[Frontend] Reduce chat template warmup logging levels (#37062)

Signed-off-by: Nick Hill <nickhill123@gmail.com>

vllm/renderers/base.py

@@ -179,17 +179,17 @@ class BaseRenderer(ABC, Generic[_T]):
         from vllm.entrypoints.chat_utils import ChatTemplateResolutionError
 
         try:
-            logger.info("Warming up chat template processing...")
+            logger.debug("Warming up chat template processing...")
             start_time = time.perf_counter()
 
             self.render_chat([[{"role": "user", "content": "warmup"}]], chat_params)
 
             elapsed = time.perf_counter() - start_time
-            logger.info("Chat template warmup completed in %.3fs", elapsed)
+            logger.debug("Chat template warmup completed in %.3fs", elapsed)
         except ChatTemplateResolutionError:
-            logger.info("This model does not support chat template.")
+            logger.debug("This model does not support chat template.")
         except Exception:
-            logger.exception("Chat template warmup failed")
+            logger.warning("Chat template warmup failed", exc_info=True)
 
         if self.mm_processor:
             from vllm.multimodal.processing import TimingContext
@@ -200,7 +200,7 @@ class BaseRenderer(ABC, Generic[_T]):
             mm_limits = processor.info.allowed_mm_limits
 
             try:
-                logger.info("Warming up multi-modal processing...")
+                logger.debug("Warming up multi-modal processing...")
                 start_time = time.perf_counter()
 
                 processor_inputs = processor.dummy_inputs.get_dummy_processor_inputs(
@@ -209,14 +209,13 @@ class BaseRenderer(ABC, Generic[_T]):
                     mm_options=mm_config.limit_per_prompt,
                 )
                 _ = processor.apply(
-                    processor_inputs,
-                    timing_ctx=TimingContext(enabled=False),
+                    processor_inputs, timing_ctx=TimingContext(enabled=False)
                 )
 
                 elapsed = time.perf_counter() - start_time
                 logger.info("Multi-modal warmup completed in %.3fs", elapsed)
             except Exception:
-                logger.exception("Multi-modal warmup failed")
+                logger.warning("Multi-modal warmup failed")
             finally:
                 self.clear_mm_cache()