[8/N] Fix unused loop variables in tests (pytorch#166921)

This PR continues to fix or remove unused loop variables in tests.

Pull Request resolved: pytorch#166921
Approved by: https://github.com/mlazos

Author: cyyever

test/quantization/core/test_workflow_ops.py

@@ -368,8 +368,8 @@ def _test_forward_per_tensor_cachemask_impl(self, device):
         float_types = (torch.float32, torch.float16, torch.float64, torch.bfloat16)
         torch_types = (torch.qint8, torch.quint8)
         Xs = (torch.randn(4, 8, device=device), torch.randn(4, 16, device=device)[:, ::2])
-        tensor_qparam = (True, False)
-        for float_type, torch_type, X, tensor_qparams in itertools.product(float_types, torch_types, Xs, tensor_qparam):
+        tensor_qparams = (True, False)
+        for float_type, torch_type, X, tensor_qparam in itertools.product(float_types, torch_types, Xs, tensor_qparams):
             # pick the scale + zp so that some values get clipped
             X = X.to(float_type)
             obs = torch.ao.quantization.MinMaxObserver(torch_type)

test/quantization/fx/test_quantize_fx.py

@@ -8807,7 +8807,7 @@ def forward(self, indices, offsets):
 
         # check it works in None and static qconfig
         for qconfig in [None, default_qconfig]:
-            qconfig_dict = {"": default_qconfig}
+            qconfig_dict = {"": qconfig}
             m = M().eval()
             m = prepare_fx(model, qconfig_dict, example_inputs=example_inputs)
             self.checkGraphModuleNodes(m, expected_node_occurrence={

test/test_datapipe.py

@@ -1136,7 +1136,7 @@ def test_fork_iterdatapipe(self):
                     )
                 break
         with warnings.catch_warnings(record=True) as wa:
-            for i, (n1, n2) in enumerate(zip(dp1, dp2)):
+            for n1, n2 in zip(dp1, dp2):
                 output1.append(n1)
                 output2.append(n2)
             self.assertEqual(len(wa), 1)

test/test_jit_fuser_te.py

@@ -1682,11 +1682,8 @@ def apply(fn):
         ]
         dtypes = ["int", "float", "bool"]
         values = {"int": [10, 3], "float": [12.34, 2.78], "bool": [True, False]}
-        devices = self.devices
-        for dtype_x, dtype_y, op, device in product(
-            dtypes, dtypes, binary_ops, devices
-        ):
-            code = ir_template.format(**locals())
+        for dtype_x, dtype_y, op in product(dtypes, dtypes, binary_ops):
+            code = ir_template.format(dtype_x=dtype_x, dtype_y=dtype_y, op=op)
 
             # Interpret the graph
             try:
@@ -1701,9 +1698,7 @@ def apply(fn):
             try:
                 k = torch._C._te.TensorExprKernel(graph)
             except Exception as e:
-                raise RuntimeError(
-                    " ".join(["Compilation failed:", device, str(code)])
-                ) from e
+                raise RuntimeError(" ".join(["Compilation failed:", str(code)])) from e
 
             # Run the graph
             for x, y in product(values[dtype_x], values[dtype_y]):
@@ -1713,9 +1708,7 @@ def apply(fn):
                     self.assertEqual(ref, res)
                 except Exception as e:
                     raise RuntimeError(
-                        " ".join(
-                            ["Failed at runtime:", device, str(x), str(y), str(code)]
-                        )
+                        " ".join(["Failed at runtime:", str(x), str(y), str(code)])
                     ) from e
 
     def test_matmul(self):

torch/_export/serde/serialize.py

@@ -617,7 +617,7 @@ def get_triton_kernel_and_cache_entry(node: torch.fx.Node):
         return actual_kernel, matching_entries[0][1]
 
     if is_autotuner:
-        for sig_key, cache_entry in matching_entries:
+        for _sig_key, cache_entry in matching_entries:
             entry_metadata = cache_entry.metadata
             # pyrefly: ignore [missing-attribute]
             for config in kernel.configs: