update benchmark value

README.md

@@ -88,8 +88,8 @@ The compression pipeline operates in two stages:
 | Qwen-Image | 0.539 | **0.548** |
 | Z-Image | **0.546** | 0.535 |
 | Ovis-Image | 0.530 | 0.521 |
-| **Amber-Image-10B** | 0.489 | 0.470 |
-| **Amber-Image-6B** | 0.477 | 0.456 |
+| **Amber-Image-10B** | 0.504 | 0.502 |
+| **Amber-Image-6B** | 0.491 | 0.486 |
 
 ### Text Rendering
 
@@ -103,7 +103,7 @@ The compression pipeline operates in two stages:
 | Z-Image | 0.935 | 0.936 |
 | Ovis-Image | 0.922 | **0.964** |
 | **Amber-Image-10B** | 0.911 | 0.915 |
-| **Amber-Image-6B** | 0.838 | 0.828 |
+| **Amber-Image-6B** | 0.870 | 0.876 |
 
 **CVTG-2K** — Complex visual text generation. Amber-Image-10B achieves the highest CLIPScore among all compared models, indicating strong semantic alignment. Word accuracy remains stable across increasing region counts.