Dockerfile.habana

FROM vault.habana.ai/gaudi-docker/1.15.1/ubuntu22.04/habanalabs/pytorch-installer-2.2.0:latest

ENV LANG=en_US.UTF-8

WORKDIR /root/llm-on-ray

COPY ./pyproject.toml .
COPY ./MANIFEST.in .

# Create llm_on_ray package directory to bypass the following 'pip install -e' command
RUN mkdir ./llm_on_ray

RUN pip install -e . && \
    pip install --upgrade-strategy eager optimum[habana] && \
    pip install git+https://github.com/HabanaAI/DeepSpeed.git@1.15.1

# Install vllm habana env
RUN pip install -v git+https://github.com/HabanaAI/vllm-fork.git@cf6952d
# Reinstall ray because vllm downgrades the ray version
RUN pip install "ray>=2.10" "ray[serve,tune]>=2.10"

# Optinal. Comment out if you are not using UI
COPY ./dev/scripts/install-ui.sh /tmp

RUN /tmp/install-ui.sh

RUN sed -i 's/#PermitRootLogin prohibit-password/PermitRootLogin yes/' /etc/ssh/sshd_config && \
    service ssh restart

ENV no_proxy=localhost,127.0.0.1

# Required by DeepSpeed
ENV RAY_EXPERIMENTAL_NOSET_HABANA_VISIBLE_MODULES=1

ENV PT_HPU_LAZY_ACC_PAR_MODE=0

ENV PT_HPU_ENABLE_LAZY_COLLECTIVES=true