Cerebras runs Llama 3.1 70B at 2,100 tokens per second, live demo available (twitter.com) 6 points by modeless 2y ago ↗ HN
1 comment
[ 3.3 ms ] story [ 14.0 ms ] thread