Improved Workspace Collaboration for Machine Learning Teams
Check out our new improved Workspace features to help foster even more seamless collaboration among your ML team.
Created on March 5|Last edited on March 6
Comment
Machine learning development is a team sport, as much a coordination and collaboration problem between people and teams as it is a technical problem.
When working on disparate experiments within projects, ML practitioners need to be able to:
- Analyze and evaluate experiments and results using customizable charts and visualizations
- Rely on a system of record that tracks all their experiment results and insights
- Communicate and share those insights with other team members and stakeholders
- Work on their own evaluations in concert with other teammates’ analyses, without disrupting each other’s workflows
To that end, we’re very excited to introduce an improved and enhanced Workspace Collaboration experience in Weights & Biases, including a new Saved Views feature and an updated, more intuitive and accessible UI. These improvements will make it much easier for teammates to share results and work together on analysis, without impinging on personal work and workflows.
Personal Workspaces are exclusive to each user, their very own scratchpad. These customizable spaces provide ample options and opportunities for in-depth analysis of models, tables and data visualizations, giving ML practitioners a powerful tool for evaluating results and deriving insights.
Users can also view other teammate’s workspaces via the new dropdown menu at the top of the Workspace bar. Each user has full edit control over their own workspace only; other teammates can view their workspace configurations but cannot alter them permanently or save changes.
Now, users can save a workspace as a new View and then fork it over to their own workspace for personal evaluation and analysis.
This new Saved Views feature represents collaborative snapshots of the workspace, which are fully viewable and usable by all project collaborators. Saved Views represent a fixed reference of a particular workspace state or configuration, and make it much easier for collective and collaborative review and discussion.
Any user can create a new View or save as a new View, to share specific snapshots that they want teammates to be able to analyze themselves and chime in on. Once users save a new view, that will now appear in the workspace navigation menu, providing seamless navigation between different workspaces and views. Users can also share a View’s url directly to any user across their team or organization.
These new improved Workspace features should greatly improve collaboration across your ML team. Whenever an ML practitioner discovers an interesting insight or needs another set of eyes to poke into their analysis, having Saved Views will make that snapshot and collaborative process much smoother.
Add a comment
Iterate on AI agents and models faster. Try Weights & Biases today.