adding new benchmarks

backend/config.py

@@ -30,6 +30,12 @@ TASKS: List[Tuple[str, str, str]] = [
         "MedArabiQ",
     ),
     ("qimma-ArablegalEval-QA|0", "BERTScore-F", "ArabLegalQA"),
+    ("acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)", "qimma-GAT:_average|0", "GAT"),
+    ("acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)", "qimma-MedAraBench:_average|0", "MedAraBench"),
+    ("humaneval_scores", "humaneval_plus", "HumanEval+"),
+    ("mbpp_scores", "mbpp_plus", "MBPP+"),
+    ("fann", "fann", "FANN"),
+    ("flop", "flop", "FLOP"),
 ]
 
 # --- 3. Model Metadata ---