Paul Xiong

My Jetson AGX Xavier randomly died during my coding-debugging…It is kind of dead but not fully dead so I want to have a bash script automatically reboot it when it is dead.

  1. generate your local ssh key by
$ ssh-gen 

2. copy to local .ssh by

$ cat .ssh/id_rsa.pub >>…

--

--

Let add a new Textbox to Gradio framework, call it Textbo1

  • Add new class Textbo1 in compnents.py:
  • make the new copy name of Textbo1 of the directory named “Textbox”
$ cp -r 
github/my_/gradio/ui/packages/app/src/components/Textbox

github/my_/gradio/ui/packages/app/src/components/Textbo1
$ cp -r /mnt/docker/github/gradio/ui/packages/app/src/components/input/Textbox/mnt/docker/github/gradio/ui/packages/app/src/components/input/Textbo1
  • Edit following file __init__.py:

--

--

4 ways to install

$ pip install gradio
  • from local network:
$ pip install 192.168.1.11/gradio
  • from local machine:
$ pip install ../gradio
  • from local and auto sync modification:
$ pip install -e .

Uninstall for pip installed

$ pip uninstall gradio
  • Uninstall for local installed:
$ rm  /usr/local/lib/python3.8/dist-packages/easy-install.pth
$ rm /usr/local/bin/gradio
$ rm -r /usr/local/lib/python3.8/dist-packages/gradio
$ rm /usr/local/lib/python3.8/dist-packages/gradio.egg-link

Build from source

  • install pnpm, (Note: pnpm latest version won’t work with build gui, use 6.x instead.)
$ wget -qO- https://get.pnpm.io/install.sh | PNPM_VERSION=6.32.11 sh

Add path to .bashrc

export PATH=$PATH:/root/.local/share/pnpm/

then

$ source ~/.bashrc
  • install backend:
$ bash scripts/install_gradio.sh
  • build front end:
$ bash scripts/build_frontend.sh

reference guide:

https://github.com/gradio-app/gradio/blob/main/CONTRIBUTING.md

--

--

As we know, most AI’s coding line is so little when comparing to non-AI’s application… document here for further reference.

Where the model is loaded from

How the inference is executed: source -> result

--

--

Paul Xiong

Paul Xiong

Coding, implementing, optimizing ML annotation with self-supervised learning, TLDR: doctor’s labeling is the 1st priority for our Cervical AI project.