Flint: Efficiently Leveraging High Bandwidth Flash for LLM Inference

2 pointsposted 5 hours ago
by metrofun
(arxiv.org)

No comments yet