[BE]: FURB142 - Remove set mutations. Use set update (#124551)

Uses set mutation methods instead of manually reimplementing (update, set_difference etc).

Pull Request resolved: https://github.com/pytorch/pytorch/pull/124551
Approved by: https://github.com/ezyang

benchmarks/dynamo/timm_models.py

@@ -167,11 +167,9 @@ def refresh_model_names():
         del all_models_family[key]
 
     chosen_models = set()
-    for value in docs_models_family.values():
+    chosen_models.update(value[0] for value in docs_models_family.values())
-        chosen_models.add(value[0])
 
-    for key, value in all_models_family.items():
+    chosen_models.update(value[0] for key, value in all_models_family.items())
-        chosen_models.add(value[0])
 
     filename = "timm_models_list.txt"
     if os.path.exists("benchmarks"):

benchmarks/operator_benchmark/benchmark_utils.py

@@ -345,8 +345,9 @@ def get_operator_range(chars_range):
             ops_start_chars_set.add(item.lower())
             continue
         start, end = item.split("-")
-        for c in range(ord(start), ord(end) + 1):
+        ops_start_chars_set.update(
-            ops_start_chars_set.add(chr(c).lower())
+            chr(c).lower() for c in range(ord(start), ord(end) + 1)
+        )
     return ops_start_chars_set
 
 

test/distributed/_composable/fully_shard/test_fully_shard_init.py

@@ -144,10 +144,12 @@ class TestInitialization(FSDPTest):
         # Check that the composable module does not add any wrapper class
         local_module_classes = set()
         composable_module_classes = set()
-        for submodule in local_model.modules():
+        local_module_classes.update(
-            local_module_classes.add(type(submodule))
+            type(submodule) for submodule in local_model.modules()
-        for submodule in composable_module.modules():
+        )
-            composable_module_classes.add(type(submodule))
+        composable_module_classes.update(
+            type(submodule) for submodule in composable_module.modules()
+        )
         self.assertEqual(local_module_classes, composable_module_classes)
 
         # Check that the composable module has the same FSDP states with the
@@ -310,14 +312,14 @@ class TestInitialization(FSDPTest):
         ]
         for data_structure_name in data_structure_names:
             all_structures = set()
-            for module in (
+            all_structures.update(
-                composable_module.u1,
+                id(getattr(fully_shard.state(module), data_structure_name))
-                composable_module.u2,
+                for module in (
-                composable_module,
+                    composable_module.u1,
-            ):
+                    composable_module.u2,
-                all_structures.add(
+                    composable_module,
-                    id(getattr(fully_shard.state(module), data_structure_name))
                 )
+            )
             self.assertEqual(len(all_structures), 1)
 
 

test/distributed/fsdp/test_wrap.py

@@ -945,8 +945,7 @@ class TestWrapUtils(TestCase):
         ignored_params = set()
         for module_name, module in model.named_modules():
             if "lora_A" in module_name:
-                for param in module.parameters():
+                ignored_params.update(module.parameters())
-                    ignored_params.add(param)
         _validate_frozen_params(model, modules_to_wrap, ignored_params, use_orig_params)
 
 

test/dynamo/test_higher_order_ops.py

@@ -1375,8 +1375,7 @@ def forward(self, getitem, const):
 
         cond_gm = backend.graphs[0]
         name_set = set()
-        for name, _ in cond_gm.named_modules():
+        name_set.update(name for name, _ in cond_gm.named_modules())
-            name_set.add(name)
         self.assertEqual(
             name_set,
             {
@@ -1735,8 +1734,7 @@ def forward(self):
         self.assertEqual(result, x + y + x)
         wrap_gm = backend.graphs[0]
         names = set()
-        for mod_name, _ in wrap_gm.named_modules():
+        names.update(mod_name for mod_name, _ in wrap_gm.named_modules())
-            names.add(mod_name)
         self.assertEqual(
             names,
             {

test/functorch/discover_coverage.py

@@ -365,8 +365,7 @@ def get_all_tested_ops():
     result = set({})
     for op in get_covered_ops(overridable_outplace_we_care_about).values():
         opinfos = op_to_opinfo[op]
-        for opinfo in opinfos:
+        result.update(opinfo.name for opinfo in opinfos)
-            result.add(opinfo.name)
     return result
 
 

test/package/test_digraph.py

@@ -79,8 +79,7 @@ class TestDiGraph(PackageTestCase):
         g.add_node(3)
 
         nodes = set()
-        for n in g:
+        nodes.update(g)
-            nodes.add(n)
 
         self.assertEqual(nodes, {1, 2, 3})
 

test/test_dataloader.py

@@ -1617,8 +1617,7 @@ except RuntimeError as e:
         dataset = SynchronizedSeedDataset(num_workers, batch_size, num_workers)
         dataloader = self._get_data_loader(dataset, batch_size=batch_size, num_workers=num_workers)
         seeds = set()
-        for batch in dataloader:
+        seeds.update(batch[0] for batch in dataloader)
-            seeds.add(batch[0])
         self.assertEqual(len(seeds), num_workers)
 
     def test_worker_seed_reproducibility(self):

test/test_torch.py

@@ -9523,8 +9523,7 @@ tensor([[[1.+1.j, 1.+1.j, 1.+1.j,  ..., 1.+1.j, 1.+1.j, 1.+1.j],
 
         device_set = {'cpu', 'cpu:0', 'cuda', 'cuda:0', 'cuda:1', 'cuda:10', 'cuda:100'}
         device_hash_set = set()
-        for device in device_set:
+        device_hash_set.update(hash(torch.device(device)) for device in device_set)
-            device_hash_set.add(hash(torch.device(device)))
         self.assertEqual(len(device_set), len(device_hash_set))
 
         def get_expected_device_repr(device):

torch/_dynamo/trace_rules.py

@@ -3233,17 +3233,19 @@ if torch.distributed.is_available():
 
 @functools.lru_cache(None)
 def get_legacy_mod_inlinelist():
-    inlinelist = set()
+    inlinelist = {
-    for m in LEGACY_MOD_INLINELIST:
+        _module_dir(torch) + m[len("torch.") :].replace(".", "/")
-        inlinelist.add(_module_dir(torch) + m[len("torch.") :].replace(".", "/"))
+        for m in LEGACY_MOD_INLINELIST
+    }
     return inlinelist
 
 
 @functools.lru_cache(None)
 def get_mod_inlinelist():
-    inlinelist = set()
+    inlinelist = {
-    for m in MOD_INLINELIST:
+        _module_dir(torch) + m[len("torch.") :].replace(".", "/")
-        inlinelist.add(_module_dir(torch) + m[len("torch.") :].replace(".", "/"))
+        for m in MOD_INLINELIST
+    }
     return inlinelist
 
 

torch/_functorch/partitioners.py

@@ -744,8 +744,7 @@ def min_cut_rematerialization_partition(
             if node.op == "placeholder" and "tangents" in node.target:
                 required_bw_nodes.add(node)
             if node in required_bw_nodes:
-                for user in node.users:
+                required_bw_nodes.update(node.users)
-                    required_bw_nodes.add(user)
 
         primal_inputs = list(filter(_is_primal, joint_module.graph.nodes))
         fwd_seed_offset_inputs = list(

torch/_inductor/codegen/cpp.py

@@ -3623,8 +3623,7 @@ class CppScheduling(BaseScheduling):
                             if var_ranges is None:
                                 var_ranges = v
                             assert var_ranges == v, (var_ranges, v, node.snodes)
-                            for expr in exprs:
+                            indexing_exprs.update(exprs)
-                                indexing_exprs.add(expr)
                         return var_ranges, list(indexing_exprs)
                     else:
                         assert isinstance(node, SchedulerNode)

torch/_inductor/graph.py

@@ -635,8 +635,7 @@ class GraphLowering(torch.fx.Interpreter):
         # - sebotnet33ts_256
         for n in self.module.graph.nodes:
             if n in output_set:
-                for child in n.users:
+                output_set.update(n.users)
-                    output_set.add(child)
 
         return output_set
 

torch/_inductor/lowering.py

@@ -89,8 +89,9 @@ def add_needs_realized_inputs(fn):
         return [add_needs_realized_inputs(x) for x in fn]
     needs_realized_inputs.add(fn)
     if isinstance(fn, torch._ops.OpOverloadPacket):
-        for overload in fn.overloads():
+        needs_realized_inputs.update(
-            needs_realized_inputs.add(getattr(fn, overload))
+            getattr(fn, overload) for overload in fn.overloads()
+        )
 
 
 def add_layout_constraint(fn, constraint):

torch/_inductor/scheduler.py

@@ -2292,9 +2292,7 @@ class Scheduler:
         Populate node.last_usage recursively (also for the nodes within a FusedSchedulerNode)
         """
 
-        future_used_buffers = set()
+        future_used_buffers = set(V.graph.get_output_names())
-        for node_name in V.graph.get_output_names():
-            future_used_buffers.add(node_name)
 
         for node in reversed(self.nodes):
             node.set_last_usage(future_used_buffers, self.mutation_real_name)

torch/_library/custom_ops.py

@@ -223,9 +223,10 @@ class CustomOpDef:
                     def backend_impl(*args, **kwargs):
                         # Checks the assumption that outputs cannot alias
                         # inputs or other outputs.
-                        storages = set()
+                        storages = {
-                        for tensor in iter_tensors(args, kwargs):
+                            id(tensor.untyped_storage())
-                            storages.add(id(tensor.untyped_storage()))
+                            for tensor in iter_tensors(args, kwargs)
+                        }
 
                         result = self._backend_fns[device_type](*args, **kwargs)
 

torch/ao/ns/fx/n_shadows_utils.py

@@ -742,8 +742,7 @@ def create_add_loggers_graph(
         insert_submodule_copy = False
         if maybe_subgraph is not None:
             first_node, last_node = maybe_subgraph[0], maybe_subgraph[-1]
-            for node_to_skip in maybe_subgraph:
+            nodes_to_skip.update(maybe_subgraph)
-                nodes_to_skip.add(node_to_skip)
             qconfig = node_name_to_qconfig[first_node.name]
             if qconfig is not None:
                 insert_submodule_copy = True
@@ -873,8 +872,7 @@ def create_add_loggers_graph(
         maybe_subgraph = _get_subgraph_containing_node(n, subgraphs_dedup)
         if maybe_subgraph is not None:
             first_node, last_node = maybe_subgraph[0], maybe_subgraph[-1]
-            for node_to_skip in maybe_subgraph:
+            nodes_to_skip.update(maybe_subgraph)
-                nodes_to_skip.add(node_to_skip)
         else:
             first_node, last_node = n, n
 

torch/ao/quantization/quantizer/embedding_quantizer.py

@@ -45,9 +45,9 @@ class EmbeddingQuantizer(Quantizer):
 
     @classmethod
     def get_supported_quantization_configs(cls) -> List[QuantizationConfig]:
-        op_configs: Set[QuantizationConfig] = set({})
+        op_configs: Set[QuantizationConfig] = {
-        for spec, _ in cls.get_supported_operators():
+            spec for spec, _ in cls.get_supported_operators()
-            op_configs.add(spec)
+        }
         return list(op_configs)
 
     @classmethod

torch/ao/quantization/quantizer/x86_inductor_quantizer.py

@@ -286,9 +286,9 @@ class X86InductorQuantizer(Quantizer):
 
     @classmethod
     def get_supported_quantization_configs(cls) -> List[QuantizationConfig]:
-        op_configs: Set[QuantizationConfig] = set({})
+        op_configs: Set[QuantizationConfig] = {
-        for spec, _ in cls.supported_config_and_operators:
+            spec for spec, _ in cls.supported_config_and_operators
-            op_configs.add(spec)
+        }
         return list(op_configs)
 
     @classmethod

torch/ao/quantization/quantizer/xnnpack_quantizer.py

@@ -305,9 +305,9 @@ class XNNPACKQuantizer(Quantizer):
 
     @classmethod
     def get_supported_quantization_configs(cls) -> List[QuantizationConfig]:
-        op_configs: Set[QuantizationConfig] = set({})
+        op_configs: Set[QuantizationConfig] = {
-        for spec, _ in cls.supported_config_and_operators:
+            spec for spec, _ in cls.supported_config_and_operators
-            op_configs.add(spec)
+        }
         return list(op_configs)
 
     @classmethod

torch/distributed/_tensor/ops/basic_strategy.py

@@ -1,7 +1,7 @@
 import itertools
 from dataclasses import dataclass
 
-from typing import List, Tuple
+from typing import List, Set, Tuple
 
 from torch.distributed._tensor.op_schema import OpStrategy, PlacementStrategy
 from torch.distributed._tensor.placement_types import (
@@ -44,10 +44,9 @@ class EinsumDims:
         Parse the dims and extract the contracting, batch, and free dimensions
         for the left and right hand sides.
         """
-        dim_char_set = set()
+        dim_char_set: Set[str] = set()
         for input_dim in input_dims:
-            for input_char in list(input_dim):
+            dim_char_set.update(input_dim)
-                dim_char_set.add(input_char)
 
         # get a determinisitc order of all dim chars
         all_dim_chars = sorted(dim_char_set)

torch/distributed/checkpoint/state_dict.py

@@ -218,7 +218,7 @@ def _verify_options(
             fqn_param_mapping[fqn] = param
             all_fqns.add(fqn)
 
-    submodule_prefixes = set()
+    submodule_prefixes: Set[str] = set()
     if submodules:
         submodules = set(submodules)
         for name, module in model.named_modules():
@@ -226,8 +226,7 @@ def _verify_options(
                 continue
             fqns = _get_fqns(model, name)
             assert len(fqns) == 1, "Submodule FQN should only have 1 instance"
-            for fqn in fqns:
+            submodule_prefixes.update(f"{fqn}." for fqn in fqns)
-                submodule_prefixes.add(f"{fqn}.")
 
     fsdp_modules = FSDP.fsdp_modules(model)
     state_dict_config: StateDictConfig

torch/fx/graph_module.py

@@ -112,9 +112,7 @@ def _format_import_statement(name: str, obj: Any, importer: Importer) -> str:
 
 
 def _format_import_block(globals: Dict[str, Any], importer: Importer):
-    import_strs: Set[str] = set()
+    import_strs: Set[str] = {_format_import_statement(name, obj, importer) for name, obj in globals.items()}
-    for name, obj in globals.items():
-        import_strs.add(_format_import_statement(name, obj, importer))
     # Sort the imports so we have a stable import block that allows us to
     # hash the graph module and get a consistent key for use in a cache.
     return "\n".join(sorted(import_strs))

torch/fx/subgraph_rewriter.py

@@ -294,8 +294,7 @@ def _replace_pattern(
         # Copy the replacement graph over
         user_nodes: Set[Node] = set()
         for n in match.returning_nodes:
-            for user in n.users:
+            user_nodes.update(n.users)
-                user_nodes.add(user)
         assert user_nodes, "The returning_nodes should have at least one user node"
 
         if len(user_nodes) == 1:

torch/profiler/_memory_profiler.py

@@ -930,8 +930,9 @@ class MemoryProfile:
                 self._is_gradient(*i) or i in used_for_gradient
                 for i in node.outputs.items()
             ):
-                for key, (_, version) in node.inputs.items():
+                used_for_gradient.update(
-                    used_for_gradient.add((key, version))
+                    (key, version) for key, (_, version) in node.inputs.items()
+                )
         candidate_parameters.intersection_update(used_for_gradient)
 
         # and depends on a gradient.

torch/utils/data/datapipes/_hook_iterator.py

@@ -34,9 +34,7 @@ def _strip_datapipe_from_name(name: str) -> str:
 def _generate_input_args_string(obj):
     """Generate a string for the input arguments of an object."""
     signature = inspect.signature(obj.__class__)
-    input_param_names = set()
+    input_param_names = set(signature.parameters.keys())
-    for param_name in signature.parameters.keys():
-        input_param_names.add(param_name)
     result = []
     for name, value in inspect.getmembers(obj):
         if name in input_param_names:

torch/utils/tensorboard/_caffe2_graph.py

@@ -578,10 +578,8 @@ def _compute_in_out(ops):
     out_blobs = set()
 
     for op in ops:
-        for input_blob in op.input:
+        in_blobs.update(op.input)
-            in_blobs.add(input_blob)
+        out_blobs.update(op.output)
-        for output_blob in op.output:
-            out_blobs.add(output_blob)
 
     input_blobs = list(in_blobs.difference(out_blobs))
     output_blobs = list(out_blobs.difference(in_blobs))
@@ -700,8 +698,7 @@ def _operators_to_graph_def(
             else [_operator_to_node(shapes, op)]
         )  # .extend() expects an iterable
         current_graph.node.extend(nodes_from_op)
-        for input_blob in op.input:
+        blobs.update(op.input)
-            blobs.add(input_blob)
         for i, output_blob in enumerate(op.output):
             blobs.add(output_blob)
             producing_ops.setdefault(output_blob, []).append((op, i))

torchgen/gen.py

@@ -2125,7 +2125,7 @@ def gen_headers(
     )
 
     def gen_aten_interned_strings() -> Dict[str, str]:
-        attrs = set()  # All function argument names
+        attrs: Set[str] = set()  # All function argument names
         names = set()  # All ATen function names
         for func in native_functions:
             names.add(str(func.func.name.name))
@@ -2133,8 +2133,7 @@ def gen_headers(
             # symbol without the underscore
             names.add(func.func.name.name.base)
 
-            for arg in func.func.schema_order_arguments():
+            attrs.update(arg.name for arg in func.func.schema_order_arguments())
-                attrs.add(arg.name)
 
         # These are keywords in C++, so aren't valid symbol names
         # https://en.cppreference.com/w/cpp/language/operator_alternative