Resolve annotators and pipelines consuming lots of driver RAM

[saif-ellafi]

Description

Annotators, specifically the Vivekn Sentiment Analysis, is consuming lots of driver RAM due to standard scala collections containing model information. This becomes a storage both inside pipelines and when reading the models back. We need to let this information flow from disk instead

Expected Behavior

Current Behavior

Possible Solution

Steps to Reproduce

Context

Your Environment

• Version used:
• Browser Name and version:
• Operating System and version (desktop or mobile):
• Link to your project:

[aleksei-ai]

@saifjsl Please try my fix with Kryo Serialization. Let me know is it works for you https://github.com/JohnSnowLabs/spark-nlp/tree/vivekn_sentiment_model_kryo_serialization

[saif-ellafi]

#78