Update to v2 dataset script with dynamic filtering

permutation-groups.py

@@ -16,7 +16,7 @@ class PermutationGroupsConfig(datasets.BuilderConfig):
         min_order=None,
         max_order=None,
         min_len=3,
-        max_len=512,
+        max_len=1024,
         **kwargs
     ):
         """
@@ -53,7 +53,7 @@ class PermutationGroups(datasets.ArrowBasedBuilder):
     VERSION = datasets.Version("4.0.0")
     
     # Define available group types
-    GROUP_TYPES = ["symmetric", "alternating", "cyclic", "dihedral", "special"]
+    GROUP_TYPES = ["symmetric", "alternating", "cyclic", "dihedral", "psl25", "f20"]
     
     BUILDER_CONFIGS = []
     
@@ -88,7 +88,7 @@ class PermutationGroups(datasets.ArrowBasedBuilder):
         "z3": ("cyclic", 3, 3), "z4": ("cyclic", 4, 4), "z5": ("cyclic", 5, 5), "z6": ("cyclic", 6, 6),
         "d3": ("dihedral", 3, 3), "d4": ("dihedral", 4, 4), "d5": ("dihedral", 5, 5),
         "d6": ("dihedral", 6, 6), "d7": ("dihedral", 7, 7), "d8": ("dihedral", 8, 8),
-        "psl25": ("special", 6, 6), "f20": ("special", 5, 5),
+        "psl25": ("psl25", 6, 6), "f20": ("f20", 5, 5),
     }
     
     for name, (group_type, min_deg, max_deg) in LEGACY_GROUPS.items():
@@ -133,9 +133,9 @@ class PermutationGroups(datasets.ArrowBasedBuilder):
     def _split_generators(self, dl_manager):
         # Determine which datasets to load
         if self.config.group_type:
-            if self.config.group_type == "special":
-                # Special groups are stored separately
-                datasets_to_load = ["psl25_data", "f20_data"]
+            if self.config.group_type in ["psl25", "f20"]:
+                # PSL25 and F20 are stored as individual datasets
+                datasets_to_load = [f"{self.config.group_type}_data"]
             else:
                 # Load the superset for this group type
                 datasets_to_load = [f"{self.config.group_type}_superset"]