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 10, 22:45 UTC
[cli] complete vllm cli help message (#31226)
Signed-off-by: Andy Xie <andy.xning@gmail.com>
Ning Xie committed
Dec 24, 2025 at 23:45 UTC
5d9308968649c81ee5903fc2a77377d738ed2f6d
2 files changed
+6
-1
vllm/entrypoints/cli/benchmark/main.py
+1
@@ -32,6 +32,7 @@ class BenchmarkSubcommand(CLISubcommand):
32
) -> FlexibleArgumentParser:
33
bench_parser = subparsers.add_parser(
34
self.name,
35
+ help=self.help,
36
description=self.help,
37
usage=f"vllm {self.name} <bench_type> [options]",
38
)
vllm/entrypoints/cli/serve.py
+5
-1
@@ -66,7 +66,11 @@ class ServeSubcommand(CLISubcommand):
66
self, subparsers: argparse._SubParsersAction
67
) -> FlexibleArgumentParser:
68
serve_parser = subparsers.add_parser(
69
- self.name, description=DESCRIPTION, usage="vllm serve [model_tag] [options]"
69
+ self.name,
70
+ help="Launch a local OpenAI-compatible API server to serve LLM "
71
+ "completions via HTTP.",
72
+ description=DESCRIPTION,
73
+ usage="vllm serve [model_tag] [options]",
74
)
75
76
serve_parser = make_arg_parser(serve_parser)