[Frontend] Add vllm bench sweep to CLI (#27639)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

vllm/entrypoints/cli/__init__.py

@@ -2,10 +2,12 @@
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 from vllm.entrypoints.cli.benchmark.latency import BenchmarkLatencySubcommand
 from vllm.entrypoints.cli.benchmark.serve import BenchmarkServingSubcommand
+from vllm.entrypoints.cli.benchmark.sweep import BenchmarkSweepSubcommand
 from vllm.entrypoints.cli.benchmark.throughput import BenchmarkThroughputSubcommand
 
 __all__: list[str] = [
     "BenchmarkLatencySubcommand",
     "BenchmarkServingSubcommand",
+    "BenchmarkSweepSubcommand",
     "BenchmarkThroughputSubcommand",
 ]

vllm/entrypoints/cli/benchmark/base.py

@@ -6,7 +6,7 @@ from vllm.entrypoints.cli.types import CLISubcommand
 
 
 class BenchmarkSubcommandBase(CLISubcommand):
-    """The base class of subcommands for vllm bench."""
+    """The base class of subcommands for `vllm bench`."""
 
     help: str
 

vllm/entrypoints/cli/benchmark/latency.py

@@ -7,7 +7,7 @@ from vllm.entrypoints.cli.benchmark.base import BenchmarkSubcommandBase
 
 
 class BenchmarkLatencySubcommand(BenchmarkSubcommandBase):
-    """The `latency` subcommand for vllm bench."""
+    """The `latency` subcommand for `vllm bench`."""
 
     name = "latency"
     help = "Benchmark the latency of a single batch of requests."

vllm/entrypoints/cli/benchmark/serve.py

@@ -7,7 +7,7 @@ from vllm.entrypoints.cli.benchmark.base import BenchmarkSubcommandBase
 
 
 class BenchmarkServingSubcommand(BenchmarkSubcommandBase):
-    """The `serve` subcommand for vllm bench."""
+    """The `serve` subcommand for `vllm bench`."""
 
     name = "serve"
     help = "Benchmark the online serving throughput."

vllm/entrypoints/cli/benchmark/sweep.py

@@ -0,0 +1,21 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+import argparse
+
+from vllm.benchmarks.sweep.cli import add_cli_args, main
+from vllm.entrypoints.cli.benchmark.base import BenchmarkSubcommandBase
+
+
+class BenchmarkSweepSubcommand(BenchmarkSubcommandBase):
+    """The `sweep` subcommand for `vllm bench`."""
+
+    name = "sweep"
+    help = "Benchmark for a parameter sweep."
+
+    @classmethod
+    def add_cli_args(cls, parser: argparse.ArgumentParser) -> None:
+        add_cli_args(parser)
+
+    @staticmethod
+    def cmd(args: argparse.Namespace) -> None:
+        main(args)

vllm/entrypoints/cli/benchmark/throughput.py

@@ -7,7 +7,7 @@ from vllm.entrypoints.cli.benchmark.base import BenchmarkSubcommandBase
 
 
 class BenchmarkThroughputSubcommand(BenchmarkSubcommandBase):
-    """The `throughput` subcommand for vllm bench."""
+    """The `throughput` subcommand for `vllm bench`."""
 
     name = "throughput"
     help = "Benchmark offline inference throughput."