update

index.html

@@ -127,7 +127,7 @@ <h2 class="title is-3">Introduction</h2>
             <p>
               We introduce <u>S</u>tructured <u>S</u>parse
               <u>F</u>ine-<u>T</u>uning (<b>S<sup>2</sup>FT</b>), the first PEFT method for LLMs that achieves high
-              quality, efficient training, and scalable serving simutaneously.
+              quality, efficient training, and scalable serving simultaneously.
               S<sup>2</sup>FT accomplishes this by <b>“selecting sparsely and computing densely”</b>. It selects a few
               heads and channels in the MHA and FFN modules for each Transformer block,
               respectively. Next, it co-permutes weight matrices on both sides of the coupled