[Misc] Clean logs for image gen task (#3414)

Signed-off-by: wuhang <wuhang6@huawei.com>

vllm_omni/diffusion/diffusion_engine.py

@@ -147,7 +147,7 @@ class DiffusionEngine:
             preprocess_start_time = time.perf_counter()
             request = self.pre_process_func(request)
             preprocess_time = time.perf_counter() - preprocess_start_time
-            logger.info(f"Pre-processing completed in {preprocess_time:.4f} seconds")
+            logger.debug("Pre-processing completed in %.4f seconds", preprocess_time)
 
         exec_start_time = time.perf_counter()
         output = await self.async_add_req_and_wait_for_response(request)
@@ -157,7 +157,7 @@ class DiffusionEngine:
             raise DiffusionRequestAbortedError(output.abort_message or "Diffusion request aborted.")
         if output.error:
             raise RuntimeError(output.error)
-        logger.info("Generation completed successfully.")
+        logger.debug("Generation completed successfully.")
 
         if output.output is None:
             logger.warning("Output is None, returning empty OmniRequestOutput")
@@ -204,10 +204,10 @@ class DiffusionEngine:
             model_fps = outputs.get("fps")
             outputs = outputs.get("video", outputs)
         postprocess_time = time.perf_counter() - postprocess_start_time
-        logger.info(f"Post-processing completed in {postprocess_time:.4f} seconds")
+        logger.debug("Post-processing completed in %.4f seconds", postprocess_time)
 
         step_total_ms = (time.perf_counter() - diffusion_engine_start_time) * 1000
-        logger.info(
+        logger.debug(
             "DiffusionEngine.step breakdown: preprocess=%.2f ms, "
             "add_req_and_wait=%.2f ms, postprocess=%.2f ms, total=%.2f ms",
             preprocess_time * 1000,

vllm_omni/diffusion/inline_stage_diffusion_client.py

@@ -86,7 +86,7 @@ class InlineStageDiffusionClient:
         sampling_params: OmniDiffusionSamplingParams,
         kv_sender_info: dict[int, dict[str, Any]] | None = None,
     ) -> None:
-        logger.info(
+        logger.debug(
             "[InlineStageDiffusionClient] stage-%s [rep-%s] add request: %s",
             self.stage_id,
             self.replica_id,
@@ -144,7 +144,7 @@ class InlineStageDiffusionClient:
         sampling_params: OmniDiffusionSamplingParams,
         kv_sender_info: dict[int, dict[str, Any]] | None = None,
     ) -> None:
-        logger.info(
+        logger.debug(
             "[InlineStageDiffusionClient] stage-%s [rep-%s] add batch request: %s (%d prompts)",
             self.stage_id,
             self.replica_id,

vllm_omni/diffusion/models/z_image/pipeline_z_image.py

@@ -536,8 +536,10 @@ class ZImagePipeline(nn.Module, DiffusionPipelineProfilerMixin):
                         image = PIL.Image.open(raw_image) if isinstance(raw_image, str) else raw_image
 
         # strength is currently only applicable for Z-Image I2I; other pipelines ignore this parameter
-        strength = req.sampling_params.strength if req.sampling_params.strength is not None else strength
-        if strength is not None and image is None:
+        explicit_strength = req.sampling_params.strength is not None
+        if explicit_strength:
+            strength = req.sampling_params.strength
+        if explicit_strength and image is None:
             logger.warning(
                 "strength parameter (%.2f) is only applicable for image-to-image (I2I) generation. "
                 "It will be ignored for text-to-image (T2I) generation.",

vllm_omni/diffusion/worker/diffusion_model_runner.py

@@ -212,7 +212,7 @@ class DiffusionModelRunner(OmniConnectorModelRunnerMixin):
         peak_allocated_gb = peak_allocated_bytes / (1024**3)
         pool_overhead_gb = peak_reserved_gb - peak_allocated_gb
 
-        logger.info(
+        logger.debug(
             "Peak GPU memory (this request): %.2f GB reserved, %.2f GB allocated, %.2f GB pool overhead (%.1f%%)",
             peak_reserved_gb,
             peak_allocated_gb,