Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Fixes #24 with newer container base, newer Python required by instructlab-dolomite, fixed version for instructlab/training and use it's requirements, comment out extra packages. #28

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fixes #24 with newer container base, newer Python required by instructlab-dolomite, fixed version for instructlab/training and use it's requirements, comment out extra packages. #28

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
29 changes: 16 additions & 13 deletions training/deepspeed/Containerfile
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,17 +1,20 @@
# Containerfile for running deepspeed training
FROM nvcr.io/nvidia/cuda:12.1.1-cudnn8-devel-ubi9

RUN dnf install -y python python-devel git
RUN python -m ensurepip --upgrade
RUN pip3 install torch==2.1.2 --index-url https://download.pytorch.org/whl/cu121
RUN pip3 install packaging wheel
RUN pip3 install flash-attn==2.5.7
RUN pip3 install deepspeed==0.14.2
RUN pip3 install transformers==4.40.1
RUN pip3 install ipdb jupyterlab gpustat matplotlib hydra-core datasets rich numba
RUN git clone https://github.com/instructlab/training.git
RUN mkdir -p /ilab-data/training_output

# Available versions in
# https://gitlab.com/nvidia/container-images/cuda/blob/master/doc/supported-tags.md#ubi9:
FROM nvcr.io/nvidia/cuda:12.2.2-cudnn8-devel-ubi9
ENV PYTHON_VERSION=3.11
# TODO: adding libaio-devel here could be a good idea for DeepSpeed, but isn't available con UBI9,
# see https://bugzilla.redhat.com/show_bug.cgi?id=1840667:
RUN dnf install -y python${PYTHON_VERSION} python${PYTHON_VERSION}-devel \
python${PYTHON_VERSION}-pip git && \
dnf clean all && \
ln -s /usr/bin/python${PYTHON_VERSION} /usr/local/bin/python
# These aren't required for ilab train, leaving commented out just in case:
#RUN pip3 install ipdb jupyterlab gpustat matplotlib hydra-core datasets rich numba
RUN git clone -b v0.1.0 https://github.com/instructlab/training.git && \
mkdir -p /ilab-data/training_output
WORKDIR /training

RUN pip${PYTHON_VERSION} install --no-cache-dir . && \
pip${PYTHON_VERSION} install --no-cache-dir .[cuda]
CMD ["/bin/bash"]