3 ms·

I'm curious about why the performance gains mentioned were so substantial for Qwen vs Llama?

• AlekseiSavin 1y ago
  it looks like llama.cpp has some performance issues with bf16