AI300
  1. Sholto Douglas / Researcher at Anthropic
    3451.
    One day we’ll be able to decompose the loss curve of a neural net into all of the quanta it learns along the way - this is one of my fav streams of fundamental research.›
    LLM score 80 · 10 months ago
  2. Demis Hassabis / CEO of DeepMind
    3452.
    For AI to be truly useful, it needs to understand you.›
    LLM score 40 · 10 months ago
  3. Andrej Karpathy / AI researcher
    3453.
    @varunneal Worked great out of the box on nanochat too, beat standard weight decay in a solid sweep.
    LLM score 70 · 10 months ago
  4. Sholto Douglas / Researcher at Anthropic
    3454.
    @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 (also some notion of, if we can crack things at the complexity level of IMO using a sufficiently general algo, the same
    LLM score 70 · 10 months ago
  5. Demis Hassabis / CEO of DeepMind
    3455.
    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.›
    LLM score 20 · 10 months ago
  6. Roland Gavrilescu / ex MTS at xAI
    3456.
    Building and maintaining in-house at the cost of SaaS is the biggest disruptor we'll see in software https://t.co/3CitMHIu4w
    LLM score 92 · 10 months ago
  7. Tim Dettmers / Research Scientist at Ai2
    3457.
    @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.›
    LLM score 20 · 10 months ago
  8. Andrew Lampinen / Research Scientist at DeepMind
    3458.
    See https://t.co/wvpcKIKk6u for in-depth discussion of three examples.›
    LLM score 75 · 10 months ago
  9. Tim Dettmers / Research Scientist at Ai2
    3459.
    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.›
    LLM score 70 · 10 months ago
  10. Michael Elabd / DeepMind Researcher
    3460.
    🚀 New paper drop on the path to continual learning: Entropy-Adaptive Fine-Tuning (EAFT)›
    LLM score 80 · 10 months ago
  11. Sholto Douglas / Researcher at Anthropic
    3461.
    Claude code for all other knowledge work.›
    LLM score 60 · 10 months ago
  12. Kevin Murphy / DeepMind Research Scientist
    3462.
    Temporal abstraction is a key missing ingredient for long-term RL (to avoid the curse of the one-step model, as @RichardSSutton likes to say).›
    LLM score 85 · 10 months ago
  13. Andrej Karpathy / AI researcher
    3463.
    @patrickc This repo shows a way that works well for me:›
    LLM score 60 · 10 months ago
  14. Keller Jordan / OpenAI Researcher
    3464.
    @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 (dynamic-exponential) years, since >99.99% of future-history will be in some other static post-exponential state
    LLM score 70 · 10 months ago
  15. Keller Jordan / OpenAI Researcher
    3465.
    @yonashav This fact makes the present seem less interesting.›
    LLM score 20 · 10 months ago
  16. Emmanuel Ameisen / Anthropic Interpretability Researcher
    3466.
    We actually understand very few mechanisms inside LLMs, so the ones we do are a treat to explore.›
    LLM score 85 · 10 months ago
  17. Karol Hausman / Physical Intelligence Cofounder
    3467.
    Thanks - this was fun Tobi brings a very grounded perspective on how learning from experience translates to long-horizon robot performance in the real world.›
    LLM score 65 · 10 months ago
  18. Noam Brown / OpenAI Research Scientist
    3468.
    @MikCampanelli @mobiuspoker Making a poker solver is out of distribution, not poker itself.›
    LLM score 75 · 10 months ago
  19. Igor Babuschkin / Cofounder of xAI
    3469.
    @DavidSHolz Most APIs implement speculative decoding where a small model or a simple heuristic is used to predict a few tokens at a time.›
    LLM score 80 · 10 months ago
  20. Noam Brown / OpenAI Research Scientist
    3470.
    @giffmana It could be a combo of traditional spam/phishing detection and LLM detection
    LLM score 30 · 10 months ago
  21. Igor Babuschkin / Cofounder of xAI
    3471.
    @jadenitripp That’s where continual learning comes in.›
    LLM score 30 · 10 months ago
  22. Noam Brown / OpenAI Research Scientist
    3472.
    A family friend recently lost $1,000 to a phishing email.›
    LLM score 30 · 10 months ago
  23. Igor Babuschkin / Cofounder of xAI
    3473.
    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).›
    LLM score 80 · 10 months ago
  24. Sholto Douglas / Researcher at Anthropic
    3474.
    @deredleritt3r I still think it requires massive effort to not end up in that world - e.g.›
    LLM score 10 · 10 months ago
  25. Andrej Karpathy / AI researcher
    3475.
    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 dial (the compute you wish to spend) to achieve monotonically better results.›
    LLM score 80 · 10 months ago
  26. Chelsea Finn / Physical Intelligence Cofounder
    3476.
    Studying generalist reward models is hard: robot datasets focus on successful demos, not failures.›
    LLM score 95 · 10 months ago
  27. Noam Brown / OpenAI Research Scientist
    3477.
    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 · 10 months ago
  28. Andrej Karpathy / AI researcher
    3478.
    @thecsguy Slope of slope
    LLM score 70 · 10 months ago
  29. Jonas Geiping / AI Researcher in Tübingen
    3479.
    @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 Bayesian reasoning is a key capability that is not learnt from yet another code/math RL environment
    LLM score 80 · 10 months ago
  30. Jonas Geiping / AI Researcher in Tübingen
    3480.
    We just published a new open-source model that we trained with RL to be capable of open-ended forecasting!›
    LLM score 20 · 10 months ago
More