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, 22:35 UTC
[ez] Remove a trailing space from compilation/decorators.py (#22028)
Zhengxu Chen committed
Jul 31, 2025 at 12:46 UTC
7349d5268bf70b7a530c1e649884e4f926615f8e
1 file changed
+1
-1
vllm/compilation/decorators.py
+1
-1
@@ -108,7 +108,7 @@ def support_torch_compile(
108
During runtime, when we actually mark dimensions of tensors,
109
it depends on the value of arguments:
110
111
- - if it is a single integer (can be negative), the corresponding dimension
111
+ - if it is a single integer (can be negative), the corresponding dimension
112
of the argument will be marked as dynamic.
113
- if it is `None`, ignored.
114
- if it is `IntermediateTensors`, all the tensors in the intermediate