Minor update

Author: nmd2k

leaderboards/codemmlu/index.html

@@ -141,7 +141,7 @@
       <h1 class="text-nowrap mt-5">CodeMMLU Leaderboard</h1>
       <h3 class="fw-light text-nowrap">
         <small id="warning"> 
-          <p>CodeMMLU: A Multi-Task Benchmark for Assessing Code Understanding Capabilities of CodeLLMs</p>
+          <p>A Multi-Task Benchmark for Assessing Code Understanding Capabilities of CodeLLMs</p>
         <br/></small>
       </h3>
       <div class="d-flex flex-row justify-content-center gap-3">
@@ -151,7 +151,7 @@ <h3 class="fw-light text-nowrap">
             alt="blog"
             class="img-fluid"
         /></a>
-        <a href="https://arxiv.org/abs/2410.01999v1#:~:text=View%20a%20PDF%20of%20the%20paper%20titled%20CodeMMLU:%20A%20Multi-Task"
+        <a href="https://arxiv.org/abs/2410.01999"
           ><img
             src="https://img.shields.io/badge/2410.01999-red?style=for-the-badge&label=arXiv"
             alt="leaderboard"
@@ -186,7 +186,7 @@ <h3 class="fw-light text-nowrap">
           id="Complete"
           checked
         />
-        <label class="btn btn-outline-primary" for="Complete">Complete</label>
+        <!-- <label class="btn btn-outline-primary" for="Complete">Complete</label> -->
         <!-- <input type="radio" class="btn-check" name="btnradio" id="Instruct" />
         <label class="btn btn-outline-primary" for="Instruct">Instruct</label> -->
       </div>

leaderboards/codemmlu/results.json

@@ -9,7 +9,7 @@
         "realtask_accuracy": 38.26,
         "syntactic_accuracy": 67.22,
         "semantic_accuracy": 66.08,
-        "prompted": false,
+        "prompted": true,
         "size": null,
         "direct_complete": false,
         "lazy": false,
@@ -25,7 +25,7 @@
         "realtask_accuracy": 77.18,
         "syntactic_accuracy": 60.41,
         "semantic_accuracy": 57.81,
-        "prompted": false,
+        "prompted": true,
         "size": null,
         "direct_complete": false,
         "lazy": false,
@@ -41,7 +41,7 @@
         "realtask_accuracy": 45.26,
         "syntactic_accuracy": 61.68,
         "semantic_accuracy": 53.65,
-        "prompted": false,
+        "prompted": true,
         "size": null,
         "direct_complete": false,
         "lazy": false,