Update README.md

README.md

@@ -28,7 +28,7 @@ license: llama3.1
 - **Model Developers:** Neural Magic
 
 Quantized version of [Meta-Llama-3.1-405B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-405B-Instruct).
-It achieves an average score of x.x on the [OpenLLM](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard) benchmark (version 1), whereas the unquantized model achieves x.x.
+It achieves an average score of 86.01 on the [OpenLLM](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard) benchmark (version 1), whereas the unquantized model achieves 86.63.
 
 ### Model Optimizations
 
@@ -146,11 +146,11 @@ This version of the lm-evaluation-harness includes versions of ARC-Challenge, GS
   <tr>
    <td>MMLU (5-shot)
    </td>
-   <td>xx.xx
+   <td>86.25
    </td>
-   <td>xx.xx
+   <td>85.97
    </td>
-   <td>xx.xx%
+   <td>99.67%
    </td>
   </tr>
   <tr>
@@ -176,37 +176,41 @@ This version of the lm-evaluation-harness includes versions of ARC-Challenge, GS
   <tr>
    <td>Hellaswag (10-shot)
    </td> 
-   <td>xx.xx
+   <td>88.33
    </td>
-   <td>xx.xx%
+  <td>88.16
+   </td>
+   <td>99.80%
    </td>
   </tr>
   <tr>
    <td>Winogrande (5-shot)
    </td>
-   <td>xx.xx
+   <td>87.21
+   </td>
+  <td>85.95
    </td>
-   <td>xx.xx%
+   <td>98.55%
    </td>
   </tr>
   <tr>
    <td>TruthfulQA (0-shot)
    </td>
-   <td>xx.xx
+   <td>64.64
    </td>
-   <td>xx.xx
+   <td>64.75
    </td>
-   <td>xx.xx%
+   <td>100.17%
    </td>
   </tr>
   <tr>
    <td><strong>Average</strong>
    </td>
-   <td><strong>xx.xx</strong>
+   <td><strong>86.63</strong>
    </td>
-   <td><strong>xx.xx</strong>
+   <td><strong>86.01</strong>
    </td>
-   <td><strong>xx.xx%</strong>
+   <td><strong>99.28%</strong>
    </td>
   </tr>
 </table>