Skip to content

Latest commit

 

History

History
51 lines (49 loc) · 2.45 KB

cloud_setup.md

File metadata and controls

51 lines (49 loc) · 2.45 KB
Raw

Important things to keep in mind

  • Delete GPU from config before starting the instance if GPU isn't necessary for that particular session. This will save a lot of credits.
  • Stop your instance when not in use. Maybe keep a reminder twice a day to stop it (to save credits when instance is not in use).

Redeem Educational Credits

  • Use this link to avail the credits using the coupon

Video Tutorial on Youtube

Create Project

  • Use suitable name
  • Use billing account from above redeemed credits

Create VM

  • Enable Compute Engine API for the project (takes a while..)
  • If you encounter no billing account linked dialog.. just refresh the page. Twitter thread talking about the bug.
  • Create instance using marketplace, select Deep Learning VM.

  • Increase Quota

    • You will encounter a Quota error. Click on link to go to the GPU quotas page.
    • Submit request to increase quota and wait for approval email.
    • Refresh quota page until increased quota is reflected after approval of quota.
  • Use default config for machine specs and GPU..
  • Enable these unticked options
    • Install GPU Drivers automatically..
    • Enable access to Jupyter lab..
  • Deploy
    • Takes a while.. wait

Create Static External Ip

  • VPC Network -> IP Address -> Reserve External Static IP address

Create firewall rule

  • Use network tag of VM for scope
  • Sources: 0.0.0.0/0
  • tcp: 8888

Connect to VM using SSH

  • SSH in browser (easier)
  • SSH from local (involves SSH key setup)

Setup Jupyter

  • jupyter notebook --generate-config
  • vi ~/.jupyter/jupyter_notebook_config.py
  • Add the following at the end of the file 
    c = get_config()
c.NotebookApp.ip = '*' 
c.NotebookApp.open_browser = False 
c.NotebookApp.port = 8888

Run Jupyter in background

  • jupyter notebook >> jup_notebook.log 2>&1 & tail -f jup_notebook.log
  • Access from browser `http://:8888.. token from above method

Clone repository and open notebook

  • git clone https://github.com/ma08/nlp_summer22_hw4.git
  • Open jupyter from the link in previouse section and navigate to nlp_summer22_hw4/Assignment4.ipynb to get started with installing the dependencies which don't come with the Deep Learning VM image and ran a few sample methods related to the hw4.