AI300
  1. Yoshua Bengio
    1561.
    Il est essentiel de développer à la fois une réglementation efficace, des garde-fous sociaux et des solutions techniques comme celles que nous développons à @LawZero_ pour éviter les risques de l’IA avancée tout en développant ses applications positives.›
    LLM score 40 · 4 months ago
  2. Shane Legg / Chief AGI Scientist at DeepMind
    1562.
    @kevinkshah @korbencopy I'd give it a 50% probability of happening within 5 years of Minimal AGI.›
    LLM score 70 · 4 months ago
  3. Shane Legg / Chief AGI Scientist at DeepMind
    1563.
    @GDalegend @korbencopy We get to 2028 and have something quite close to Minimal AGI, but not quite.
    LLM score 20 · 4 months ago
  4. Shane Legg / Chief AGI Scientist at DeepMind
    1564.
    @tablefourthree @BogdanIonutCir2 @korbencopy It's actually surprising to me that my probability hasn't shifted yet, given that 2028 is so close.›
    LLM score 20 · 4 months ago
  5. Shane Legg / Chief AGI Scientist at DeepMind
    1565.
    @korbencopy 50% chance of Minimal AGI by 2028.›
    LLM score 70 · 4 months ago
  6. John Carmack
    1566.
    #PaperADay 9 floq: Training Critics via Flow-Matching for Scaling Compute in Value-Based RL›
    LLM score 90 · 4 months ago
  7. Kevin Lu / Thinking Machines Researcher
    1567.
    encouraging progress on continued test-time adaptation beyond model deployment!›
    LLM score 65 · 4 months ago
  8. Karan Dalal / Stanford SAIL Researcher
    1568.
    Test-time training is inevitable.›
    LLM score 70 · 4 months ago
  9. Ronak Malde / DeepMind Researcher
    1569.
    At Deepmind ever since we maxed out SWE-bench , I’ve felt that there hasn’t been a difficult enough benchmark that is aligned with how people want to use AI.›
    LLM score 75 · 4 months ago
  10. Sergey Levine / Physical Intelligence Cofounder
    1570.
    Benjie posted a reply to the Physical Intelligence "Olympics" attempt: https://t.co/sPkc5DK4ls›
    LLM score 70 · 4 months ago
  11. Siyan Zhao / UCLA PhD
    1571.
    Introducing 💡On-Policy Self-Distillation💡, a simple method that enables LLM to teach itself with dense per-token feedback on its own on-policy generations—achieving 4-8x more token efficiency vs.›
    LLM score 92 · 4 months ago
  12. Noam Brown / OpenAI Research Scientist
    1572.
    @_sholtodouglas @jekbradbury @GoogleDeepMind @andy_l_jones @OpenAI @kevin_wang3290 A word on comp: I know folks who become quants to make money but 5 years later ask what they're doing with their life.›
    LLM score 20 · 4 months ago
  13. Noam Brown / OpenAI Research Scientist
    1573.
    @_sholtodouglas @jekbradbury @GoogleDeepMind @andy_l_jones Labs like @OpenAI also hire researchers straight out of undergrad, like @kevin_wang3290, though the bar is high.›
    LLM score 60 · 4 months ago
  14. Noam Brown / OpenAI Research Scientist
    1574.
    @_sholtodouglas @jekbradbury @GoogleDeepMind @andy_l_jones was a semi-retired quant when he wrote a paper comparing the impact of scaling pretraining vs scaling test-time compute (before TTC scaling was cool).›
    LLM score 70 · 4 months ago
  15. Demis Hassabis / CEO of DeepMind
    1575.
    We’re excited to be working with @JNJInnovation to accelerate the path to new medicines.›
    LLM score 60 · 4 months ago
  16. John Carmack
    1576.
    #PaperADay 8 Beyond Gradient Averaging in Parallel Optimization: Improved Robustness through Gradient Agreement Filtering›
    LLM score 85 · 4 months ago
  17. Igor Babuschkin / Cofounder of xAI
    1577.
    @MillionInt The engineer presented a single grain of etched silicon.›
    LLM score 65 · 4 months ago
  18. Lewis Tunstall / Hugging Face Researcher
    1578.
    This is a big deal because it shows that (a) Claude can solve hard math problems when given an appropriate scaffold with tools (Lean) and (b) the framework generalises beyond competitive mathematics to enable autoformalization at scale›
    LLM score 85 · 4 months ago
  19. John Carmack
    1579.
    #PaperADay 7 Cautious Weight Decay https://t.co/D0u9KL5zsS›
    LLM score 95 · 5 months ago
  20. Keller Jordan / OpenAI Researcher
    1580.
    @supreme__sharma There's no good answer.›
    LLM score 20 · 5 months ago
  21. Keller Jordan / OpenAI Researcher
    1581.
    @eliebakouch In terms of hypothetical disagreement-winrates between these different cultures, I’d estimate that my (heavily muted) FYP has +500 ELO vs.›
    LLM score 30 · 5 months ago
  22. Clive Chan / OpenAI Hardware
    1582.
    nothing is free in computer architecture.›
    LLM score 85 · 5 months ago
  23. Clive Chan / OpenAI Hardware
    1583.
    Dojo1 was ultra ambitious and really pushed a lot of technologies (packaging, power delivery, system design, clock distribution, even number formats), but execution and focus could've been better.›
    LLM score 85 · 5 months ago
  24. Keller Jordan / OpenAI Researcher
    1584.
    The internet has become real life only in the sense that *public* cultural systems have been largely replaced by it›
    LLM score 20 · 5 months ago
  25. Boris Cherny / Creator of Claude Code
    1585.
    @aref_vc The formula for the auto-compact buffer is really simple: (contextWindow - maxOutputTokens) - (small constant)›
    LLM score 80 · 5 months ago
  26. Boris Cherny / Creator of Claude Code
    1586.
    @Wind_AndSnow Ah, you want permissions: { allow: "Read(~/.config/runmenu/*)"}
    LLM score 80 · 5 months ago
  27. Boris Cherny / Creator of Claude Code
    1587.
    @aref_vc Reduce your CLAUDE_CODE_MAX_OUTPUT_TOKENS or rm the env var, and you'll decrease the buffer to 45k
    LLM score 70 · 5 months ago
  28. Boris Cherny / Creator of Claude Code
    1588.
    @DominikPeters Claude usually only uses Plan agents.›
    LLM score 70 · 5 months ago
  29. Boris Cherny / Creator of Claude Code
    1589.
    Now in Claude Code: when you accept a plan, Claude automatically clears your context, so your plan gets a fresh context window.›
    LLM score 70 · 5 months ago
  30. Sander Dieleman / DeepMind Research Scientist
    1590.
    Great interactive blog post on reflow, which reduces the number of steps required to sample from diffusion models by "straightening" paths from noise to data.›
    LLM score 80 · 5 months ago
More