Update README.md

README.md

@@ -67,9 +67,9 @@ This provides strong legal-domain grounding before fine-tuning.
 | Max summary length | 256 tokens |
 | FP16 | Yes |
 | Warmup | 500 steps |
-| Scheduler | Linear |
+| Logging-steps | 50 steps |
 
-Training was performed on **Kaggle T4 GPU (16GB)**.
+Training was performed on **Kaggle P100 GPU (16GB)**.
 
 ---
 
@@ -78,16 +78,16 @@ Training was performed on **Kaggle T4 GPU (16GB)**.
 ## **ROUGE Scores (Test Set)**  
 | Metric | F1 |
 |--------|------|
-| ROUGE-1 | ~0.55 |
-| ROUGE-2 | ~0.35 |
-| ROUGE-L | ~0.41 |
+| ROUGE-1 | ~0.5554 |
+| ROUGE-2 | ~0.3531 |
+| ROUGE-L | ~0.4178 |
 
 ## **BERTScore (Semantic Similarity)**  
 | Metric | Score |
 |--------|--------|
-| Precision | 0.884 |
-| Recall | 0.894 |
-| F1 | 0.886 |
+| Precision | 0.8841 |
+| Recall | 0.8943 |
+| F1 | 0.8864 |
 
 **BERTScore** is emphasized since legal summarization requires semantic preservation rather than lexical overlap.