[Mini-VLLM] Cut non-OpenAI surfaces, dead kernels, Mistral stack
- Remove Anthropic, Cohere, SageMaker, decisions, structured-decisions, rl, render-launcher and batch entrypoints; keep scale-out token routes - Drop the Swagger static bundle and the /docs, /redoc routes - Delete device-keyed tuning configs for fused_moe, quant utils, mamba select-state and helion; no sm86 entry ever existed for any of them - Remove ten unused rope variants plus the fbgemm_fp8 and fp_quant quantizers and their config, metrics and loader plumbing - Cut the CPU device build: csrc/cpu, cpu_extension.cmake, cpu requirements and the setup.py cpu target path - Delete 29 environment variables nothing reads and 14 chat-template fallbacks for models this fork no longer registers - Remove nine dead CUDA ops (deepseek_v4, kimi_k3, minimax, muse_spark, dsv3 gemm) with their kernels, bindings, wrappers and CMake entries - Delete the connector test suites that targeted removed connectors - Cut the Mistral tokenizer and parser stack, its tests, and the mistral_common and ijson dependency pins - Restore the Qwen4Exp registry rows and nvidia/mtp.py that an earlier cut deleted while keeping the model tree Co-authored-by: Hinata <hinata@idihore.id> Signed-off-by: MOVZX <movzx@yahoo.com>
This commit is contained in:
@@ -2,11 +2,9 @@
|
||||
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
||||
|
||||
import json
|
||||
from types import SimpleNamespace
|
||||
|
||||
import pytest
|
||||
|
||||
import vllm.entrypoints.launchers.cli_args as cli_args_module
|
||||
from tests.utils import VLLM_PATH
|
||||
from vllm.engine.arg_utils import AsyncEngineArgs
|
||||
from vllm.entrypoints.launchers.cli_args import (
|
||||
@@ -180,73 +178,6 @@ def test_multiple_valid_inputs(serve_parser):
|
||||
|
||||
|
||||
### Tests for serve argument validation that run prior to loading
|
||||
@pytest.mark.parametrize(
|
||||
"port, extra_args",
|
||||
[
|
||||
(50051, []),
|
||||
(0, ["--port", "0"]),
|
||||
(50051, ["--port", "50051", "--uds", "/tmp/vllm.sock"]),
|
||||
],
|
||||
)
|
||||
def test_rust_grpc_port_accepts_frontend_launch(
|
||||
vllm_parser, monkeypatch, port, extra_args
|
||||
):
|
||||
monkeypatch.setenv("VLLM_USE_RUST_FRONTEND", "1")
|
||||
args = vllm_parser.parse_args(
|
||||
[
|
||||
"serve",
|
||||
"--grpc-port",
|
||||
str(port),
|
||||
"--data-parallel-size",
|
||||
"8",
|
||||
"--data-parallel-size-local",
|
||||
"4",
|
||||
"--data-parallel-start-rank",
|
||||
"4",
|
||||
"--data-parallel-hybrid-lb",
|
||||
*extra_args,
|
||||
]
|
||||
)
|
||||
validate_parsed_serve_args(args)
|
||||
assert args.grpc_port == port
|
||||
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"rust_enabled, extra_args, port",
|
||||
[
|
||||
(False, [], 50051),
|
||||
(False, ["--grpc"], 50051),
|
||||
(True, ["--grpc"], 50051),
|
||||
(True, ["--headless"], 50051),
|
||||
(True, ["--api-server-count", "0"], 50051),
|
||||
(True, ["--api-server-count", "-1"], 50051),
|
||||
(True, ["--data-parallel-multi-port-external-lb"], 50051),
|
||||
(True, ["--port", "50051"], 50051),
|
||||
(True, [], -1),
|
||||
(True, [], 65536),
|
||||
],
|
||||
)
|
||||
def test_rust_grpc_port_rejects_incompatible_launch(
|
||||
vllm_parser, monkeypatch, rust_enabled, extra_args, port
|
||||
):
|
||||
monkeypatch.setenv("VLLM_USE_RUST_FRONTEND", "1" if rust_enabled else "0")
|
||||
args = vllm_parser.parse_args(["serve", "--grpc-port", str(port), *extra_args])
|
||||
error = (
|
||||
"--grpc and --grpc-port are mutually exclusive"
|
||||
if "--grpc" in extra_args
|
||||
else "--grpc-port"
|
||||
)
|
||||
with pytest.raises(ValueError, match=error):
|
||||
validate_parsed_serve_args(args)
|
||||
|
||||
|
||||
def test_rust_grpc_port_requires_serve_subcommand(monkeypatch):
|
||||
"""Direct Python render/API entrypoints must reject a Rust-only listener."""
|
||||
monkeypatch.setenv("VLLM_USE_RUST_FRONTEND", "1")
|
||||
parser = make_arg_parser(FlexibleArgumentParser())
|
||||
args = parser.parse_args(["--grpc-port", "50051"])
|
||||
with pytest.raises(ValueError, match="--grpc-port requires"):
|
||||
validate_parsed_serve_args(args)
|
||||
|
||||
|
||||
def test_enable_auto_choice_fails_without_tool_call_parser(serve_parser):
|
||||
@@ -263,7 +194,7 @@ def test_enable_auto_choice_passes_with_tool_call_parser(serve_parser):
|
||||
args=[
|
||||
"--enable-auto-tool-choice",
|
||||
"--tool-call-parser",
|
||||
"mistral",
|
||||
"qwen3_coder",
|
||||
]
|
||||
)
|
||||
validate_parsed_serve_args(args)
|
||||
@@ -280,7 +211,7 @@ def test_auto_tool_choice_with_reasoning_parser_passes(serve_parser):
|
||||
args=[
|
||||
"--enable-auto-tool-choice",
|
||||
"--tool-call-parser",
|
||||
"mistral",
|
||||
"qwen3_coder",
|
||||
"--reasoning-parser",
|
||||
"deepseek_r1",
|
||||
]
|
||||
@@ -333,48 +264,6 @@ def test_per_request_metrics_requires_log_stats(serve_parser):
|
||||
validate_parsed_serve_args(args)
|
||||
|
||||
|
||||
def _build_launch_render_parser():
|
||||
"""Mirror `vllm launch render`.
|
||||
|
||||
`vllm/entrypoints/cli/main.py` parses subcommands with ``dest="subparser"``,
|
||||
and `LaunchSubcommandBase.add_cli_args` builds the component parser with
|
||||
`make_arg_parser`, so a launch component carries serve args under
|
||||
``args.subparser == "launch"``.
|
||||
"""
|
||||
vllm_parser = FlexibleArgumentParser()
|
||||
subparsers = vllm_parser.add_subparsers(required=False, dest="subparser")
|
||||
launch_parser = subparsers.add_parser("launch")
|
||||
launch_subparsers = launch_parser.add_subparsers(
|
||||
required=True, dest="launch_component"
|
||||
)
|
||||
render_parser = launch_subparsers.add_parser("render")
|
||||
make_arg_parser(render_parser)
|
||||
return vllm_parser
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
def launch_render_parser():
|
||||
return _build_launch_render_parser()
|
||||
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"cli_args, error",
|
||||
[
|
||||
(["--enable-auto-tool-choice"], TypeError),
|
||||
(["--grpc-port", "50051"], ValueError),
|
||||
],
|
||||
)
|
||||
def test_launch_render_validates_serve_args(
|
||||
launch_render_parser, monkeypatch, cli_args, error
|
||||
):
|
||||
"""`vllm launch render` reuses the serve parser, so it gets the serve checks"""
|
||||
monkeypatch.setenv("VLLM_USE_RUST_FRONTEND", "1")
|
||||
args = launch_render_parser.parse_args(args=["launch", "render", *cli_args])
|
||||
assert args.subparser == "launch"
|
||||
with pytest.raises(error):
|
||||
validate_parsed_serve_args(args)
|
||||
|
||||
|
||||
def test_subcommand_without_serve_args_skips_validation():
|
||||
"""A subcommand that does not use the serve parser is still skipped"""
|
||||
vllm_parser = FlexibleArgumentParser()
|
||||
@@ -456,26 +345,6 @@ def test_default_chat_template_kwargs_default_none(serve_parser):
|
||||
assert args.default_chat_template_kwargs is None
|
||||
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"default_kwargs,cohere_format,expected",
|
||||
[
|
||||
(None, "cmd3", {"cohere_format": "cmd3"}),
|
||||
(
|
||||
{"cohere_format": "cmd4", "custom": True},
|
||||
"cmd3",
|
||||
{"cohere_format": "cmd4", "custom": True},
|
||||
),
|
||||
],
|
||||
)
|
||||
def test_resolve_default_chat_template_kwargs(default_kwargs, cohere_format, expected):
|
||||
args = SimpleNamespace(
|
||||
default_chat_template_kwargs=default_kwargs,
|
||||
cohere_format=cohere_format,
|
||||
)
|
||||
|
||||
assert cli_args_module.resolve_default_chat_template_kwargs(args) == expected
|
||||
|
||||
|
||||
def test_default_chat_template_kwargs_invalid_json(serve_parser):
|
||||
"""Ensure invalid JSON raises an error."""
|
||||
with pytest.raises(SystemExit):
|
||||
|
||||
Reference in New Issue
Block a user