Deploy models

Deploy models

2021.02.26 (Fri)


Model Deployment Overview

In addition to Test Inference, a deploy function is available to perform inference on models created with VARISTA.
Deploying allows you to do the following

  • Fast form inference
  • Fast batch inference
  • Inference via API

Deploy List

With any project open, do the following

Sidebar › Deploy

VARISTA Documents Model Deploy 01

Once the deployment is done, it will appear in the deployment list.
Objects that are in the process of being deployed will have an orange indicator, which will change to a green check mark when complete.

Deployment Procedure

To deploy a model, with any project open, do the following

Sidebar › Modeling › Any model › (top right of screen) Deploy

VARISTA Documents Model Report 00

A dialog box for setting the deployment name will appear, so set an arbitrary name.
The number of deployments that can be created depends on the plan, so please check the Pricing Page for details.
VARISTA Documents Model Report 7
After setting the name and selecting OK, it will appear in the Deploy list.

Deployment details

When the deployment is completed, the following functions will be available.

Deploy Info.

VARISTA Documents Model Deploy 02
You can see the user who did the deployment, the date and time, the status, the model information, and stop and resume the deployment.

Form Prediction.

VARISTA Documents Model Deploy 03
In the same way as Test Inference, inference is performed from the browser using the deployed model.
The inference speed is faster than test inference.

Batch Prediction

VARISTA Documents Model Deploy 04
This function allows you to upload files and perform inference continuously, similar to test prediction. Up to 10 files can be uploaded at a time.

__Supported files

  • Number of simultaneous uploads: 10 files
  • Supported files: csv, tsv, xls, xlsx
  • File size: 1024MB / file


VARISTA Documents Model Deploy 05

After deploying the model, a URL will be issued, and you can send a request to this URL using the POST method to perform inference.

  • API URL: The URL of the API. It cannot be changed.
  • API Key: Key information to be used when authenticating the API URL.
  • Project ID: The project ID.
  • Sample Script: A sample script for making a request in Python.
Made with
by VARISTA Team.
© COLLESTA, Inc. 2021. All rights reserved.