[Refactor][Frontend] Keep all logic about reasoning into one class (#14428)

Signed-off-by: Ce Gao <cegao@tensorchord.ai>
2025-03-28 15:23:30 +08:00
parent 2d9045fce8
commit 32b14baf8a
18 changed files with 171 additions and 200 deletions
--- a/vllm/engine/llm_engine.py
+++ b/vllm/engine/llm_engine.py
@@ -2080,8 +2080,9 @@ class LLMEngine:
            guided_decoding.backend = guided_decoding.backend or \
                self.decoding_config.guided_decoding_backend

-            logger.debug("Reasoning backend: %s",
-                         self.decoding_config.reasoning_backend)
+            if self.decoding_config.reasoning_backend is not None:
+                logger.debug("Building with reasoning backend %s",
+                             self.decoding_config.reasoning_backend)

            processor = get_local_guided_decoding_logits_processor(
                guided_params=guided_decoding,