520-release-candidate

[maziyarpanahi]

• ONNX support for T5
• use_cache support for newly exported T5 in TF (to make performance close to ONNX version)
• ONNX support for Marian
• new notebook fro structure streaming
• fix a bug for Spark NLP version in Scala
• fix Whisper models from being downloaded due to bad lib version
• Fixing chunk construction when an entity is found [SPARKNLP-937] Fixing chunk construction when an entity is found #14047
• ONNX Support for BertSentenceEmbeddings and XlmRoBertaSentenceEmbeddings SPARKNLP-920: ONNX Support for BertSentenceEmbeddings and XlmRoBertaSentenceEmbeddings #14048
• fix sentence-baed exception in E5 and MPNet - add the missing pool average in both annotators in ONNX SPARKNLP-938 E5 and MPNet embeddings crash on a sentence basis - missing pool average #14051
• add ONNX support to CamemBertForQuestionAnswering, CamemBertForTokenClassification, and CamemBertForSequenceClassification annotators [SPARKNLP-939] Adding ONNX support for CamemBert transformers #14052
• Introducing a new DocumentTokenSplitter annotator SPARKNLP-925 DocumentTokenSplitter #14053
• improve zip util code and add tests for both ZipArchiveUtil and OnnxWrapper improve zip util code and add tests for both ZipArchiveUtil ane OnnxW… #14056
• Refactor ONNX feature and add OnnxSession to broadcast [SPARKNLP-941] Adding OnnxSession to broadcast onnx options #14078
• Update ONNX Runtime to 1.16.3
• Fix and refactor saving best model based on given metrics during NerDL training
• CLIPForZeroShotClassification SPARKNLP-635: CLIPForZeroShotClassification #14083

[danilojsl]

Test in Colab ✅
Test in EMR ✅
Test in Databricks ✅