Fast Inference of Mixture-of-Experts Language Models with Offloading

The paper explores strategies for running large sparse Mixture-of-Experts (MoE) language models on consumer hardware with limited accelerator memory, proposing a novel offloading strategy that allows for efficient execution on desktop hardware and free-tier Google Colab instances.

https://arxiv.org/abs//2312.17238

YouTube: @arxivpapers

TikTok: TikTok: arxiv_papers

Apple Podcasts: https://podcasts.apple.com/us/podcast...

Spotify: https://podcasters.spotify.com/pod/sh...