Pith Wave Signal

2 stories tagged #Mixture-of-Experts

Newest Most read
  1. Trillion-Parameter AI Runs on Budget GPU with Legacy Memory
    tech

    Trillion-Parameter AI Runs on Budget GPU with Legacy Memory

    A Chinese enthusiast runs Moonshot AI's Kimi K2.5 on an RTX 3060 using Intel Optane memory, achieving 4 tokens per second.

    last wk. 1 min read
  2. Tencent Unveils Hy3: The Efficient Chinese AI Model Outperforming Rivals
    tech

    Tencent Unveils Hy3: The Efficient Chinese AI Model Outperforming Rivals

    Tencent's new Hy3 AI model, a 295 billion parameter Mixture-of-Experts architecture, achieves remarkable efficiency and performance gains, particularly in coding and reasoning tasks.

    last mo. 1 min read