connect_colab.md

How to use colab from local vscode?

TODO: paste portmap.io instructions

1. setup colab VM

a. notebook swift ssh

• open notebook swift ssh on colab,
• make sure runtime type is gpu
• run notebook up to ssh tunnel cell

2. ssh/terminal

0. setup ssh connection ~/.ssh/config

# ============= colab via portmap.io ===============
# ssh -i ~/.ssh/wojtekcz.first.pem wojtekcz.first@wojtekcz-22423.portmap.io -N -R 22423:localhost:22
Host colabp
HostName wojtekcz-22423.portmap.io
User root
Port 22423
StrictHostKeyChecking no
UserKnownHostsFile /dev/null
IdentityFile ~/.ssh/private_key.pem
LocalForward 6006 localhost:6006

b. ssh into colabp

ssh colabp

c. run tmux in iterm

tmux -CC

d. set git user name & email

git config --global user.name "Wojtek Czarnowski"
git config --global user.email "wojtek.czarnowski@gmail.com"

e. run Lang2motion script

cd /content/language2motion.gt
swift run -c release Lang2motion

f. (optional) scp rsa key into colabp

scp ~/.ssh/id_rsa_wcz\@MacBook-Wojtka.local colabp:/root/.ssh

g. (optional) load github credentials

chmod 600 ~/.ssh/id_rsa_wcz\@MacBook-Wojtka.local
ssh-agent bash
ssh-add /root/.ssh/id_rsa_wcz\@MacBook-Wojtka.local

3. vscode integration

0. install extension

• Remote-SSH

a. Remote-SSH: Connect to Host... colabp

b. install extensions (in container)

• CodeLLDB
• Maintained Swift Development Environment

c. File/Open workspace...

/content/language2motion.gt/l2m.code-workspace

4. (optional) tensorbord

a. start tensorboard

cd /content/language2motion.gt/data
tensorboard --bind_all --logdir runs/Lang2motion/