Collaboration in Neptune

Neptune helps your team collaborate on ML model building in a few different ways:
​Central ML metadata store:  run ML workflows anywhere and have all the models and experiment results in a single place 
​Shared table for all your ML runs: see, compare, and visualize all the model training your team does 
​Persistent links for Neptune UI:  share any dashboard, table view, or comparison from the UI with a link
​Share buttons: click the Share button to copy, email, or tweet a link to any page in the Neptune UI
To collaborate on private projects in Neptune you need a Team Plan. 
We charge per usage on a Team Plan so that you can invite 2 or 100 people to your Projects and pay the same price. You can try Neptune Team for 14 days for free.
For more, see Neptune plans.  
Central ML metadata store
You and your teammates can run model training on a laptop, cloud environment, or a computation cluster and log all model building metadata to a central metadata store.  
Think of Neptune as a Database + Dashboard that was built for model building metadata. You can log ML metadata from anywhere, organize, display and compare it in the UI, and query anything you want programmatically.     
Central ML metadata Store
You can download models and metadata from your entire team with Neptune API:
import neptune.new as neptune
​
run = neptune.init(
    project='<YOUR_WORKSPACE/YOUR_PROJECT>',
    api_token='<YOUR_API_TOKEN>',
    run='<RUN_ID>' # for example 'SAN-123'
    )
    
# query a field
epoch_nr = run['epoch_nr'].fetch() 
run['model_weights'].download()
For more, see Querying and downloading metadata.
Shared table for all your ML runs
Neptune UI gives your team a shared Runs Table where you can see all the models and experiments your team runs. Specifically:
compare models from many people on your team
see the progress your team is doing at any time     
monitor all active model training runs
Neptune UI was built to be used by many people at the same time (multi-tenant):
Customize Runs Table:  everyone can see what they think is important by choosing columns (like metrics or parameters) and row filters (like metric values).
Save Runs Table Views: everyone can save different Views of the Runs Table to have a quick lookup of best-performing runs or all experiments for a particular model type in saved Views.    
Saving custom Runs Table Views
Persistent links for Neptune UI
You can share the work you do in the Neptune UI with a persistent link. 
Whenever you do something in the Neptune UI, like filter the Runs Table or compare the Runs the Neptune URL changes. When you share this URL with other people it will take them to the very same view of the Neptune UI you are seeing.
Neptune URLs are persistent, which means that they will always work and you can safely use them in other tools and reports. 
Sharing Neptune UI with a persistent link
Share buttons
Similar to persistent links, Share buttons let you send a persistent link to any page in the Neptune UI. 
You can share Run comparisons, live charts of learning curves, hardware consumption dashboards, and more. 
Using Share button in Neptune
What's next?
Querying and downloading metadata
Projects
Neptune UI
Integrations and Supported Tools

You should know - Previous

Connection modes

Next - How-to guides

Experiment tracking

Last modified 1yr ago

Export as PDF

Copy link

Outline

Central ML metadata store

Shared table for all your ML runs

Persistent links for Neptune UI

Share buttons

What's next?