Seldon Server Versions Save

This release contains a gRPC interface for prediction calls to complement the REST APIs. Details can be found here along with a benchmarking test.

This release contains a breaking change in the prediction REST and JS API calls for the format to pass in data.

There are also changes to the seldon scripts including a new script to start REST and RPC microservices and one to start locust load tests.

• bug fix release for Flask microservices
• scikit example for iris datatset

• Adds initial A/B testing for prediction endpoint.
• Bug fixes for Luigi item-similarity job

• Added new demo: digits recogniser based on tensorflow’s tutorial "deep MNIST for experts"
• seldon-cli updates

A minor release adding a static recommender that pulls precomputed recommendations from a DB.

This a maintenance release

This a maintenance release

Updates to allow use with Kubernetes minikube for local testing

• New sentiment prediction demo
• Easier modification of resources and settings for Tomcat, Mysql for benchmarking

• Updated analytics jobs using Kafka Streams
• New Movielens 10 Million Demo
• Guide to benchmarking Seldon with load testing for production deployment

For more information please read our 1.3.3 release blog post.