7x speed improvement for LLaMA in less than 10 lines of code #1 Post by hack_ml » Mon, Jul 10, 2023, 8:29 PM UTC 7x speed improvement for LLaMA in less than 10 lines of codegithub.com
Re: 7x speed improvement for LLaMA in less than 10 lines of code #2 Post by brucethemoose2 » Mon, Jul 10, 2023, 9:34 PM UTC Is that 5s per token?