Tips

Search Engine

• lookao
• dogedoge

Visual featuremap

• Debugging and Visualisation in PyTorch using Hooks
• visualising-cnn-models-using-pytorch
• pytorch-cnn-visualizations

Scripts

• json2pts

Vscode

• Better Comments
• vscode-icons
• kite | TabNine
• black
• Linting
• Code Runner | Output Colorizer
• vscode-fileheader
• Auto Import
• GitLens
• RemoteHub
• Rainbow Brackets
• indent-rainbow

Shell

• bash Anaconda3-2019.03-Linux-x86_64.sh -u
• sudo mount -t nfs 10.180.2.101:/data0 /data0
• history | grep mount
• ps aux | grep mount
• sudo fuser -v /dev/nvidia*
• sudo chgrp -R ahs data
• sudo chown -R ahs data
• tmux new -s zhangcc
• tmux a -t zhangcc
• (tmux)[http://louiszhai.github.io/2017/09/30/tmux/#新建会话]

Docker

nvidia-docker run -it \
-p 16006:6006 \
-p 10241:10241 \
-p 10243:10243 \
-p 10244:10244 \
-p 10242:22 \
--name=zhangcc \
--shm-size=64G \
--privileged=true \
-v /data/zhangcc/:/data/zhangcc/ \
-v /data0/datasets/:/data0/datasets/ \
-v /data0/zhangcc/:/data0/zhangcc/ \
zccyman/deepframe:stable /bin/bash

nvidia-docker run -it \
-p 16006:6006 \
-p 10221:10221 \
-p 10223:10223 \
-p 10224:10224 \
-p 10222:22 \
--name=Henson \
--shm-size=64G \
--privileged=true \
-v /home/aihuishou/zhangcc/:/home/aihuishou/zhangcc/ \
-v /home/sda/ftp/pub/aihuishou/files/:/home/sda/ftp/pub/aihuishou/files/ \
zccyman/pytorch:base /bin/bash

apt update
apt install openssh-server
mkdir /var/run/sshd
echo 'root:passwd' | chpasswd
sed 's@session\s*required\s*pam_loginuid.so@session optional pam_loginuid.so@g' -i /etc/pam.d/sshd
echo "export VISIBLE=now" >> /etc/profile
vim /etc/ssh/sshd_config，添加：PermitRootLogin yes

cd /root
mkdir .ssh
vim authorized_keys #copy windows id_rsa.pub

service ssh restart

Python

• @staticmethod 静态方法无需实例化, 也可以实例化后调用
• @property 负责把一个方法变成属性调用的

  class Student(object):

    @property
    def score(self):
        return self._score

    @score.setter
    def score(self, value):
        self._score = value

   >>> s = Student()
   >>> s.score = 60 # OK，实际转化为s.set_score(60)
   >>> s.score # OK，实际转化为s.get_score()
       60

Pytorch

• meshgrid

    grid_x = grid_x_tmp = (torch.arange(0, width) + 1).unsqueeze(dim=0)
    for i in range(1, height):
        grid_x = torch.cat([grid_x, grid_x_tmp], dim=0)
    grid_x = grid_x.expand_as(refine_out).cuda(
        device=torch.cuda.current_device()).float()

    grid_y = grid_y_tmp = (torch.arange(0, height) + 1).unsqueeze(dim=1)
    for i in range(1, width):
        grid_y = torch.cat([grid_y, grid_y_tmp], dim=1)
    grid_y = grid_y.expand_as(refine_out).cuda(
        device=torch.cuda.current_device()).float()

    grid_x, grid_y = torch.meshgrid(
        [torch.arange(0, height) + 1, torch.arange(0, width) + 1])
    grid_x = grid_x.expand_as(refine_out).cuda(
        device=torch.cuda.current_device()).float()
    grid_y = grid_y.expand_as(refine_out).cuda(
        device=torch.cuda.current_device()).float()