“Now we're going to be way faster than flash attention…” — Latent Space: The AI Engineer Podcast at 13:30
“Now we're going to be way faster than flash attention because we're actually going to find the right way to configure the kernel to call it on that specific input shape. Great. And so people now can just do pip install retention, basically. And that's it. It's a pretty plug and play. Yeah, it's completely a drop-in…”
0:00
32:59