[BugFix][V1] Fix overhead related to bad_words sampling when not in use (vllm-project#14894)

Signed-off-by: Nick Hill <[email protected]>
Signed-off-by: Mu Huai <[email protected]>

Author: njhill

tests/v1/worker/test_gpu_input_batch.py

@@ -124,8 +124,9 @@ def _construct_expected_sampling_metadata(
         if req.sampling_params.allowed_token_ids:
             allowed_token_ids_mask[index_in_input_batch][
                 req.sampling_params.allowed_token_ids] = True
-        bad_words_token_ids[
-            index_in_input_batch] = req.sampling_params.bad_words_token_ids
+        if req.sampling_params.bad_words_token_ids:
+            bad_words_token_ids[
+                index_in_input_batch] = req.sampling_params.bad_words_token_ids
 
     return SamplingMetadata(
         temperature=torch.tensor(temperature, dtype=torch.float,

vllm/sampling_params.py

@@ -235,7 +235,7 @@ class SamplingParams(
 
     # Fields used for bad words
     bad_words: Optional[list[str]] = None
-    _bad_words_token_ids: list[list[int]] = msgspec.field(default_factory=list)
+    _bad_words_token_ids: Optional[list[list[int]]] = None
 
     @staticmethod
     def from_optional(
@@ -464,8 +464,9 @@ def update_from_generation_config(
                     self.stop_token_ids = list(eos_ids)
 
     def update_from_tokenizer(self, tokenizer: AnyTokenizer) -> None:
-        if self.bad_words is None:
+        if not self.bad_words:
             return
+        self._bad_words_token_ids = []
         for bad_word in self.bad_words:
             # To prohibit words both at the beginning
             # and in the middle of text
@@ -516,7 +517,7 @@ def all_stop_token_ids(self) -> set[int]:
         return self._all_stop_token_ids
 
     @property
-    def bad_words_token_ids(self) -> list[list[int]]:
+    def bad_words_token_ids(self) -> Optional[list[list[int]]]:
         # For internal use only. Backward compatibility not guaranteed
         return self._bad_words_token_ids
 

vllm/v1/worker/gpu_input_batch.py

@@ -324,8 +324,9 @@ def add_request(
             self.allowed_token_ids_mask_cpu_tensor[req_index][
                 sampling_params.allowed_token_ids] = False
 
-        self.bad_words_token_ids[
-            req_index] = sampling_params.bad_words_token_ids
+        if sampling_params.bad_words_token_ids:
+            self.bad_words_token_ids[
+                req_index] = sampling_params.bad_words_token_ids
 
         # Add request lora ID
         if request.lora_request: