Merge pull request #13078 from RasaHQ/prepare-release-3.6.21

Rasa 3.6.21 (2025-01-13)

Bugfixes

• #1424: Replace pickle and joblib with safer alternatives, e.g. json, safetensors, and skops, for
  serializing components.

  Note: This is a model breaking change. Please retrain your model.

  If you have a custom component that inherits from one of the components listed below and modified the persist or
  load method, make sure to update your code. Please contact us in case you encounter any problems.

  Affected components:

  • CountVectorFeaturizer
  • LexicalSyntacticFeaturizer
  • LogisticRegressionClassifier
  • SklearnIntentClassifier
  • DIETClassifier
  • CRFEntityExtractor
  • TrackerFeaturizer
  • TEDPolicy
  • UnexpectedIntentTEDPolicy

3.6.20

Rasa 3.6.20 (2024-04-18)

Miscellaneous internal changes

• #13030, #13031

3.6.19

Rasa 3.6.19 (2024-03-04)

Bugfixes

• #13019: Changed the ordering of returned events to order by ID (previously timestamp) in SQL Tracker Store

3.6.18

Rasa 3.6.18 (2024-02-23)

Bugfixes

• #13017: Flush messages when Kafka producer is closed. This is to ensure that all messages in the producer's internal queue are sent to the broker.

3.6.17

Rasa 3.6.17 (2024-02-13)

Miscellaneous internal changes

• #13007

3.6.16

Rasa 3.6.16 (2024-01-19)

Bugfixes

• #12983: Upgrade Cryptography to fix improper certificate validation.
• #12998: Fixes a bug that caused the full_retrieval_intent_name key to be missing in the published event. Rasa Analytics makes use of this key to get the Retrieval Intent Name

Miscellaneous internal changes

• #712

3.5.17

Rasa 3.5.17 (2023-12-05)

Improvements

• #12851: Added username to the connection parameters for RedisLockStore and RedisTrackerStore
• #1493: Telemetry data is only send for licensed users.

3.6.15

Rasa 3.6.15 (2023-12-01)

Bugfixes

• #12965: Fixed connection timeout to action server by setting KEEP_ALIVE_TIMEOUT to 120, and reverting changes introduced in #12886.

3.6.14

Rasa 3.6.14 (2023-11-17)

Bugfixes

• #12948: Fixed UnexpecTEDIntentlessPolicy training errors that resulted from a change to batching behavior. Changed the batching behavior back to the original for all components. Made the changed batching behavior accessible in DietClassifier using drop_small_last_batch: True.

3.6.13

Rasa 3.6.13 (2023-10-23)

Bugfixes

• #12927: Fix wrong conflicts that occur when rasa validate stories is run with slots that have active_loop set to null in mapping conditions.