[BE]: Update flake8-comprehensions and enable C420 (#130699)

Uses `dict.fromkeys` whenever possible as covered by flake8-comprehensions rule C420. While the ruff rule RUF025 is still in preview, flake8-comprehensions have added a new rule which covers this. Use dict.fromkeys is faster when the value being added to the dictionary is the same at every iteration and is immutable, it also removes an unnecessary dict comprehension.

This rule will be enabled with our current ruleset in RUF in 0.6 as C420.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/130699
Approved by: https://github.com/lezcano, https://github.com/ezyang

.lintrunner.toml

@@ -38,7 +38,7 @@ init_command = [
     '--dry-run={{DRYRUN}}',
     'flake8==6.1.0',
     'flake8-bugbear==23.3.23',
-    'flake8-comprehensions==3.12.0',
+    'flake8-comprehensions==3.15.0',
     'flake8-executable==2.1.3',
     'flake8-logging-format==0.9.0',
     'flake8-pyi==23.3.1',

test/inductor/test_max_autotune.py

@@ -322,7 +322,7 @@ class TestMaxAutotune(TestCase):
                 return None
 
         fake_choices = [FakeChoiceCaller() for i in range(10)]
-        fake_lookup_result = {choice: 0.123 for choice in fake_choices}
+        fake_lookup_result = dict.fromkeys(fake_choices, 0.123)
 
         def no_lookup(
             choices: List[ChoiceCaller],

test/jit/test_list_dict.py

@@ -1645,9 +1645,9 @@ class TestDict(JitTestCase):
 
         def test_dictcomprehension_is_typed_from_annotation():
             metasyntactics = ["foo", "bar", "baz"]
-            x: Dict[str, Optional[int]] = {
+            x: Dict[str, Optional[int]] = {  # noqa: C420, RUF025
                 word: None for word in metasyntactics
-            }  # noqa: RUF025
+            }
             return x
 
         self.checkScript(test_dictcomprehension_is_typed_from_annotation, ())

torch/_dynamo/variables/nn_module.py

@@ -921,7 +921,7 @@ class UnspecializedNNModuleVariable(UserDefinedObjectVariable):
             params_list = collect_parameters(self, recurse=recurse)
 
             # Account for duplicated params
-            deduplicated_params = list({param: None for param in params_list}.keys())
+            deduplicated_params = list(dict.fromkeys(params_list).keys())
 
             return variables.ListIteratorVariable(
                 deduplicated_params, mutable_local=MutableLocal()

torch/_inductor/dependencies.py

@@ -69,7 +69,7 @@ class MemoryDep(Dep):
         """
         Return the offset by setting every variable to be 0.
         """
-        return sympy_subs(self.index, {v: 0 for v in self.var_names})
+        return sympy_subs(self.index, dict.fromkeys(self.var_names, 0))
 
     def normalize_with_stride_order(self, prefix="t"):
         r"""

torch/distributed/_tools/mem_tracker.py

@@ -439,7 +439,7 @@ class MemTracker(TorchDispatchMode):
         maybe_zero = False
         # Ensure the device entry exists in the current memory snapshot, initializing if necessary.
         dev_snap = self._curr_mem_snap.setdefault(
-            winfo.device, {reftype: 0 for reftype in self._ref_class}
+            winfo.device, dict.fromkeys(self._ref_class, 0)
         )
         dev_snap.setdefault(_TOTAL_KEY, 0)
         # Handle different types of updates based on the update type (`u_type`).

torch/fx/passes/infra/partitioner.py

@@ -18,7 +18,7 @@ logger.setLevel(logging.WARNING)
 class Partition:
     def __init__(self, id: Optional[int] = None, nodes: Optional[Iterable[Node]] = None):
         self.id = id
-        self.nodes = {node: None for node in nodes} if nodes is not None else {}
+        self.nodes = dict.fromkeys(nodes) if nodes is not None else {}
 
     def __repr__(self) -> str:
         return str(self.nodes)