matthoffner commited on
Commit
6db6156
·
1 Parent(s): dc394b0

Update Dockerfile

Browse files
Files changed (1) hide show
  1. Dockerfile +38 -14
Dockerfile CHANGED
@@ -1,28 +1,52 @@
1
- FROM nvidia/cuda:11.6.0-base-ubuntu20.04
2
- RUN export PATH="/usr/local/cuda/bin:$PATH"
 
 
 
 
 
 
 
3
  RUN apt update && \
4
  apt install --no-install-recommends -y build-essential python3 python3-pip wget curl git && \
5
  apt clean && rm -rf /var/lib/apt/lists/*
6
 
7
- EXPOSE 8000
8
-
9
  WORKDIR /app
10
 
11
- RUN export PATH="/usr/local/cuda/bin:$PATH"
12
- RUN wget -qO- "https://cmake.org/files/v3.17/cmake-3.17.0-Linux-x86_64.tar.gz" | tar --strip-components=1 -xz -C /usr/local
 
13
  RUN CMAKE_ARGS="-DLLAMA_CUBLAS=on" FORCE_CMAKE=1 pip install git+https://github.com/abetlen/llama-cpp-python --no-cache-dir
14
 
15
  COPY requirements.txt ./
16
- RUN pip install --upgrade pip && \
17
- pip install -r requirements.txt
18
 
19
- #RUN curl -s https://packagecloud.io/install/repositories/github/git-lfs/script.deb.sh | bash
20
- #RUN apt-get install git-lfs -y
21
- #RUN git clone https://huggingface.co/TheBloke/robin-13B-v2-GGML
22
- #RUN mv robin-13B-v2-GGML/robin-13b.ggmlv3.q3_K_M.bin .
23
- #RUN rm -rf robin-13B-v2-GGML/
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
24
 
25
- COPY . .
26
  RUN ls -al
27
 
 
 
28
  CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]
 
1
+ FROM nvidia/cuda:11.0.3-base-ubuntu20.04
2
+
3
+ ENV MODEL_NAME="robin-13B-v2-GGML"
4
+ ENV DEFAULT_MODEL_FILE="robin-13b.ggmlv3.q4_K_M.bin"
5
+ ENV MODEL_USER="TheBloke"
6
+ ENV DEFAULT_MODEL_BRANCH="main"
7
+ ENV MODEL_URL="https://huggingface.co/${MODEL_USER}/${MODEL_NAME}/resolve/${DEFAULT_MODEL_BRANCH}/${DEFAULT_MODEL_FILE}"
8
+ ENV PATH="/usr/local/cuda/bin:$PATH"
9
+
10
  RUN apt update && \
11
  apt install --no-install-recommends -y build-essential python3 python3-pip wget curl git && \
12
  apt clean && rm -rf /var/lib/apt/lists/*
13
 
 
 
14
  WORKDIR /app
15
 
16
+ RUN apt-get install -y wget && \
17
+ wget -qO- "https://cmake.org/files/v3.18/cmake-3.18.0-Linux-x86_64.tar.gz" | tar --strip-components=1 -xz -C /usr/local
18
+
19
  RUN CMAKE_ARGS="-DLLAMA_CUBLAS=on" FORCE_CMAKE=1 pip install git+https://github.com/abetlen/llama-cpp-python --no-cache-dir
20
 
21
  COPY requirements.txt ./
 
 
22
 
23
+ RUN pip3 install --upgrade pip && \
24
+ pip3 install -r requirements.txt
25
+
26
+ WORKDIR /app
27
+
28
+ RUN echo ${MODEL_URL} && \
29
+ wget -O /app/${DEFAULT_MODEL_FILE} ${MODEL_URL}
30
+
31
+ RUN useradd -m -u 1000 user
32
+
33
+ RUN mkdir -p /home/user/app && \
34
+ mv /app/${DEFAULT_MODEL_FILE} /home/user/app
35
+
36
+ RUN chown -R user:user /home/user/app
37
+
38
+ USER user
39
+ ENV HOME=/home/user \
40
+ PATH=/home/user/.local/bin:$PATH \
41
+ MODEL_NAME=${MODEL_NAME} \
42
+ MODEL_FILE=/home/user/app/${DEFAULT_MODEL_FILE}
43
+
44
+ WORKDIR $HOME/app
45
+
46
+ COPY --chown=user . .
47
 
 
48
  RUN ls -al
49
 
50
+ EXPOSE 8000
51
+
52
  CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]