From bf85bf3aa0dccc4b9383d4c8481e45d85a26f2b4 Mon Sep 17 00:00:00 2001
From: laoliu5280 <haokunliu@uchicago.edu>
Date: Wed, 20 Nov 2024 15:13:05 -0600
Subject: [PATCH] change wording

---
 index.html | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/index.html b/index.html
index 90ec0b2..f62a175 100644
--- a/index.html
+++ b/index.html
@@ -358,7 +358,7 @@ <h2 class="title is-3">Results</h2>
         
         <h3 class="subtitle is-4">Automatic Evaluation</h3>
         <p>
-          Combining literature-based and data-driven methods produced the best performance across tasks. The integrated approach improved improvement of 11.92% over few-shot methods and 16.54% over literature-based methods for GPT, and 6.03% over few-shot methods and 14.97% over literature-based methods for Llama.
+          Combining literature-based and data-driven methods produced the best performance across tasks. The integrated approach achieved an accuracy improvement of 11.92% over few-shot methods and 16.54% over literature-based methods for GPT, and 6.03% over few-shot methods and 14.97% over literature-based methods for Llama.
         </p>
         <p>
           Our framework with literature + data showed significant advantages over zero-shot, few-shot, data-driven and literature-based methods, particularly in handling OOD datasets.