Skip to content

Commit

Permalink
update Phi numbers
Browse files Browse the repository at this point in the history
  • Loading branch information
sanjay920 committed Jul 9, 2024
1 parent 5ea9b0e commit 80284ef
Show file tree
Hide file tree
Showing 2 changed files with 37 additions and 35 deletions.
50 changes: 26 additions & 24 deletions docs/docs/models/Phi.md
Original file line number Diff line number Diff line change
Expand Up @@ -43,32 +43,34 @@ Phi-3 is a state of the art, lightweight model. It performs exceptionally well d
<tr>
<td>Phi-3 Mini 128k Instruct</td>
<td>-</td>
<td>68.17</td>
<td>25.90</td>
<td>80.44</td>
<td>28.12</td>
<td>7.92</td>
<td>51</td>
<td>45</td>
<td>64</td>
<td>0.31875</td>
<td>0.28125</td>
<td><strong>0.51875</strong></td>
<td>69.36</td>
<td>27.01</td>
<td>83.7</td>
<td>32.92</td>
<td>8.02</td>
<td>21</td>
<td>72</td>
<td>67</td>
<td>0.13125</td>
<td>0.45000</td>
<td>0.340625</td>
</tr>
<tr>
<td>Rubra Enhanced Phi-3 Mini 128k Instruct</td>
<td>65.71%</td>
<td>66.66</td>
<td>29.24</td>
<td>74.09</td>
<td>26.84</td>
<td>7.45</td>
<td>45</td>
<td>51</td>
<td>64</td>
<td>0.28125</td>
<td>0.31875</td>
<td>0.48125</td>
<td>70.0%</td>
<td>67.87</td>
<td>29.69</td>
<td>79.45</td>
<td>30.80 </td>
<td>8.21</td>
<td>72</td>
<td>21</td>
<td>67</td>
<td>0.45000</td>
<td>0.13125</td>
<td><strong>0.659375</strong></td>
</tr>
</tbody>
</table>
</table>

* The above is based on the Phi-3 Mini that was updated by Microsoft in June 2024. The original Phi-3 mini came out in April and the Rubra enhanced model has been trained on both versions
22 changes: 11 additions & 11 deletions docs/src/components/BenchmarkTable.js
Original file line number Diff line number Diff line change
Expand Up @@ -117,21 +117,21 @@ const data = [
model: 'Phi-3 Mini 128k Instruct',
params: 3.82,
functionCalling: '-',
mmlu: '68.17',
gpqa: '25.90',
gsm8k: '80.44',
math: '28.12',
mtBench: '7.92',
mmlu: '69.36',
gpqa: '27.01',
gsm8k: '83.7',
math: '32.92',
mtBench: '8.02',
},
{
model: 'Rubra Phi-3 Mini 128k Instruct',
params: 4.27,
params: 4.73,
functionCalling: '70.00%',
mmlu: '66.66',
gpqa: '29.24',
gsm8k: '74.09',
math: '26.84',
mtBench: '7.45',
mmlu: '67.87',
gpqa: '29.69',
gsm8k: '79.45',
math: '30.80',
mtBench: '8.21',
},
{
model: 'Qwen2-7B-Instruct',
Expand Down

0 comments on commit 80284ef

Please sign in to comment.