-
Notifications
You must be signed in to change notification settings - Fork 4
/
Dockerfile
40 lines (34 loc) · 1.34 KB
/
Dockerfile
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
FROM continuumio/miniconda3
WORKDIR /root
ARG CONDA_DIR=/opt/conda
ENV PATH $CONDA_DIR/bin:$PATH
########### OPTION FOR RUNNING LOCALLY WITH JUPYTER NOTEBOOK -- 2.35GB (1.89GB without jupyter notebook)
#
RUN conda install --yes --quiet -c conda-forge jsonnet
RUN conda install --yes --quiet nomkl
RUN conda install --yes --quiet -c conda-forge pytorch-cpu
RUN conda install --yes --quiet --freeze-installed matplotlib
RUN conda install --yes --quiet --freeze-installed jupyter \
&& mkdir /opt/notebooks
RUN conda clean -afy \
&& find /opt/conda/ -follow -type f -name '*.a' -delete \
&& find /opt/conda/ -follow -type f -name '*.pyc' -delete \
&& find /opt/conda/ -follow -type f -name '*.js.map' -delete
#
###########
########### OPTION FOR PACKAGING AS A CUSTOM CONTAINER FOR GCLOUD AI PLATFORM HYPERTUNE
#
# RUN conda install --yes --quiet -c conda-forge pytorch-cpu
# RUN pip install google-cloud-storage
# RUN pip install cloudml-hypertune
# # Path configuration
# ENV PATH $PATH:/root/tools/google-cloud-sdk/bin
# # Make sure gsutil will use the default service account
# RUN echo '[GoogleCompute]\nservice_account = default' > /etc/boto.cfg
# #Sets up the entry point to invoke the trainer.
# ENTRYPOINT ["python", "./trainer/task_hyperparam_search.py"]
#
###########
COPY . /root/echo
WORKDIR /root/echo
RUN pwd