perf: benchmark singletons — they're slower, keep them off

complex_tokenization/examples/bne.py

@@ -19,7 +19,6 @@ def train_bne_tokenizer(texts: list[str],
 
     GraphSettings.ONLY_MINIMAL_MERGES = True
     GraphSettings.MAX_MERGE_SIZE = n
-    GraphSettings.USE_SINGLETONS = False
 
     graphs = tuple(words(text, connected=connected, units=units) for text in texts)
 

complex_tokenization/graph.py

@@ -15,17 +15,6 @@ def dot_escape(s: str) -> str:
 
 
 class GraphVertex:
-    _instances = {}  # Singleton pattern
-
-    def __new__(cls, *args, **kwargs):
-        if not GraphSettings.USE_SINGLETONS:
-            return super().__new__(cls)
-
-        key = (cls, args, tuple(sorted(kwargs.items())))
-        if key not in cls._instances:
-            cls._instances[key] = super().__new__(cls)
-        return cls._instances[key]
-
     def __bytes__(self):
         raise NotImplementedError
 
@@ -36,12 +25,6 @@ def __str__(self):
             return token_replacement
         return self_str
 
-    def __eq__(self, other):
-        return self is other
-
-    def __hash__(self):
-        return id(self)
-
     def dot(self, level=0) -> Iterable[str]:
         raise NotImplementedError
 

complex_tokenization/graphs/settings.py

@@ -1,4 +1,3 @@
 class GraphSettings:
-    USE_SINGLETONS = False  # speeds up computation but hurts visualization
     MAX_MERGE_SIZE = 2
     ONLY_MINIMAL_MERGES = True

tests/conftest.py

@@ -6,19 +6,9 @@
 @pytest.fixture(autouse=True)
 def reset_graph_settings():
     original = {
-        "USE_SINGLETONS": GraphSettings.USE_SINGLETONS,
         "MAX_MERGE_SIZE": GraphSettings.MAX_MERGE_SIZE,
         "ONLY_MINIMAL_MERGES": GraphSettings.ONLY_MINIMAL_MERGES,
     }
     yield
-    GraphSettings.USE_SINGLETONS = original["USE_SINGLETONS"]
     GraphSettings.MAX_MERGE_SIZE = original["MAX_MERGE_SIZE"]
     GraphSettings.ONLY_MINIMAL_MERGES = original["ONLY_MINIMAL_MERGES"]
-
-
-@pytest.fixture(autouse=True)
-def clear_singleton_cache():
-    from complex_tokenization.graph import GraphVertex
-    GraphVertex._instances.clear()
-    yield
-    GraphVertex._instances.clear()