PowerInfer is a new inference engine designed to run large language models more quickly on a single consumer-grade GPU, achieving speeds up to 11.69x faster than llama.cpp. Read the paper on Hugging Face.
via Hugging Face.
Read next
Rating: ★★★★☆
Upcoming features and improvements:
[x] Saved places integration
[ ] Report feature
[ ] Add Twitter likes and Bookmarks
[x] Goodreads integration
For updating all current Substack posts in your inbox to "Archived" - copy and paste this into the console on your browser:
let isArchiving = false;
let lastProcessedPost = null;
function archiveVisiblePosts() {
if (isArchiving) return;
isArchiving = true;
const posts = document.querySelectorAll('._linkRow_1qqfh_28.reader2-post-container');
let delay = 0;
const delayIncrement = 300;Subscribe to Newsletter
Stay up to date with email notifications of new posts.