Added support for include_plotlyjs="cdn" in upload() and as_html(). Added expand parameter to stringify_unsupported(), which logs lists of values in a proper folder structure.
Web application
The model registry feature is deprecated. The Models section in the web app and Model objects of the API will be removed in April 2025. You can instead use runs to manage model lifecycle and metadata.
GPU power consumption is a new system metric that's automatically logged in the monitoring namespace. Added docstrings for the pop() method and improved the warning about automatic tracking in interactive sessions.
Single- and multi-run exploration is now unified: When multiple runs are visible, the display tabs dynamically update to show multiple runs at once. We also added new features:
Run groups and quick toggle between list and group mode
Reports for saving and sharing analyses of specific runs
Better support for distributed, async workflows. Enhanced pipeline execution status tracking. Fixed node input parameters not being logged to Neptune if passed as a list.
This release brings the new UI to on-prem. It includes the fresh look and all recently released web app features, such as scatter plot widgets, improved runs table search, keyboard shortcuts, distraction-free mode, and enhanced source-code display. Learn more →
When fetching runs with Project.fetch_runs_table(), we removed the limit where a maximum of 10 000 runs could be returned. We also improved the performance, making the fetching up to twice as fast as before. (Also applies to model-fetching methods: fetch_models_table() and fetch_model_versions_table()).
We changed the behavior of tracking system metrics so that no monitoring namespace is created if all related options are disabled. This fixes an issue with extra fields being created especially if resuming a run multiple times.
Clarified the message shown when exceeding a limit related to the target workspace or project.
Improved the visibility of docstrings in some IDEs when initializing Neptune objects.
We reworked disk usage checking and introduced two related environment variables: NEPTUNE_MAX_DISK_USAGE and NEPTUNE_RAISE_ERROR_ON_DISK_USAGE_EXCEEDED. We also improved handling of non-standard float values, among numerous other fixes and changes.
Increased the version requirement of Plotly and removed the Matplotlib version requirement.
Web application
We made some improvements to the runs table: When modifying a search query, you can now edit the operator and value without changing the other parts. You can also click on a username in the Owner column to instantly create a filter for runs created by that account.
Project owners can now set storage limits for projects individually.
We added the list_fileset_files() function, which you can use to query metadata from FileSet fields. We also fixed a couple of issues: stringify_unsupported() now catches a broader MutableMapping class, and the retry request now works properly if an initial request to the Neptune servers fails.
You can now disable Git tracking by passing git_ref=False when initializing a run. We also added a new environment variable NEPTUNE_DATA_DIRECTORY, which lets you specify where the .neptune folder should be created instead of the current working directory.
Web application
In charts displaying multiple runs, we now identify a run more clearly when you hover over its graph.
Logging dicts and Namespaces that contain an empty string "" key now leads to a warning and the entry being skipped, instead of an exception being raised.
If a workspace invite has not been accepted, admins can now resend the invite. We also added rendering of .patch files (a diff.patch file is logged automatically if there are uncommitted changes).
In the Copy link to project modal, the link copy button now works as expected. We fixed a couple of issues with dashboards: the minimum value of a float series is now displayed correctly, and panning an image works properly in a custom image widget.
We added the invite_to_workspace() function which lets you invite people to workspaces programmatically. We also added support for the newly released PyTorch integration, and improved the docstrings of our API's modules and classes.
When filtering runs or other objects by owner, service accounts are now included in the listed owners. Fixed an issue with search when an object contained more than 10 000 fields of the same type.
Web application
Our redesigned web app is live for everyone. You can read about the changes and next steps on the Neptune website .
We reworked how Git tracking works: You can now specify the path where Neptune should look for a Git repository, or disable the logging of Git information completely.
Python client library
The package name has been updated from "neptune-client" to "neptune" also on conda.
We've been working on a redesign of our web application. You can now opt into the beta version and give it a try. Log in to the app and click the link in the banner, or access it later from the user menu in the top-right corner.
Web application
Improved performance of generating thumbnails for image series.
Fixed an issue with the scrollbar when the legend tooltip is visible in the chart widget.
Web application
Improved rendering of several widgets (charts, image galleries) on a single page. Before this change, large widgets would cause the browser to hang, whereas they now load instantly. Scrolling performance is also smoother, and the local state of widgets is preserved while scrolling.
Fixed an issue where grouped model versions don't expand.
Enabled compression in table view API requests, which improves the loading of the runs table.
With our first major release of the Neptune client library, the "new" API is now officially the main one. See Upgrade the Neptune client library for how to get the new package, and check out the full list of changes in the neptune 1.0 upgrade guide.
Web application
In image widgets, you can now scroll the image if it's larger than the widget preview size. In interactive CSV tables, the data is now rendered properly for column names that contain dots.
Web application
We fixed a couple of issues: a saved dashboard now retains the y-axis scaling if it has been changed to logarithmic, and deleted projects no longer cause the app to crash when viewing usage statistics. We also further tweaked the image gallery.
As a workspace member, you no longer lose access to a project you created if you set the visibility to "workspace" and then back to "private". We also ensured that data is deleted more thoroughly when deleting projects, dashboards, objects, and other things that contain user data.
In our first release of the Neptune-AWS integration, we expose the init_run() function that reads the Neptune API token and project name from AWS Secrets.
We've deprecated the name argument in init_project() as well as all management functions where the name of an existing project or workspace is passed. Use the project or workspace arguments instead:
The image gallery has received further improvements. In the Source code view, we tweaked the Git info a bit – for example, remote URLs are now rendered as clickable links. As for the table view, due to a technical limitation, we now limit the number of indexed characters in String fields to 1024 characters. This means that for runs table search, only the first 1024 characters will be considered.
When you've grouped runs in the table view, you can now bulk-select all runs of a group for comparison. We've also updated the image gallery with multiple enhancements and new features, such as the carousel view mode.
We want to make the log() method more intuitive, so we've released the methods that will eventually replace it: append() and extend(). They're used to create a series of values, appending either a single value or a collection of values at a time, respectively. (It was always possible to append one or multiple values with log(), but we're now separating those into their own methods.)
New integrations
Two new integrations have been released:
The Neptune-skorch integration has been updated to work with the new Neptune API as well as the legacy one.
Integration with ZenML – Neptune is now supported as an Experiment Tracker flavor of the ZenML framework.
Web application
In the model versions table, we fixed an issue where suggested columns would not load properly.
In the runs table, we fixed a Firefox scrollbar issue and disabled the "Apply" button if you go to manage tags but no changes have actually been made. In the run and comparison views, we fixed a bug where unauthorized users would be able to save changes to dashboards. We also tweaked the image gallery a bit.
We updated the create_booster_summary() function to log tree dataframes in CSV instead of HTML format, and changed the default value of the log_trees_as_dataframe parameter to False. We also made some package and import related fixes.
Integrations (multiple)
We made some import fixes and package-related updates to several of our integration libraries: neptune-prophet 0.4.1, neptune-tensorflow-keras 1.2.0, neptune-fastai 0.10.4, neptune-xgboost 0.10.1, neptune-sacred 0.10.1, kedro-neptune 0.1.4, and neptune-sklearn 0.9.7.
Methods that fetch objects in table format (such as fetch_runs_table()) no longer query trashed objects.
The way we detect the entrypoint file, source files, and Git repositories is now more strict and consistent than before.
Due to recent PyTorch Lightning changes, we've removed the jsonschema requirements.
The following alternate ways to install neptune-client now work properly: pip install git+https://github.com/neptune-ai/neptune-client.git and pip install . (after cloning the neptune-client repository).
Ahead of the 1.0 release, we've marked the following as deprecated:
Implicit casting of unsupported types to String when using log() or assign() (a.k.a. =). What does this mean? So far, if you attempt to log a value that is not among the supported neptune.types (such as a tuple) Neptune silently logs it as a string. As of 1.x, you're expected to use str(value) if you want it to be logged as a string; otherwise an exception will be thrown in the case of unsupported types.
Web application
We've revamped the image gallery! To analyze images logged as series, you now have more options for displaying and navigating the samples throughout the training process.
Integration: neptune-tensorflow-keras 1.1.0
We fixed a bug that crashed the integration when the input layer of a neural network didn't have the input_shape parameter defined or the model was not built.
We've improved the Neptune integration with 🤗 Transformers, which was released with version 4.22.0 . For the updated functionality, see the 🤗 Transformers guide.
As part of the management API, you can now move runs or other objects to trash with the trash_objects() function.
Python client library: neptune-client 0.16.7
When resuming a run or other object, we've introduced a new, unified parameter for specifying the ID of the object: with_id.
When fetching a table of objects (for example, with project.fetch_runs_table() you can use the new columns parameter to specify which fields to include as columns in the table, rather than having all fields returned.
Neptune no longer hangs indefinitely if the synchronization thread has stopped when wait() is called. The process also finishes faster when you call stop().
Integration: neptune-tensorflow-keras 1.0.0
We've added a bunch of new functionality to our Keras integration, such as saving the model summary and parameters used for fitting the model. See the integration changelog for details.
You can now create service accounts to automate your workflow. To get started, head to your Neptune workspace and select the Service accounts tab.
Web application
Fixed an issue with scrolling when viewing the metadata of a model version.
We now support emoji! 🎉 You can use them in tags and strings.
Fixed an issue where the counter of compared runs did not reset when removing the last run from comparison.
When filtering projects by owner, the query returns members of workspace projects accurately.
We made it easier to resize columns when comparing images side by side.
When comparing graphs of multiple run groups in the Charts view, you can toggle a switch to see the average of each group represented as one graph. For details, see Comparing grouped runs.
In the table view, sorting and filtering that involves integer comparison now works properly.
Dates are displayed correctly when plotting time on the x axis of charts.
Fixed an issue where the runs table would re-fetch the data when changing something about the columns while using Group by.
Python client library: neptune-client 0.16.5
You can now set the connection mode as an environment variable named NEPTUNE_MODE. To learn more, see Environment variables.
When creating a project through the API, providing a project key is now optional. If omitted, Neptune will generate a key for you based on the project name.
If you set a custom run ID, it's now stored in the system namespace of the run. You can access it normally by using the fetch() method on the "sys/custom_run_id" field.
Web application
When comparing artifacts, the diff view was adjusted to detect and highlight differences of fewer characters than before.
Fixed issue with opening files with upperCase extension.
Friendlier search/filtering options for object size ("sys/size") and model version stage ("sys/stage").
In the People tab of the workspace settings, the project access counter now includes both users who have direct access and users who have access through the "workspace" visibility setting.
We now avoid rendering JSON files with lines too long to preview.
Python client library: neptune-client 0.16.4
Fixed issue with uploading in-memory files lager than 5 MB.
Fixed issue with fetch_extension() not working for File fields.
You can now fetch model objects from a project with the new fetch_models_table() method, and we changed the fetch_runs_table() behavior to match all listed tags instead of any.
There was bug in multi-part upload and an issue with the processing of deletion followed by file or artifact upload in a single batch, both of which are now fixed.
Web application
Table previews are now interactive. In the runs table, we fixed the views dropdown z-index. The "run label", which determines which fields are displayed in legends, was renamed "pinned columns".
We also improved the visual feedback on compare selection, fixed runs table CSV downloads so that they no longer include trashed runs, and fixed fuzzy search exact match being listed first.
Python client library: neptune-client 0.16.2
You now have the option to sync only offline runs inside the .neptune folder. We also fixed the handling of server errors for init_model() and init_model_version().
Ahead of the