update result captions

Author: Tianjiao Yu

projects/ece/index.html

@@ -249,10 +249,8 @@ <h2 class="title is-3">Results</h2>
               <div style="text-align: center; padding: 0 0 20px 0;">
               <img src="./static/images/r2.png" class="interpolation-image" alt="ece results." width="95%" />
               <p class="has-text-justified">
-                <b>ECE and AUROC across models and Execution Policies.</b>  
-                Each row contains metrics for <b>ECE</b> (top) and <b>AUROC</b> (bottom) for three different models: GPT-4V, MineLLM, and STEVE,  
-                under different uncertainty elicitation and execution policies.  
-                <span style="color: red;">Red lines</span> represent the average ECE and AUROC with no execution policy applied.
+                 <b>ECE and AUROC across Models and Execution Policies.</b> Bars present ECE (top, lower is better) and AUROC (bottom, higher is better)
+                 under different elicitation strategies. <span style="color: rgb(255, 140, 0); font-weight: bold;">Red dashed lines</span> are metrics for Vanilla elicitation with no execution policy applied.
               </p>
             </div>
           </div>