Comment by rnxrx

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