[Frontend] Add logits_processors as an extra completion argument (#11150)

Signed-off-by: Brad Hilton <brad.hilton.nw@gmail.com>
2024-12-14 09:46:42 -07:00
parent 3cb5769883
commit 9c3dadd1c9
6 changed files with 127 additions and 39 deletions
--- a/vllm/entrypoints/openai/serving_completion.py
+++ b/vllm/entrypoints/openai/serving_completion.py
@@ -123,7 +123,8 @@ class OpenAIServingCompletion(OpenAIServing):
                        default_max_tokens)
                else:
                    sampling_params = request.to_sampling_params(
-                        default_max_tokens)
+                        default_max_tokens,
+                        self.model_config.logits_processor_pattern)

                request_id_item = f"{request_id}-{i}"