Application Demo of ML Model CI

Please go to demo website for more details.

Sample applications that can be served and profiled by ML Model CI.

The web application contains two parts (tabs):

• BERT Application on Descriptive Text Sentiment Analysis
• Mask R-CNN Application on Image Object Detection

Quick Start

Download Model

Click the following links to download the models.

• BERT, IMDB Dataset
• Mask R-CNN.

Serve Model by ModelCI

You can build a script to start the server or using command line tool.

Build a Script to Start

Please refer to the ModelCI doc.

Using Script to Start

By using the serve.py in ModelCI, we can start the inference server easily, but you need to register the model first.

python serving.py name --m MRCNN -f tensorflow -e tfs --device cuda:1

The same as BERT model.

Connect to the Web Application

You need to modify the API address in the application source code, to start the services.

Address location:

• BERT
• Mask R-CNN

After all of these, you can start the web application to see the serving and inference results by:

npm install
npm start

Screenshots

BERT Application on Descriptive Text Sentiment Analysis	Mask R-CNN Application on Image Object Detection

For more details about the screenshots.