A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 22:15 UTC
[Misc][Chore] Clean Up `AsyncOutputProcessing` Logs (#13780)
Robert Shaw committed
Feb 24, 2025 at 19:39 UTC
f61528d46d4f2e2b180a861ece4ca7311606882f
1 file changed
-11
vllm/config.py
-11
@@ -710,8 +710,6 @@ class ModelConfig:
710
return
711
712
if parallel_config.pipeline_parallel_size > 1:
713
- logger.warning("Async output processing can not be enabled "
714
- "with pipeline parallel")
713
self.use_async_output_proc = False
714
return
715
@@ -719,15 +717,10 @@ class ModelConfig:
717
# If the feature combo become valid
718
from vllm.platforms import current_platform
719
if not current_platform.is_async_output_supported(self.enforce_eager):
722
- logger.warning(
723
- "Async output processing is not supported on the "
724
- "current platform type %s.", current_platform.device_type)
720
self.use_async_output_proc = False
721
return
722
723
if envs.VLLM_USE_RAY_SPMD_WORKER:
729
- logger.warning(
730
- "Async output processing can not be enabled with ray spmd")
724
self.use_async_output_proc = False
725
return
726
@@ -739,8 +732,6 @@ class ModelConfig:
732
# Reminder: Please update docs/source/features/compatibility_matrix.md
733
# If the feature combo become valid
734
if speculative_config:
742
- logger.warning("Async output processing is not supported with"
743
- " speculative decoding currently.")
735
self.use_async_output_proc = False
736
737
def verify_with_parallel_config(
@@ -768,8 +759,6 @@ class ModelConfig:
759
"Supported models implement the `SupportsPP` interface.")
760
761
if self.use_async_output_proc:
771
- logger.warning("Async output processor is not supported with "
772
- "pipeline parallelism currently. Disabling it.")
762
self.use_async_output_proc = False
763
764
def get_hf_config_sliding_window(