Update README.md

README.md

@@ -30,7 +30,7 @@ We also provide goal vectors (`delta_*` or `target_*`) for all goal dimensions.
 
 ## Results
 
-Here, we will retain the most recent versions of SteerBench for top-performing models. Shown here: steering error of recent models (`mean (std. dev.)`).
+Here, we will retain the most recent versions of SteerBench for top-performing models. Shown here: steering error of recent models (`median (IQR)`).
 <table>
   <tr>
     <td><b>Model family</b></td>
@@ -40,75 +40,75 @@ Here, we will retain the most recent versions of SteerBench for top-performing m
   <tr>
     <td rowspan=5><b>Llama3</b></td>
     <td>Llama3-8B</td>
-    <td>0.53 (0.22)</td>
+    <td>0.495 (0.252)</td>
   </tr>
   <tr>
     <td>Llama3.1-8B</td>
-    <td>0.50 (0.20)</td>
+    <td>0.452 (0.256)</td>
   </tr>
   <tr>
     <td>Llama3-70B</td>
-    <td>0.50 (0.21)</td>
+    <td>0.452 (0.239)</td>
   </tr>
   <tr>
     <td>Llama3.1-70B</td>
-    <td>0.48 (0.21)</td>
+    <td>0.452 (0.239)</td>
   </tr>
   <tr>
     <td>Llama3.3-70B</td>
-    <td>0.49 (0.21)</td>
+    <td>0.452 (0.256)</td>
   </tr>
   <tr>
     <td rowspan=4><b>GPT</b></td>
     <td>GPT-3.5 turbo</td>
-    <td>0.55 (0.23)</td>
+    <td>0.535 (0.251)</td>
   </tr>
   <tr>
     <td>GPT-4 turbo</td>
-    <td>0.56 (0.22)</td>
+    <td>0.515 (0.266)</td>
   </tr>
   <tr>
     <td>GPT-4o</td>
-    <td>0.50 (0.20)</td>
+    <td>0.474 (0.239)</td>
   </tr>
   <tr>
     <td>GPT-4.1</td>
-    <td>0.46 (0.19)</td>
+    <td>0.429 (0.203)</td>
   </tr>
   <tr>
     <td rowspan=2><b>OpenAI o-series</b></td>
     <td>o1-mini</td>
-    <td><i>0.53 (0.21)*</i></td>
+    <td><i>0.495 (0.261)*</i></td>
   </tr>
   <tr>
     <td>o3-mini</td>
-    <td><i>0.55 (0.22)*</i></td>
+    <td><i>0.515 (0.232)*</i></td>
   </tr>
   <tr>
     <td rowspan=2><b>Deepseek-R1</b></td>
     <td>Deepseek-R1-Distill-Llama-8B</td>
-    <td>0.57 (0.24)</td>
+    <td>0.535 (0.281)</td>
   </tr>
   <tr>
     <td>Deepseek-R1-Distill-Llama-70B</td>
-    <td>0.52 (0.22)</td>
+    <td>0.474 (0.256)</td>
   </tr>
   <tr>
     <td rowspan=4><b>Qwen3</b></td>
     <td>Qwen-32B (no thinking)</td>
-    <td>0.57 (0.23)</td>
+    <td>0.535 (0.271)</td>
   </tr>
   <tr>
     <td>Qwen-32B (thinking)</td>
-    <td>0.57 (0.23)</td>
+    <td>0.535 (0.271)</td>
   </tr>
   <tr>
     <td>Qwen-30B-A3B (no thinking)</td>
-    <td>0.52 (0.22)</td>
+    <td>0.495 (0.273)</td>
   </tr>  
   <tr>
     <td>Qwen-30B-A3B (thinking)</td>
-    <td>0.52 (0.22)</td>
+    <td>0.495 (0.2273</td>
   </tr>
 </table>
 * >1% invalid response rate