Mirrors
/
hf_text-generation-inference
mirror of https://github.com/huggingface/text-generation-inference.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Using prebuilt.

This commit is contained in:
Nicolas Patry 2024-08-17 00:42:51 +02:00
parent 9d4c5d39fe
commit f2bdc65098
No known key found for this signature in database
GPG Key ID: 64AF4752B2967863
2 changed files with 4 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Dockerfile
View File

@ -188,7 +188,7 @@ RUN make build-all
FROM kernel-builder AS flashinfer-builder FROM kernel-builder AS flashinfer-builder
WORKDIR /usr/src WORKDIR /usr/src
COPY server/Makefile-flashinfer Makefile COPY server/Makefile-flashinfer Makefile
RUN make build-flashinfer RUN make install-flashinfer
# Text Generation Inference base image # Text Generation Inference base image
FROM nvidia/cuda:12.1.0-base-ubuntu22.04 AS base FROM nvidia/cuda:12.1.0-base-ubuntu22.04 AS base
@ -242,7 +242,7 @@ COPY --from=vllm-builder /usr/src/vllm/build/lib.linux-x86_64-cpython-310 /opt/c
# Copy build artifacts from mamba builder # Copy build artifacts from mamba builder
COPY --from=mamba-builder /usr/src/mamba/build/lib.linux-x86_64-cpython-310/ /opt/conda/lib/python3.10/site-packages COPY --from=mamba-builder /usr/src/mamba/build/lib.linux-x86_64-cpython-310/ /opt/conda/lib/python3.10/site-packages
COPY --from=mamba-builder /usr/src/causal-conv1d/build/lib.linux-x86_64-cpython-310/ /opt/conda/lib/python3.10/site-packages COPY --from=mamba-builder /usr/src/causal-conv1d/build/lib.linux-x86_64-cpython-310/ /opt/conda/lib/python3.10/site-packages
COPY --from=flashinfer-builder /usr/src/flashinfer/build/lib.linux-x86_64-cpython-310/ /opt/conda/lib/python3.10/site-packages COPY --from=flashinfer-builder /opt/conda/lib/python3.10/site-packages/flashinfer/ /opt/conda/lib/python3.10/site-packages/flashinfer/
# Install flash-attention dependencies # Install flash-attention dependencies
RUN pip install einops --no-cache-dir RUN pip install einops --no-cache-dir

14
server/Makefile-flashinfer
View File

@ -1,12 +1,2 @@
flashinfer_commit := v0.1.5 install-flashinfer:
pip install flashinfer==0.1.5 -i https://flashinfer.ai/whl/cu124/torch2.4
build-flashinfer:
git clone https://github.com/flashinfer-ai/flashinfer.git flashinfer && \
cd flashinfer && git fetch && git checkout $(flashinfer_commit) && \
git submodule update --init --recursive && \
cd python/ && \
CUDA_ARCH_LIST="8.0;9.0a" NVCC_GENCODE="-gencode=arch=compute_80,code=sm_80 -gencode=arch=compute_90a,code=sm_90a" TORCH_CUDA_ARCH_LIST="8.0;9.0a" python setup.py build
install-flashinfer: build-flashinfer
cd flashinfer/python/ && \
CUDA_ARCH_LIST="8.0;9.0a" NVCC_GENCODE="-gencode=arch=compute_80,code=sm_80 -gencode=arch=compute_90a,code=sm_90a" TORCH_CUDA_ARCH_LIST="8.0;9.0a" python setup.py install