Reproduce CVPR 2018 LearningToCompare_FSL environment on Moment Pool Cloud

This is a few-shot learning paper from CVPR 2018: Learning to Compare: Relation Network for Few-Shot Learning

Source address: https://github.com/floodsung/LearningToCompare_FSL

The environment uses Tensorflow 1.4 because it is cuda8.

switch conda source

 bash /public/script/switch_conda_source.sh

Create a virtual python environment

 conda create -n py27 python=2.7

 conda deactivate
conda activate py27

Install torch 0.3

The next task is to find the whl installation package for torch 0.3, which I found from the link below

https://download.pytorch.org/whl/cu80/torch_stable.html

Here I am directly pip, just copy the following command.

 pip install https://download.pytorch.org/whl/cu80/torch-0.3.0.post4-cp27-cp27mu-linux_x86_64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple some- package
pip install torchvision==0.2.1
pip install matplotlib scipy

 pip list

Pull the github library

 git clone https://github.com/brendenlake/omniglot.git

I use a github mirror here to complete

 git clone https://hub.fastgit.org/floodsung/LearningToCompare_FSL.git
cd LearningToCompare_FSL/
ls

Unzip the file and test run

 cd /LearningToCompare_FSL/datas
unzip omniglot_28x28.zip
cd /LearningToCompare_FSL/omniglot
python omniglot_train_one_shot.py -w 5 -s 1 -b 19 

Check if gpu is used

 nvidia-smi -l 5

View article

How to speed up GitHub downloads on Moment Pool Cloud?

What should I do if the download is very slow when executing conda install on Moment Pool Cloud? How to switch sources?

https://pytorch.org/get-started/previous-versions/