AI300
  1. Jason Wei / AI Researcher at Meta
    1291.
    Fun nine months! My first week i remember we had a long dinner in the cafeteria daydreaming about the cool research directions to pursue, then going to back to our desks to write a basic script to inference llama.›
    LLM score 85 · 4 months ago
  2. Yoshua Bengio
    1292.
    Today, our group at @Mila_Quebec and the lab of @francesarnold at @Caltech just released a new paper I contributed to, exploring how multimodal generative modeling could accelerate protein sciences! ⬇️ https://t.co/Lw53KsOu4J
    LLM score 85 · 4 months ago
  3. Hyung Won Chung / Research Scientist at Meta
    1293.
    We are releasing Muse Spark today.›
    LLM score 92 · 4 months ago
  4. Shuchao Bi / Meta Researcher
    1294.
    Muse Spark represents the beginning of a significant collaborative team effort.›
    LLM score 75 · 4 months ago
  5. Shuchao Bi / Meta Researcher
    1295.
    Meet Muse Spark. This is one step towards personal superintelligence and a great team effort involving numerous nights of training, debugging, improving instability, and fighting reward hacking.›
    LLM score 92 · 4 months ago
  6. Jeff Dean / Chief Scientist at DeepMind
    1296.
    Hedged requests (apparently inspired by the Tail at Scale paper by myself and Luiz Barroso) applied within a single machine to replicating data across DRAM channels and issuing reads to all channels, using the one that comes back first.›
    LLM score 95 · 4 months ago
  7. Merve Noyan / Hugging Face ML Engineer
    1297.
    why local models matter for reproducibility, proven time after time›
    LLM score 85 · 4 months ago
  8. Cameron Wolfe / Researcher at Netflix
    1298.
    I was reading the writeup on economic benchmarks (e.g., GDPval or RLI) from epoch ai last night, and there are some really great practical takeaways on creating realistic benchmarks for agents:›
    LLM score 92 · 4 months ago
  9. Jason Weston / Meta Research Scientist
    1299.
    🏋️Thinking Mid-training: RL of Interleaved Reasoning🎗️›
    LLM score 92 · 4 months ago
  10. Jie Tang / Z.ai Cofounder
    1300.
    GLM 5.1 is coming https://t.co/rSnHYW95Dk.›
    LLM score 85 · 4 months ago
  11. Zixuan Li / Lead Z.ai
    1301.
    If you’ve encountered garbled output like this while using GLM-5 or GLM-5.1 on our official service, the issue is now resolved.›
    LLM score 92 · 4 months ago
  12. Dan Fu / VP of Kernels at Together
    1302.
    Super cool! Small models are super undertrained relative to their test-time scaling properties.›
    LLM score 75 · 4 months ago
  13. Boris Cherny / Creator of Claude Code
    1303.
    Mythos is very powerful, and should feel terrifying.›
    LLM score 85 · 4 months ago
  14. Ben Burtenshaw / Hugging Face Researcher
    1304.
    question: if I want to optimize kernels for apple silicon, how do I benchmark them with macos on apple silicon (as-a-service)?›
    LLM score 92 · 4 months ago
  15. Thomas Wolf / Hugging Face Cofounder
    1305.
    Releasing one of our *largest* robotics project yet in the open›
    LLM score 92 · 4 months ago
  16. Sander Dieleman / DeepMind Research Scientist
    1306.
    Continuous language diffusion strikes back! Flow maps are really starting to come their own as a viable method for language modelling with very fast inference.›
    LLM score 85 · 4 months ago
  17. Fei-Fei Li
    1307.
    Making improvements one step at a time for Marble.›
    LLM score 85 · 4 months ago
  18. Tim Dettmers / Research Scientist at Ai2
    1308.
    I was going crazy because I could not replicate TurboQuant.›
    LLM score 92 · 4 months ago
  19. Merve Noyan / Hugging Face ML Engineer
    1309.
    no shade but @huggingface YT channel has so many gem technical content & podcasts that are watched more 😄 https://t.co/oDzrhHaePg
    LLM score 75 · 4 months ago
  20. Eric Jang / ex VP of AI at 1X Robotics
    1310.
    from my recent blog post: "Based on my own usage patterns, it's beginning to dawn on me how much inference compute we will need in the coming years.›
    LLM score 85 · 4 months ago
  21. Mehtaab Sawhney / OpenAI for Science
    1311.
    In a beautiful recent paper, Vishesh Jain and Clayton Mizgerd used GPT-5.4 Pro to prove a striking result in the theory of Markov chains: https://t.co/KU8nGjHiB7›
    LLM score 92 · 4 months ago
  22. Cameron Wolfe / Researcher at Netflix
    1312.
    Really wonderful paper that perfectly demonstrates a practice people should care more about: setting strong baselines.›
    LLM score 92 · 4 months ago
  23. John Carmack
    1313.
    So many judging tasks could be improved by aggregating partial orderings, and in the limit, just ordering pairs.›
    LLM score 72 · 4 months ago
  24. Thomas Wolf / Hugging Face Cofounder
    1314.
    We’re very excited to deepen our work with the @SAIRfoundation co-founded by Terence Tao.›
    LLM score 75 · 4 months ago
  25. Sander Dieleman / DeepMind Research Scientist
    1315.
    Since we're talking about high-dimensional Gaussians today: human intuition tends to break down in high dimensions (it's firmly rooted in 3D), so I wanted to quantify the incorrect distribution that we think of instead.›
    LLM score 92 · 4 months ago
  26. Lewis Tunstall / Hugging Face Researcher
    1316.
    Terence Tao's SAIR foundation is doing some really cool work on enabling AI4Maths to be open and collaborative›
    LLM score 82 · 4 months ago
  27. Tri Dao / Chief Scientist at Together
    1317.
    Fast muon optimizer coming to consumer cards.›
    LLM score 92 · 4 months ago
  28. Karol Hausman / Physical Intelligence Cofounder
    1318.
    Great introductory video on history of recent robot learning advances: https://t.co/01TUOgRubS starting from SayCan to recent advancements at @physical_int›
    LLM score 92 · 4 months ago
  29. Merve Noyan / Hugging Face ML Engineer
    1319.
    tip: Gemma 4 exposes a thought channel for reasoning›
    LLM score 92 · 4 months ago
  30. Damek Davis / Assoc. Professor Wharton Stats
    1320.
    Ramon van Handel recently proved that every centered 1-subgaussian X is dominated in convex order by cG for a universal c>0, where G∼N(0,1):›
    LLM score 98 · 4 months ago
More