update readme eval

README.md

@@ -191,7 +191,7 @@ print("IDs   :", enc["input_ids"][0].tolist())
 
 Under identical parameter budgets and training settings:
 
-- **NER (ARMAN + PEYMA):** TooKaBERT achieves the highest F1 (95.5), our model is competitive (94.08) and close to FABERT but slightly lower on F1 | نزدیک به FABERT اما کمی پایین‌تر روی F1 (94.7 in P/R, F1 94.7).  
+- **NER (ARMAN + PEYMA):** TooKaBERT achieves the highest F1 (95.5), our model is competitive (94.08) and close to FABERT but slightly lower on F1 .
 - **Relation Extraction (PERLEX):** Our model (F1=90) surpasses FABERT (88) and is slightly below TooKaBERT (91).
 
 These results suggest the tokenizer/backbone choices here are strong for RE and competitive for NER, especially considering the compact backbone.