Updated on 2024-12-13

Author: lxaw

index.html

@@ -74,7 +74,7 @@ <h1>Where?</h1>
         </p>
         <h1>When?</h1>
         <p>
-        Last time this was edited was 2024-12-12 (YYYY/MM/DD).
+        Last time this was edited was 2024-12-13 (YYYY/MM/DD).
         </p>
         <small><a href="misc.html">misc</a></small>
     </div>

papers/list.json

@@ -1,4 +1,13 @@
 [
+  {
+    "title": "Rho-1: Not All Tokens Are What You Need",
+    "author": "Zhenghao Lin et al",
+    "year": "2024",
+    "topic": "tokens, reference model",
+    "venue": "NeurIPS",
+    "description": "This paper scores tokens using a reference model and then trains a language model to focus on the tokens with higher scores. They find that they can improve performance while training on less tokens.",
+    "link": "https://arxiv.org/pdf/2404.07965"
+  },
   {
     "title": "QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs",
     "author": "Saleh Ashkboos et al",

papers_read.html

@@ -75,10 +75,10 @@ <h1>Here's where I keep a list of papers I have read.</h1>
         I typically use this to organize papers I found interesting. Please feel free to do whatever you want with it. Note that this is not every single paper I have ever read, just a collection of ones that I remember to put down.
     </p>
     <p id="paperCount">
-        So far, we have read 189 papers. Let's keep it up!
+        So far, we have read 190 papers. Let's keep it up!
     </p> 
     <small id="searchCount">
-        Your search returned 189 papers. Nice! 
+        Your search returned 190 papers. Nice! 
     </small>
 
     <div class="search-inputs">
@@ -105,6 +105,16 @@ <h1>Here's where I keep a list of papers I have read.</h1>
         </thead>
         <tbody>
 
+            <tr>
+                <td>Rho-1: Not All Tokens Are What You Need</td>
+                <td>Zhenghao Lin et al</td>
+                <td>2024</td>
+                <td>tokens, reference model</td>
+                <td>NeurIPS</td>
+                <td>This paper scores tokens using a reference model and then trains a language model to focus on the tokens with higher scores. They find that they can improve performance while training on less tokens.</td>
+                <td><a href="https://arxiv.org/pdf/2404.07965" target="_blank">Link</a></td>
+            </tr>
+        
             <tr>
                 <td>QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs</td>
                 <td>Saleh Ashkboos et al</td>