Update Dockerfile

Dockerfile

@@ -1,51 +1,60 @@
-FROM ubuntu:22.04
+FROM nvidia/cuda:11.0.3-base-ubuntu20.04
 
-ENV DEBIAN_FRONTEND=noninteractive
+ENV MODEL_NAME="Nous-Hermes-13B-GGUF"
+ENV DEFAULT_MODEL_FILE="Nous-Hermes-13B.Q4_K_M.gguf"
+ENV MODEL_USER="TheBloke"
+ENV DEFAULT_MODEL_BRANCH="main"
+ENV MODEL_URL="https://huggingface.co/${MODEL_USER}/${MODEL_NAME}/resolve/${DEFAULT_MODEL_BRANCH}/${DEFAULT_MODEL_FILE}"
+ENV PATH="/usr/local/cuda/bin:$PATH"
 
-# Update and install necessary dependencies
 RUN apt update && \
-    apt install --no-install-recommends -y \
-        build-essential \
-        python3 \
-        python3-pip \
-        wget \
-        curl \
-        git \
-        cmake \
-        zlib1g-dev \
-        libblas-dev && \
-    apt clean && \
-    rm -rf /var/lib/apt/lists/*
+    apt install --no-install-recommends -y build-essential python3 python3-pip wget curl git && \
+    apt clean && rm -rf /var/lib/apt/lists/*
 
 WORKDIR /app
 
-# Download ggml and mmproj models from HuggingFace
-RUN wget https://huggingface.co/mys/ggml_bakllava-1/resolve/main/ggml-model-q4_k.gguf && \
-    wget https://huggingface.co/mys/ggml_bakllava-1/resolve/main/mmproj-model-f16.gguf 
+RUN apt-get install -y wget && \
+    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
 
-# Clone and build llava-server with CUDA support
-RUN git clone https://github.com/ggerganov/llama.cpp.git && \
-    cd llama.cpp && \
-    git submodule init && \
-    git submodule update && \
-    make
+COPY requirements.txt ./
 
-# Create a non-root user for security reasons
-RUN useradd -m -u 1000 user && \
-    mkdir -p /home/user/app && \
-    cp /app/ggml-model-q4_k.gguf /home/user/app && \
-    cp /app/mmproj-model-f16.gguf /home/user/app
+RUN pip3 install --upgrade pip && \
+    pip3 install -r requirements.txt
 
-RUN chown user:user /home/user/app/ggml-model-q4_k.gguf && \
-    chown user:user /home/user/app/mmproj-model-f16.gguf
+# Cloning the llama-cpp-python repository
+#RUN git clone https://github.com/abetlen/llama-cpp-python.git
+#WORKDIR llama-cpp-python
+# Initializing and updating the llama.cpp submodule
+#RUN git submodule init && git submodule update
+
+# Building the llama-cpp-python
+# RUN FORCE_CMAKE=1 pip install .
+
+# Moving back to /app directory
+WORKDIR /app
+
+RUN echo ${MODEL_URL} && \
+    wget -O /app/${DEFAULT_MODEL_FILE} ${MODEL_URL}
+
+RUN useradd -m -u 1000 user
+
+RUN mkdir -p /home/user/app && \
+    mv /app/${DEFAULT_MODEL_FILE} /home/user/app
+
+RUN chown -R user:user /home/user/app
 
 USER user
-ENV HOME=/home/user
+ENV HOME=/home/user \
+	PATH=/home/user/.local/bin:$PATH \
+    MODEL_NAME=${MODEL_NAME} \
+    MODEL_FILE=/home/user/app/${DEFAULT_MODEL_FILE} 
 
 WORKDIR $HOME/app
 
-# Expose the port
-EXPOSE 8080
+COPY --chown=user . .
+
+RUN ls -al
+
+EXPOSE 8000
 
-# Start the llava-server with models
-CMD ["/app/llama.cpp/server", "--model", "ggml-model-q4_k.gguf", "--mmproj", "mmproj-model-f16.gguf", "--host", "0.0.0.0", "--threads", "10"]
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]