Ruifeng Zheng created SPARK-59152:
-------------------------------------
Summary: Create CountVectorizer broadcast per transform
Key: SPARK-59152
URL: https://issues.apache.org/jira/browse/SPARK-59152
Project: Spark
Issue Type: Sub-task
Components: MLlib
Affects Versions: 5.0.0
Reporter: Ruifeng Zheng
CountVectorizerModel currently retains a broadcast dictionary as model state.
Create the dictionary and its broadcast for each transform call so
transformation-only state follows the returned lazy DataFrame lifecycle, does
not remain attached to the model, and is not reused across SparkContexts.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
