Update index.html

index.html

@@ -292,10 +292,10 @@ <h3 class="title is-3">Forget Quality</h3>
                         <p>Here's the deal with \( p \)-values: a high \( p \)-value is like the test saying, "I can't really tell these two apart." This means the model is doing a great job at forgetting – the Truth Ratios for the unlearned and retain models look pretty similar. On the flip side, a low \( p \)-value is the test's way of saying, "These are definitely not the same." That's not great news for us, because it means there's still some memory of what it's supposed to forget, indicating a leak in privacy and a thumbs-down in unlearning.</p>
                     </div>
 
-                <h3>MAPO: Multiplicative Aggregation of Performance on Unlearning</h3>
+<!--                 <h3>MAPO: Multiplicative Aggregation of Performance on Unlearning</h3>
                 <p>To characterize an unlearning algorithm's efficacy, we aggregate these metrics using the MAPO (Multiplicative Aggregation of Performance On Unlearning) score, calculated across four datasets: forget set, retain set, real authors, and world facts.</p>
                 
-                <p>Metrics are normalized to fall within [0, 1], ensuring higher values indicate better performance. This includes adjusting probabilities and ROUGE scores for the forget set and handling truth ratios differently for each dataset.</p>
+                <p>Metrics are normalized to fall within [0, 1], ensuring higher values indicate better performance. This includes adjusting probabilities and ROUGE scores for the forget set and handling truth ratios differently for each dataset.</p> -->
 
 
               </div>