fix: fix rebase format issue

Signed-off-by: Qixiang Lin <[email protected]>

Author: qixiang-99

tensorrt_llm/_torch/attention_backend/trtllm.py

@@ -706,7 +706,8 @@ def forward(
         use_paged_context_fmha = (
             metadata.runtime_features.chunked_prefill
             or metadata.runtime_features.cache_reuse
-            or metadata.runtime_features.has_speculative_draft_tokens) if metadata.runtime_features else False
+            or metadata.runtime_features.has_speculative_draft_tokens
+        ) if metadata.runtime_features else False
 
         if use_paged_context_fmha and self.has_fp8_kv_cache:
             # NOTE: W4A8_AWQ can be included too, exclude for now since

tensorrt_llm/_torch/pyexecutor/model_engine.py

@@ -556,7 +556,8 @@ def no_cuda_graph():
                              resource_manager=resource_manager)
                 torch.cuda.synchronize()
 
-    def _set_up_attn_metadata(self, kv_cache_manager: KVCacheManager,
+    def _set_up_attn_metadata(self,
+                              kv_cache_manager: KVCacheManager,
                               is_dummy_forward: bool = False):
         # is_dummy_forward is used to indicate whether the forward is
         # a dummy forward for memory estimation OR