From f58c59f131ac82fe259432d38d6b939ecd490543 Mon Sep 17 00:00:00 2001 From: Da Yin <42200725+WadeYin9712@users.noreply.github.com> Date: Wed, 8 Nov 2023 15:14:29 -0800 Subject: [PATCH] Update index.html --- docs/index.html | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) diff --git a/docs/index.html b/docs/index.html index 33ba7d9..37ee39a 100644 --- a/docs/index.html +++ b/docs/index.html @@ -3,7 +3,7 @@ <head> <meta charset="utf-8"> <meta name="viewport" content="width=device-width, initial-scale=1"> - <title>Lumos: Language Agents with Unified Data Formats, Modular Design, and Open-Source LLMs</title> + <title>🪄 Lumos: Language Agents with Unified Data Formats, Modular Design, and Open-Source LLMs</title> <!-- Global site tag (gtag.js) - Google Analytics --> <script async src="https://www.googletagmanager.com/gtag/js?id=G-PYVRSFMDRL"></script> @@ -369,7 +369,7 @@ <h2 class="title is-3">Comparison with Baseline Formulations</h2> <img src="static/images/lumos_results_2.png" class="center"> </p> <p> - We compare <strong>Lumos</strong> formulation with other baseline formulations to train open-source agents. The baseline formulations are Vanilla Training, Chain-of-Thought Training, + We compare <strong>Lumos</strong> formulation with other baseline formulations to train open-source agents. The baseline formulations are Chain-of-Thought Training and Integrated Agent Training. </p> <p> @@ -399,8 +399,7 @@ <h2 class="title is-3">Generalizability of Lumos</h2> </p> <p> We find that after the unified training, <strong>Lumos</strong> would have slightly higher - performance on web and complex QA tasks. We also observe that <strong>Lumos</strong> can bring an improvement over domain-specific - agents 5-10 reward improvement, and also better performance than larger agents with 13B and 30B sizes. + performance on web and complex QA tasks. We also observe that <strong>Lumos</strong> can bring an improvement over domain-specific agents 5-10 reward improvement, and also better performance than larger agents with 13B and 30B sizes. </p> </div> </div> @@ -425,7 +424,7 @@ <h2 class="title is-3">Further Analysis on Annotations</h2> We also conduct deeper analysis about annotation quality and the choice of annotation formats. We answer the following questions: <ul> <li><strong>Q1: How good is our converted training annotations?</strong></li> - <li><strong>Q2: Would it be better if we adopt low-level subgoals instead of our proposed high-level subgoals? </strong></li> + <li><strong>Q2: Would it be better if we adopt low-level subgoals instead of our proposed high-level subgoals?</strong></li> </ul> </p> <p>