From cd8314a0f5343c7c8936c1e28443555ca4c1834f Mon Sep 17 00:00:00 2001
From: Matt Post <mattpost@microsoft.com>
Date: Fri, 1 Nov 2024 12:48:38 -0400
Subject: [PATCH] Fix stray tag

---
 data/xml/2024.findings.xml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/data/xml/2024.findings.xml b/data/xml/2024.findings.xml
index 786d74b5cd..0415ae52a6 100644
--- a/data/xml/2024.findings.xml
+++ b/data/xml/2024.findings.xml
@@ -21760,7 +21760,7 @@
       <author><first>Xingdi</first><last>Yuan</last><affiliation>Microsoft Research</affiliation></author>
       <author><first>Bang</first><last>Liu</last><affiliation>University of Montreal</affiliation></author>
       <pages>3534-3568</pages>
-      <abstract>Large language models (LLMs), trained on vast amounts of internet data, have developed a broad understanding of the world, enhancing the decision-making capabilities of embodied agents. This success is largely due to the comprehensive and in-depth domain knowledge within their training datasets. However, the extent of this knowledge can vary across different domains, and existing methods often assume that LLMs have a complete understanding of their environment, overlooking potential gaps in their grasp of actual world dynamics. To address this gap, we introduce <b>Di</b>scover, <b>V</b>erify, and <b>E</b>volve (<b/>), a framework that <b>discovers</b> world dynamics from a small number of demonstrations, <b>verifies</b> the correctness of these dynamics, and <b>evolves</b> new, advanced dynamics tailored to the current situation. Through extensive evaluations, we analyze the impact of each component on performance and compare the automatically generated dynamics from with human-annotated world dynamics. Our results demonstrate that LLMs guided by can make better decisions, achieving rewards comparable to human players in the Crafter environment.</abstract>
+      <abstract>Large language models (LLMs), trained on vast amounts of internet data, have developed a broad understanding of the world, enhancing the decision-making capabilities of embodied agents. This success is largely due to the comprehensive and in-depth domain knowledge within their training datasets. However, the extent of this knowledge can vary across different domains, and existing methods often assume that LLMs have a complete understanding of their environment, overlooking potential gaps in their grasp of actual world dynamics. To address this gap, we introduce <b>Di</b>scover, <b>V</b>erify, and <b>E</b>volve (DiVE), a framework that <b>discovers</b> world dynamics from a small number of demonstrations, <b>verifies</b> the correctness of these dynamics, and <b>evolves</b> new, advanced dynamics tailored to the current situation. Through extensive evaluations, we analyze the impact of each component on performance and compare the automatically generated dynamics from with human-annotated world dynamics. Our results demonstrate that LLMs guided by can make better decisions, achieving rewards comparable to human players in the Crafter environment.</abstract>
       <url hash="100611d1">2024.findings-emnlp.202</url>
       <bibkey>sun-etal-2024-enhancing-agent</bibkey>
     </paper>