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 14, 05:30 UTC
[CI] Fix linting error on main (#23835)
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
Thomas Parnell committed
Aug 28, 2025 at 15:52 UTC
1f096f9b9536aa8f520f89b178a518da294a7dce
1 file changed
+2
-1
vllm/v1/cudagraph_dispatcher.py
+2
-1
@@ -11,7 +11,8 @@ logger = init_logger(__name__)
11
12
class CudagraphDispatcher:
13
"""
14
- Runtime cudagraph dispatcher to dispatch keys for multiple set of cudagraphs.
14
+ Runtime cudagraph dispatcher to dispatch keys for multiple set of
15
+ cudagraphs.
16
17
The dispatcher stores two sets of dispatch keys, one for PIECEWISE and one
18
for FULL cudagraph runtime mode. The keys are initialized depending on