diff --git a/.ci/helion/install.sh b/.ci/helion/install.sh
index 30d889bbf..b9f366fee 100644
--- a/.ci/helion/install.sh
+++ b/.ci/helion/install.sh
@@ -37,4 +37,4 @@ cd ${tritonbench_dir}
 
 python install.py --helion
 # Helion requires tritonbench installed as a library
-pip install -e .
\ No newline at end of file
+pip install -e .
diff --git a/benchmarks/gen_metadata/run.py b/benchmarks/gen_metadata/run.py
index e8b91784a..d570df404 100644
--- a/benchmarks/gen_metadata/run.py
+++ b/benchmarks/gen_metadata/run.py
@@ -87,7 +87,7 @@ def run(args: argparse.Namespace):
         DTYPE_OPERATORS[op] = op_bench.DEFAULT_PRECISION
         if baseline := op_bench.has_baseline():
             BASELINE_OPERATORS[op] = baseline
-        if op_bench.has_metric("tflops") and not op in TFLOPS_SKIP_OPERATORS:
+        if op_bench.has_metric("tflops") and op not in TFLOPS_SKIP_OPERATORS:
             TFLOPS_OPERATORS.append(op)
         if op_bench.has_bwd():
             BACKWARD_OPERATORS.append(op)
diff --git a/benchmarks/mojo_matmul/run.py b/benchmarks/mojo_matmul/run.py
index f981ed496..9c36f54d6 100644
--- a/benchmarks/mojo_matmul/run.py
+++ b/benchmarks/mojo_matmul/run.py
@@ -4,14 +4,10 @@
 pip install --pre modular --index-url https://dl.modular.com/public/nightly/python/simple/
 """
 
-import argparse
-import json
-import logging
 import os
 import sys
 
 from os.path import abspath, exists
-from typing import Dict, List
 
 
 def setup_tritonbench_cwd():
@@ -36,15 +32,13 @@ def setup_tritonbench_cwd():
 from typing import Callable
 
 import max.graph as mg
-import torch
 
 from max import driver, engine
-from max.graph import DeviceRef, Graph, ops, TensorType, TensorValue
-from max.graph.type import DType, Shape, ShapeLike
+from max.graph import DeviceRef, ops, TensorType
+from max.graph.type import DType
 
 from tritonbench.operators import load_opbench_by_name
 from tritonbench.utils.parser import get_parser
-from tritonbench.utils.triton_op import register_benchmark
 
 
 def promote_mojo_tensor_to_fp32(mojo_tensor, dtype):
diff --git a/benchmarks/nightly/gen.py b/benchmarks/nightly/gen.py
index 1f9f383b5..1ec5acd37 100644
--- a/benchmarks/nightly/gen.py
+++ b/benchmarks/nightly/gen.py
@@ -59,7 +59,7 @@ def gen_run(operators: List[str], bwd: bool = False) -> Dict[str, Any]:
             cmd.append("--bwd")
         # add backends
         run_backends = list(TRITON_OPS[op].keys())
-        if _has_meaningful_baseline(op) and not BASELINE_OPS[op] in run_backends:
+        if _has_meaningful_baseline(op) and BASELINE_OPS[op] not in run_backends:
             run_backends.append(BASELINE_OPS[op])
         cmd.extend(["--only", ",".join(run_backends)])
         out[run_name] = {}
diff --git a/benchmarks/nightly/run.py b/benchmarks/nightly/run.py
index 831c6bdd6..dcd53e3a0 100644
--- a/benchmarks/nightly/run.py
+++ b/benchmarks/nightly/run.py
@@ -107,7 +107,7 @@ def run():
     logger.info(f"[nightly] logging result json file to {result_json_file}.")
     if args.log_scuba:
         log_benchmark(aggregated_obj)
-        logger.info(f"[nightly] logging results to scuba.")
+        logger.info("[nightly] logging results to scuba.")
 
 
 if __name__ == "__main__":
diff --git a/benchmarks/power_analysis/run.py b/benchmarks/power_analysis/run.py
index 179f928a3..8cc2757ff 100644
--- a/benchmarks/power_analysis/run.py
+++ b/benchmarks/power_analysis/run.py
@@ -7,7 +7,6 @@
 import os
 import sys
 
-import torch
 
 CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
 
diff --git a/pyproject.toml b/pyproject.toml
index a5c7264be..cd1e47df0 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -15,6 +15,10 @@ dependencies = [
 [tool.setuptools.packages.find]
 include = ["tritonbench*"]
 
+[tool.ruff]
+fix = true
+exclude = ["submodules"]
+
 [tool.ufmt]
 formatter = "ruff-api"
 sorter = "usort"
diff --git a/tools/flash_attn/install.py b/tools/flash_attn/install.py
index 8179db670..c804fed64 100644
--- a/tools/flash_attn/install.py
+++ b/tools/flash_attn/install.py
@@ -1,6 +1,5 @@
 import os
 import subprocess
-import sys
 
 from pathlib import Path
 
diff --git a/tritonbench/components/power/charts.py b/tritonbench/components/power/charts.py
index cfb907a36..38a340c1a 100644
--- a/tritonbench/components/power/charts.py
+++ b/tritonbench/components/power/charts.py
@@ -1,9 +1,6 @@
 import csv
 import logging
 import os
-import signal
-import subprocess
-import time
 
 import matplotlib.pyplot as plt
 
diff --git a/tritonbench/components/power/power_manager.py b/tritonbench/components/power/power_manager.py
index 0ea01440d..cd1e4c71c 100644
--- a/tritonbench/components/power/power_manager.py
+++ b/tritonbench/components/power/power_manager.py
@@ -1,5 +1,4 @@
 import csv
-import dataclasses
 import os
 import threading
 import time
@@ -14,12 +13,10 @@
     NVML_CLOCK_SM,
     NVML_FI_DEV_POWER_CURRENT_LIMIT,
     NVML_FI_DEV_POWER_INSTANT,
-    NVML_SUCCESS,
     NVML_TEMPERATURE_GPU,
     nvmlDeviceGetClock,
     nvmlDeviceGetFieldValues,
     nvmlDeviceGetHandleByIndex,
-    nvmlDeviceGetPerformanceState,
     nvmlDeviceGetTemperature,
     nvmlInit,
     nvmlShutdown,
diff --git a/tritonbench/components/tasks/manager.py b/tritonbench/components/tasks/manager.py
index 84872c70b..ea70c74dc 100644
--- a/tritonbench/components/tasks/manager.py
+++ b/tritonbench/components/tasks/manager.py
@@ -35,11 +35,8 @@ def make_instance(
         class_name: str,
     ) -> None:
         import importlib
-        import os
-        import traceback
 
         # required as this is in child process
-        from tritonbench.components.power.power_manager import PowerManager
 
         module = importlib.import_module(module_path, package=package)
         Ctor = getattr(module, class_name)
diff --git a/tritonbench/kernels/blackwell_attention_utils.py b/tritonbench/kernels/blackwell_attention_utils.py
index f22707da1..3e6ddc133 100644
--- a/tritonbench/kernels/blackwell_attention_utils.py
+++ b/tritonbench/kernels/blackwell_attention_utils.py
@@ -3,9 +3,6 @@
 generic attention kernels.
 """
 
-import os
-from functools import lru_cache
-
 import torch
 import triton
 
diff --git a/tritonbench/kernels/blackwell_triton_fused_attention.py b/tritonbench/kernels/blackwell_triton_fused_attention.py
index 03174c616..e432eb29c 100644
--- a/tritonbench/kernels/blackwell_triton_fused_attention.py
+++ b/tritonbench/kernels/blackwell_triton_fused_attention.py
@@ -24,7 +24,6 @@
     is_blackwell,
     is_cuda,
     is_hip,
-    is_hopper,
     supports_host_descriptor,
 )
 
diff --git a/tritonbench/kernels/gluon_attention_forward.py b/tritonbench/kernels/gluon_attention_forward.py
index d9e5336f0..f39ba3ada 100644
--- a/tritonbench/kernels/gluon_attention_forward.py
+++ b/tritonbench/kernels/gluon_attention_forward.py
@@ -1,5 +1,3 @@
-import itertools
-
 import torch
 import triton
 import triton.language as tl
diff --git a/tritonbench/kernels/gluon_attention_persistent_forward.py b/tritonbench/kernels/gluon_attention_persistent_forward.py
index a03de4492..0b0770d5e 100644
--- a/tritonbench/kernels/gluon_attention_persistent_forward.py
+++ b/tritonbench/kernels/gluon_attention_persistent_forward.py
@@ -1,5 +1,4 @@
 import copy
-import itertools
 
 import torch
 import triton
diff --git a/tritonbench/operators/addmm/hstu.py b/tritonbench/operators/addmm/hstu.py
index 2aeed9f85..3be3355c3 100644
--- a/tritonbench/operators/addmm/hstu.py
+++ b/tritonbench/operators/addmm/hstu.py
@@ -1,5 +1,3 @@
-import importlib
-
 from typing import Tuple
 
 import torch
diff --git a/tritonbench/operators/addmm/operator.py b/tritonbench/operators/addmm/operator.py
index e5f9f8dc8..493f722e5 100644
--- a/tritonbench/operators/addmm/operator.py
+++ b/tritonbench/operators/addmm/operator.py
@@ -16,7 +16,6 @@
 except ImportError:
     streamk_matmul = None
 
-from tritonbench.operators.gemm import stream_k
 from tritonbench.utils.triton_op import (
     BenchmarkOperator,
     BenchmarkOperatorMetrics,
diff --git a/tritonbench/operators/bf16xint16_gemm/bf16xint16_gemm.py b/tritonbench/operators/bf16xint16_gemm/bf16xint16_gemm.py
index 17ce423ea..528d3d690 100644
--- a/tritonbench/operators/bf16xint16_gemm/bf16xint16_gemm.py
+++ b/tritonbench/operators/bf16xint16_gemm/bf16xint16_gemm.py
@@ -10,13 +10,11 @@
 """
 
 import argparse
-import statistics
 
 from typing import Any, List, Optional
 
 import torch
 import triton
-import triton.language as tl
 
 from tritonbench.utils.triton_op import (
     BenchmarkOperator,
diff --git a/tritonbench/operators/blackwell_attentions/operator.py b/tritonbench/operators/blackwell_attentions/operator.py
index e82578221..60eb79845 100644
--- a/tritonbench/operators/blackwell_attentions/operator.py
+++ b/tritonbench/operators/blackwell_attentions/operator.py
@@ -187,7 +187,7 @@ def _is_sdpa_cudnn_attention_available():
     try:
         _sdpa_cudnn_attention(q, k, v)
         return True
-    except RuntimeError as e:
+    except RuntimeError:
         return False
 
 
diff --git a/tritonbench/operators/flex_attention/operator.py b/tritonbench/operators/flex_attention/operator.py
index d8fbc1b58..9166b7dc7 100644
--- a/tritonbench/operators/flex_attention/operator.py
+++ b/tritonbench/operators/flex_attention/operator.py
@@ -26,7 +26,6 @@
 except ImportError:
     pass
 
-from tritonbench.utils.input import input_filter
 from tritonbench.utils.triton_op import (
     BenchmarkOperator,
     BenchmarkOperatorMetrics,
@@ -379,7 +378,7 @@ def flash_v3(
             )
         elif mod_type == "document_mask":
             # Document mask requires special handling with varlen function
-            print(f"[SKIP] Flash Attention v3 document_mask not implemented yet")
+            print("[SKIP] Flash Attention v3 document_mask not implemented yet")
             raise NotImplementedError(
                 "Flash Attention v3 document_mask not implemented yet"
             )
diff --git a/tritonbench/operators/fp8_attention/operator.py b/tritonbench/operators/fp8_attention/operator.py
index c5384e81b..c7fdce036 100644
--- a/tritonbench/operators/fp8_attention/operator.py
+++ b/tritonbench/operators/fp8_attention/operator.py
@@ -6,7 +6,7 @@
 import argparse
 import math
 
-from typing import Any, Callable, Generator, List, Optional, Tuple
+from typing import Any, Callable, Generator, List, Optional
 
 import torch
 
diff --git a/tritonbench/operators/fp8_gemm/persistent.py b/tritonbench/operators/fp8_gemm/persistent.py
index 9c44e5a69..ad280df6f 100644
--- a/tritonbench/operators/fp8_gemm/persistent.py
+++ b/tritonbench/operators/fp8_gemm/persistent.py
@@ -1,5 +1,3 @@
-from functools import lru_cache
-
 from typing import Optional
 
 import torch
diff --git a/tritonbench/operators/fp8_gemm_blockwise/operator.py b/tritonbench/operators/fp8_gemm_blockwise/operator.py
index 8ce3a2810..e771adca9 100644
--- a/tritonbench/operators/fp8_gemm_blockwise/operator.py
+++ b/tritonbench/operators/fp8_gemm_blockwise/operator.py
@@ -42,14 +42,10 @@ def parse_args(args: List[str]) -> argparse.Namespace:
 HAS_CUTLASS = False
 if is_cuda():
     try:
-        import fbgemm_gpu.experimental.gen_ai
-
         cutlass_fp8_block = torch.ops.llama_cpp.fp8_blockwise_matmul
         HAS_CUTLASS = True
     except:
         try:
-            import fbgemm_gpu.experimental.gen_ai
-
             cutlass_fp8_block = torch.ops.fbgemm.f8f8bf16_blockwise
             HAS_CUTLASS = True
         except:
diff --git a/tritonbench/operators/fp8_gemm_rowwise_grouped/operator.py b/tritonbench/operators/fp8_gemm_rowwise_grouped/operator.py
index 1cd8e44b6..913aa3a15 100644
--- a/tritonbench/operators/fp8_gemm_rowwise_grouped/operator.py
+++ b/tritonbench/operators/fp8_gemm_rowwise_grouped/operator.py
@@ -48,14 +48,11 @@
 
 # Import necessary libraries and modules
 import argparse
-import random
 from typing import Any, Callable, Generator, List, Optional, Tuple
 
 import torch
 import triton
 
-from tritonbench.utils.data_utils import get_production_shapes
-
 from tritonbench.utils.triton_op import (
     BenchmarkOperator,
     BenchmarkOperatorMetrics,
diff --git a/tritonbench/operators/gdpa/gdpa.py b/tritonbench/operators/gdpa/gdpa.py
index ffb59c39c..a0fea4d1c 100644
--- a/tritonbench/operators/gdpa/gdpa.py
+++ b/tritonbench/operators/gdpa/gdpa.py
@@ -19,7 +19,6 @@
 from typing import Tuple
 
 import torch
-import torch.nn.functional as F
 
 import triton  # @manual=//triton:triton
 import triton.language as tl  # @manual=//triton:triton
diff --git a/tritonbench/operators/gdpa/gdpa_blackwell_tlx.py b/tritonbench/operators/gdpa/gdpa_blackwell_tlx.py
index 3cc5be1da..acfd2abb3 100644
--- a/tritonbench/operators/gdpa/gdpa_blackwell_tlx.py
+++ b/tritonbench/operators/gdpa/gdpa_blackwell_tlx.py
@@ -10,7 +10,7 @@
 from triton.tools.tensor_descriptor import TensorDescriptor
 
 from .gdpa_utils import get_num_sms
-from .math import activation_string_to_int, fast_gelu_grad, gelu, gelu_grad
+from .math import activation_string_to_int, gelu, gelu_grad
 
 
 def _host_descriptor_pre_hook(nargs):
diff --git a/tritonbench/operators/gdpa/gdpa_utils.py b/tritonbench/operators/gdpa/gdpa_utils.py
index 77ef95948..35d2e3ffa 100644
--- a/tritonbench/operators/gdpa/gdpa_utils.py
+++ b/tritonbench/operators/gdpa/gdpa_utils.py
@@ -3,7 +3,7 @@
 # pyre-strict
 import math
 from functools import lru_cache
-from typing import Any, List, Optional
+from typing import Any, Optional
 
 import torch
 import triton  # @manual=//triton:triton
diff --git a/tritonbench/operators/gdpa/operator.py b/tritonbench/operators/gdpa/operator.py
index 8fa69e538..41c287282 100644
--- a/tritonbench/operators/gdpa/operator.py
+++ b/tritonbench/operators/gdpa/operator.py
@@ -133,7 +133,7 @@ def parse_args(args):
         "--kv_len",
         default=None,
         type=int,
-        help=f"Sequence length for K/V, if None, the tensor will be jagged and have the same length as Q",
+        help="Sequence length for K/V, if None, the tensor will be jagged and have the same length as Q",
     )
     parser.add_argument(
         "--activation",
diff --git a/tritonbench/operators/grouped_gemm/cutedsl/kernels.py b/tritonbench/operators/grouped_gemm/cutedsl/kernels.py
index 93952f269..133bce5a0 100644
--- a/tritonbench/operators/grouped_gemm/cutedsl/kernels.py
+++ b/tritonbench/operators/grouped_gemm/cutedsl/kernels.py
@@ -1996,7 +1996,7 @@ def compile_cutedsl_grouped_gemm(
         C_cpu = torch.zeros((m, n, 1), dtype=torch.float32)
         torch_fp32_tensors_abc_seed.append([A_cpu, B_cpu, C_cpu])
 
-    print(f"Running Blackwell Grouped GEMM test with:")
+    print("Running Blackwell Grouped GEMM test with:")
     print(f"{num_groups} groups")
     for i, (m, n, k, l) in enumerate(problem_sizes_mnkl):
         print(f"Group {i}: {m}x{n}x{k}x{l}")
diff --git a/tritonbench/operators/int4_gemm/int4_gemm.py b/tritonbench/operators/int4_gemm/int4_gemm.py
index de1040940..095231190 100644
--- a/tritonbench/operators/int4_gemm/int4_gemm.py
+++ b/tritonbench/operators/int4_gemm/int4_gemm.py
@@ -6,13 +6,11 @@
 """
 
 import argparse
-import statistics
 
 from typing import Any, List, Optional
 
 import torch
 import triton
-import triton.language as tl
 
 from tritonbench.utils.triton_op import (
     BenchmarkOperator,
@@ -21,7 +19,7 @@
     register_metric,
 )
 
-from .kernel import _group_quantize_tensor, matmul, matmul_kernel, pack_2xint4
+from .kernel import _group_quantize_tensor, matmul, pack_2xint4
 
 
 class Operator(BenchmarkOperator):
diff --git a/tritonbench/operators/jagged_layer_norm/operator.py b/tritonbench/operators/jagged_layer_norm/operator.py
index e43a9af3d..632818f38 100644
--- a/tritonbench/operators/jagged_layer_norm/operator.py
+++ b/tritonbench/operators/jagged_layer_norm/operator.py
@@ -1,8 +1,5 @@
 import argparse
-import itertools
-import math
 import os
-import random
 from typing import Callable, Generator, List, Optional, Tuple
 
 import torch
diff --git a/tritonbench/operators/jagged_softmax/operator.py b/tritonbench/operators/jagged_softmax/operator.py
index 57c3a166a..b64ad511c 100644
--- a/tritonbench/operators/jagged_softmax/operator.py
+++ b/tritonbench/operators/jagged_softmax/operator.py
@@ -1,8 +1,5 @@
 import argparse
-import itertools
-import math
 import os
-import random
 from typing import Callable, Generator, List, Optional, Tuple
 
 import torch
diff --git a/tritonbench/operators/launch_latency/operator.py b/tritonbench/operators/launch_latency/operator.py
index a2099ce40..a800945e3 100644
--- a/tritonbench/operators/launch_latency/operator.py
+++ b/tritonbench/operators/launch_latency/operator.py
@@ -1,16 +1,9 @@
-import triton.language as tl
 from torch import zeros
-from torch._C import _cuda_getCurrentRawStream as get_raw_stream
 
 from torch._inductor.utils import triton_version_uses_attrs_dict
 from triton.compiler import CompiledKernel
 
-from tritonbench.utils.triton_op import (
-    BenchmarkOperator,
-    BenchmarkOperatorMetrics,
-    register_benchmark,
-    register_metric,
-)
+from tritonbench.utils.triton_op import BenchmarkOperator, register_benchmark
 
 from .kernels import get_trivial_add_kernel, nop_kernel, nop_with_args_kernel
 
diff --git a/tritonbench/operators/low_mem_dropout/kernels.py b/tritonbench/operators/low_mem_dropout/kernels.py
index 3278b9464..c01a43d88 100644
--- a/tritonbench/operators/low_mem_dropout/kernels.py
+++ b/tritonbench/operators/low_mem_dropout/kernels.py
@@ -1,6 +1,3 @@
-import tabulate
-import torch
-
 import triton
 import triton.language as tl
 
diff --git a/tritonbench/operators/mamba2_chunk_scan/operator.py b/tritonbench/operators/mamba2_chunk_scan/operator.py
index c5cddcf1d..3795891d8 100644
--- a/tritonbench/operators/mamba2_chunk_scan/operator.py
+++ b/tritonbench/operators/mamba2_chunk_scan/operator.py
@@ -1,22 +1,17 @@
 import argparse
-import functools
 import itertools
 import os
 import sys
-from contextlib import nullcontext
-from itertools import chain
 
-from typing import Any, Callable, Generator, List, Optional
+from typing import Any, Generator, List, Optional
 
 import torch
 
-from tritonbench.utils.input import input_filter
 from tritonbench.utils.python_utils import try_import
 
 from tritonbench.utils.triton_op import (
     BenchmarkOperator,
     BenchmarkOperatorMetrics,
-    Mode as BenchmarkMode,
     register_benchmark,
     register_metric,
     register_x_val,
diff --git a/tritonbench/operators/mamba2_chunk_state/operator.py b/tritonbench/operators/mamba2_chunk_state/operator.py
index 0e2376c55..949f4c0c4 100644
--- a/tritonbench/operators/mamba2_chunk_state/operator.py
+++ b/tritonbench/operators/mamba2_chunk_state/operator.py
@@ -3,20 +3,16 @@
 import itertools
 import os
 import sys
-from contextlib import nullcontext
-from itertools import chain
 
-from typing import Any, Callable, Generator, List, Optional
+from typing import Any, Generator, List, Optional
 
 import torch
 
-from tritonbench.utils.input import input_filter
 from tritonbench.utils.python_utils import try_import
 
 from tritonbench.utils.triton_op import (
     BenchmarkOperator,
     BenchmarkOperatorMetrics,
-    Mode as BenchmarkMode,
     register_benchmark,
     register_metric,
     register_x_val,
diff --git a/tritonbench/operators/op_task.py b/tritonbench/operators/op_task.py
index 413025b69..d4a84d315 100644
--- a/tritonbench/operators/op_task.py
+++ b/tritonbench/operators/op_task.py
@@ -96,10 +96,6 @@ def worker(self) -> subprocess_worker.SubprocessWorker:
     @base_task.run_in_worker(scoped=True)
     @staticmethod
     def _maybe_import_operator(package: str, op_name: str) -> Dict[str, Any]:
-        import importlib
-        import os
-        import traceback
-
         from tritonbench.operators import load_opbench_by_name
 
         Operator = load_opbench_by_name(op_name)
diff --git a/tritonbench/operators/ragged_attention/operator.py b/tritonbench/operators/ragged_attention/operator.py
index ff2b25664..ec68e3045 100644
--- a/tritonbench/operators/ragged_attention/operator.py
+++ b/tritonbench/operators/ragged_attention/operator.py
@@ -1,12 +1,11 @@
 import argparse
 
-from typing import Any, Callable, List, Optional
+from typing import List, Optional
 
 import torch
 
 from tritonbench.utils.env_utils import get_nvidia_gpu_model, is_cuda, is_fbcode
 
-from tritonbench.utils.input import input_filter
 from tritonbench.utils.triton_op import (
     BenchmarkOperator,
     BenchmarkOperatorMetrics,
diff --git a/tritonbench/operators/softmax/operator.py b/tritonbench/operators/softmax/operator.py
index e258630e6..665bef75d 100644
--- a/tritonbench/operators/softmax/operator.py
+++ b/tritonbench/operators/softmax/operator.py
@@ -1,5 +1,5 @@
 import argparse
-from typing import Generator, List, Optional
+from typing import List, Optional
 
 import torch
 import triton
diff --git a/tritonbench/operators/sum/kernels.py b/tritonbench/operators/sum/kernels.py
index 65b865ef6..eb7795bfd 100644
--- a/tritonbench/operators/sum/kernels.py
+++ b/tritonbench/operators/sum/kernels.py
@@ -1,6 +1,5 @@
 import itertools
 
-import torch
 import triton
 import triton.language as tl
 
diff --git a/tritonbench/operators/sum/operator.py b/tritonbench/operators/sum/operator.py
index 6ea86f01a..ecad273bb 100644
--- a/tritonbench/operators/sum/operator.py
+++ b/tritonbench/operators/sum/operator.py
@@ -2,11 +2,10 @@
 import itertools
 import math
 import os
-from typing import Callable, Generator, List, Optional, Tuple
+from typing import Callable, Generator, List, Optional
 
 import torch
 import triton
-import triton.language as tl
 
 from tritonbench.utils.triton_op import (
     BenchmarkOperator,
@@ -18,7 +17,6 @@
 from .kernels import (
     triton_sum_kernel_1D_result_buffer_then_sum,
     triton_sum_kernel_1D_result_sum_then_buffer,
-    triton_sum_kernel_2D_result_dim_1,
     triton_sum_kernel_2D_result_dim_1_sum_then_buffer,
     triton_sum_kernel_scalar_result,
 )
diff --git a/tritonbench/operators/template_attention/operator.py b/tritonbench/operators/template_attention/operator.py
index eff94ca8e..efd4b6962 100644
--- a/tritonbench/operators/template_attention/operator.py
+++ b/tritonbench/operators/template_attention/operator.py
@@ -1,20 +1,10 @@
 import argparse
-import csv
-import os
-import statistics
-from typing import Any, Callable, Generator, List, Optional
+from typing import Callable, Generator, List, Optional
 
-import numpy
 import torch
-import triton
 from torch._dynamo.testing import rand_strided, same
 
-from tritonbench.utils.triton_op import (
-    BenchmarkOperator,
-    BenchmarkOperatorMetrics,
-    register_benchmark,
-    register_metric,
-)
+from tritonbench.utils.triton_op import BenchmarkOperator, register_benchmark
 
 from .triton_attention import (
     triton_attention_no_exp2 as triton_test_no_exp2,
diff --git a/tritonbench/operators/template_attention/triton_attention.py b/tritonbench/operators/template_attention/triton_attention.py
index 45c7864f4..e0eb5129a 100644
--- a/tritonbench/operators/template_attention/triton_attention.py
+++ b/tritonbench/operators/template_attention/triton_attention.py
@@ -7,9 +7,6 @@
 
 import triton
 import triton.language as tl
-from torch._C import _cuda_getCurrentRawStream as get_raw_stream
-from torch._inductor.runtime import triton_helpers, triton_heuristics
-from torch._inductor.runtime.triton_helpers import libdevice
 
 empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
 reinterpret_tensor = torch.ops.inductor._reinterpret_tensor
diff --git a/tritonbench/operators/vector_add/kernels.py b/tritonbench/operators/vector_add/kernels.py
index 40aa3734f..28e0b2513 100644
--- a/tritonbench/operators/vector_add/kernels.py
+++ b/tritonbench/operators/vector_add/kernels.py
@@ -1,4 +1,3 @@
-import torch
 import triton
 import triton.language as tl
 
diff --git a/tritonbench/operators/vector_exp/operator.py b/tritonbench/operators/vector_exp/operator.py
index 97160955a..1967d00b7 100644
--- a/tritonbench/operators/vector_exp/operator.py
+++ b/tritonbench/operators/vector_exp/operator.py
@@ -1,4 +1,4 @@
-from typing import Any, Callable, Generator, List
+from typing import Callable, Generator, List
 
 import torch
 import triton
diff --git a/tritonbench/operators/welford/operator.py b/tritonbench/operators/welford/operator.py
index 6deb43bad..c170cb3c1 100644
--- a/tritonbench/operators/welford/operator.py
+++ b/tritonbench/operators/welford/operator.py
@@ -1,14 +1,10 @@
 import argparse
-from typing import Any, Callable, Generator, List, Optional
+from typing import Callable, Generator, List, Optional
 
 import torch
 from torch._dynamo.testing import rand_strided
 
-from tritonbench.utils.triton_op import (
-    BenchmarkOperator,
-    register_benchmark,
-    register_metric,
-)
+from tritonbench.utils.triton_op import BenchmarkOperator, register_benchmark
 
 from .triton_welford import (
     fused_native_layer_norm as triton_welford,
diff --git a/tritonbench/operators/welford/triton_welford.py b/tritonbench/operators/welford/triton_welford.py
index b248cfe71..648d666c0 100644
--- a/tritonbench/operators/welford/triton_welford.py
+++ b/tritonbench/operators/welford/triton_welford.py
@@ -8,7 +8,7 @@
 import triton
 import triton.language as tl
 from torch._C import _cuda_getCurrentRawStream as get_raw_stream
-from torch._inductor.runtime import triton_helpers, triton_heuristics
+from torch._inductor.runtime import triton_helpers
 from torch._inductor.runtime.triton_helpers import libdevice
 
 empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
diff --git a/tritonbench/utils/list_operator_details.py b/tritonbench/utils/list_operator_details.py
index d11accb42..ebd29cb15 100644
--- a/tritonbench/utils/list_operator_details.py
+++ b/tritonbench/utils/list_operator_details.py
@@ -2,7 +2,6 @@
 Utilities for listing operator details including metrics and backends in tritonbench.
 """
 
-import sys
 from dataclasses import fields
 from typing import Dict, List, Optional
 
@@ -10,7 +9,6 @@
 INDENT = "  "  # Base indentation unit (2 spaces)
 INDENT2 = INDENT * 2  # Double indentation (4 spaces)
 
-from tritonbench.operators_collection import list_operators_by_collection
 from tritonbench.utils.operator_utils import (
     batch_load_operators,
     get_backends_for_operator,
@@ -18,7 +16,7 @@
     get_overridden_metrics_for_operator,
     is_operator_loaded,
 )
-from tritonbench.utils.triton_op import BenchmarkOperatorMetrics, REGISTERED_BENCHMARKS
+from tritonbench.utils.triton_op import BenchmarkOperatorMetrics
 
 
 def get_builtin_metrics() -> List[str]:
diff --git a/tritonbench/utils/python_utils.py b/tritonbench/utils/python_utils.py
index 568e7d9fd..d7b477a2d 100644
--- a/tritonbench/utils/python_utils.py
+++ b/tritonbench/utils/python_utils.py
@@ -9,5 +9,5 @@ def try_import(cond_name: str):
     try:
         yield
         _caller_globals[cond_name] = True
-    except (ImportError, ModuleNotFoundError) as e:
+    except (ImportError, ModuleNotFoundError):
         _caller_globals[cond_name] = False
diff --git a/tritonbench/utils/run_utils.py b/tritonbench/utils/run_utils.py
index df598eafc..5006362c7 100644
--- a/tritonbench/utils/run_utils.py
+++ b/tritonbench/utils/run_utils.py
@@ -96,7 +96,7 @@ def run_in_helion(op: str, op_args: Dict[str, str], extra_envs: Dict[str, str]):
     environ.update(extra_envs)
     cmd = [sys.executable, "benchmarks/run.py"] + op_args
     print(
-        f"[tritonbench] Running helion benchmark: " + " ".join(cmd),
+        "[tritonbench] Running helion benchmark: " + " ".join(cmd),
         flush=True,
     )
     subprocess.check_call(
@@ -303,7 +303,7 @@ def load_operator_by_args(task_args: List[str]):
 def run_one_operator(task_args: List[str], with_bwd: bool = False):
     op = load_operator_by_args(task_args)
     op.run()
-    if with_bwd and op.has_bwd() and not op.name in FWD_ONLY_OPS:
+    if with_bwd and op.has_bwd() and op.name not in FWD_ONLY_OPS:
         op_name = copy.deepcopy(op.name)
         del op
         if op_name in BWD_ARGS_OPS:
diff --git a/tritonbench/utils/triton_op.py b/tritonbench/utils/triton_op.py
index eaefeb2f2..5448b24e1 100644
--- a/tritonbench/utils/triton_op.py
+++ b/tritonbench/utils/triton_op.py
@@ -819,8 +819,8 @@ def __post__init__(self):
                 )
             if self._input_sample_mode == "equally-spaced-k":
                 raise ValueError(
-                    f"Cannot use --input-sample-mode equally-spaced-k with multiple input IDs. "
-                    f"Either specify multiple IDs directly or use equally-spaced-k with --num-inputs."
+                    "Cannot use --input-sample-mode equally-spaced-k with multiple input IDs. "
+                    "Either specify multiple IDs directly or use equally-spaced-k with --num-inputs."
                 )
             # Validate that all IDs are within range
             invalid_ids = [