Hacker GitHub

← All repos

turboderp/exllama

View on GitHub ↗ ★ 2,946 Python

A more memory-efficient rewrite of the HF transformers implementation of Llama for use with quantized weights.

18 mentions on Hacker News by 11 people: 1 story, 17 comments · first 2023-05-23 · latest 2025-03-24

Mentions per year 18 mentions · busiest year 2023 (15)
20082026
AllStoriesComments NewestTopOldest
  1. ▲ 3
    ExLlama: Memory efficient way to run Llama
    story by Palmik · 2023-08-16 · 0 comments