CAD: Disaggregating Core Attention for Efficient Long-Context LLM Training

3 points | by ginda307 2 hours ago

1 comments

  • 2 hours ago
    [deleted]