Tag: Training

  1. Scaling laws look precise, but the way you fit them is fragile (lilianweng.github.io)
    AI · · June 30, 2026
  2. How frontier post-training got complicated, and why distillation now sits at the center (interconnects.ai)
    AI · · June 20, 2026
  3. Async RL gets cheap to ship when 99% of weights do not change (huggingface.co)
    AI · · May 27, 2026
  4. Muon, the optimizer everyone switched to, quietly kills neurons (importai.substack.com)
    AI · · May 22, 2026
  5. Decoupled DiLoCo trains a 12B model across four regions, 20x faster (deepmind.google)
    Engineering · · April 23, 2026

← all tags