Team Collaboration
Collaboration Feature in Grid
When running machine learning workflows, it is beneficial to share and collaborate with fellow
team members. The collaboration feature in Grid platform enables multiple people to view each
other’s runs, sessions and datastores, download artifacts and logs and leverage work done by
others in their team.
Please note: Adding users and credits is not a self-service functionality at this time
Grid.ai needs the following from you to enable collaboration features:
  1. 1.
    Name of the team: What should the team be called
  2. 2.
    Roles: Soon we will have an admin and owner designation, indicate person that would be admin and owner; currently everyone is a user
  3. 3.
    User names (email address or Github username )that will use the collaboration feature as part of a team
    1. 1.
      Please make sure they are registered to use the Grid platform
    2. 2.
      They will need to register the same way as an individual user; using the grid.ai website and perform credit card verification when using GPUs

How to enable the collaboration features?

Once you have provided above information to us, we will notify that the team is created and
collaboration features are now enabled
You don't have to do anything special to enable it.

Using the CLI for collaboration

Start a session or run or create datastore as you would normally in the Grid platform, per docs:
https://docs.grid.ai/start-here/typical-workflow-cli-user
or
https://docs.grid.ai/start-here/typical-workflow-web-user
  • Create/Update/Stop/Pause/Resume/Delete just as you would normally.
  • Datastores are shared globally, so you can use a datastore that is created by someone else in the team
  • Sessions are global: You can access a session via Jupyterlab created by someone else in the team, please be mindful as Github logins are integrated with sessions and github credential of the person that created the session is automatic.
  • Runs are individual: You can see the runs and experiments from everyone in the team.
You can download logs and artifacts from other runs
Use the following command for checking out other sessions in the team:
grid session --global
Your own sessions can be listed with simply
grid session
Use the following command for checking out runs and experiments in the team
grid status --global
Your own status can be listed with simply
grid status
Use the following format to fetch artifacts of an experiment or run created by somebody else in your team
grid artifacts <username>:<experiment_name>
grid artifacts <username>:<run_name>
Use the following format to fetch logs of an experiment created by somebody else in your team
grid logs <username>:<experiment_name>
Use the following command for checking out datastores in the team
grid datastore --global
Use the following command to get information on users in the team
grid teams <team-name>
In Summary, following new CLI commands are available
grid session --global
grid status --global
grid session --global <username>
grid artifacts <username>:<experiment_name>
grid artifacts <username>:<run_name>
grid logs <username>:<experiment_name>
grid datastore --global
grid teams <team-name>

How to use Web application

Web application shows team member's Runs, experiments, sessions, datastores
As part of a collaborative team; following are some things users can do:
  • share, clone each other's runs, experiments
  • download artifacts from any run created by someone
  • access Jupyterlab session created by any member of the team
  • use datastores created by any member of the team to attach to a new session or run
Users cannot stop or delete someone else's runs, interactive sessions or datastores

How do I report a problem

Please contact your business partner in Grid; our support team is available to help as well [email protected]

FAQ

  1. 1.
    Can I delete a session or run or datastore created by someone else?
    1. 1.
      No, you will get a warning message
  2. 2.
    Can I use the Jupyterlab link from my friend's session?
    1. 1.
      Yes
  3. 3.
    Can I attach a datastore created by someone else to my run or session
    1. 1.
      Yes
  4. 4.
    How do I know what is my remaining balance?
    1. 1.
      As individual user part of team, you can use Web UI to see remaining balance
Last modified 1mo ago