Update README.md
Browse files
README.md
CHANGED
|
@@ -258,7 +258,9 @@ cross-entropy loss to align the student with gold-label financial sentiment.
|
|
| 258 |
</tbody>
|
| 259 |
</table>
|
| 260 |
|
| 261 |
-
|
|
|
|
|
|
|
| 262 |
|
| 263 |
<ul>
|
| 264 |
<li><strong>KL Divergence</strong> between teacher soft targets and student logits (soft label transfer)</li>
|
|
|
|
| 258 |
</tbody>
|
| 259 |
</table>
|
| 260 |
|
| 261 |
+
<h3>Loss Function</h3>
|
| 262 |
+
|
| 263 |
+
The Stage 1 total loss combines:
|
| 264 |
|
| 265 |
<ul>
|
| 266 |
<li><strong>KL Divergence</strong> between teacher soft targets and student logits (soft label transfer)</li>
|