LLM inference, visually explained: why a 13B model reads 26 GB per token (youtube.com) 1 points by ryansakhala 5d ago ↗ HN
1 comment
[ 185 ms ] story [ 237 ms ] thread