Update Dockerfile

Dockerfile

@@ -22,6 +22,7 @@ RUN apt-get update && \
     libxmlsec1-dev \
     libffi-dev \
     liblzma-dev \
+    golang-go \
     golang-1.22-go \
     nvidia-driver-550 \
     ffmpeg
@@ -30,9 +31,10 @@ ENV USER='user'
 RUN useradd -m -u 1000 ${USER}
 USER ${USER}
 ENV HOME=/home/${USER} \
-    PATH=${HOME}/.local/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:${PATH} 
-WORKDIR ${HOME}/app
-
+    PATH=${HOME}/.local/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:${PATH} \
+    APPDIR=${HOME}/app
+    
+WORKDIR ${APPDIR}
 
 ENV NVIDIA_VISIBLE_DEVICES=all
 RUN curl https://pyenv.run | bash
@@ -44,14 +46,12 @@ RUN pyenv install ${PYTHON_VERSION} && \
     pip install --no-cache-dir -U pip setuptools wheel && \
     pip install "huggingface-hub" "hf-transfer" "gradio[oauth]>=4.28.0" "gradio_huggingfacehub_search==0.0.7" "APScheduler"    
     
-COPY --chown=1000 . ${HOME}/app
-
+COPY --chown=1000 . ${APPDIR}
 RUN git clone https://github.com/ollama/ollama 
 RUN git clone https://github.com/ggerganov/llama.cpp
-RUN pip install -r llama.cpp/requirements.txt
 COPY groups_merged.txt ${HOME}/app/llama.cpp/.
     
-ENV PYTHONPATH=${HOME}/app \
+ENV PYTHONPATH=${APPDIR} \
     PYTHONUNBUFFERED=1 \
     HF_HUB_ENABLE_HF_TRANSFER=1 \
     GRADIO_ALLOW_FLAGGING=never \
@@ -66,5 +66,17 @@ ENV PYTHONPATH=${HOME}/app \
     NVIDIA_VISIBLE_DEVICES=all \
     OLLAMA_HOST=0.0.0.0
 
+WORKDIR ${APPDIR}/ollama
+RUN OLLAMA_CUSTOM_CPU_DEFS="-DGGML_AVX=on -DGGML_AVX2=on -DGGML_F16C=on -DGGML_FMA=on" go generate ./... --verbose \
+    go build . --verbose \
+    go install . --verbose
+    
+WORKDIR ${APPDIR}/llama.cpp
+RUN pip install -r requirements.txt \
+    LLAMA_CUDA=0 make -j llama-quantize --verbose
+    
+WORKDIR ${APPDIR}
+
 # EXPOSE map[11434/tcp:{}]
-ENTRYPOINT /bin/sh start.sh
+ENTRYPOINT ollama serve --verbose & sleep 5 && \
+    python app.py --verbose