update readme

README.md

@@ -78,7 +78,6 @@ The model was trained on 1 million carefully curated single-frame samples from p
 ## Benchmarks
 
 Performance metrics on our internal evaluation set:
-
 | Model | Avg Judge Score | ROUGE-1 | ROUGE-2 | ROUGE-L | BLEU |
 |-------|-----------------|---------|---------|---------|------|
 | cliptagger_12b | **3.53** | **0.674** | **0.404** | **0.520** | **0.267** |
@@ -87,14 +86,16 @@ Performance metrics on our internal evaluation set:
 
 ### Benchmark Visualizations
 
-<p align="center">
-  <img src="./assets/judge-score.png" alt="Average Judge Score Comparison" width="49%" />
-  <img src="./assets/rouge-1.png" alt="ROUGE-1 Score Comparison" width="49%" />
-</p>
-<p align="center">
-  <img src="./assets/rouge-L.png" alt="ROUGE-L Score Comparison" width="49%" />
-  <img src="./assets/bleu.png" alt="BLEU Score Comparison" width="49%" />
-</p>
+<table>
+  <tr>
+    <td><img src="./assets/judge-score.png" alt="Average Judge Score Comparison" width="100%" /></td>
+    <td><img src="./assets/rouge-1.png" alt="ROUGE-1 Score Comparison" width="100%" /></td>
+  </tr>
+  <tr>
+    <td><img src="./assets/rouge-L.png" alt="ROUGE-L Score Comparison" width="100%" /></td>
+    <td><img src="./assets/bleu.png" alt="BLEU Score Comparison" width="100%" /></td>
+  </tr>
+</table>
 
 FP8 quantization showed no measurable quality degradation compared to bf16 precision.