Spaces:
Running on Zero
Running on Zero
Test standalone llama server container
Browse files- Dockerfile +36 -0
Dockerfile
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
FROM python:3.10-slim
|
| 2 |
+
|
| 3 |
+
ENV DEBIAN_FRONTEND=noninteractive \
|
| 4 |
+
CUDA_VISIBLE_DEVICES="" \
|
| 5 |
+
GGML_CUDA=0 \
|
| 6 |
+
OMP_NUM_THREADS=16 \
|
| 7 |
+
OMP_THREAD_LIMIT=16 \
|
| 8 |
+
HF_HOME=/data/.cache/huggingface
|
| 9 |
+
|
| 10 |
+
RUN apt-get update && apt-get install -y --no-install-recommends \
|
| 11 |
+
build-essential cmake git ca-certificates \
|
| 12 |
+
&& rm -rf /var/lib/apt/lists/*
|
| 13 |
+
|
| 14 |
+
WORKDIR /opt
|
| 15 |
+
RUN git clone --depth=1 https://github.com/ggml-org/llama.cpp.git \
|
| 16 |
+
&& cmake -S llama.cpp -B llama.cpp/build \
|
| 17 |
+
-DGGML_NATIVE=OFF \
|
| 18 |
+
-DGGML_OPENMP=OFF \
|
| 19 |
+
-DGGML_CUDA=OFF \
|
| 20 |
+
-DGGML_AVX=ON \
|
| 21 |
+
-DGGML_AVX2=ON \
|
| 22 |
+
-DGGML_FMA=ON \
|
| 23 |
+
-DGGML_AVX512=OFF \
|
| 24 |
+
-DBUILD_SHARED_LIBS=OFF \
|
| 25 |
+
&& cmake --build llama.cpp/build --config Release --target llama-server -j16 \
|
| 26 |
+
&& find llama.cpp/build -type f -name llama-server -perm -111 -exec cp {} /usr/local/bin/llama-server \; \
|
| 27 |
+
&& test -x /usr/local/bin/llama-server \
|
| 28 |
+
&& rm -rf /opt/llama.cpp
|
| 29 |
+
|
| 30 |
+
RUN pip install --no-cache-dir huggingface-hub
|
| 31 |
+
|
| 32 |
+
COPY server_entrypoint.sh /opt/server_entrypoint.sh
|
| 33 |
+
RUN chmod +x /opt/server_entrypoint.sh
|
| 34 |
+
|
| 35 |
+
EXPOSE 7860
|
| 36 |
+
CMD ["/opt/server_entrypoint.sh"]
|