From 26cb8f444ccfa4976c57f0cea2465b0ccacd20b1 Mon Sep 17 00:00:00 2001 From: mark <51519902+mark-lord@users.noreply.github.com> Date: Thu, 26 Sep 2024 16:52:37 +0100 Subject: [PATCH] Update README.md description of changes --- README.md | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/README.md b/README.md index bd180975..e4de4cf9 100644 --- a/README.md +++ b/README.md @@ -1,3 +1,12 @@ +# This repo features changes to MLX_LM. +Specfically main changes are from line 318 in utils.py; I kind of shoe-horned in the same kv-cache saving system as cache_prompt.py. + +Then had to change generate.py to take --save-kv-cache as an argument which was just a matter of changing the response handling on line 232 of generate.py. + +Messy, but seems to work! + +/// + # MLX Examples This repo contains a variety of standalone examples using the [MLX