[Frontend] improve vllm serve --help display (#18643)

Signed-off-by: reidliu41 <reid201711@gmail.com>
Co-authored-by: reidliu41 <reid201711@gmail.com>

vllm/entrypoints/cli/main.py

@@ -9,7 +9,7 @@ import vllm.entrypoints.cli.collect_env
 import vllm.entrypoints.cli.openai
 import vllm.entrypoints.cli.serve
 import vllm.version
-from vllm.entrypoints.utils import cli_env_setup
+from vllm.entrypoints.utils import VLLM_SERVE_PARSER_EPILOG, cli_env_setup
 from vllm.utils import FlexibleArgumentParser
 
 CMD_MODULES = [
@@ -32,7 +32,10 @@ def register_signal_handlers():
 def main():
     cli_env_setup()
 
-    parser = FlexibleArgumentParser(description="vLLM CLI")
+    parser = FlexibleArgumentParser(
+        description="vLLM CLI",
+        epilog=VLLM_SERVE_PARSER_EPILOG,
+    )
     parser.add_argument('-v',
                         '--version',
                         action='version',

vllm/entrypoints/cli/serve.py

@@ -11,6 +11,8 @@ from vllm.entrypoints.cli.types import CLISubcommand
 from vllm.entrypoints.openai.api_server import run_server
 from vllm.entrypoints.openai.cli_args import (make_arg_parser,
                                               validate_parsed_serve_args)
+from vllm.entrypoints.utils import (VLLM_SERVE_PARSER_EPILOG,
+                                    show_filtered_argument_or_group_from_help)
 from vllm.logger import init_logger
 from vllm.usage.usage_lib import UsageContext
 from vllm.utils import FlexibleArgumentParser, get_tcp_uri
@@ -77,7 +79,10 @@ class ServeSubcommand(CLISubcommand):
             "https://docs.vllm.ai/en/latest/serving/openai_compatible_server.html#cli-reference"
         )
 
-        return make_arg_parser(serve_parser)
+        serve_parser = make_arg_parser(serve_parser)
+        show_filtered_argument_or_group_from_help(serve_parser)
+        serve_parser.epilog = VLLM_SERVE_PARSER_EPILOG
+        return serve_parser
 
 
 def cmd_init() -> list[CLISubcommand]:

vllm/entrypoints/utils.py

@@ -13,6 +13,13 @@ from vllm.logger import init_logger
 
 logger = init_logger(__name__)
 
+VLLM_SERVE_PARSER_EPILOG = (
+    "Tip: Use `vllm serve --help=<keyword>` to explore arguments from help.\n"
+    "   - To view a argument group:     --help=ModelConfig\n"
+    "   - To view a single argument:    --help=max-num-seqs\n"
+    "   - To search by keyword:         --help=max\n"
+    "   - To list all groups:           --help=listgroup")
+
 
 async def listen_for_disconnect(request: Request) -> None:
     """Returns if a disconnect message is received"""
@@ -158,3 +165,55 @@ def _validate_truncation_size(
             tokenization_kwargs["max_length"] = truncate_prompt_tokens
 
     return truncate_prompt_tokens
+
+
+def show_filtered_argument_or_group_from_help(parser):
+    import sys
+    for arg in sys.argv:
+        if arg.startswith('--help='):
+            search_keyword = arg.split('=', 1)[1]
+
+            # List available groups
+            if search_keyword == 'listgroup':
+                print("\nAvailable argument groups:")
+                for group in parser._action_groups:
+                    if group.title and not group.title.startswith(
+                            "positional arguments"):
+                        print(f"  - {group.title}")
+                        if group.description:
+                            print("    " + group.description.strip())
+                        print()
+                sys.exit(0)
+
+            # For group search
+            formatter = parser._get_formatter()
+            for group in parser._action_groups:
+                if group.title and group.title.lower() == search_keyword.lower(
+                ):
+                    formatter.start_section(group.title)
+                    formatter.add_text(group.description)
+                    formatter.add_arguments(group._group_actions)
+                    formatter.end_section()
+                    print(formatter.format_help())
+                    sys.exit(0)
+
+            # For single arg
+            matched_actions = []
+
+            for group in parser._action_groups:
+                for action in group._group_actions:
+                    # search option name
+                    if any(search_keyword.lower() in opt.lower()
+                           for opt in action.option_strings):
+                        matched_actions.append(action)
+
+            if matched_actions:
+                print(f"\nParameters matching '{search_keyword}':\n")
+                formatter = parser._get_formatter()
+                formatter.add_arguments(matched_actions)
+                print(formatter.format_help())
+                sys.exit(0)
+
+            print(f"\nNo group or parameter matching '{search_keyword}'")
+            print("Tip: use `--help=listgroup` to view all groups.")
+            sys.exit(1)

vllm/utils.py

@@ -33,7 +33,8 @@ import uuid
 import warnings
 import weakref
 from argparse import (Action, ArgumentDefaultsHelpFormatter, ArgumentParser,
-                      ArgumentTypeError, _ArgumentGroup)
+                      ArgumentTypeError, RawDescriptionHelpFormatter,
+                      _ArgumentGroup)
 from asyncio import FIRST_COMPLETED, AbstractEventLoop, Task
 from collections import UserDict, defaultdict
 from collections.abc import (AsyncGenerator, Awaitable, Generator, Hashable,
@@ -1323,7 +1324,8 @@ class StoreBoolean(Action):
                              "Expected 'true' or 'false'.")
 
 
-class SortedHelpFormatter(ArgumentDefaultsHelpFormatter):
+class SortedHelpFormatter(ArgumentDefaultsHelpFormatter,
+                          RawDescriptionHelpFormatter):
     """SortedHelpFormatter that sorts arguments by their option strings."""
 
     def _split_lines(self, text, width):