How to permanently install package in GoogleColab using conda?

user_11627002 0
Viewed 3528

I am trying yo use a PyTorch library SparseConvNet (https://github.com/facebookresearch/SparseConvNet) in Google Colaboratory. In order to install it properly, you need to first install Conda, and then using Conda install the SparseConvNet package. Here is the code I am using (following the instructions from scn readme file): 

!wget -c https://repo.continuum.io/archive/Anaconda3-5.1.0-Linux-x86_64.sh
!chmod +x Anaconda3-5.1.0-Linux-x86_64.sh
!bash ./Anaconda3-5.1.0-Linux-x86_64.sh -b -f -p /usr/local

import sys
sys.path.append('/usr/local/lib/python3.6/site-packages/')

!conda install pytorch torchvision cudatoolkit=10.0 -c pytorch
!conda install google-sparsehash -c bioconda
!conda install -c anaconda pillow

!git clone https://github.com/facebookresearch/SparseConvNet.git
!cd SparseConvNet/
!bash develop.sh

When I run this it is working and I can successfully import sparseconvnet package, but I need to do it every time I enter the notebook or restart runtime, and it's taking a lot of time. Is it possible to install these packages permanently?

There is one similar question, and the answer suggest that I should install it on my drive, but I don't know how to do it using conda. Thanks!

pytorch conda google-colaboratory python-3.x python
3 Answers

You can specify the directory for conda to install to using

conda install -p path_to_your_dir

So, you can mount your google drive and conda install there to make it permanent.

user_67290100

The whole environment that Google Colaboratory runs your notebooks is not permanent, it is one of their premises. If you need a persistent environment consider running Jupyter directly on a Google Cloud Compute Engine VM, they have pre-built images with everything configured here or Google Cloud Datalab (which runs on a GCE VM, but is managed)

user_98653930

Unfortunately no. Google colab machine will erase after some time. It is a docker inside, and every time you start the GC it will start a new docker image. But you can connect to your local machine via colab. Check the option on Connect button.

user_58849550
Related