Flint: Efficiently Leveraging High Bandwidth Flash for LLM Inference

 (arxiv.org)

2 points | by metrofun 1 hour ago

0 comments