AI300
  1. Andrej Karpathy / AI researcher
    61.
    @varunneal Worked great out of the box on nanochat too, beat standard weight decay in a solid sweep.
    LLM score 70 · 15 days ago
  2. Sholto Douglas / Researcher at Anthropic
    62.
    @terracotta_hawk It’s a mix of - a lot of the best people working on ai have math ability as the bar they feel most deeply about, so they work on making the models good at math...
    LLM score 70 · 15 days ago
  3. Demis Hassabis / CEO of DeepMind
    63.
    We’ve made lots of improvements with Veo 3.1! It’s more expressive, supports portrait mode and SOTA video upscaling to 1080p and 4K. Rolling out starting today - try it in the...
    LLM score 20 · 16 days ago
  4. Tim Dettmers / Research Scientist at Ai2
    64.
    @main_horse @pangramlabs Interesting! Not sure how good their tool is, but I guess it supports my point that AI-generated text can just be as good as human-written text. I think...
    LLM score 20 · 16 days ago
  5. Andrew Lampinen / Research Scientist at DeepMind
    65.
    See https://t.co/wvpcKIKk6u for in-depth discussion of three examples. But in brief, I think it’s quite common for impossibility proofs to sneak in an assumption about the...
    LLM score 75 · 16 days ago
  6. Tim Dettmers / Research Scientist at Ai2
    66.
    My newest blog post is about my experience of using coding agents in the past 8 months to automate my own work as much as possible. This post introduces concepts I found useful...
    LLM score 70 · 16 days ago
  7. Michael Elabd / DeepMind Researcher
    67.
    New EAFT paper tackles catastrophic forgetting in continual learning by entropy-adaptive fine-tuning.
    LLM score 80 · 17 days ago
  8. Sholto Douglas / Researcher at Anthropic
    68.
    Claude code for all other knowledge work. Many of our best engineers no longer manually write code, they multiplex across multiple cc sessions - soon this will be true for...
    LLM score 60 · 17 days ago
  9. Andrej Karpathy / AI researcher
    69.
    @patrickc This repo shows a way that works well for me: https://t.co/L3K42MU4wF Basically I use epub (not pdf), the code then parses it into text. I usually go chapter by chapter,...
    LLM score 60 · 18 days ago
  10. Keller Jordan / OpenAI Researcher
    70.
    @yonashav By “interesting,” I mean that, e.g., if we ask “was it a good idea for humanity to do science?”, then it doesn’t make sense to judge based on only the last few hundred...
    LLM score 70 · 19 days ago
  11. Keller Jordan / OpenAI Researcher
    71.
    @yonashav This fact makes the present seem less interesting. The ever-changing way things are now (“modernity”) is driven by exponential processes and is thus transitory. The more...
    LLM score 20 · 19 days ago
  12. Noam Brown / OpenAI Research Scientist
    72.
    @MikCampanelli @mobiuspoker Making a poker solver is out of distribution, not poker itself. There’s very little open source code on how to make a good poker bot
    LLM score 75 · 20 days ago
  13. Igor Babuschkin / Cofounder of xAI
    73.
    @DavidSHolz Most APIs implement speculative decoding where a small model or a simple heuristic is used to predict a few tokens at a time. This greatly speeds up a bunch of things,...
    LLM score 80 · 21 days ago
  14. Noam Brown / OpenAI Research Scientist
    74.
    @giffmana It could be a combo of traditional spam/phishing detection and LLM detection
    LLM score 30 · 21 days ago
  15. Igor Babuschkin / Cofounder of xAI
    75.
    @jadenitripp That’s where continual learning comes in. The model should be training in between inference sessions.
    LLM score 30 · 21 days ago
  16. Noam Brown / OpenAI Research Scientist
    76.
    A family friend recently lost $1,000 to a phishing email. Afterward, I ran it through ChatGPT and it easily identified it as a scam. I hope Gmail prioritizes better phishing...
    LLM score 30 · 21 days ago
  17. Igor Babuschkin / Cofounder of xAI
    77.
    I suspect the reason Claude Code doesn’t work as well for large codebases is that they post-trained it mostly on smaller repos (big corp sized repos are rare). To perform really...
    LLM score 80 · 21 days ago
  18. Sholto Douglas / Researcher at Anthropic
    78.
    @deredleritt3r I still think it requires massive effort to not end up in that world - e.g. we should be doing everything we can to accelerate the development of the robotic supply...
    LLM score 10 · 22 days ago
  19. Andrej Karpathy / AI researcher
    79.
    New post: nanochat miniseries v1 The correct way to think about LLMs is that you are not optimizing for a single specific model but for a family models controlled by a single...
    LLM score 80 · 22 days ago
  20. Noam Brown / OpenAI Research Scientist
    80.
    Got this DM: I appreciate that you posted this - increasingly my twitter feed feels out of whack, especially with people claiming Claude Code makes them 1000000x more efficient....
    LLM score 30 · 23 days ago
  21. Andrej Karpathy / AI researcher
    81.
    @thecsguy Slope of slope
    LLM score 70 · 23 days ago
  22. Jonas Geiping / AI Researcher in Tübingen
    82.
    @ziv_ravid Yeah I think forecasting is one of the few domains where we can really evaluate reasoning under uncertainty, and this may be a bit of contrarian take, I do think...
    LLM score 80 · 23 days ago
  23. Jonas Geiping / AI Researcher in Tübingen
    83.
    We just published a new open-source model that we trained with RL to be capable of open-ended forecasting! Open-endedness is a really interesting capability for a forecasting...
    LLM score 20 · 23 days ago
  24. Andrej Karpathy / AI researcher
    84.
    @ChrisPainterYup All of the following are true: I came out sounding too critical in that moment on the pod and accidentally created a misrepresenting sound bite, I was positively...
    LLM score 20 · 23 days ago
  25. Noam Brown / OpenAI Research Scientist
    85.
    @zachcpa I'd give the win to Codex due to the way better optimizations. But full disclosure: I work at OpenAI.
    LLM score 20 · 23 days ago
  26. Noam Brown / OpenAI Research Scientist
    86.
    @mobiuspoker One of the reasons I chose to make a poker bot is because I thought it would be pretty out of distribution for the models, and most real-world tasks are a little out...
    LLM score 70 · 24 days ago
  27. Demis Hassabis / CEO of DeepMind
    87.
    We’re making great progress with our Gemini Robotics work in bringing AI to the physical world - a critical aspect of AGI. As part of our next steps, super excited to announce our...
    LLM score 40 · 24 days ago
  28. Noam Brown / OpenAI Research Scientist
    88.
    @swyx I don't remember the prompt but the initial prompt wasn't very detailed. The hard part for the agents was making it easy for the user to select ranges without clicking every...
    LLM score 20 · 24 days ago
  29. Noam Brown / OpenAI Research Scientist
    89.
    @_thomasip I think it's similar to the hallucination problem. A lot of folks are working on this and to be fair a lot of progress has been made. I think we'll continue to see a...
    LLM score 20 · 24 days ago
  30. Noam Brown / OpenAI Research Scientist
    90.
    This solver is a project I've wanted to do for a while. I hope people find value from it, and if there's interest I might implement more advanced poker solvers with Codex, like...
    LLM score 40 · 24 days ago
More