[3/N] Fix unused loop variables (#166509)

This PR removes unused loop variables in tests.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/166509
Approved by: https://github.com/Lucaskabela, https://github.com/Skylion007

test/distributed/_composable/fsdp/test_fully_shard_training.py

@@ -827,7 +827,7 @@ class TestFullyShardShardPlacementFnMultiProcess(FSDPTest):
 
         torch.manual_seed(42 + self.rank)
         inp = torch.randint(0, model_args.vocab_size, (2, 16), device=device_type.type)
-        for iter_idx in range(5):
+        for _ in range(5):
             ref_loss = ref_model(inp).sum()
             loss = model(inp).sum()
             self.assertEqual(ref_loss, loss)

test/dynamo/test_modules.py

@@ -262,7 +262,7 @@ class ConstLoop(torch.nn.Module):
         self.count = 3
 
     def forward(self, x):
-        for i in range(self.count):
+        for _ in range(self.count):
             x = torch.sigmoid(self.linear1(x))
         return x
 
@@ -509,7 +509,7 @@ class CfgModule(torch.nn.Module):
         self.layer = torch.nn.Linear(10, 10)
 
     def forward(self, x):
-        for i in range(self.cfg.count):
+        for _ in range(self.cfg.count):
             x = self.layer(x + self.cfg.val)
         return x
 
@@ -781,7 +781,7 @@ class ParametersModule5(torch.nn.Module):
 
     def forward(self, x):
         counter = 0
-        for param in self.parameters():
+        for _param in self.parameters():
             counter += 1
 
         return x * self.scale * counter
@@ -841,7 +841,7 @@ class EnumValues(torch.nn.ModuleDict):
 
     def forward(self, init_features):
         features = [init_features]
-        for idx, layer in enumerate(self.values()):
+        for layer in self.values():
             new_features = layer(features)
             features.append(new_features)
         return torch.cat(features, 1)
@@ -2161,7 +2161,7 @@ class OptimizedModuleTest(torch._dynamo.test_case.TestCase):
 
         cnts = torch._dynamo.testing.CompileCounterWithBackend("eager")
         opt_mod = torch.compile(fn, backend=cnts)
-        for i in range(8):
+        for _ in range(8):
             mod = Mod()
             opt_mod(torch.randn(5, 5), mod)
 
@@ -2516,7 +2516,7 @@ class OptimizedModuleTest(torch._dynamo.test_case.TestCase):
         compiled_model = torch.compile(model, backend="aot_eager")
 
         activations = compiled_activations
-        for i in range(2):
+        for _ in range(2):
             # second iteration is key, hooks would have fired during aot trace
             # on first iter
             compiled_activations.clear()
@@ -2526,7 +2526,7 @@ class OptimizedModuleTest(torch._dynamo.test_case.TestCase):
             loss.backward()
 
         activations = eager_activations
-        for i in range(2):
+        for _ in range(2):
             # second iteration is key, hooks would have fired during aot trace
             # on first iter
             eager_activations.clear()
@@ -2575,12 +2575,12 @@ class OptimizedModuleTest(torch._dynamo.test_case.TestCase):
         def save_activations(mod, inp, out):
             activations.append(inp)
 
-        for name, module in model.named_modules():
+        for module in model.modules():
             module.register_forward_hook(save_activations)
 
         cnt = torch._dynamo.testing.CompileCounter()
         model = torch.compile(model, backend=cnt, fullgraph=True)
-        for i in range(2):
+        for _ in range(2):
             # second iteration is key, hooks would have fired during aot trace
             # on first iter
             activations.clear()
@@ -2703,7 +2703,7 @@ class OptimizedModuleTest(torch._dynamo.test_case.TestCase):
 
         model = torch.compile(model, backend="aot_eager")
 
-        for i in range(2):
+        for _ in range(2):
             # second iteration is key, hooks would have fired during aot trace
             # on first iter
             x = torch.randn((20, 10))

test/functorch/test_aotdispatch.py

@@ -7555,7 +7555,7 @@ metadata incorrectly.
             (_inp, _tg3),
         ]
 
-        for i, (inp_fn, tg_fn) in enumerate(TEST_CASES):
+        for inp_fn, tg_fn in TEST_CASES:
             ref_x = inp_fn()
             x = ref_x.detach().clone().requires_grad_()
 

test/higher_order_ops/test_with_effects.py

@@ -491,9 +491,7 @@ def forward(self, arg0_1, arg1_1, arg2_1):
             def ins_dense():
                 return torch.tensor([1.0, 2.0, 3.0]), torch.tensor([4.0, 5.0, 6.0])
 
-            for i, (ins_fn, expected_fw_count) in enumerate(
+            for ins_fn, expected_fw_count in zip([ins_sc, ins_dense], [2, 1]):
-                zip([ins_sc, ins_dense], [2, 1])
-            ):
                 reset_counter()
                 ref_out = fn(*ins_fn())
                 assert_counter(expected_fw_count, 0)
@@ -524,16 +522,14 @@ def forward(self, arg0_1, arg1_1, arg2_1):
                     ),
                 )
 
-            for i, (
+            for (
                 ins_fn_req_grad,
                 (
                     expected_fw_count,
                     expected_fw_count_after_bw,
                     expected_bw_count_after_bw,
                 ),
-            ) in enumerate(
+            ) in zip([ins_dense_req_grad, ins_sc_req_grad], [(1, 1, 1), (2, 2, 2)]):
-                zip([ins_dense_req_grad, ins_sc_req_grad], [(1, 1, 1), (2, 2, 2)])
-            ):
                 ref_ins = ins_fn_req_grad()
                 reset_counter()
                 ref_out = fn(*ref_ins)

test/jit/test_async.py

@@ -35,7 +35,7 @@ class TestAsync(JitTestCase):
     def test_async_future_type_python(self):
         def foo(inp):
             futures = torch.jit.annotate(List[torch.jit.Future[torch.Tensor]], [])
-            for i in range(5):
+            for _ in range(5):
                 futures.append(torch.jit.fork(lambda x: x, inp))
             all_outputs = []
             for future in futures:
@@ -458,7 +458,7 @@ class TestAsync(JitTestCase):
         class TestListFutureModule(nn.Module):
             def forward(self, input):
                 input_list = []
-                for i in range(3):
+                for _ in range(3):
                     input_list.append(input)
 
                 fut_list: List[Future[torch.Tensor]] = []

test/jit/test_autodiff.py

@@ -68,7 +68,7 @@ class TestAutodiffJit(JitTestCase):
 
         fn_s = torch.jit.script(fn)
 
-        for i in range(4):
+        for _ in range(4):
             x, y = fn_s(a, b, c)
             self.assertFalse(x.requires_grad)
             self.assertTrue(y.requires_grad)
@@ -90,7 +90,7 @@ class TestAutodiffJit(JitTestCase):
         b = torch.rand((10, 10), requires_grad=False)
         c = torch.rand((10, 10), requires_grad=True)
 
-        for i in range(4):
+        for _ in range(4):
             x_s, y_s, z_s = fn_s(a, b, c)
             x, y, z = fn(a, b, c)
 
@@ -115,7 +115,7 @@ class TestAutodiffJit(JitTestCase):
         b = torch.rand((10, 10), requires_grad=False)
         c = torch.rand((10, 10), requires_grad=True)
 
-        for i in range(4):
+        for _ in range(4):
             x_s, y_s, z_s = fn_s(a, b, c)
             x, y, z = fn(a, b, c)
 
@@ -141,7 +141,7 @@ class TestAutodiffJit(JitTestCase):
         b = torch.rand((10, 10), requires_grad=True)
         c = torch.rand((10, 10), requires_grad=True)
 
-        for i in range(4):
+        for _ in range(4):
             x_s, y_s, z_s = fn_s(a, b, c)
             x, y, z = fn(a, b, c)
 

test/jit/test_list_dict.py

@@ -2989,7 +2989,7 @@ class TestScriptList(JitTestCase):
         test_script.segments_groupby_col
 
         # Smoketest for flakiness. Takes around 2s.
-        for i in range(300):
+        for _ in range(300):
             test = Test()
             test_script = torch.jit.script(test)
 

test/jit/test_logging.py

@@ -19,7 +19,7 @@ class TestLogging(JitTestCase):
         class ModuleThatLogs(torch.jit.ScriptModule):
             @torch.jit.script_method
             def forward(self, x):
-                for i in range(x.size(0)):
+                for _ in range(x.size(0)):
                     x += 1.0
                     torch.jit._logging.add_stat_value("foo", 1)
 
@@ -33,7 +33,7 @@ class TestLogging(JitTestCase):
         old_logger = torch.jit._logging.set_logger(logger)
         try:
             mtl = ModuleThatLogs()
-            for i in range(5):
+            for _ in range(5):
                 mtl(torch.rand(3, 4, 5))
 
             self.assertEqual(logger.get_counter_val("foo"), 15)
@@ -60,7 +60,7 @@ class TestLogging(JitTestCase):
         class ModuleThatTimes(torch.jit.ScriptModule):
             def forward(self, x):
                 tp_start = torch.jit._logging.time_point()
-                for i in range(30):
+                for _ in range(30):
                     x += 1.0
                 tp_end = torch.jit._logging.time_point()
                 torch.jit._logging.add_stat_value("mytimer", tp_end - tp_start)
@@ -80,7 +80,7 @@ class TestLogging(JitTestCase):
             @torch.jit.script_method
             def forward(self, x):
                 tp_start = torch.jit._logging.time_point()
-                for i in range(30):
+                for _ in range(30):
                     x += 1.0
                 tp_end = torch.jit._logging.time_point()
                 torch.jit._logging.add_stat_value("mytimer", tp_end - tp_start)
@@ -97,7 +97,7 @@ class TestLogging(JitTestCase):
 
     def test_counter_aggregation(self):
         def foo(x):
-            for i in range(3):
+            for _ in range(3):
                 torch.jit._logging.add_stat_value("foo", 1)
             return x + 1.0
 

test/jit/test_misc.py

@@ -518,7 +518,7 @@ class TestMisc(JitTestCase):
         ref = fn(x)
 
         script_fn = torch.jit.script(fn)
-        for i in range(4):
+        for _ in range(4):
             res = script_fn(x)
 
         self.assertEqual(ref, res)

test/jit/test_tracer.py

@@ -2025,7 +2025,7 @@ class TestTracer(JitTestCase):
         module = torch.jit.trace_module(n, inputs)
 
         check_inputs = []
-        for i in range(2):
+        for _ in range(2):
             check_weight = torch.rand(1, 1, 3, 3)
             check_forward_input = torch.rand(1, 1, 3, 3)
             check_inputs.append(

test/jit/test_types.py

@@ -341,7 +341,7 @@ class TestTypesAndAnnotation(JitTestCase):
                     self.x = x
 
                 def set(self, val: int):
-                    for i in range(3):
+                    for _ in range(3):
                         self.x: int = val
 
         # Type annotation in __init__, should not fail

test/test_autograd.py

@@ -582,14 +582,14 @@ class TestAutograd(TestCase):
         ctx_1 = torch.autograd.graph.saved_tensors_hooks(lambda x: x, unpack)
         ctx_2 = torch.autograd.graph.saved_tensors_hooks(lambda x: x, lambda x: x)
 
-        for i in range(10):
+        for _ in range(10):
             with ctx_2:
                 ctx_1.__enter__()
                 x = torch.randn(3, 3, requires_grad=True)
                 x.sin().sum().backward()
 
         # Clean up
-        for i in range(10):
+        for _ in range(10):
             ctx_1.__exit__()
 
         # Validate there are no more hooks on the stack
@@ -2989,7 +2989,7 @@ class TestAutograd(TestCase):
         state = set()
         with torch.enable_grad():
             coro = coro_no_grad(state)
-            for i in range(5):
+            for _ in range(5):
                 next(coro)
 
             coro.close()
@@ -2998,7 +2998,7 @@ class TestAutograd(TestCase):
         state = set()
         with torch.no_grad():
             coro = coro_enable_grad(state)
-            for i in range(5):
+            for _ in range(5):
                 next(coro)
 
             coro.close()
@@ -5293,7 +5293,7 @@ Done""",
         rnn = torch.nn.LSTM(10, 20, 2)
         total_time_s = 0
         with profile(record_shapes=True, use_kineto=kineto_available()) as prof:
-            for i in range(20):
+            for _ in range(20):
                 input = torch.randn(5, 3, 10)
                 h = torch.randn(2, 3, 20)
                 c = torch.randn(2, 3, 20)
@@ -5925,7 +5925,7 @@ Done""",
         self.assertTrue(p_a == p_g or p_b == p_g)
 
         # Run backwards multiple times to ensure accumulation works.
-        for i in range(10):
+        for _ in range(10):
             loss.backward(retain_graph=True)
 
         # non-contiguous indices and value, we should trigger a copy.
@@ -5943,7 +5943,7 @@ Done""",
         self.assertFalse(p_b == p_g)
 
         # Run backwards multiple times to ensure accumulation works.
-        for i in range(10):
+        for _ in range(10):
             loss.backward(retain_graph=True)
 
     def test_gradcheck_single_input(self):
@@ -7132,7 +7132,7 @@ for shape in [(1,), ()]:
         )
 
         feat_combined = []
-        for r in range(num_inp):
+        for _ in range(num_inp):
             data_r = torch.empty(1, nz_inp)
             data_r.uniform_()
             data_r.requires_grad = True
@@ -7202,7 +7202,7 @@ for shape in [(1,), ()]:
                 self.use_checkpoint = use_checkpoint
                 self.use_reentrant = use_reentrant
                 self.layers = nn.ModuleList()
-                for i in range(self.n):
+                for _ in range(self.n):
                     layer = nn.Sequential(
                         nn.Linear(256, 256), nn.Linear(256, 256), nn.Linear(256, 256)
                     )
@@ -7513,7 +7513,7 @@ for shape in [(1,), ()]:
 
         feat_combined = []
         feat_combined_no_checkpoint = []
-        for r in range(num_inp):
+        for _ in range(num_inp):
             data_r = torch.empty(1, nz_inp)
             data_r.uniform_()
             data_r.requires_grad = input_requires_grad
@@ -11714,7 +11714,7 @@ class TestAutogradDeviceType(TestCase):
     def test_parameter_resize(self, device):
         asd = torch.nn.Parameter(torch.ones(16, dtype=torch.double, device=device))
 
-        for i in range(2):
+        for _ in range(2):
             with torch.no_grad():
                 asd.set_(asd[1:])
                 asd.grad = None
@@ -11942,7 +11942,7 @@ class TestAutogradDeviceType(TestCase):
 
         # Child gpu graph (much longer than parent graph).
         prev = t2 * t2
-        for i in range(10):
+        for _ in range(10):
             prev = prev * t2
         reentrant_root = prev
 

test/test_cuda.py

@@ -4784,7 +4784,7 @@ print(torch.cuda.get_allocator_backend())
                 total -= x.numel()
 
             choices = [alloc, free, torch.cuda.memory.empty_cache]
-            for i in range(N):
+            for _ in range(N):
                 while total >= 1024 * 1024 * 1024 / (4 * 10):
                     free()
                 (action,) = random.choices(choices, weights=[1, 1 if mem else 0, 0.1])

test/test_fake_tensor.py

@@ -512,7 +512,7 @@ class FakeTensorTest(TestCase):
     def test_upsample_bilinear_small_channels(self):
         out = []
         mode = FakeTensorMode()
-        for i, context in enumerate([contextlib.nullcontext, lambda: mode]):
+        for context in [contextlib.nullcontext, lambda: mode]:
             with context():
                 arg0_1 = torch.empty_strided(
                     (3, 427, 640), (1, 1920, 3), dtype=torch.float32, device="cuda"

test/test_foreach.py

@@ -318,13 +318,11 @@ class TestForeach(TestCase):
                 return arg
 
         scalar_self_arg_test_complete = False
-        for i, sample in enumerate(
+        for sample in op.sample_inputs(
-            op.sample_inputs(
             device,
             dtype,
             noncontiguous=not is_fastpath,
             allow_higher_dtype_scalars=True,
-            )
         ):
             (rhs_arg,) = sample.args
             kwargs = {} or sample.kwargs

test/test_indexing.py

@@ -156,7 +156,7 @@ class TestIndexing(TestCase):
             torch.DoubleTensor if not device.startswith("mps") else torch.FloatTensor
         )
         tensor = _make_tensor(lst).to(device)
-        for _i in range(100):
+        for _ in range(100):
             idx1_start = random.randrange(10)
             idx1_end = idx1_start + random.randrange(1, 10 - idx1_start + 1)
             idx1_step = random.randrange(1, 8)

test/test_jit.py

@@ -2337,9 +2337,9 @@ graph(%Ra, %Rb):
                 print("stays")
             while False:
                 print("removed")
-            for _i in range(0):
+            for _ in range(0):
                 print("removed")
-            for _i in range(-4):
+            for _ in range(-4):
                 print("removed")
             return b
 
@@ -3138,7 +3138,7 @@ class TestScript(JitTestCase):
         with enable_profiling_mode_for_profiling_tests():
 
             def fct_loop(x):
-                for i in range(3):
+                for _ in range(3):
                     x = torch.cat((x, x), 0)
                 return x
 
@@ -3245,7 +3245,7 @@ class TestScript(JitTestCase):
     def test_nested_bailouts(self):
         @torch.jit.script
         def fct_loop(x):
-            for i in range(3):
+            for _ in range(3):
                 x = torch.cat((x, x), 0)
             return x
 
@@ -3907,7 +3907,7 @@ def foo(x):
                 else:
                     return f'v{idx - len(exprs)}'
 
-            for i in range(50):
+            for _ in range(50):
                 n = None
                 while n is None or n > len(exprs) + n_variables:
                     template = random.choice(templates)
@@ -3922,7 +3922,7 @@ def foo(x):
             src_lines.append('  return ({})\n'.format(''.join(f'v{i},' for i in range(n_variables))))
             return '\n'.join(src_lines)
 
-        for i in range(100):
+        for _ in range(100):
             g = {'torch': torch}
             code = gen_code()
             builtins.exec(code, g, None)
@@ -4602,7 +4602,7 @@ def foo(xyz):
         y = torch.randn(3, 3, requires_grad=True)
 
         def grad_in_loop(x, y):
-            for i in range(100):
+            for _ in range(100):
                 x = y @ x
             return x
 
@@ -5559,7 +5559,7 @@ a")
             @torch.jit.script
             def test(x):
                 after_resize_alias = torch.zeros([2])
-                for _i in range(5):
+                for _ in range(5):
                     b = x + 1
                     f = [1]
                     before_resize_alias = b.sub_(1)
@@ -5950,7 +5950,7 @@ a")
             # type: (int) -> int
             prev = 1
             v = 1
-            for i in range(x):
+            for _ in range(x):
                 save = v
                 v = v + prev
                 prev = save
@@ -7785,7 +7785,7 @@ dedent """
             while int(tensor.add_(1)) < 4:
                 if y == 1:
                     continue
-                for i in range(y):
+                for _ in range(y):
                     continue
                     ret += 1
                 ret += 1
@@ -7896,7 +7896,7 @@ dedent """
         def assign_after_break_nested(y):
             # type: (int)
             x = 0
-            for i in range(y):
+            for _ in range(y):
                 if y == 1:
                     x = 5
                     break
@@ -7916,7 +7916,7 @@ dedent """
         def may_break(y):
             # type: (int)
             x = 0
-            for i in range(y):
+            for _ in range(y):
                 if y == 1:
                     x = 5
                 else:
@@ -7988,7 +7988,7 @@ dedent """
         def test_varexit(cond):
             # type: (int)
             m = 0
-            for i in range(3):
+            for _ in range(3):
                 if cond == 2:
                     if cond == 2:
                         m = 2
@@ -8376,7 +8376,7 @@ dedent """
                 # find the last output, then all subsequent uses
                 fc.check(out_name[-1] + " : ")
                 # skip past node body
-                for i in range(contained_blocks(node)):
+                for _ in range(contained_blocks(node)):
                     fc.check("->")
                 if (node.kind() == "prim::If"):
                     fc.check("->").check("->").check("\n")
@@ -8429,7 +8429,7 @@ dedent """
             a = 1
             b = 2
             c = 3
-            for i in range(iter):
+            for _ in range(iter):
                 a = 4
                 b = 5
                 c = 6
@@ -8445,7 +8445,7 @@ dedent """
             a = 1
             b = 2
             c = 3
-            for i in range(iter):
+            for _ in range(iter):
                 c = c + 1
                 b = b + 1
                 a = a + 1
@@ -10938,7 +10938,7 @@ dedent """
 
             # Test symbolic differentiation
             # Run Forward and Backward thrice to trigger autodiff graph
-            for i in range(3):
+            for _ in range(3):
                 y = jit_module(x)
                 y.backward(grad)
             x.grad.zero_()
@@ -11030,7 +11030,7 @@ dedent """
         W.data /= 4
 
         with enable_profiling_mode_for_profiling_tests():
-            for i in range(4):
+            for _ in range(4):
                 self.assertTrue((foo(x, y, W).grad_fn is None) == (jitted_foo(x, y, W).grad_fn is None))
 
 
@@ -11822,7 +11822,7 @@ dedent """
     def test_for_in_tensors(self):
         def test_sizes(x):
             sumz = 0
-            for s in x:
+            for _ in x:
                 sumz += 1
             return sumz
         self.checkScript(test_sizes, (torch.rand(5, 4, 3, 2, 1),))
@@ -11834,7 +11834,7 @@ dedent """
             @torch.jit.script
             def test_sizes(x):
                 sumz = 0
-                for s in x:
+                for _ in x:
                     sumz += 1
                 return sumz
 
@@ -11846,7 +11846,7 @@ dedent """
             def test_sizes(x):
                 # type: (float) -> int
                 sumz = 0
-                for s in x:
+                for _ in x:
                     sumz += 1
                 return sumz
 
@@ -11856,7 +11856,7 @@ dedent """
         def test_sizes(x):
             sumz = 0
             for n in x:
-                for t in n:
+                for _ in n:
                     sumz += 1
             return sumz
 
@@ -12316,7 +12316,7 @@ dedent """
 
                 @torch.jit.script_method
                 def forward(self, x):
-                    for _i in range(4):
+                    for _ in range(4):
                         x += self.param
                     return x
 
@@ -12840,7 +12840,7 @@ dedent """
 
         # Load from filename
         tracemalloc.start()
-        for i in range(num_iters):
+        for _ in range(num_iters):
             torch._C.PyTorchFileReader(filename)
         _, peak_from_string = tracemalloc.get_traced_memory()
         tracemalloc.stop()
@@ -12848,7 +12848,7 @@ dedent """
         # Load from stream
         tracemalloc.start()
         with open(filename, 'rb') as f:
-            for i in range(num_iters):
+            for _ in range(num_iters):
                 f.seek(0)
                 torch._C.PyTorchFileReader(f)
         _, peak_from_file = tracemalloc.get_traced_memory()
@@ -13287,7 +13287,7 @@ dedent """
     def test_pass(self):
         def foo(x):
             # type: (bool) -> int
-            for _i in range(3):
+            for _ in range(3):
                 pass
             if x:
                 pass
@@ -13903,7 +13903,7 @@ dedent """
         def test_loop_no_escape(x):
             # type: (int)
             if x >= 0:
-                for i in range(x):
+                for _ in range(x):
                     raise RuntimeError("hi")
             else:
                 return 5
@@ -14116,7 +14116,7 @@ dedent """
 
         def test_will_ret(y):
             # type: (int) -> int
-            for i in range(y):
+            for _ in range(y):
                 return 2
             return 1
 
@@ -14125,8 +14125,8 @@ dedent """
 
         def test_loop_nest_ret(y):
             # type: (int) -> int
-            for i in range(y):
+            for _ in range(y):
-                for i in range(y - 2):
+                for _ in range(y - 2):
                     return 10
                 return 5
             return 0
@@ -15387,7 +15387,7 @@ dedent """
                 if isinstance(item, list):
                     return is_tensor_value(item[0])
                 return False
-            for name, value, the_type in self.get_pickle_values():
+            for name, value, _the_type in self.get_pickle_values():
                 if is_tensor_value(value):
                     continue
                 self.assertEqual(value, getattr(loaded, "_" + name))
@@ -15768,7 +15768,7 @@ dedent """
     def test_for_else(self):
         def fn():
             c = 0
-            for i in range(4):
+            for _ in range(4):
                 c += 10
             else:
                 print("In else block of for...else")

test/test_jiterator.py

@@ -115,7 +115,7 @@ class TestPythonJiterator(TestCase):
     @parametrize("num_inputs", [1, 5, 8])
     def test_various_num_inputs(self, num_inputs):
         inputs = []
-        for i in range(num_inputs):
+        for _ in range(num_inputs):
             inputs.append(torch.rand(3, device='cuda').mul(10))
 
         input_string = ",".join([f"T i{i}" for i in range(num_inputs)])

test/test_multiprocessing.py

@@ -58,7 +58,7 @@ class SubProcess(mp.Process):
 
 
 def _test_cuda_ipc_deadlock_actor(queue, iterations):
-    for i in range(iterations):
+    for _ in range(iterations):
         if not queue.empty():
             queue.get()
         time.sleep(0.01)
@@ -66,7 +66,7 @@ def _test_cuda_ipc_deadlock_actor(queue, iterations):
 
 def _test_cuda_ipc_deadlock_learner(queue, iterations):
     net = torch.nn.LSTM(1, 1).cuda()
-    for i in range(iterations):
+    for _ in range(iterations):
         if not queue.full():
             queue.put(copy.deepcopy(net.state_dict()))
         time.sleep(0.01)
@@ -138,7 +138,7 @@ def send_tensor_with_untyped_storage(queue, event):
 
 def receive_and_send_sum(queue, out_queue, event, device, dtype, count, size=5):
     s = torch.full([size], 0, device=device, dtype=dtype)
-    for i in range(count):
+    for _ in range(count):
         t = queue.get()
         s += t
     out_queue.put(s)
@@ -146,7 +146,7 @@ def receive_and_send_sum(queue, out_queue, event, device, dtype, count, size=5):
 
 
 def receive_and_send(queue, out_queue, event, count):
-    for i in range(count):
+    for _ in range(count):
         t = queue.get()
         out_queue.put(t.clone())
     event.wait()

test/test_nn.py

@@ -1238,7 +1238,7 @@ tensor(..., device='meta', size=(1,), requires_grad=True)""")
 
         def check():
             self.assertEqual(len(parameter_dict), len(parameters))
-            for i, (k1, (k2, m2)) in enumerate(zip(parameters, parameter_dict.named_parameters())):
+            for (k1, (k2, m2)) in zip(parameters, parameter_dict.named_parameters()):
                 self.assertEqual(k1, k2)
                 self.assertIs(parameters[k1], m2)
             for k1, k2 in zip(parameters, parameter_dict):
@@ -2958,7 +2958,7 @@ tensor(..., device='meta', size=(1,), requires_grad=True)""")
                                                batch_first=batch_first)
 
             # set constant weights of the model
-            for idx, p in enumerate(model.parameters()):
+            for p in model.parameters():
                 x = p.data
                 sz = x.view(-1).size(0)
                 shape = x.shape
@@ -3108,7 +3108,7 @@ tensor(..., device='meta', size=(1,), requires_grad=True)""")
                                                activation, batch_first=batch_first)
 
             # set constant weights of the model
-            for idx, p in enumerate(model.parameters()):
+            for p in model.parameters():
                 x = p.data
                 sz = x.view(-1).size(0)
                 shape = x.shape
@@ -3185,7 +3185,7 @@ tensor(..., device='meta', size=(1,), requires_grad=True)""")
 
             with torch.no_grad():
                 # set constant weights of the model
-                for idx, p in enumerate(layer.parameters()):
+                for p in layer.parameters():
                     x = p.data
                     sz = x.view(-1).size(0)
                     shape = x.shape
@@ -13129,7 +13129,7 @@ if __name__ == '__main__':
                 model = model.eval()
 
             # set constant weights of the model
-            for idx, p in enumerate(model.parameters()):
+            for p in model.parameters():
                 x = p.data
                 sz = x.view(-1).size(0)
                 shape = x.shape
@@ -13349,7 +13349,7 @@ if __name__ == '__main__':
                 model = model.eval()
 
             # set constant weights of the model
-            for idx, p in enumerate(model.parameters()):
+            for p in model.parameters():
                 x = p.data
                 sz = x.view(-1).size(0)
                 shape = x.shape

test/test_sparse.py

@@ -1483,7 +1483,7 @@ class TestSparse(TestSparseBase):
         def test_shape(num_mats, dim_i, dim_j, dim_k, nnz):
             a_list = []
             b_list = []
-            for mat_idx in range(num_mats):
+            for _ in range(num_mats):
                 a_list.append(self._gen_sparse(2, nnz, [dim_i, dim_j], dtype, device, coalesced)[0])
                 b_list.append(torch.randn([dim_j, dim_k], dtype=dtype, device=device))
 

test/test_sparse_csr.py

@@ -4251,9 +4251,9 @@ class TestSparseCompressedTritonKernels(TestCase):
         # Test warn_once when requesting non-existing tuned parameters multiple times
         f = io.StringIO()
         with redirect_stderr(f):
-            for i in range(5):
+            for _ in range(5):
                 get_meta(16, 16, 16)
-            for i in range(5):
+            for _ in range(5):
                 get_meta(16, 16, 32)
 
         msg = f.getvalue()

test/test_static_runtime.py

@@ -139,7 +139,7 @@ def fork_wait_graph_exception(input1, input2):
 
 def loop_graph(a, b, iters: int):
     c = a + b * 2
-    for i in range(iters):
+    for _ in range(iters):
         c = c + b
         c *= 2
         c -= a

test/test_tensor_creation_ops.py

@@ -81,7 +81,7 @@ def _generate_input(shape, dtype, device, with_extremal):
 # TODO: replace with make_tensor
 def _rand_shape(dim, min_size, max_size):
     shape = []
-    for i in range(dim):
+    for _ in range(dim):
         shape.append(random.randint(min_size, max_size))
     return tuple(shape)
 
@@ -942,7 +942,7 @@ class TestTensorCreation(TestCase):
                 num_tensors = random.randint(1, 5)
                 torch_input = []
                 # Create tensors with shape being different along one axis only
-                for param in range(num_tensors):
+                for _ in range(num_tensors):
                     shape[i] = random.randint(1, 5)
                     torch_input.append(_generate_input(tuple(shape), dtype, device, with_extremal=False))
 
@@ -997,7 +997,7 @@ class TestTensorCreation(TestCase):
         ops = ((torch.vstack, np.vstack), (torch.row_stack, np.vstack))
         for torch_op, np_op in ops:
             self._test_special_stacks(0, 2, torch_op, np_op, device, dtype)
-            for i in range(5):
+            for _ in range(5):
                 # Test dimension change for 1D tensor of size (N) and 2D tensor of size (1, N)
                 n = random.randint(1, 10)
                 input_a = _generate_input((n,), dtype, device, with_extremal=False)
@@ -1012,7 +1012,7 @@ class TestTensorCreation(TestCase):
     @dtypes(*all_types_and_complex_and(torch.half))
     def test_dstack(self, device, dtype):
         self._test_special_stacks(2, 3, torch.dstack, np.dstack, device, dtype)
-        for i in range(5):
+        for _ in range(5):
             # Test dimension change for 1D tensor of size (N), 2D tensor of size (1, N), and 3D tensor of size (1, N, 1)
             n = random.randint(1, 10)
             input_a = _generate_input((n,), dtype, device, with_extremal=False)
@@ -2885,7 +2885,7 @@ class TestTensorCreation(TestCase):
     @dtypesIfCUDA(torch.float, torch.double, torch.bfloat16, torch.half, torch.long)
     @dtypes(torch.float, torch.double, torch.long, torch.bfloat16, torch.float16)
     def test_kaiser_window(self, device, dtype):
-        for num_test in range(50):
+        for _ in range(50):
             self._test_signal_window_functions('kaiser', dtype, device, beta=random.random() * 30)
 
     def _test_signal_windows_functions(self, name, dtype, device, **kwargs):
@@ -2918,7 +2918,7 @@ class TestTensorCreation(TestCase):
     @unittest.skipIf(not TEST_SCIPY, "Scipy not found")
     @dtypes(torch.float, torch.double)
     def test_kaiser(self, device, dtype):
-        for num_test in range(50):
+        for _ in range(50):
             self._test_signal_windows_functions('kaiser', dtype, device, beta=random.random() * 30)
 
     def test_tensor_factories_empty(self, device):

test/test_tensorexpr.py

@@ -1216,7 +1216,7 @@ class TestTensorExprFuser(BaseTestClass):
         @torch.jit.script
         def test(x: torch.Tensor, y: torch.Tensor, z: int) -> torch.Tensor:
             b = y
-            for i in range(z):
+            for _ in range(z):
                 a = x + y
                 b = b + y
             return b

test/test_transformers.py

@@ -303,7 +303,7 @@ class TestTransformers(NNTestCase):
         encoder = nn.TransformerEncoder(layer, 2).to(device)
         optimizer = optim.SGD(encoder.parameters(), lr=0.1, momentum=0.9)
         encoder.train()
-        for i in range(iters):
+        for _ in range(iters):
             encoder.train()
             optimizer.zero_grad()
             inputs = torch.cat([torch.randn(1, 2, 2), torch.zeros(1, 2, 2)], dim=1).to(device)
@@ -537,7 +537,7 @@ class TestTransformers(NNTestCase):
 
         with torch.no_grad():
             # set constant weights of the model
-            for idx, p in enumerate(model.parameters()):
+            for p in model.parameters():
                 x = p.data
                 sz = x.view(-1).size(0)
                 shape = x.shape
@@ -587,7 +587,7 @@ class TestTransformers(NNTestCase):
 
             with torch.no_grad():
                 # set constant weights of the model
-                for idx, p in enumerate(layer.parameters()):
+                for p in layer.parameters():
                     x = p.data
                     sz = x.view(-1).size(0)
                     shape = x.shape