[Mini-VLLM] Remove LoRA, plugins, speech-to-text and scale-out surfaces

Production runs one static model behind the OpenAI-compatible API. Every
adapter path and non-OpenAI HTTP surface is unreachable there.

- Delete the LoRA stack: vllm/lora, LoRARequest on the engine wire, the
  scheduler, worker, cudagraph and MoE plumbing, LoRA metrics, the lora
  CLI group, and the LoRA tests.
- Drop the plugin loader in vllm/plugins. Platform resolution is now
  CUDA-only and the general/stat-logger plugin hooks are gone.
- Delete the speech-to-text and realtime entrypoints, the scale-out
  render routes, and vllm/config/speech_to_text.py. GenerationTask is
  generate-only.
- Cut DecodeBenchConnector and the disagg-prefill doc and image. The
  Multi, Offloading and Example connectors stay; production uses the
  CPU KV offload tier.
- Delete the ROCm-only fused KDA chunk kernel (2346 lines) with its CMake
  arch gate, declarations and op registrations. The CUDA build never
  defined the guard and no Python caller existed.
- Delete the orphan Mistral and Pixtral chat fixtures and examples/.
- Fix _check_model: removing the LoRA early return made every request
  fail with "The model ... does not exist. Valid aliases: ...".

Verified on an RTX 3090: import sweep reports 1242 modules with 7 known
optional-dependency failures; a live server answers chat, completions,
responses, tokenize and detokenize with zero log errors and 404 for the
removed surfaces; collection yields 8558 tests with the 3 pre-existing
errors; ruff adds no new finding; the incremental _C build succeeds.

Co-authored-by: Hinata <hinata@idihore.id>
Signed-off-by: MOVZX <movzx@yahoo.com>
This commit is contained in:
2026-10-19 21:06:19 +07:00
co-authored by Hinata
parent 5f3c7bd8ab
commit 5aa2ac64f8
269 changed files with 417 additions and 29932 deletions
@@ -27,15 +27,7 @@ MODEL_NAME = "HuggingFaceH4/zephyr-7b-beta"
@pytest.fixture(scope="module")
def zephyr_lora_files():
"""Download zephyr LoRA files once per test session."""
from vllm.transformers_utils.repo_utils import hf_api
return hf_api().snapshot_download(repo_id="typeof/zephyr-7b-beta-lora")
@pytest.fixture(scope="module")
def server(zephyr_lora_files):
def server():
args = [
# use half precision for speed and memory savings in CI environment
"--dtype",
@@ -43,14 +35,6 @@ def server(zephyr_lora_files):
"--max-model-len",
"8192",
"--enforce-eager",
# lora config below
"--enable-lora",
"--lora-modules",
f"zephyr-lora={zephyr_lora_files}",
"--max-lora-rank",
"64",
"--max-cpu-loras",
"2",
"--max-num-seqs",
"128",
]
@@ -67,9 +51,8 @@ async def client(server):
@pytest.mark.asyncio
@pytest.mark.parametrize(
# first test base model, then test loras
"model_name",
[MODEL_NAME, "zephyr-lora"],
[MODEL_NAME],
)
async def test_no_logprobs_chat(client: openai.AsyncOpenAI, model_name: str):
messages = [
@@ -91,9 +74,8 @@ async def test_no_logprobs_chat(client: openai.AsyncOpenAI, model_name: str):
@pytest.mark.asyncio
@pytest.mark.parametrize(
# just test 1 lora hereafter
"model_name",
[MODEL_NAME, "zephyr-lora"],
[MODEL_NAME],
)
async def test_zero_logprobs_chat(client: openai.AsyncOpenAI, model_name: str):
messages = [
@@ -119,7 +101,7 @@ async def test_zero_logprobs_chat(client: openai.AsyncOpenAI, model_name: str):
@pytest.mark.asyncio
@pytest.mark.parametrize(
"model_name",
[MODEL_NAME, "zephyr-lora"],
[MODEL_NAME],
)
async def test_some_logprobs_chat(client: openai.AsyncOpenAI, model_name: str):
messages = [
@@ -145,7 +127,7 @@ async def test_some_logprobs_chat(client: openai.AsyncOpenAI, model_name: str):
@pytest.mark.asyncio
@pytest.mark.parametrize(
"model_name",
[MODEL_NAME, "zephyr-lora"],
[MODEL_NAME],
)
async def test_too_many_chat_logprobs(client: openai.AsyncOpenAI, model_name: str):
messages = [
@@ -254,7 +236,7 @@ async def test_more_than_one_prompt_logprobs_chat(
@pytest.mark.asyncio
@pytest.mark.parametrize(
"model_name",
[MODEL_NAME, "zephyr-lora"],
[MODEL_NAME],
)
async def test_single_chat_session(client: openai.AsyncOpenAI, model_name: str):
messages = [
@@ -297,9 +279,8 @@ async def test_single_chat_session(client: openai.AsyncOpenAI, model_name: str):
@pytest.mark.asyncio
@pytest.mark.parametrize(
# just test 1 lora hereafter
"model_name",
[MODEL_NAME, "zephyr-lora"],
[MODEL_NAME],
)
async def test_chat_streaming(client: openai.AsyncOpenAI, model_name: str):
messages = [
@@ -345,7 +326,7 @@ async def test_chat_streaming(client: openai.AsyncOpenAI, model_name: str):
@pytest.mark.asyncio
@pytest.mark.parametrize(
"model_name",
["HuggingFaceH4/zephyr-7b-beta", "zephyr-lora"],
["HuggingFaceH4/zephyr-7b-beta"],
)
async def test_chat_completion_stream_options(
client: openai.AsyncOpenAI, model_name: str
@@ -9,7 +9,6 @@ import openai # use the official client for correctness check
import pytest
import pytest_asyncio
# downloading lora to test lora requests
from tests.utils import ROCM_EXTRA_ARGS, RemoteOpenAIServer
# any model with a chat template should work here