This repository contains the code for our paper SplitEasy: A Practical Approach for Training ML models on Mobile Devices in a split second.
The entire code is written using React Native. To follow the general guideline for React Native apps follow this link.
There are three parts to this code: loading the dataset on the mobile device, running the models on the mobile device and running the models on the server.
The data files will be available in this link. Move them to your assets folder. The json files are split into 4 parts each with 25 images, so that we can load the image into the phone.
Once the data is in the assets folder, go to App.js and ensure that you load the LoadImageNet component. The code will take some time to load on your device during the first time, once it's done click the load data button. The index of the data being loaded will be logged in your terminal. Then, change the i value in line 58 of imagenet_load.js to i+25 and change the name of the file to the next file. Repeat this process for 5 times and the data is loaded in your phone.
The component to load in your App.js now is the SplitNet component. The main changes to make in this component is to add the server url in the App.js file and final_implementation.js file. Based on the model, load the appropriate file for ModelA and ModelC, the files are availabe in this link. The component will have the button for training the model and once you click it the values are logged in your terminal.
The server code should preferably be run on a server with GPU support. The code requires the argument --model_name in which you mention the name of the architecture. Once you run the code, you can use the IP address of your server to run the javascript code.