Update README.md

README.md

@@ -91,6 +91,12 @@ The model was trained using **Knowledge Distillation**. The student (`TinyBERT`)
 | **Teacher** (XLM-R) | 278M | ~1 GB | ~360 samples/sec | 100% (Baseline) |
 | **Student** (TinyBERT) | **11M** | **42 MB** | **~3,300 samples/sec** | **96%** |
 
+## Evaluation Results
+
+The model shows strong separation between the routing categories, with minimal confusion between semantically distinct classes (e.g., Address vs. Regex).
+
+![Confusion Matrix](confusion_matrix.png)
+
 ## Usage
 
 This model outputs **Multi-Label Probabilities** for the 4 engines. We recommend a threshold of **0.5** to trigger a route.