We have renamed Views to Projects in Iterative Studio.
Accordingly, Views dashboard is now called Projects dashboard; View settings are now called Project settings; and so on.
When you create a project by connecting to a Git repository, the project is
added to your Iterative Studio dashboard. You can dive deep into the projects
shown in your Iterative Studio dashboard to explore all the ML experiments,
visualize and compare them, and run new experiments. For this, open the project
by clicking the project name (in this case,
An experiments table for the project will be generated as shown below. This includes metrics, hyperparameters, and information about datasets and models.
The experiments table of the project has different components that show you the complete experiment history as well as enable to you to generate plots, compare experiments, run new experiments, etc. The major components of the table are:
The branches and commits in your Git repository are displayed along with the corresponding models, metrics, hyperparameters, and DVC-tracked files.
The table contains buttons to specify filters and other preferences regarding which commits and columns to display.
You can filter the commits that you want to display by the following fields:
avg_precis greater than
avg_precchanged by more than
0.1compared to the baseline experiment.
Select the columns you want to display and hide the rest.
Additionally, you can click and drag the columns in the table to rearrange them as per your preferences.
If your project is missing some required columns or includes columns that you do not want, refer to the following troubleshooting sections to understand why this may have happened.
The following functionality are available for you to hide irrelevant commits from the table.
Iterative Studio auto-hides irrelevant commits: Iterative Studio identifies commits where metrics, files and hyperparameters did not change and hides them automatically.
Iterative Studio auto-hides commits that contain
[skip studio] in the
commit message: This is particularly useful if your workflow creates
multiple commits per experiment and you would like to hide all those commits
except the final one. Below is an example:
Suppose you submit hyper-parameter changes by creating a Git commit. Your CI job gets invoked when the commit is created, starting the model training process. Suppose your CI job creates a new Git commit with the experiment results at the end of model training. This new Git commit, therefore, contains the new values of the hyper-parameters as well as experiment results (metrics).
In Iterative Studio, you may want to display only this final commit and auto-hide the original commit which contains the the new values of the hyper-parameters but not the experiment results (metrics). In this case, what you should do is add the string
[skip studio]to the commit message when you create the original commit.
Hide commits and branches manually: You can selectively hide commits and
branches. This can be useful if there are commits that do not add much value
in your project. To hide a commit or branch, click on the 3-dot menu next to
the commit or branch name and click on
Hide commit or
Unhide commits: You can unhide commits as needed, so that you don't lose
any experimentation history. To display all hidden commits, click on the
Show hidden commits toggle (refer the above gif).
This will display all hidden commits, with a
hidden (closed eye) indicator.
To unhide any commit, click on the 3-dot menu for that commit and click on
Use this toggle switch to show/hide experiments that you have not selected.
Toggle between absolute values and difference from the first row.
Save your filters or column display preferences so that these preferences remain intact even after you log out of Iterative Studio and log back in later.
The table also contains buttons to visualize, compare and run experiments.