Mahesh, Unfortunately no. We start to support model training over caches with binary objects from Ignite 2.8.
This support was added by ticket https://issues.apache.org/jira/browse/IGNITE-10700 So, for now, this option is available only for builds from the master branch. Best regards, Yuriy Babak вс, 20 янв. 2019 г. в 16:02, mahesh76private <maheshk...@gmail.com>: > In other words, for ML functions, > > can I use IgniteBiFunction with Key and Value as *BinaryObject*, and at the > same time KEY_TYPE and VALUE_TYPE is not specified while creating the SQL > table cache? > > Unfortunately, in our product, we have data in SQL tables, and from these > we > intend to create datasets and develop some ML functions with it. These SQL > tables are generated dynamically by users. > > > > > > -- > Sent from: http://apache-ignite-users.70518.x6.nabble.com/ >