AI300
  1. Tim Dettmers / Research Scientist at Ai2
    2191.
    his work was mostly the genius of Ethan Shen.›
    LLM score 70 · 6 months ago
  2. Tim Dettmers / Research Scientist at Ai2
    2192.
    Our method became so efficient (26x vs RL; 57x vs other synth gen), that we could easily generate 1000s of trajectories for a single repo.›
    LLM score 85 · 6 months ago
  3. Tim Dettmers / Research Scientist at Ai2
    2193.
    This is very impactful: you can now distill frontier performance into small models that are specialized to private repositories.›
    LLM score 75 · 6 months ago
  4. Tim Dettmers / Research Scientist at Ai2
    2194.
    From there we could do a massiv amounts of experiments and really understand what matters for training coding agents.›
    LLM score 80 · 6 months ago
  5. Cameron Wolfe / Researcher at Netflix
    2195.
    Continual learning is a popular topic in LLM research, but it might not be as far away as we think.›
    LLM score 65 · 6 months ago
  6. Ethan Shen / Ai2 Researcher
    2196.
    Finally, we conduct an analysis of variance across SWE-Bench runs.›
    LLM score 95 · 6 months ago
  7. Ethan Shen / Ai2 Researcher
    2197.
    We also experiment with mixing data from rollouts and discover that training on SERA’s first and second rollouts complement each other to increase performance in data constrained regimes.›
    LLM score 85 · 6 months ago
  8. Ethan Shen / Ai2 Researcher
    2198.
    Even more interesting is the fact that the best data comes from truncated trajectories that just barely exceed the context limit.›
    LLM score 85 · 6 months ago
  9. Niklas Muennighoff / AI Researcher at Stanford
    2199.
    Community-built open benchmarks work really well, e.g., Terminal-Bench, HLE, MMTEB.›
    LLM score 80 · 6 months ago
  10. John Carmack
    2200.
    #PaperADay 11 Discovering state-of-the-art reinforcement learning algorithms›
    LLM score 80 · 6 months ago
  11. Noam Brown / OpenAI Research Scientist
    2201.
    Had to cut this one for space: 2019: AI can't create art—creativity is uniquely human
    LLM score 20 · 6 months ago
  12. Andrej Karpathy / AI researcher
    2202.
    @0xabi96 It feels like I’m cheating.›
    LLM score 70 · 6 months ago
  13. Andrej Karpathy / AI researcher
    2203.
    @ChiragLathiya The nearest neighbor really is some kind of a junior engineer.›
    LLM score 80 · 6 months ago
  14. Andrej Karpathy / AI researcher
    2204.
    @jeremytwei Love the word "comprehension debt", haven't encountered it so far, it's very accurate.›
    LLM score 30 · 6 months ago
  15. Andrej Karpathy / AI researcher
    2205.
    @airesearch12 💯 @ Spec-driven development It's the limit of imperative -> declarative transition, basically being declarative entirely.›
    LLM score 60 · 6 months ago
  16. Andrej Karpathy / AI researcher
    2206.
    A few random notes from claude coding quite a bit last few weeks.›
    LLM score 20 · 6 months ago
  17. Noam Brown / OpenAI Research Scientist
    2207.
    1987: AI can't win at chess—planning is uniquely human›
    LLM score 70 · 6 months ago
  18. Yoshua Bengio
    2208.
    Had a great discussion on AI's major societal impacts and emerging risks on the @RestIsPolitics podcast this past November.›
    LLM score 20 · 6 months ago
  19. Jonathan Ross / TPU Creator
    2209.
    Success in the Information Age was about being able to answer questions.›
    LLM score 20 · 6 months ago
  20. Ben Burtenshaw / Hugging Face Researcher
    2210.
    this is a blog post on claude + llama.cpp https://t.co/yej6WsNnQA
    LLM score 20 · 6 months ago
  21. Yoshua Bengio
    2211.
    Merci à @HugoDecrypte pour l'invitation à faire un tour d'horizon des risques de l'IA ainsi que des solutions nécessaires pour se diriger vers une meilleure trajectoire.›
    LLM score 20 · 6 months ago
  22. Ben Burtenshaw / Hugging Face Researcher
    2212.
    I know it's massively off trend, but I love setting up a nice quiver of models in @cursor_ai, reviewing generated tokens in the legit ui, switching models for certain parts, and just generally getting shit done.›
    LLM score 30 · 6 months ago
  23. Ronak Malde / DeepMind Researcher
    2213.
    I don't think people have realized how crazy the results are from this new TTT + RL paper from Stanford/Nvidia.›
    LLM score 85 · 6 months ago
  24. Clive Chan / OpenAI Hardware
    2214.
    latest response hit 9h25m + 2.3 million output tokens and did everything i wanted it to over the course of TWELVE context compactions https://t.co/F9ByO19kc0
    LLM score 85 · 6 months ago
  25. Cameron Wolfe / Researcher at Netflix
    2215.
    Continual learning is being positioned as a prerequisite for AGI (i.e., general systems must be adaptable).›
    LLM score 70 · 6 months ago
  26. Geoffrey Hinton
    2216.
    I just watched a really great conversation about the future of AI.›
    LLM score 30 · 6 months ago
  27. Ben Burtenshaw / Hugging Face Researcher
    2217.
    The only thing I don't get is why it has to be a Mac mini.›
    LLM score 20 · 6 months ago
  28. Chelsea Finn / Physical Intelligence Cofounder
    2218.
    Video models serve as a good pretrained backbone for robot policies.›
    LLM score 70 · 6 months ago
  29. Sander Dieleman / DeepMind Research Scientist
    2219.
    As I said in my blog post about latent representations last year: "representation learning and reconstruction are two separate tasks, and while it is convenient to do both at once, this might not be optimal."›
    LLM score 85 · 6 months ago
  30. Shane Legg / Chief AGI Scientist at DeepMind
    2220.
    @2024dion @deredleritt3r The fact that it still seems 50-50 by 2028 surprises me! I would have thought that evidence would have tipped it one way or another by now.
    LLM score 30 · 6 months ago
More