Dockerfile_train

FROM verlai/verl:base-verl0.5-preview-cu128-cudnn9.8-torch2.7.1-fa2.8.0
RUN apt update && DEBIAN_FRONTEND="noninteractive" apt install -y openssh-server ssh git tmux && apt-get autoremove -y && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

RUN mkdir -p /var/run/sshd &&\
    touch /root/.ssh/authorized_keys


COPY sshd_config /etc/ssh/sshd_config
EXPOSE 22

#RUN sed -i '/numba/d' /etc/pip/constraint.txt
#RUN sed -i '/llvmlite/d' /etc/pip/constraint.txt

ENV TORCH_CUDA_ARCH_LIST="9.0+PTX;10.0+PTX"
ENV VLLM_FA_CMAKE_GPU_ARCHES="90-real;100-real"


RUN pip install accelerate codetiming datasets dill hydra-core tensordict torchdata transformers \
    numpy pandas peft pyarrow>=15.0.0 pybind11 pylatexenc ray[all] wandb meilisearch glances gpustat

RUN pip install git+https://github.com/volcengine/verl.git --no-dependencies

RUN pip install openai vllm
RUN pip install torchmetrics plotly
WORKDIR /workdir
COPY /src .
CMD ["/usr/sbin/sshd", "-D"]