diff --git a/.github/workflows/pr_ete_test.yml b/.github/workflows/pr_ete_test.yml index 3a19ebe870..2d1c4b63f5 100644 --- a/.github/workflows/pr_ete_test.yml +++ b/.github/workflows/pr_ete_test.yml @@ -10,7 +10,7 @@ on: - "3rdparty/**" - "lmdeploy/**" - "requirements/**" - - "requirements.txt" + - "requirements_cuda.txt" - "CMakeLists.txt" - "setup.py" workflow_dispatch: @@ -68,7 +68,7 @@ jobs: export PATH=$PATH:/usr/local/openmpi/bin export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/openmpi/lib python3 -m pip install cmake packaging wheel transformers_stream_generator transformers datasets openai einops timm decord - python3 -m pip install -r requirements.txt -r requirements/test.txt -r requirements/build.txt + python3 -m pip install -r requirements_cuda.txt -r requirements/test.txt -r requirements/build.txt mkdir -p build && cd build &&\ sh ../generate.sh &&\ ninja -j$(nproc) && ninja install &&\ diff --git a/.github/workflows/unit-test.yml b/.github/workflows/unit-test.yml index ec6db0682d..ec70f61f4c 100644 --- a/.github/workflows/unit-test.yml +++ b/.github/workflows/unit-test.yml @@ -10,7 +10,7 @@ on: - "3rdparty/**" - "lmdeploy/**" - "requirements/**" - - "requirements.txt" + - "requirements_cuda.txt" - "CMakeLists.txt" - "setup.py" push: @@ -24,7 +24,7 @@ on: - "3rdparty/**" - "lmdeploy/**" - "requirements/**" - - "requirements.txt" + - "requirements_cuda.txt" - "CMakeLists.txt" - "setup.py" tags: @@ -78,7 +78,7 @@ jobs: python3 -m pip install pynvml packaging protobuf transformers_stream_generator # manually install flash attn python3 -m pip install /root/packages/flash_attn-2.6.3+cu118torch2.3cxx11abiFALSE-cp38-cp38-linux_x86_64.whl - python3 -m pip install -r requirements.txt -r requirements/test.txt + python3 -m pip install -r requirements_cuda.txt -r requirements/test.txt python3 -m pip install . - name: Check env run: | diff --git a/docker/Dockerfile_aarch64_ascend b/docker/Dockerfile_aarch64_ascend index 1c9591197b..ecc2d1334e 100644 --- a/docker/Dockerfile_aarch64_ascend +++ b/docker/Dockerfile_aarch64_ascend @@ -122,4 +122,4 @@ WORKDIR /opt/lmdeploy RUN --mount=type=cache,target=/root/.cache/pip \ sed -i '/triton/d' requirements/runtime.txt && \ - pip3 install -v --no-build-isolation -e . + LMDEPLOY_TARGET_DEVICE=ascend pip3 install -v --no-build-isolation -e . diff --git a/requirements/runtime_ascend.txt b/requirements/runtime_ascend.txt index c5d44cc995..965175faf3 100644 --- a/requirements/runtime_ascend.txt +++ b/requirements/runtime_ascend.txt @@ -16,7 +16,8 @@ safetensors sentencepiece shortuuid tiktoken -torch<=2.4.0,>=2.0.0 +torch<=2.4.0,>=2.3.1 +torch-npu==2.3.1 torchvision<=0.19.0,>=0.15.0 transformers uvicorn diff --git a/requirements/runtime.txt b/requirements/runtime_cuda.txt similarity index 100% rename from requirements/runtime.txt rename to requirements/runtime_cuda.txt diff --git a/requirements/runtime_maca.txt b/requirements/runtime_maca.txt new file mode 100644 index 0000000000..f65b3827cd --- /dev/null +++ b/requirements/runtime_maca.txt @@ -0,0 +1,22 @@ +accelerate==0.32.1 +einops +fastapi +fire +mmengine-lite +numpy<2.0.0 +openai +outlines<0.1.0 +peft<=0.11.1 +pillow +protobuf +pydantic>2.0.0 +pynvml +safetensors +sentencepiece +shortuuid +tiktoken +torch<=2.4.0,>=2.0.0 +torchvision<=0.19.0,>=0.15.0 +transformers +triton>=2.1.0; sys_platform == "linux" +uvicorn diff --git a/requirements.txt b/requirements_cuda.txt similarity index 70% rename from requirements.txt rename to requirements_cuda.txt index 91d38808f1..7c1d387dfb 100644 --- a/requirements.txt +++ b/requirements_cuda.txt @@ -1,4 +1,4 @@ -r requirements/build.txt --r requirements/runtime.txt +-r requirements/runtime_cuda.txt -r requirements/lite.txt -r requirements/serve.txt diff --git a/requirements_maca.txt b/requirements_maca.txt new file mode 100644 index 0000000000..075b132c8c --- /dev/null +++ b/requirements_maca.txt @@ -0,0 +1,4 @@ +-r requirements/build.txt +-r requirements/runtime_maca.txt +-r requirements/lite.txt +-r requirements/serve.txt diff --git a/setup.py b/setup.py index 7a08ac7919..52e180d8a2 100644 --- a/setup.py +++ b/setup.py @@ -4,18 +4,14 @@ from setuptools import find_packages, setup -npu_available = False -try: - import torch_npu - - npu_available = torch_npu.npu.is_available() -except ImportError: - pass - pwd = os.path.dirname(__file__) version_file = 'lmdeploy/version.py' +def get_target_device(): + return os.getenv('LMDEPLOY_TARGET_DEVICE', 'cuda') + + def readme(): with open(os.path.join(pwd, 'README.md'), encoding='utf-8') as f: content = f.read() @@ -154,16 +150,12 @@ def gen_packages_items(): setup_requires=parse_requirements('requirements/build.txt'), tests_require=parse_requirements('requirements/test.txt'), install_requires=parse_requirements( - 'requirements/runtime_ascend.txt' - if npu_available else 'requirements/runtime.txt'), + f'requirements/runtime_{get_target_device()}.txt'), extras_require={ 'all': - parse_requirements('requirements_ascend.txt' - if npu_available else 'requirements.txt'), - 'lite': - parse_requirements('requirements/lite.txt'), - 'serve': - parse_requirements('requirements/serve.txt') + parse_requirements(f'requirements_{get_target_device()}.txt'), + 'lite': parse_requirements('requirements/lite.txt'), + 'serve': parse_requirements('requirements/serve.txt') }, has_ext_modules=check_ext_modules, classifiers=[