e2a0c19cea
1. remove some useless test func and file 2. fix format.sh problem 3. enable full test for singlecard and multicard 4. move long term test to long_term folder. For this kind of test, it only runs by labeled and daily test. Include: spec decode、accuracy test ## After refactor: There are 4 test modules - `singlecard`: contains the test running on one NPU. It'll be run for each PR and daily test. - `multicard`: contains the test running on multi NPUs. It'll be run for each PR and daily test. - `long_term`: contains the test that cost much time(Now include `spec decode` and `accuracy` test). It'll be run for the PR with `long-term-test` labeled and daily test. - `e2e`: contains the test for doc and pd feature. It'll be run for the PR with `pd-test` labeled and daily test. ## Todo: 1. some test are skipped, they should be fixed and reenabled in the future. 2. pyhccl test for multicard doesn't work at all. It should be enabled as well. 3. ensure long-term-test pass by daily test. ### Know issue Now, `ready` labels is required to start pd test or long term test. And when `long-term-test` or `pd-test` is labeled after another one, the old labeled test will be re-run again. So the labeled test should be ran in the following step: 1. decide which test need run, then label it. `long-term-test` or `pd-test` or both. 2. add `ready-for-test` label, then the test will be ran. Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
84 lines
3.1 KiB
Python
84 lines
3.1 KiB
Python
#
|
|
# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
|
|
# Copyright 2023 The vLLM team.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
# This file is a part of the vllm-ascend project.
|
|
# Adapted from vllm/tests/basic_correctness/test_basic_correctness.py
|
|
#
|
|
"""Compare the short outputs of HF and vLLM when using greedy sampling.
|
|
|
|
Run `pytest tests/test_offline_inference.py`.
|
|
"""
|
|
import os
|
|
|
|
import pytest
|
|
import vllm # noqa: F401
|
|
from vllm.assets.image import ImageAsset
|
|
|
|
import vllm_ascend # noqa: F401
|
|
from tests.conftest import VllmRunner
|
|
|
|
MODELS = [
|
|
"Qwen/Qwen2.5-0.5B-Instruct",
|
|
"vllm-ascend/Qwen2.5-0.5B-Instruct-w8a8",
|
|
"Qwen/Qwen3-0.6B-Base",
|
|
]
|
|
MULTIMODALITY_MODELS = ["Qwen/Qwen2.5-VL-3B-Instruct"]
|
|
os.environ["PYTORCH_NPU_ALLOC_CONF"] = "max_split_size_mb:256"
|
|
|
|
|
|
@pytest.mark.parametrize("model", MODELS)
|
|
@pytest.mark.parametrize("dtype", ["half", "float16"])
|
|
@pytest.mark.parametrize("max_tokens", [5])
|
|
def test_models(model: str, dtype: str, max_tokens: int) -> None:
|
|
# 5042 tokens for gemma2
|
|
# gemma2 has alternating sliding window size of 4096
|
|
# we need a prompt with more than 4096 tokens to test the sliding window
|
|
prompt = "The following numbers of the sequence " + ", ".join(
|
|
str(i) for i in range(1024)) + " are:"
|
|
example_prompts = [prompt]
|
|
|
|
with VllmRunner(model,
|
|
max_model_len=8192,
|
|
dtype=dtype,
|
|
enforce_eager=False,
|
|
gpu_memory_utilization=0.7) as vllm_model:
|
|
vllm_model.generate_greedy(example_prompts, max_tokens)
|
|
|
|
|
|
@pytest.mark.parametrize("model", MULTIMODALITY_MODELS)
|
|
@pytest.mark.skipif(os.getenv("VLLM_USE_V1") == "1",
|
|
reason="qwen2.5_vl is not supported on v1")
|
|
def test_multimodal(model, prompt_template, vllm_runner):
|
|
image = ImageAsset("cherry_blossom") \
|
|
.pil_image.convert("RGB")
|
|
img_questions = [
|
|
"What is the content of this image?",
|
|
"Describe the content of this image in detail.",
|
|
"What's in the image?",
|
|
"Where is this image taken?",
|
|
]
|
|
images = [image] * len(img_questions)
|
|
prompts = prompt_template(img_questions)
|
|
with vllm_runner(model,
|
|
max_model_len=4096,
|
|
mm_processor_kwargs={
|
|
"min_pixels": 28 * 28,
|
|
"max_pixels": 1280 * 28 * 28,
|
|
"fps": 1,
|
|
}) as vllm_model:
|
|
vllm_model.generate_greedy(prompts=prompts,
|
|
images=images,
|
|
max_tokens=64)
|