12cae04db9
### What this PR does / why we need it? Add a `VLLMAscendQuantizer` to support w8a8 static (W8A8) and dynamic on linear and moe (W8A8_DYNAMIC), the quantizer will be enable if a model has [quantize filed](https://huggingface.co/vllm-ascend/Qwen2.5-0.5B-Instruct-w8a8/blob/main/config.json#L27). If MindIE Turbo is installed, the MindIE Turbo Quantizer will apply, otherwise will use VLLMAscendQuantizer directly. - This patch fix installation docs to make installation work - This patch enable norm quantization by patch `RMSNorm.__init__`, `RMSNorm.forward_oot`, `NPUModelRunnerBase.load_model` - Add `AscendW8A8LinearMethod` for W8A8 - Add `AscendW8A8DynamicLinearMethod` and `AscendW8A8DynamicFusedMoEMethod` for W8A8_DYNAMIC - Add a e2e test for `vllm-ascend/Qwen2.5-0.5B-Instruct-w8a8` ### Does this PR introduce _any_ user-facing change? Yes, support w8a8 quantization. After this patch supported, users can use below commands to run w8a8 models: ``` vllm serve /root/.cache/modelscope/hub/Qwen/Qwen2.5-7B-Instruct-w8a8 --served-model-name "qwen2.5-7B" ``` ### How was this patch tested? 0. CI passed: add e2e test for `vllm-ascend/Qwen2.5-0.5B-Instruct-w8a8` 1. From @Yikun: I test Qwen2.5-0.5B-Instruct-w8a8 for functional test all is well, pls refer to https://github.com/vllm-project/vllm-ascend/pull/580#issuecomment-2816747613 2. From @dingdingchaomian : Use qwen2.5-72b-instruct model and deepseek-v2-lite-chat tested, both models were quantized using Ascend's msmodelslim tool: - Qwen2.5-72b-instruct were tested twice, one for w8a8 static and one for w8a8 dynamic. - Deepseek-v2-lite-chat were tested once because its quantization used both static and dynamic w8a8. Models were tested using both off line inference and online serving, and both work well. The inference codes are exactly the same with the examples in https://vllm-ascend.readthedocs.io/en/latest/quick_start.html, with model path and tensor parallel number changed. --------- Signed-off-by: dingdingchaomian <wangce21@huawei.com> Signed-off-by: Yikun Jiang <yikunkero@gmail.com> Co-authored-by: dingdingchaomian <wangce21@huawei.com> Co-authored-by: Angazenn <zengyanjia@huawei.com> Co-authored-by: liujiaxu <liujiaxu4@huawei.com> Co-authored-by: ApsarasX <apsarax@outlook.com> Co-authored-by: ganyi1996ppo <pleaplusone.gy@gmail.com>
61 lines
2.0 KiB
Python
61 lines
2.0 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
|
|
|
|
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",
|
|
]
|
|
os.environ["VLLM_USE_MODELSCOPE"] = "True"
|
|
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)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
import pytest
|
|
pytest.main([__file__])
|