Dockerfile.Standalone

FROM nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04

ENV DEBIAN_FRONTEND=noninteractive \
    PYTHONUNBUFFERED=1 \
    PIP_NO_CACHE_DIR=on \
    SHELL=/bin/bash

SHELL ["/bin/bash", "-o", "pipefail", "-c"]

# Upgrade apt packages and install required dependencies
RUN apt update && \
    apt upgrade -y && \
    apt install -y \
      python3-dev \
      python3-pip \
      python3.10-venv \
      fonts-dejavu-core \
      rsync \
      git \
      git-lfs \
      jq \
      moreutils \
      aria2 \
      wget \
      curl \
      libglib2.0-0 \
      libsm6 \
      libgl1 \
      libxrender1 \
      libxext6 \
      ffmpeg \
      libgoogle-perftools-dev \
      procps && \
    apt-get autoremove -y && \
    rm -rf /var/lib/apt/lists/* && \
    apt-get clean -y

# Set working directory
WORKDIR /workspace

# Create venv
RUN python3 -m venv /workspace/venv

# Install Torch
RUN source /workspace/venv/bin/activate && \
    pip3 install --no-cache-dir torch==2.1.2 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118 && \
    deactivate

# Install LLaVA
RUN git clone /~https://github.com/haotian-liu/LLaVA.git llava && \
    cd /workspace/llava && \
    source /workspace/venv/bin/activate && \
    pip3 install . && \
    pip3 install ninja packaging wheel && \
    pip3 install flash-attn --no-build-isolation && \
    pip3 install transformers==4.37.2 && \
    pip3 install protobuf && \
    deactivate

# Install the worker
RUN cd /workspace && \
    git clone /~https://github.com/ashleykleynhans/runpod-worker-llava.git && \
    cd /workspace/runpod-worker-llava && \
    cd src && \
    source /workspace/venv/bin/activate && \
    pip3 install -r requirements.txt && \
    deactivate

# Download the model
ENV HF_HOME="/workspace"
ENV MODEL="liuhaotian/llava-v1.6-mistral-7b"
RUN source /workspace/venv/bin/activate && \
    python3 /workspace/runpod-worker-llava/src/download_models.py && \
    deactivate

# Copy the Serverless handler
COPY --chmod=755 src/rp_handler.py /workspace/runpod-worker-llava/src/rp_handler.py

# Copy start script
COPY --chmod=755 start_standalone.sh /start.sh

# Start the container
ENTRYPOINT /start.sh