FlashAttention-2: Making Transformers 800% faster AND exact (youtube.com)

2 points by kken 3y ago ↗ HN

0 comments

[ 5.2 ms ] story [ 12.9 ms ] thread

No comments yet.

A showcase of shard-db. Data from the Hacker News API. Refreshed every 15 minutes.

Source · Live DB stats