AI300
  1. Sayak Paul / Hugging Face Researcher
    1441.
    I think going through an important PR of a repository like PyTorch, even as an external observer, is an under-appreciated learning resource.›
    LLM score 92 · 4 months ago
  2. Ben Burtenshaw / Hugging Face Researcher
    1442.
    this is huge for the community.›
    LLM score 92 · 4 months ago
  3. Clive Chan / OpenAI Hardware
    1443.
    it's crazy how a single one of the 160 SMs in a GB200 has more FP32 compute than this 15kW fridge.›
    LLM score 92 · 4 months ago
  4. Sebastien Bubeck / OpenAI MTS
    1444.
    I've been in lots of places in my career.›
    LLM score 92 · 4 months ago
  5. Cameron Wolfe / Researcher at Netflix
    1445.
    I've been reading a lot about rubrics-as-rewards (RaR) for RL.›
    LLM score 75 · 4 months ago
  6. Andrew Lampinen / Research Scientist at DeepMind
    1446.
    Interesting results by @eghbal_hosseini on how language models representation geometry evolves during different types of in-context learning! https://t.co/yfCfQKcOt9
    LLM score 85 · 4 months ago
  7. Sander Dieleman / DeepMind Research Scientist
    1447.
    Latent representations are pervasive in modern generative modelling at scale, because iterative refinement in a compact latent space is much more cost-effective, and latents can be decoded to pixels in a single forward pass.›
    LLM score 92 · 4 months ago
  8. Andrej Karpathy / AI researcher
    1448.
    Enabled fp8 training for +4.3% improvement to "time to GPT-2", down to 2.91 hours now.›
    LLM score 92 · 4 months ago
  9. Noam Brown / OpenAI Research Scientist
    1449.
    Every ~6 months I hear people claim @OpenAI isn’t doing real research and is just incrementally improving ChatGPT.›
    LLM score 92 · 4 months ago
  10. Andy Jones / Anthropic Research Engineer
    1450.
    my guess is that in low-competition regimes AI behaves like a human because that's a strong prior on how entities should behave.›
    LLM score 92 · 4 months ago
  11. Mark Chen / OpenAI Chief Research Officer
    1451.
    Just because we focus on research doesn't mean we will pursue *all* research.›
    LLM score 82 · 4 months ago
  12. Thang Luong / DeepMind Principal Scientist
    1452.
    Another cute comic for this work: a nice collaborative story between #Aletheia and human experts for cracking Erdős problems, thanks @AwaitSeedIgnite! I wish we have ComicBanana for all papers :) First four pages here and full one in this link https://t.co/vJBHR2ZKM9.›
    LLM score 85 · 4 months ago
  13. Mark Chen / OpenAI Chief Research Officer
    1453.
    How does OpenAI balance long-term research bets with product-forward research fundamentals? ›
    LLM score 92 · 4 months ago
  14. Albert Gu / Cartesia Chief Scientist
    1454.
    one question we got a lot about H-Net is how it compares to MoE.›
    LLM score 92 · 4 months ago
  15. Jim Fan / NVIDIA Director of Robotics
    1455.
    The Second Pre-training Paradigm(blog)
    LLM score 100 · 4 months ago
  16. Ethan Shen / Ai2 Researcher
    1456.
    Excited to announce several updates to Ai2 Open Coding Agents:›
    LLM score 85 · 4 months ago
  17. Wenting Zhao / Alibaba Qwen Researcher
    1457.
    This release is an emtional one for me because I had stayed up so much for it 🥹 It has been truly amazing to see this model becomes better bit by bit through every change we make, and we have come a long way.›
    LLM score 75 · 4 months ago
  18. Omar Khattab / MIT CSAIL Asst professor
    1458.
    I guess the original interpretation of the bitter lesson is arguably at odds with LLM practice so far.›
    LLM score 92 · 4 months ago
  19. Binyuan Hui / Alibaba Qwen Research Scientist
    1459.
    I believe agents are about to enter an era of efficient execution.›
    LLM score 92 · 4 months ago
  20. alphaXiv
    1460.
    It's probably not an exaggeration that scientific figures are the soul of a paper›
    LLM score 92 · 4 months ago
  21. Ted Sanders / OpenAI Researcher
    1461.
    no one knows how good AI will get, but prospects are so exciting that legions now labor to build it, gauging (correctly) that the risk of building the wrong thing beats building nothing.›
    LLM score 82 · 4 months ago
  22. Merve Noyan / Hugging Face ML Engineer
    1462.
    came across this gem today > when modern inpainters (diffusion models + VAE) inpaint the image, it experiences spectral shifts that detectors exploit›
    LLM score 85 · 4 months ago
  23. Sayak Paul / Hugging Face Researcher
    1463.
    We shipped a new CP backend in Diffusers and seems like it's our best one so far 🔥›
    LLM score 92 · 4 months ago
  24. Roland Gavrilescu / ex MTS at xAI
    1464.
    We used to have a rule at xAI: always keep the GPUs running.›
    LLM score 85 · 4 months ago
  25. Ben Burtenshaw / Hugging Face Researcher
    1465.
    If you're doing AI in the medical domain, follow OpenMed.›
    LLM score 75 · 4 months ago
  26. Jascha Sohl-Dickstein / Anthropic Researcher
    1466.
    When AI fails, will it do so by coherently pursuing the wrong goals? Or will it fail the way humans often fail, and take incoherent actions that don't pursue any consistent goal.›
    LLM score 92 · 4 months ago
  27. Alex Zhang / MIT CSAIL PhD
    1467.
    This is bar for bar one of the core pieces of intuition behind how we came up with RLMs, why they currently work, and why they are so promising for this year!›
    LLM score 92 · 4 months ago
  28. Omar Khattab / MIT CSAIL Asst professor
    1468.
    This sounds so cool: Auto-regressive late interaction / multi-vector retrieval models.›
    LLM score 12 · 4 months ago
  29. Quoc Le / Google Fellow
    1469.
    Excited to share our latest work: "Semi-Autonomous Mathematics Discovery with Gemini." We used Gemini to systematically evaluate 700 "open" conjectures in the Erdős Problems database.›
    LLM score 92 · 4 months ago
  30. Lucas Beyer / Meta Researcher
    1470.
    Ah! Some people might know that I've long been suspicious of looking almost only at perplexity for pretrain...›
    LLM score 92 · 4 months ago
More