Plug and Play Your TensorFlow Model with Embeddings

The plug and play use case provides an opportunity for you to train your own models using your own infrastructure and then leverage Abacus.AI infrastructure and abstractions to get your models hosted and running on our scalable service with easy to integrate APIs. The plug and play TensorFlow Model with Embeddings is designed to perform vector matching using your embeddings and models trained using TensorFlow uploaded by you on our system. This document covers how to create your model with embeddings in the correct format usable by our system, how to upload it to the system, and then how you could utilize our services using our APIs and user interface.


Generating Artifacts

Generating TensorFlow model file (.tgz) and embedding files (.csv) in the correct formats to upload to Abacus.AI. Model verification file and default items file could also be uploaded.


Model Deployment and Prediction API

Creating deployment versions and generating predictions from the deployment