diff --git a/onnxruntime/python/tools/transformers/bert_perf_test.py b/onnxruntime/python/tools/transformers/bert_perf_test.py
index 022ee076770be..16f376bafeeae 100644
--- a/onnxruntime/python/tools/transformers/bert_perf_test.py
+++ b/onnxruntime/python/tools/transformers/bert_perf_test.py
@@ -173,7 +173,7 @@ def onnxruntime_inference_with_io_binding(session, all_inputs, output_names, tes
     results = []
     latency_list = []
     device = "cuda" if test_setting.use_gpu else "cpu"
-    for test_case_id, inputs in enumerate(all_inputs):
+    for inputs in all_inputs:
         result = session.run(output_names, inputs)
         results.append(result)
         outputs = {}
@@ -201,7 +201,7 @@ def onnxruntime_inference(session, all_inputs, output_names):
 
     results = []
     latency_list = []
-    for test_case_id, inputs in enumerate(all_inputs):
+    for inputs in all_inputs:
         start_time = timeit.default_timer()
         result = session.run(output_names, inputs)
         latency = timeit.default_timer() - start_time
@@ -240,14 +240,12 @@ def run_one_test(model_setting, test_setting, perf_results, all_inputs, intra_op
 
     all_latency_list = []
     if test_setting.use_io_binding:
-        for i in range(test_setting.test_times):
-            results, latency_list = onnxruntime_inference_with_io_binding(
-                session, all_inputs, output_names, test_setting
-            )
+        for _ in range(test_setting.test_times):
+            _, latency_list = onnxruntime_inference_with_io_binding(session, all_inputs, output_names, test_setting)
             all_latency_list.extend(latency_list)
     else:
-        for i in range(test_setting.test_times):
-            results, latency_list = onnxruntime_inference(session, all_inputs, output_names)
+        for _ in range(test_setting.test_times):
+            _, latency_list = onnxruntime_inference(session, all_inputs, output_names)
             all_latency_list.extend(latency_list)
 
     # latency in miliseconds
diff --git a/onnxruntime/python/tools/transformers/models/gpt2/convert_to_onnx.py b/onnxruntime/python/tools/transformers/models/gpt2/convert_to_onnx.py
index 78e718e6e80c4..cb8e1a337de68 100644
--- a/onnxruntime/python/tools/transformers/models/gpt2/convert_to_onnx.py
+++ b/onnxruntime/python/tools/transformers/models/gpt2/convert_to_onnx.py
@@ -30,14 +30,14 @@
 
 sys.path.append(os.path.join(os.path.dirname(__file__), "..", ".."))
 
-from benchmark_helper import (
+from benchmark_helper import (  # noqa: E402
     Precision,
     create_onnxruntime_session,
     get_ort_environment_variables,
     prepare_environment,
     setup_logger,
 )
-from quantize_helper import QuantizeHelper
+from quantize_helper import QuantizeHelper  # noqa: E402
 
 logger = logging.getLogger("")
 
diff --git a/onnxruntime/python/tools/transformers/models/gpt2/gpt2_helper.py b/onnxruntime/python/tools/transformers/models/gpt2/gpt2_helper.py
index 250aa52a479c6..95f6a7dcaa000 100644
--- a/onnxruntime/python/tools/transformers/models/gpt2/gpt2_helper.py
+++ b/onnxruntime/python/tools/transformers/models/gpt2/gpt2_helper.py
@@ -13,7 +13,7 @@
 import tempfile
 import time
 from pathlib import Path
-from typing import Dict, List, Tuple, Union
+from typing import Collection, Dict, List, Tuple, Union
 
 import numpy
 import onnx
@@ -22,12 +22,12 @@
 
 sys.path.append(os.path.join(os.path.dirname(__file__), "..", ".."))
 
-from benchmark_helper import Precision
-from float16 import float_to_float16_max_diff
-from fusion_options import AttentionMaskFormat
-from io_binding_helper import IOBindingHelper
-from onnx_model import OnnxModel
-from torch_onnx_export_helper import torch_onnx_export
+from benchmark_helper import Precision  # noqa: E402
+from float16 import float_to_float16_max_diff  # noqa: E402
+from fusion_options import AttentionMaskFormat  # noqa: E402
+from io_binding_helper import IOBindingHelper  # noqa: E402
+from onnx_model import OnnxModel  # noqa: E402
+from torch_onnx_export_helper import torch_onnx_export  # noqa: E402
 
 logger = logging.getLogger(__name__)
 
@@ -551,24 +551,26 @@ def optimize_onnx(
     @staticmethod
     def auto_mixed_precision(
         onnx_model: OnnxModel,
-        op_block_list: List[str] = [
-            "Add",
-            "LayerNormalization",
-            "SkipLayerNormalization",
-            "FastGelu",
-            "EmbedLayerNormalization",
-        ],
+        op_block_list: Collection[str] = frozenset(
+            (
+                "Add",
+                "LayerNormalization",
+                "SkipLayerNormalization",
+                "FastGelu",
+                "EmbedLayerNormalization",
+            )
+        ),
     ):
         """Convert GPT-2 model to mixed precision.
            It detects whether original model has fp16 weights, and set parameters for float16 conversion automatically.
         Args:
-            onnx_model (OnnxModel): optimized ONNX model
-            op_block_list (List[str], optional): operators to compute in fp32. Defaults to ["Add", "LayerNormalization",
-                                                 "SkipLayerNormalization", "FastGelu", "EmbedLayerNormalization"]
+            onnx_model: Optimized ONNX model
+            op_block_list: Operators to compute in fp32. Defaults to {"Add", "LayerNormalization",
+                "SkipLayerNormalization", "FastGelu", "EmbedLayerNormalization"}.
         Returns:
-            parameters(dict): a dictionary of parameters used in float16 conversion
+            A dictionary of parameters used in float16 conversion.
         """
-        op_full_set = set([node.op_type for node in onnx_model.nodes()])
+        op_full_set = set(node.op_type for node in onnx_model.nodes())
         fp32_op_set = set(op_block_list)
         fp16_op_set = op_full_set.difference(fp32_op_set)
         logger.info(f"fp32 op: {fp32_op_set} fp16 op: {fp16_op_set}")
diff --git a/onnxruntime/python/tools/transformers/models/gpt2/gpt2_tester.py b/onnxruntime/python/tools/transformers/models/gpt2/gpt2_tester.py
index be303b4e188bf..b6e9a1406fb5c 100644
--- a/onnxruntime/python/tools/transformers/models/gpt2/gpt2_tester.py
+++ b/onnxruntime/python/tools/transformers/models/gpt2/gpt2_tester.py
@@ -17,7 +17,7 @@
 
 sys.path.append(os.path.join(os.path.dirname(__file__), "..", ".."))
 
-from benchmark_helper import Precision
+from benchmark_helper import Precision  # noqa: E402
 
 logger = logging.getLogger(__name__)
 
@@ -151,7 +151,7 @@ def __init__(
             0,
             hidden_size // num_attention_heads,
         ]
-        for i in range(num_layer):
+        for _ in range(num_layer):
             empty_past = torch.empty(past_shape).type(torch.float16 if is_fp16 else torch.float32)
             self.past.append(empty_past.to(device))
 
@@ -194,7 +194,7 @@ def add_tensor(input_tensors, torch_tensor, name):
                 f.write(tensor.SerializeToString())
 
         output_names = [output.name for output in session.get_outputs()]
-        for i, name in enumerate(output_names):
+        for i, _ in enumerate(output_names):
             tensor = numpy_helper.from_array(
                 output[i] if isinstance(output[i], numpy.ndarray) else output[i].clone().cpu().numpy()
             )
diff --git a/onnxruntime/python/tools/transformers/models/gpt2/parity_check_helper.py b/onnxruntime/python/tools/transformers/models/gpt2/parity_check_helper.py
index d57ca9bca9a5b..87dc766628aab 100644
--- a/onnxruntime/python/tools/transformers/models/gpt2/parity_check_helper.py
+++ b/onnxruntime/python/tools/transformers/models/gpt2/parity_check_helper.py
@@ -19,7 +19,7 @@
 
 sys.path.append(os.path.join(os.path.dirname(__file__), "..", ".."))
 
-from benchmark_helper import create_onnxruntime_session
+from benchmark_helper import create_onnxruntime_session  # noqa: E402
 
 NON_ZERO_VALUE = str(1)
 ZERO_VALUE = str(0)
@@ -137,8 +137,8 @@ def post_processing(outputs_path, outputs_path_other):
     dummy_inputs_fp32 = dummy_inputs_fp16.to_fp32()
 
     # Get GPT-2 model from huggingface using convert_to_onnx.py
-    os.system("python convert_to_onnx.py -m gpt2 --output gpt2_fp32.onnx -o -p fp32 --use_gpu")
-    os.system("python convert_to_onnx.py -m gpt2 --output gpt2_fp16.onnx -o -p fp16 --use_gpu")
+    os.system("python convert_to_onnx.py -m gpt2 --output gpt2_fp32.onnx -o -p fp32 --use_gpu")  # noqa: DUO106
+    os.system("python convert_to_onnx.py -m gpt2 --output gpt2_fp16.onnx -o -p fp16 --use_gpu")  # noqa: DUO106
 
     # Specify the directory to dump the node's I/O
     outputs_path_fp32_gpu = "./fp32_gpu"
diff --git a/onnxruntime/python/tools/transformers/models/t5/past_helper.py b/onnxruntime/python/tools/transformers/models/t5/past_helper.py
index fe113491067fd..d6b1d50f5a47f 100644
--- a/onnxruntime/python/tools/transformers/models/t5/past_helper.py
+++ b/onnxruntime/python/tools/transformers/models/t5/past_helper.py
@@ -38,7 +38,7 @@ def group_by_self_or_cross(present_key_values):
         """
         present_self = []
         present_cross = []
-        for i, present_layer_i in enumerate(present_key_values):
+        for present_layer_i in present_key_values:
             assert len(present_layer_i) == 4, f"Expected to have four items. Got {len(present_layer_i)}"
             (
                 present_key_self,
diff --git a/onnxruntime/python/tools/transformers/models/t5/t5_helper.py b/onnxruntime/python/tools/transformers/models/t5/t5_helper.py
index 4d853a6544ef0..56f19c7407931 100644
--- a/onnxruntime/python/tools/transformers/models/t5/t5_helper.py
+++ b/onnxruntime/python/tools/transformers/models/t5/t5_helper.py
@@ -8,7 +8,7 @@
 import os
 import sys
 from pathlib import Path
-from typing import Dict, List, Union
+from typing import Collection, Dict, Union
 
 import torch
 from t5_decoder import T5Decoder, T5DecoderHelper, T5DecoderInit
@@ -19,9 +19,9 @@
 from onnxruntime import InferenceSession
 
 sys.path.append(os.path.join(os.path.dirname(__file__), "..", ".."))
-from float16 import float_to_float16_max_diff
-from onnx_model import OnnxModel
-from optimizer import optimize_model
+from float16 import float_to_float16_max_diff  # noqa: E402
+from onnx_model import OnnxModel  # noqa: E402
+from optimizer import optimize_model  # noqa: E402
 
 logger = logging.getLogger(__name__)
 
@@ -150,26 +150,31 @@ def export_onnx(
     @staticmethod
     def auto_mixed_precision(
         onnx_model: OnnxModel,
-        op_block_list: List[str] = [
-            "Pow",
-            "ReduceMean",
-            "Add",
-            "Sqrt",
-            "Div",
-            "Mul",
-            "Softmax",
-            "Relu",
-        ],
+        op_block_list: Collection[str] = frozenset(
+            (
+                "Pow",
+                "ReduceMean",
+                "Add",
+                "Sqrt",
+                "Div",
+                "Mul",
+                "Softmax",
+                "Relu",
+            )
+        ),
     ):
         """Convert model to mixed precision.
-           It detects whether original model has fp16 precision weights, and set parameters for float16 conversion automatically.
+
+        It detects whether original model has fp16 precision weights,
+        and set parameters for float16 conversion automatically.
+
         Args:
-            onnx_model (OnnxModel): optimized ONNX model
-            op_block_list (List[str], optional): . Defaults to ["Pow", "ReduceMean", "Add", "Sqrt", "Div", "Mul", "Softmax", "Relu"]
+            onnx_model: Optimized ONNX model
+            op_block_list: Defaults to {"Pow", "ReduceMean", "Add", "Sqrt", "Div", "Mul", "Softmax", "Relu"}
         Returns:
-            parameters(dict): a dictionary of parameters used in float16 conversion
+            parameters: A dictionary of parameters used in float16 conversion
         """
-        op_full_set = set([node.op_type for node in onnx_model.nodes()])
+        op_full_set = set(node.op_type for node in onnx_model.nodes())
         fp32_op_set = set(op_block_list)
         fp16_op_set = op_full_set.difference(fp32_op_set)
         logger.info(f"fp32 op: {fp32_op_set} fp16 op: {fp16_op_set}")
diff --git a/onnxruntime/python/tools/transformers/onnx_exporter.py b/onnxruntime/python/tools/transformers/onnx_exporter.py
index c42e286922bbb..edb10a17832ba 100644
--- a/onnxruntime/python/tools/transformers/onnx_exporter.py
+++ b/onnxruntime/python/tools/transformers/onnx_exporter.py
@@ -19,7 +19,7 @@
 from transformers import AutoConfig, AutoTokenizer, LxmertConfig, TransfoXLConfig
 
 sys.path.append(os.path.join(os.path.dirname(__file__), "models", "gpt2"))
-from gpt2_helper import PRETRAINED_GPT2_MODELS, GPT2ModelNoPastState, TFGPT2ModelNoPastState
+from gpt2_helper import PRETRAINED_GPT2_MODELS, GPT2ModelNoPastState, TFGPT2ModelNoPastState  # noqa: E402
 
 os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2"
 
diff --git a/onnxruntime/python/tools/transformers/onnx_model_bart.py b/onnxruntime/python/tools/transformers/onnx_model_bart.py
index 33db231c52332..df5c841938b90 100644
--- a/onnxruntime/python/tools/transformers/onnx_model_bart.py
+++ b/onnxruntime/python/tools/transformers/onnx_model_bart.py
@@ -97,7 +97,7 @@ def fuse(self, normalize_node, input_name_to_nodes, output_name_to_node):
             return
 
         other_inputs = []
-        for i, input in enumerate(normalize_node.input):
+        for input in normalize_node.input:
             if input not in output_name_to_node:
                 continue
             if input == qkv_nodes[0].output[0]:
diff --git a/onnxruntime/python/tools/transformers/onnx_model_bert.py b/onnxruntime/python/tools/transformers/onnx_model_bert.py
index 590f00045b2d0..bb3218a5c0af0 100644
--- a/onnxruntime/python/tools/transformers/onnx_model_bert.py
+++ b/onnxruntime/python/tools/transformers/onnx_model_bert.py
@@ -235,7 +235,6 @@ def use_dynamic_axes(self, dynamic_batch_dim="batch_size", dynamic_seq_len="max_
             casted=True
         ) + self.get_graph_inputs_from_fused_nodes(casted=False)
 
-        {}
         for input in self.model.graph.input:
             if input.name in bert_graph_inputs:
                 dim_proto = input.type.tensor_type.shape.dim[0]
diff --git a/onnxruntime/python/tools/transformers/onnx_model_tnlr.py b/onnxruntime/python/tools/transformers/onnx_model_tnlr.py
index a464c4af5d49c..5947672a43845 100644
--- a/onnxruntime/python/tools/transformers/onnx_model_tnlr.py
+++ b/onnxruntime/python/tools/transformers/onnx_model_tnlr.py
@@ -5,14 +5,13 @@
 import logging
 from typing import Union
 
+import numpy as np
 from fusion_attention import AttentionMask, FusionAttention
 from fusion_utils import NumpyHelper
 from onnx import NodeProto, TensorProto, helper, numpy_helper
 from onnx_model import OnnxModel
 from onnx_model_bert import BertOnnxModel
 
-import numpy as np
-
 logger = logging.getLogger(__name__)
 
 
diff --git a/onnxruntime/python/tools/transformers/torch_onnx_export_helper.py b/onnxruntime/python/tools/transformers/torch_onnx_export_helper.py
index 119455684cea1..997461befd198 100644
--- a/onnxruntime/python/tools/transformers/torch_onnx_export_helper.py
+++ b/onnxruntime/python/tools/transformers/torch_onnx_export_helper.py
@@ -4,9 +4,9 @@
 # --------------------------------------------------------------------------
 
 import torch
+from packaging.version import Version
 
 TrainingMode = torch.onnx.TrainingMode
-from packaging.version import Version
 
 
 def torch_onnx_export(
diff --git a/tools/ci_build/github/android/build_aar_package.py b/tools/ci_build/github/android/build_aar_package.py
index fac04797cfff1..5f6f753d1b33c 100644
--- a/tools/ci_build/github/android/build_aar_package.py
+++ b/tools/ci_build/github/android/build_aar_package.py
@@ -155,9 +155,9 @@ def _build_aar(args):
     # clean, build, and publish to a local directory
     subprocess.run(gradle_command + ["clean"], env=temp_env, shell=use_shell, check=True, cwd=JAVA_ROOT)  # noqa: DUO116
     subprocess.run(gradle_command + ["build"], env=temp_env, shell=use_shell, check=True, cwd=JAVA_ROOT)  # noqa: DUO116
-    subprocess.run(
+    subprocess.run(  # noqa: DUO116
         gradle_command + ["publish"], env=temp_env, shell=use_shell, check=True, cwd=JAVA_ROOT
-    )  # noqa: DUO116
+    )
 
 
 def parse_args():
diff --git a/tools/doc/rename_folders.py b/tools/doc/rename_folders.py
index d65b8a350eed1..56b2a39357411 100644
--- a/tools/doc/rename_folders.py
+++ b/tools/doc/rename_folders.py
@@ -13,7 +13,7 @@ def rename_folder(root):
     Returns the list of renamed folders.
     """
     found = []
-    for r, dirs, files in os.walk(root):
+    for r, dirs, _ in os.walk(root):
         for name in dirs:
             if name.startswith("_"):
                 found.append((r, name))
@@ -35,7 +35,7 @@ def replace_files(root, renamed):
     subs = {r[1]: r[2] for r in renamed}
     reg = re.compile('(\\"[a-zA-Z0-9\\.\\/\\?\\:@\\-_=#]+\\.([a-zA-Z]){2,6}' '([a-zA-Z0-9\\.\\&\\/\\?\\:@\\-_=#])*\\")')
 
-    for r, dirs, files in os.walk(root):
+    for r, _, files in os.walk(root):
         for name in files:
             if os.path.splitext(name)[-1] != ".html":
                 continue
diff --git a/tools/python/example_operator_perf_test.py b/tools/python/example_operator_perf_test.py
index 50a3edd5c9b27..41c0d605bc636 100644
--- a/tools/python/example_operator_perf_test.py
+++ b/tools/python/example_operator_perf_test.py
@@ -99,7 +99,7 @@ def run_test():
         # run the model and measure time after 'iters' calls
         while total < num_seconds:
             start = time.time_ns()
-            for i in range(iters):
+            for _ in range(iters):
                 # ignore the outputs as we're not validating them in a performance test
                 sess.run(None, inputs)
             end = time.time_ns()
diff --git a/tools/python/gen_contrib_doc.py b/tools/python/gen_contrib_doc.py
index 15e7f65d093d9..6604a433ac528 100644
--- a/tools/python/gen_contrib_doc.py
+++ b/tools/python/gen_contrib_doc.py
@@ -361,7 +361,7 @@ def main(output_path: str, domain_filter: [str]):
             fout.write(s)
 
             for _, namemap in supportmap:
-                for n, schema, versions in namemap:
+                for n, schema, _ in namemap:
                     s = '  * {}<a href="#{}">{}</a>\n'.format(
                         support_level_str(schema.support_level),
                         format_name_with_domain(domain, n),
diff --git a/tools/python/util/run.py b/tools/python/util/run.py
index c3a389233ff72..98724ae956402 100644
--- a/tools/python/util/run.py
+++ b/tools/python/util/run.py
@@ -46,7 +46,7 @@ def run(
     def output(is_stream_captured):
         return subprocess.PIPE if is_stream_captured else (subprocess.DEVNULL if quiet else None)
 
-    completed_process = subprocess.run(
+    completed_process = subprocess.run(  # noqa: DUO116
         cmd,
         cwd=cwd,
         check=check,