huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-06-19 15:52:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

Revert dummy modifications.

Browse Source
This commit is contained in:
Nicolas Patry 2025-02-05 11:55:17 +01:00
parent 951eb62b56
commit a1c78adc19
No known key found for this signature in database
GPG Key ID: D2920555C90F704C
2 changed files with 48 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
Dockerfile_amd
View File

@ -1,5 +1,5 @@
# Rust builder
FROM lukemathwalker/cargo-chef:latest-rust-1.80.1 AS chef
FROM lukemathwalker/cargo-chef:latest-rust-1.84.0 AS chef
WORKDIR /usr/src
ARG CARGO_REGISTRIES_CRATES_IO_PROTOCOL=sparse
@ -104,7 +104,7 @@ RUN case ${TARGETPLATFORM} in \
/opt/conda/bin/conda clean -ya
# Install flash-attention, torch dependencies
RUN python3 -m pip install --upgrade pip && pip install numpy einops ninja joblib msgpack cmake --no-cache-dir && rm -rf /var/lib/apt/lists/*
RUN python3 -m pip install --upgrade pip uv && pip install numpy einops ninja joblib msgpack cmake --no-cache-dir && rm -rf /var/lib/apt/lists/*
RUN conda install mkl=2021
ENV LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/rocm/lib/:/opt/conda/lib/python3.11/site-packages/torch/lib:/opt/conda/lib/
@ -234,6 +234,7 @@ FROM kernel-builder AS vllm-builder
WORKDIR /usr/src
COPY server/Makefile-vllm Makefile
RUN pip install setuptools_scm
# Build specific version of vllm
RUN make build-vllm-rocm
@ -267,6 +268,24 @@ COPY server/exllamav2_kernels/ .
RUN python setup.py build
FROM kernel-builder AS marlin-kernels
WORKDIR /usr/src
ENV MARLIN_KERNELS_BRANCH=v0.3.6
ENV VLLM_TARGET_DEVICE=rocm
RUN git clone https://github.com/danieldk/marlin-kernels.git && \
cd marlin-kernels && \
git checkout ${MARLIN_KERNELS_BRANCH} && \
python setup.py install
FROM kernel-builder AS moe-kernels
WORKDIR /usr/src
ENV MOE_KERNELS_BRANCH=v0.8.2
ENV VLLM_TARGET_DEVICE=rocm
RUN git clone https://github.com/danieldk/moe-kernels.git && \
cd moe-kernels && \
git checkout ${MOE_KERNELS_BRANCH} && \
python setup.py install
FROM install_deps AS base-copy
# Text Generation Inference base env
@ -289,14 +308,21 @@ COPY --from=exllama-kernels-builder /usr/src/build/lib.linux-x86_64-cpython-311
# Copy build artifacts from exllamav2 kernels builder
COPY --from=exllamav2-kernels-builder /usr/src/build/lib.linux-x86_64-cpython-311 /opt/conda/lib/python3.11/site-packages
# Copy build artifacts from marlin kernels
COPY --from=marlin-kernels /usr/src/marlin-kernels/build/lib.linux-x86_64-cpython-311 /opt/conda/lib/python3.11/site-packages
# Copy build artifacts from moe kernels
COPY --from=moe-kernels /usr/src/moe-kernels/build/lib.linux-x86_64-cpython-311 /opt/conda/lib/python3.11/site-packages
# Install server
COPY proto proto
COPY server server
COPY server/Makefile server/Makefile
ENV UV_SYSTEM_PYTHON=1
RUN cd server && \
make gen-server && \
pip install -r requirements_rocm.txt --no-cache-dir
# pip install ".[accelerate, compressed-tensors, peft, outlines]" --no-cache-dir
pip install -U pip uv && \
uv pip install -r requirements_rocm.txt --no-cache-dir
# Install benchmarker
COPY --from=builder /usr/src/target/release-opt/text-generation-benchmark /usr/local/bin/text-generation-benchmark

31
Dockerfile_intel
View File

@ -1,6 +1,6 @@
ARG PLATFORM=xpu
FROM lukemathwalker/cargo-chef:latest-rust-1.80.1 AS chef
FROM lukemathwalker/cargo-chef:latest-rust-1.84.0 AS chef
WORKDIR /usr/src
ARG CARGO_REGISTRIES_CRATES_IO_PROTOCOL=sparse
@ -97,11 +97,10 @@ ENV HF_HOME=/data \
WORKDIR /usr/src
RUN pip install https://intel-extension-for-pytorch.s3.us-east-1.amazonaws.com/ipex_dev/xpu/torch-2.5.0a0%2Bgite84e33f-cp311-cp311-linux_x86_64.whl --no-cache-dir
RUN pip install https://intel-extension-for-pytorch.s3.us-east-1.amazonaws.com/ipex_dev/xpu/torchaudio-2.5.0a0%2B56bc006-cp311-cp311-linux_x86_64.whl --no-cache-dir
RUN pip install https://intel-extension-for-pytorch.s3.us-east-1.amazonaws.com/ipex_dev/xpu/torchvision-0.20.0a0%2B8e8a208-cp311-cp311-linux_x86_64.whl --no-cache-dir
RUN pip install https://intel-extension-for-pytorch.s3.us-east-1.amazonaws.com/ipex_dev/xpu/intel_extension_for_pytorch-2.5.10%2Bgit9d489a8-cp311-cp311-linux_x86_64.whl --no-cache-dir
RUN pip install https://intel-extension-for-pytorch.s3.us-east-1.amazonaws.com/ipex_dev/xpu/oneccl_bind_pt-2.5.0%2Bxpu-cp311-cp311-linux_x86_64.whl --no-cache-dir
RUN pip install https://intel-optimized-pytorch.s3.cn-north-1.amazonaws.com.cn/ipex_dev/xpu/torch-2.5.0a0%2Bgite84e33f-cp311-cp311-linux_x86_64.whl --no-cache-dir
RUN pip install https://intel-optimized-pytorch.s3.cn-north-1.amazonaws.com.cn/ipex_dev/xpu/torchaudio-2.5.0a0%2B56bc006-cp311-cp311-linux_x86_64.whl --no-cache-dir
RUN pip install https://intel-optimized-pytorch.s3.cn-north-1.amazonaws.com.cn/ipex_dev/xpu/torchvision-0.20.0a0%2B8e8a208-cp311-cp311-linux_x86_64.whl --no-cache-dir
RUN pip install https://intel-optimized-pytorch.s3.cn-north-1.amazonaws.com.cn/ipex_dev/xpu/oneccl_bind_pt-2.5.0%2Bxpu-cp311-cp311-linux_x86_64.whl --no-cache-dir
RUN pip install triton-xpu==3.0.0b2 --no-cache-dir
@ -109,15 +108,20 @@ RUN pip install triton-xpu==3.0.0b2 --no-cache-dir
COPY proto proto
COPY server server
COPY server/Makefile server/Makefile
ENV UV_SYSTEM_PYTHON=1
RUN cd server && \
make gen-server && \
pip install -r requirements_intel.txt && \
pip install ".[accelerate, compressed-tensors, peft, outlines]" --no-cache-dir
pip install -U pip uv && \
uv pip install -r requirements_intel.txt --no-cache-dir
ENV LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/intel/oneapi/pti/0.9/lib:/opt/conda/lib
ENV CCL_ZE_IPC_EXCHANGE=sockets
#ENV TORCH_LLM_ALLREDUCE=1
#ENV CCL_TOPO_FABRIC_VERTEX_CONNECTION_CHECK=0
ENV TORCH_DEVICE_BACKEND_AUTOLOAD=0
RUN git clone https://github.com/intel/intel-extension-for-pytorch && cd intel-extension-for-pytorch && git checkout 1ccf72b2d11cd00b47aef6d6cd054c088aa6f083
RUN cd intel-extension-for-pytorch && git submodule update --init --recursive && USE_AOT_DEVLIST='pvc,ats-m150' BUILD_SEPARATE_OPS=OFF BUILD_WITH_CPU=OFF USE_XETLA=ON python setup.py install && rm -rf /usr/src/intel-extension-for-pytorch
# Install benchmarker
COPY --from=builder /usr/src/target/release-opt/text-generation-benchmark /usr/local/bin/text-generation-benchmark
@ -209,10 +213,11 @@ ENV LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/opt/conda/lib/"
COPY proto proto
COPY server server
COPY server/Makefile server/Makefile
ENV UV_SYSTEM_PYTHON=1
RUN cd server && \
make gen-server && \
uv pip install -r requirements_intel.txt --no-cache-dir
# pip install ".[accelerate, compressed-tensors, peft, outlines]" --no-cache-dir
pip install -U pip uv && \
uv pip install -e ".[accelerate, compressed-tensors, peft, outlines]" --no-cache-dir
# Install benchmarker
COPY --from=builder /usr/src/target/release-opt/text-generation-benchmark /usr/local/bin/text-generation-benchmark
@ -222,9 +227,9 @@ COPY --from=builder /usr/src/target/release-opt/text-generation-router /usr/loca
COPY --from=builder /usr/src/target/release-opt/text-generation-launcher /usr/local/bin/text-generation-launcher
FROM ${PLATFORM} AS final
ENV ATTENTION=paged
ENV PREFIX_CACHING=0
ENV PREFILL_CHUNKING=0
ENV ATTENTION=flashdecoding-ipex
ENV PREFIX_CACHING=1
ENV PREFILL_CHUNKING=1
ENV CUDA_GRAPHS=0
ENTRYPOINT ["text-generation-launcher"]
CMD ["--json-output"]