From 200cb125045401413a16c4605c05ad8b87b87d3f Mon Sep 17 00:00:00 2001 From: Paul Gauthier Date: Sun, 17 Dec 2023 16:46:02 -0800 Subject: [PATCH] copy --- docs/unified-diffs.md | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/docs/unified-diffs.md b/docs/unified-diffs.md index 44eb27fdc..e5cedf8d3 100644 --- a/docs/unified-diffs.md +++ b/docs/unified-diffs.md @@ -25,10 +25,8 @@ designed to both provoke and quantify lazy coding. It consists of 39 python refactoring tasks, which ask GPT to remove a non-trivial method from a class and make it a stand alone function. -GPT-4 Turbo is prone to being lazy on this sort of task, because it's mostly a -"cut & paste" of code from one place in a file to another. -Rather than writing out the code, GPT often just leaves -a comment like +GPT-4 Turbo is prone to being lazy on this sort of copy/paste task, +by leaving a comment like "...include the body of the original method..." This new laziness benchmark produced the following results with `gpt-4-1106-preview`: