[TRTLLM-11289][fix] Fix pre-commit formatting violations

Apply ruff format/lint fixes:
- Convert multi-line docstrings to single-line where appropriate (D200)
- Remove f-string prefix on strings without placeholders (F541)
- Remove unused import
- Use consistent double-quote docstrings instead of single-quotes
- Fix indentation in docstrings

Signed-off-by: Peace He <peaceh@nvidia.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: peaceh-nv

tensorrt_llm/_torch/custom_ops/cute_dsl_custom_ops.py

@@ -331,8 +331,6 @@ def get_dense_gemm_approximate_cta_nums(
         STATE_SIZE as CLUSTER_TOPK_STATE_SIZE
     from ..cute_dsl_kernels.blackwell.top_k.single_pass_multi_cta_radix_topk_cluster import (
         SinglePassMultiCTARadixTopKClusterKernel, _query_max_cluster_size)
-    from ..cute_dsl_kernels.blackwell.dense_gemm_persistent import \
-        PersistentDenseGemmKernel
     from ..cute_dsl_kernels.blackwell.utils import make_ptr
 
     class CuteDSLNVFP4BlackwellRunner(TunableRunner):
@@ -700,7 +698,7 @@ def forward(
                     max_active_clusters,
                     stream,
                     swap_ab,
-                    options=f"--opt-level 2 --enable-tvm-ffi"
+                    options="--opt-level 2 --enable-tvm-ffi"
                     if self.use_tvm_ffi else "--opt-level 2",
                 )
 
@@ -2413,7 +2411,7 @@ def forward(
                     c_cute_tensor,
                     max_active_clusters=max_active_clusters,
                     stream=stream,
-                    options=f"--opt-level 2 --enable-tvm-ffi"
+                    options="--opt-level 2 --enable-tvm-ffi"
                     if self.use_tvm_ffi else "--opt-level 2",
                 )
                 self.__class__.kernel_cache[cache_key] = compiled_gemm
@@ -2727,7 +2725,7 @@ def forward(
                     c_cute_tensor,
                     max_active_clusters=max_active_clusters,
                     stream=stream,
-                    options=f"--opt-level 2 --enable-tvm-ffi"
+                    options="--opt-level 2 --enable-tvm-ffi"
                     if self.use_tvm_ffi else "--opt-level 2",
                 )
                 self.__class__.kernel_cache[cache_key] = compiled_gemm
@@ -4402,7 +4400,7 @@ def forward(
                     a_stride_batch,
                     max_active_clusters=max_active_clusters,
                     stream=stream,
-                    options=f"--opt-level 2 --enable-tvm-ffi"
+                    options="--opt-level 2 --enable-tvm-ffi"
                     if self.use_tvm_ffi else "--opt-level 2",
                 )
                 self.__class__.kernel_cache[cache_key] = compiled_gemm
@@ -4681,7 +4679,7 @@ def forward(
                     c_cute_tensor,
                     max_active_clusters=max_active_clusters,
                     stream=stream,
-                    options=f"--opt-level 2 --enable-tvm-ffi"
+                    options="--opt-level 2 --enable-tvm-ffi"
                     if self.use_tvm_ffi else "--opt-level 2",
                 )
                 self.__class__.kernel_cache[cache_key] = compiled_gemm

tensorrt_llm/_torch/model_config.py

@@ -296,7 +296,7 @@ def load_modelopt_quant_config(quant_config_file, checkpoint_dir,
                     json_extended_quant_configs = json.load(fm)
             except Exception:
                 logger.info(
-                    f"No quant_cfg.json found for layer quant info, using hf_quant_config.json."
+                    "No quant_cfg.json found for layer quant info, using hf_quant_config.json."
                 )
             json_quant_configs.update(json_extended_quant_configs)
             # kv_cache_quant_algo is global regardless of MIXED_PRECISION

tensorrt_llm/_torch/models/modeling_deepseekv3.py

@@ -1817,7 +1817,7 @@ def __init__(self, model_config: ModelConfig[PretrainedConfig]):
         # at the end of __init__.
         if model_config.mapping.has_cp_helix():
             print(
-                f"[DeepseekV3ForCausalLM::__init__] Repurposing KVP ranks to TP while keeping other details the same."
+                "[DeepseekV3ForCausalLM::__init__] Repurposing KVP ranks to TP while keeping other details the same."
             )
             self.mapping_with_cp = copy.deepcopy(model_config.mapping)
             # Repurpose KVP ranks to TP while keeping other details the same.
@@ -1880,8 +1880,7 @@ def __init__(self, model_config: ModelConfig[PretrainedConfig]):
         # Undo any manipulations done to mapping.
         if self.mapping_with_cp is not None:
             print(
-                f"[DeepseekV3ForCausalLM::__init__] Restoring original mapping."
-            )
+                "[DeepseekV3ForCausalLM::__init__] Restoring original mapping.")
             model_config._frozen = False
             model_config.mapping = self.mapping_with_cp
             model_config._frozen = True