Update backend/config.py

backend/config.py

@@ -20,7 +20,7 @@ TASKS: List[Tuple[str, str, str]] = [
     ("qimma-ArabicMMLU:_average|0",         "acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)",     "ArabicMMLU"),
     ("qimma-ArabCulture:_average|0",        "acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)",     "ArabCulture"),
     ("qimma-SyntheticQA:_average|0",        "acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)",     "SyntheticQA"),
-    ("qimma-PalmX|0",                  "acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)",     "PALMX"),
+    ("qimma-PalmX:_average|0|0",                  "acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)",     "PALMX"),
     ("qimma-NativeQA|0",                    "acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)",     "NativeQA"),
     ("qimma-AraTrust|0",                    "acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)",     "AraTrust"),
     ("qimma-Mizan|0", "normalized_mc_prob_norm",     "MizanQA"),
@@ -29,7 +29,7 @@ TASKS: List[Tuple[str, str, str]] = [
         [("acc:logprob_normalization=LogProbCharNorm(name='norm', ignore_first_space=False)", 0.5), ("BERTScore-F", 0.5)],
         "MedArabiQ",
     ),
-    ("qimma-ArablegalEval-QA-QA|0", "BERTScore-F", "ArabLegalQA"),
+    ("qimma-ArablegalEval-QA|0", "BERTScore-F", "ArabLegalQA"),
 ]
 
 # --- 3. Model Metadata ---