The following code is only tested on Apple Silicon.

1. Config Git

   1 2 3 4 5

   git config user.name <your_name> git config user.email <your@email.address> git config -l # list git config ssh-keygen -t rsa -C "your@email.address" cat ~/.ssh/id_rsa.pub

   Copy rsa public key to GitHub->Settings->SSH and GPG keys.

2. Install Nodejs

   1. In China, you may consider install cnpm

      1	sudo npm install -g cnpm --registry=https://registry.npm.taobao.org

   2. In the following steps, I will use cnpm, same for npm.

Here is some tips for tensorflow1.x.

1. print tensor

   1 2 3

   self.p = tf.Print(<input tensor>, ['<message>', <printed tensor>], summarize=100) sess.run(model.p)

2. print model weight. find the corresponding op name in tensorboard

Here is some tips for pytorch1.x.

tensor

1
2
3
4
5
6
7
8
9
10
11
12
13
14

>>> import torch
>>> torch.tensor([1]).dtype
torch.int64
>>> torch.tensor([1], dtype=torch.float32).dtype
torch.float32
>>> torch.tensor([1.]).dtype
torch.float32
>>> torch.Tensor([1]).dtype
torch.float32
>>> torch.tensor([True]).dtype
torch.bool
>>> torch.Tensor([True]).dtype
torch.float32
>>> exit()

note: torch.tensor will auto guess the type; torch.Tensor is same with dtype=float32

1. Use lshw (list hardware) to show the computer hardware information.

2. Use uanme -a to show system information.

3. tar

   1 2 3 4 5 6

   tar –xvf file.tar # unzip .tar tar -xzvf file.tar.gz # unzip .gz tar -xjvf file.tar.bz2 # unzip .bz2 tar –xZvf file.tar.Z # unzip .Z unrar e file.rar # unzip .rar unzip file.zip # unzip .zip

4. Show the structure of current directory.

   1. sudo apt install tree
   2. tree -L N show \(N^{th}\) depth.

EasyRec

This is a public repository of recommender. It will be replaced by DeepRec.

1. check local modification

   1. git stash temporally store modification in stash.
   2. git stash pop restore modification from stash.
   3. git checkout . cancel all local modification.
   4. git reset --hard HASH reset to certain HASH commit, losing all local modification.

2. ssh -T git@github.com test your ssh connection with github.

1. install

2. test installation: docker container run hello-world

3. switch docker image repository

   1. sudo mkdir -p /etc/docker

   2. vim /etc/docker/daemon.json Replace the mirror with your own.

      1.    {
              "registry-mirrors": ["https://yxzrazem.mirror.aliyuncs.com"]
            }

   3. sudo systemctl daemon-reload; sudo systemctl restart docker

4. add user to docker group (you can run docker commands without sudo)

   1. sudo groupadd docker
   2. sudo usermod -aG docker $USER
   3. newgrp docker
   4. verify: docker run hello-world

5. restart docker service sudo service docker restart

6. list docker images: docker images or docker image ls

7. list containers: docker ps; list all containers: docker ps -a

8. remove containers: docker rm <container id 1> <container id 2> ...; remove all stopped containers: docker rm $(docker ps -a -q status=exited) or docker container prune

9. run container: docker run <container id>/<image id> <command>; run container with interface: docker run -it <container id>/<image id> bash

cpplint

cpplint is a static code checker for C++, following Google C++ styles.

– check code format

1
2

pip install cpplint
cpplint.py --verbose=3 test.cpp # only consider warning level > 3

C++ code style

This could be complicated. But it will help. Optimization Algorithms on Matrix Manifold.

More details will be post later. (Maybe...)

1. install miniconda in M1
2. create and enter new environment of conda
3. conda install -c conda-forge opencv
   1. opencv will be available in this environment.
4. test opencv: python -c "import cv2; print(cv2.__version__)"