Weekly Release

• Improvements to training error messages in the UI
• Improvements for policy serving (but the feature is still turned off in production)

Weekly release - Enhanced error messages and code cleanup

• Enhanced error messages phase 1
• Policy server is disabled on prod with a feature flag
• Bug fixes and code cleanup

Weekly Update - AL 8.7.5

Highlights include:

• Support for AL 8.7.5
• Much more responsive UI for experiment lists and experiment view for newly run experiments.

Weekly Update

• Stripe integration changed(Cancel date)
• Turn off early stopper (at 50 iteration)
• Restrict Action Masking for free user

Thanks @kepricon for doing the release this week.

Weekly Update

Many bug fixes and improved speed for experiments UI

Weekly Update - Pro Subscriptions, Ray 1.3, Call center demo

This is another big release. This finally enables the self-serve subscription page.

Free users get unlimited experiments, but only one can be running at a time. They also can't run models with action masking.

After months of debugging, we finally upgraded from Ray 1.0.1 to 1.3. In theory this makes resuming training smoother and enables other features like PB2 (not enabled yet)

Big Update - New AnyLogic Version, New NativeRL

Releases were blocked for a few weeks while we tried debugging any version of Ray > 1.0. This is X weeks of commits and improvements (258 + 16 commits) that are being released.

I'll be adding to this list, but we can start with:

• Support PLE uploads for AL 8.7.4

Hotfix - Fix Chrome 90 CSS conflict bug #3061

A small change for a big bug that broke all confirmation dialogs in our app. #3061

Hotfix - Add support for AnyLogic 8.7.3

Small update to support the latest version of AnyLogic.

Hotfix - Increase limit for more agents and observations

This allows Felipe and others to train models with more agents and observations.

#2939 increases the column limit for the CSV Parser