago
  6. Dan Fu / VP of Kernels at Together
    126.
    My response to @Tim_Dettmers great post last week that we won't reach AGI because of resource limitations. My take - there's a ton of headroom in today's systems, it's too early...
    LLM score 65 · about 2 months ago
  7. Noam Brown / OpenAI Research Scientist
    127.
    @tszzl Yeah, the Claude 3 announcement from March 2024 still listed GSM8K as one of the benchmarks
    LLM score 70 · about 2 months ago
  8. Demis Hassabis / CEO of DeepMind
    128.
    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
  9. Noam Brown / OpenAI Research Scientist
    129.
    An important lesson that ARC-AGI has internalized, but not many others have, is that benchmark perf is a function of test-time compute. @OpenAI publishes single-number benchmark...
    LLM score 85 · about 2 months ago
  10. Noam Brown / OpenAI Research Scientist
    130.
    IMO GDPVal is the most important result from our @OpenAI GPT-5.2 launch. We outperform in-domain experts and are SOTA among *all* models on GPDVal, which measures performance on...
    LLM score 80 · about 2 months ago
  11. Noam Brown / OpenAI Research Scientist
    131.
    I'm also really happy that @OpenAI was willing to publish the original GDPVal results showing Claude ahead of ChatGPT. Excellent work on the eval from @tejalpatwardhan and her...
    LLM score 60 · about 2 months ago
  12. Demis Hassabis / CEO of DeepMind
    132.
    The UK is an amazing place for science & innovation. Thrilled to deepen our partnership with the UK Government to turbocharge scientific discovery with AI - giving scientists...
    LLM score 50 · about 2 months ago
  13. Demis Hassabis / CEO of DeepMind
    133.
    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...
    LLM score 20 · about 2 months ago
  14. Demis Hassabis / CEO of DeepMind
    134.
    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...
    LLM score 45 · about 2 months ago
  15. Andrej Karpathy / AI researcher
    135.
    Quick new post: Auto-grading decade-old Hacker News discussions with hindsight I took all the 930 frontpage Hacker News article+discussion of December 2015 and asked the GPT 5.1...
    LLM score 80 · about 2 months ago
  16. Tim Dettmers / Research Scientist at Ai2
    136.
    Many people think AI will continue improve towards AGI. In my new blog post, I argue that we will not reach AGI due to physical reasons. Key items discussed: The physical reality...
    LLM score 80 · about 2 months ago
  17. Tim Dettmers / Research Scientist at Ai2
    137.
    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
  18. Andrej Karpathy / AI researcher
    138.
    In today's episode of programming horror... In the Python docs of random.seed() def, we're told "If a is an int, it is used directly." [1] But if you seed with 3 or -3, you...
    LLM score 40 · about 2 months ago
  19. Noam Brown / OpenAI Research Scientist
    139.
    From inception to release, the journal publication process can easily take over a year. @OpenAI o1 was made available only a year ago. https://t.co/ax7pC4OfxE
    LLM score 70 · about 2 months ago
  20. Igor Babuschkin / Cofounder of xAI
    140.
    SGLang is the best inference framework for LLMs. We heavily used it at xAI because it allowed us to run large models faster than anyone else. The developers have now started...
    LLM score 20 · about 2 months ago
  21. Andrej Karpathy / AI researcher
    141.
    @ChrSzegedy I could certainly imagine that "nesting" the simulation might be too "effortful" for the model, compute or data density wise. My results with it are not too bad so imo...
    LLM score 60 · about 2 months ago
  22. Noam Brown / OpenAI Research Scientist
    142.
    @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
  23. Andrej Karpathy / AI researcher
    143.
    @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...
    LLM score 30 · about 2 months ago
  24. Andrej Karpathy / AI researcher
    144.
    Don't think of LLMs as entities but as simulators. For example, when exploring a topic, don't ask: "What do you think about xyz"? There is no "you". Next time try: "What would...
    LLM score 85 · about 2 months ago
  25. Demis Hassabis / CEO of DeepMind
    145.
    Gemini has always had exceptionally strong multimodal capabilities. Gemini 3 Pro is an incredible vision AI model and is SOTA across all main vision & multimodal benchmarks....
    LLM score 15 · about 2 months ago
  26. Noam Brown / OpenAI Research Scientist
    146.
    @deredleritt3r Ah yeah that could have been worded better. The IMO model was included in the scaffold for ICPC and got 11/12 problems by itself.
    LLM score 75 · about 2 months ago
  27. Noam Brown / OpenAI Research Scientist
    147.
    @deredleritt3r We didn’t use the same exact model for the IMO, IOI, and ICPC. It was the same family of models, but with minor differences. The IMO model by itself would have done...
    LLM score 70 · about 2 months ago
  28. Demis Hassabis / CEO of DeepMind
    148.
    Gemini 3 Deep Think is now available for Google AI Ultra subscribers in the @GeminiApp, incorporating our gold medal winning IMO and ICPC technologies! 🏅With its parallel...
    LLM score 35 · about 2 months ago
  29. Noam Shazeer
    149.
    Gemini 3 Deep Think mode is live for Ultra users today. When using parallel lines of thought in this mode, Gemini shows meaningful improvement on key reasoning benchmarks such as...
    LLM score 75 · about 2 months ago
  30. Noam Brown / OpenAI Research Scientist
    150.
    @fchollet Do you consider any humans to have an understanding of differential equations?
    LLM score 40 · about 2 months ago
More