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 6, 10:45 UTC
[Refactor] Relocate endpoint tests to mirror serving code directory structure (#37504)
Signed-off-by: sfeng33 <4florafeng@gmail.com>
Flora Feng committed
Mar 19, 2026 at 03:19 UTC
b21d3843048001101713e70597fc9484332a5f7e
17 files changed
+14
-17
.buildkite/scripts/hardware_ci/run-amd-test.sh
+1
-2
@@ -336,11 +336,10 @@ apply_rocm_test_overrides() {
336
--ignore=entrypoints/openai/chat_completion/test_audio.py \
337
--ignore=entrypoints/openai/completion/test_shutdown.py \
338
--ignore=entrypoints/openai/test_completion.py \
339
- --ignore=entrypoints/openai/test_models.py \
339
+ --ignore=entrypoints/openai/models/test_models.py \
340
--ignore=entrypoints/openai/test_lora_adapters.py \
341
--ignore=entrypoints/openai/test_return_tokens_as_ids.py \
342
--ignore=entrypoints/openai/chat_completion/test_root_path.py \
343
- --ignore=entrypoints/openai/test_tokenization.py \
343
--ignore=entrypoints/openai/completion/test_prompt_validation.py "}
344
fi
345
pyproject.toml
+1
-1
@@ -121,7 +121,7 @@ python = "./.venv"
121
# these files may be written in non english words
122
extend-exclude = ["tests/models/fixtures/*", "tests/prompts/*", "tests/tokenizers_/*",
123
"benchmarks/sonnet.txt", "tests/lora/data/*", "examples/pooling/token_embed/*", "build/*",
124
- "vllm/third_party/*", "vllm/entrypoints/serve/instrumentator/static/*", "tests/entrypoints/openai/test_transcription_validation.py",
124
+ "vllm/third_party/*", "vllm/entrypoints/serve/instrumentator/static/*", "tests/entrypoints/openai/speech_to_text/test_transcription_validation.py",
125
"docs/governance/process.md", "tests/v1/engine/test_fast_incdec_prefix_err.py", ".git/*"]
126
ignore-hidden = false
127
tests/entrypoints/anthropic/test_messages.py
renamed
+1
-1
@@ -5,7 +5,7 @@ import anthropic
5
import pytest
6
import pytest_asyncio
7
8
-from ...utils import RemoteOpenAIServer
8
+from tests.utils import RemoteOpenAIServer
9
10
MODEL_NAME = "Qwen/Qwen3-0.6B"
11
tests/entrypoints/openai/models/__init__.py
tests/entrypoints/openai/models/test_models.py
renamed
+1
-1
@@ -5,7 +5,7 @@ import openai # use the official client for correctness check
5
import pytest
6
import pytest_asyncio
7
8
-from ...utils import RemoteOpenAIServer
8
+from tests.utils import RemoteOpenAIServer
9
10
# any model with a chat template should work here
11
MODEL_NAME = "Qwen/Qwen3-0.6B"
tests/entrypoints/openai/realtime/__init__.py
tests/entrypoints/openai/realtime/test_realtime_validation.py
renamed
+2
-3
@@ -11,11 +11,10 @@ import pybase64 as base64
11
import pytest
12
import websockets
13
14
+from tests.entrypoints.openai.conftest import add_attention_backend
15
+from tests.utils import ROCM_ENV_OVERRIDES, ROCM_EXTRA_ARGS, RemoteOpenAIServer
16
from vllm.assets.audio import AudioAsset
17
16
-from ...utils import ROCM_ENV_OVERRIDES, ROCM_EXTRA_ARGS, RemoteOpenAIServer
17
-from .conftest import add_attention_backend
18
-
18
MISTRAL_FORMAT_ARGS = [
19
"--tokenizer_mode",
20
"mistral",
tests/entrypoints/openai/responses/test_mcp_tools.py
+1
-1
@@ -42,7 +42,7 @@ class TestMCPToolServerUnit:
42
Note: The wildcard "*" is normalized to None by
43
_extract_allowed_tools_from_mcp_requests before reaching this layer,
44
so we only test None and specific tool filtering here.
45
- See test_serving_responses.py for "*" normalization tests.
45
+ See responses/test_serving_responses.py for "*" normalization tests.
46
"""
47
48
def test_get_tool_description(self):
tests/entrypoints/openai/responses/test_serving_responses.py
renamed
tests/entrypoints/openai/speech_to_text/__init__.py
tests/entrypoints/openai/speech_to_text/test_transcription_validation.py
renamed
+2
-2
@@ -6,8 +6,8 @@ import json
6
7
import pytest
8
9
-from ...utils import ROCM_ENV_OVERRIDES, ROCM_EXTRA_ARGS, RemoteOpenAIServer
10
-from .conftest import add_attention_backend
9
+from tests.entrypoints.openai.conftest import add_attention_backend
10
+from tests.utils import ROCM_ENV_OVERRIDES, ROCM_EXTRA_ARGS, RemoteOpenAIServer
11
12
MISTRAL_FORMAT_ARGS = [
13
"--tokenizer_mode",
tests/entrypoints/openai/speech_to_text/test_transcription_validation_whisper.py
renamed
+1
-1
@@ -13,7 +13,7 @@ import pytest
13
import pytest_asyncio
14
import soundfile as sf
15
16
-from ...utils import RemoteOpenAIServer
16
+from tests.utils import RemoteOpenAIServer
17
18
MODEL_NAME = "openai/whisper-large-v3-turbo"
19
tests/entrypoints/openai/speech_to_text/test_translation_validation.py
renamed
+2
-2
@@ -14,8 +14,8 @@ import pytest
14
import pytest_asyncio
15
import soundfile as sf
16
17
-from ...utils import RemoteOpenAIServer
18
-from .conftest import add_attention_backend
17
+from tests.entrypoints.openai.conftest import add_attention_backend
18
+from tests.utils import RemoteOpenAIServer
19
20
SERVER_ARGS = ["--enforce-eager"]
21
tests/entrypoints/serve/__init__.py
tests/entrypoints/serve/tokenize/__init__.py
tests/entrypoints/serve/tokenize/test_tokenization.py
renamed
+1
-2
@@ -5,10 +5,9 @@ import pytest
5
import pytest_asyncio
6
import requests
7
8
+from tests.utils import RemoteOpenAIServer
9
from vllm.tokenizers import get_tokenizer
10
10
-from ...utils import RemoteOpenAIServer
11
-
11
# any model with a chat template should work here
12
MODEL_NAME = "HuggingFaceH4/zephyr-7b-beta"
13
tests/entrypoints/serve/tokenize/test_tokenization_vlm.py
renamed
+1
-1
@@ -13,7 +13,7 @@ import json
13
import pytest
14
import requests
15
16
-from ...utils import RemoteOpenAIServer
16
+from tests.utils import RemoteOpenAIServer
17
18
MODEL_NAME = "Qwen/Qwen2.5-VL-3B-Instruct"
19