AI300
  1. Horace He / Thinking Machines Founding Engineer
    1441.
    A conversation I've had several times lately: Them: "LLMs are so good for learning whatever you want — there's never been a better time to learn!"›
    LLM score 92 · 4 months ago
  2. Jason Weston / Meta Research Scientist
    1442.
    🌐Unified Post-Training via On-Policy-Trained LM-as-RM🔧›
    LLM score 92 · 4 months ago
  3. Sharon Zhou / AMD VP of AI
    1443.
    codegen is cheap now. performance isn’t: most generated kernels are kinda mid.›
    LLM score 92 · 4 months ago
  4. Lewis Tunstall / Hugging Face Researcher
    1444.
    You can now pretrain LLMs entirely on the HF Hub 💥›
    LLM score 92 · 4 months ago
  5. Zixuan Li / Lead Z.ai
    1445.
    Last time Chinese models missed the SWE-rebench top 10, they got slammed for "benchmaxing." Now, GLM-5 is back in a very "interesting" spot.›
    LLM score 82 · 4 months ago
  6. Yoshua Bengio
    1446.
    Evidence of deceptive behavior has already appeared in widely used AI systems, and the risk is expected to grow as AI becomes more capable, more autonomous, and more embedded in everyday decision-making.›
    LLM score 92 · 4 months ago
  7. Ben Burtenshaw / Hugging Face Researcher
    1447.
    share you agent harness on the hub. built a hf cli extension that backs up and shares coding agent workspaces on the hub.›
    LLM score 92 · 4 months ago
  8. Jonathan Ross / TPU Creator
    1448.
    AI will split companies into two camps. Value Preservation — using AI to cut costs.›
    LLM score 12 · 4 months ago
  9. Sayak Paul / Hugging Face Researcher
    1449.
    We released 🤗 Kernels 0.12.3 to support Flash-Attention 4! This means we now support kernels written in `cutlass.cute`.›
    LLM score 85 · 4 months ago
  10. Skyler Miao / MiniMax Head of Engineering
    1450.
    We’ve pushed performance quite a bit here — M2.7 is showing strong results across both MLE and semi-autonomous RL.›
    LLM score 92 · 4 months ago
  11. Skyler Miao / MiniMax Head of Engineering
    1451.
    This is an outstanding project.›
    LLM score 72 · 4 months ago
  12. Skyler Miao / MiniMax Head of Engineering
    1452.
    Yes, we performed extensive optimizations and even established a dedicated benchmark for it.›
    LLM score 85 · 4 months ago
  13. Cameron Wolfe / Researcher at Netflix
    1453.
    Very well-written post. Whether you think takeoff is imminent or not, it's worth deeply considering this trajectory and the bottlenecks that could arise.›
    LLM score 92 · 4 months ago
  14. Sharon Zhou / AMD VP of AI
    1454.
    Mood: agents optimizing kernels Claude won on kernel optimization: gemm_bf16 at 1.19x vs Codex's 0.94x.›
    LLM score 92 · 4 months ago
  15. Alex Zhang / MIT CSAIL PhD
    1455.
    soon all benchmarks of this form will be obsolete :) https://t.co/gzwPimFngT https://t.co/Hr1GH9COyS
    LLM score 72 · 4 months ago
  16. Andrew White / Edison Scientific Cofounder
    1456.
    The names of molecular representations are kind of insane:›
    LLM score 75 · 4 months ago
  17. Skyler Miao / MiniMax Head of Engineering
    1457.
    my team built a big part of this. we’ll keep shipping more — especially the skills we use internally.›
    LLM score 75 · 4 months ago
  18. Skyler Miao / MiniMax Head of Engineering
    1458.
    M2.7 open weights coming in ~2 weeks. still actively iterating›
    LLM score 92 · 4 months ago
  19. Zixuan Li / Lead Z.ai
    1459.
    Whether driven by luck or analytical capabilities, GLM-5 is currently the only model outperforming the Human Baseline on https://t.co/n3xOlzDjrA.›
    LLM score 85 · 4 months ago
  20. Cameron Wolfe / Researcher at Netflix
    1460.
    Awesome paper. TL;DR: you can 3-4x gains during RL by running well-tuned midtraining beforehand.›
    LLM score 92 · 4 months ago
  21. Sergey Levine / Physical Intelligence Cofounder
    1461.
    Data-driven optimization works best when we can leverage structure.›
    LLM score 92 · 4 months ago
  22. Ben Burtenshaw / Hugging Face Researcher
    1462.
    for open source ai to fully compete, models need to be used like linux.›
    LLM score 82 · 4 months ago
  23. Ronak Malde / DeepMind Researcher
    1463.
    I think people are understating Cursor’s technical achievements for Composer 2.›
    LLM score 92 · 4 months ago
  24. Igor Babuschkin / Cofounder of xAI
    1464.
    It seems inevitable that LLMs will soon surpass human physicists at creating new theories.›
    LLM score 85 · 4 months ago
  25. Skyler Miao / MiniMax Head of Engineering
    1465.
    one more thing from MiniMax-M2.7 with stronger EQ + character consistency,›
    LLM score 92 · 4 months ago
  26. Ben Burtenshaw / Hugging Face Researcher
    1466.
    Or... be cursor and ship high value products without starting from scratch.›
    LLM score 75 · 4 months ago
  27. Roland Gavrilescu / ex MTS at xAI
    1467.
    Yep yep :) been building on this thesis for the past few months, sharing more soon https://t.co/DAtkM22tMu
    LLM score 75 · 4 months ago
  28. Sergey Levine / Physical Intelligence Cofounder
    1468.
    A while ago we figured out that structure enables data-driven design: if we have data of designs + rewards, we can find a design with *higher* reward if we learn a structured function: https://t.co/MZs2chOuoS›
    LLM score 92 · 4 months ago
  29. Boris Cherny / Creator of Claude Code
    1469.
    Join our team https://t.co/VtiL0d4Ylm
    LLM score 75 · 4 months ago
  30. Boris Cherny / Creator of Claude Code
    1470.
    Desktop and https://t.co/h4Wlj4OWot should be feeling faster https://t.co/VHhF3fghu0
    LLM score 85 · 4 months ago
More