AIBAT - Adaptive Intelligent Behavioral Assessment Tool

This is a full stack application built with Next.js and Django implementing the adatest library, to allow for a more user-friendly and complete end-to-end experience.

Using an AI grader and generator, users can generate, grade, and review results of adaptive test cases. We use Llama3 - grader and Mistral - generator.

Getting started

Clone the repository

Create a .env file in /backend with the following content:

MODEL=modelname # mistral or openai
HUGGING_FACE_TOKEN=your_hugging_face_token # only needed if modelname=mistral
OPENAI_API_KEY=your_openai_api_key # only needed if modelname=openai

If using Docker:

• run the following command to start the application: docker-compose up
• Stop the application with the following command: docker-compose down

If not using Docker:

Backend:

• in /backend, set up a conda environment and run the following commands to install packages:

pip install --upgrade pip
pip install notebook==6.1.5
pip install -r ${MODEL}_requirements.txt

Frontend:

• in /frontend, run the following commands to install packages:

npm install

Running the application:

• in /scripts, run the following command to run the application:

bash start-application.sh

• in /scripts, run the following command to stop the application:

bash stop-application.sh

Docker or not, the application will be available at http://localhost:8000

Running the Application on the VM

To connect to the vm, first retrieve the ssh key "adatest.pem" from a team member. Then, on a terminal in a folder that contains adatest.pem on your local computer, give it permissions with the command chmod 600 adatest.pem

Next, go to the Lambda Cloud team page to start up an instance (any instance type is fine but A10 is best, Region: Viriginia-East, Attach Adatest Filesystem, and make it use adatest key) Once the instance is launched, grab the ssh command (should look like ssh ubuntu@1280.136.135) copy the address (without the ssh) and replace it with the 'address' in the code below.

Then ssh into the vm with the command:

ssh -i adatest.pem ubuntu@<address> -L localhost:8000:localhost:8000

Once in the vm, you will be able to navigate into the Adatest folder and run the vm_setup script. This installs and give permissions to everything for the app and Docker to run.

cd Adatest/Adatest-App/scripts
bash vm-setup.sh
source ~/.bashrc

In the same directory, stop run the application, run following command:

bash start-application.sh

When done, to stop the application, run the following command:

bash stop-application.sh

If ngrok is set up on the current virtual machine/filesystem, you can run this command to generate a temporary URL:

ngrok http http://localhost:8000

The application will be available at http://localhost:8000

Retrieving Log Files

Once finished, click "End Session", it will save log files to the /Adatest/Adatest-App/backend folder. To retrieve these files, you will need to use the scp command.

In the same directory as the adatest.pem file, run the following command:

scp -i adatest.pem ubuntu@<address>:/home/ubuntu/Adatest/Adatest-App/backend/log.csv <destination in local>

You will need to replace the "address" with the address of the VM and "destination in local" with the location you want to save the log file to on your local computer. This command saves the file "log.csv" to the location you specify,

Repeat for "perturbations.csv" and "tests.csv," using the same command except replacing "log.csv"

In Case vm_setup.sh fails

The VM won't have any packages installed, use the following commands to download docker-compose:

sudo apt update -y
sudo apt install -y docker
sudo service docker start
sudo usermod -a -G docker ubuntu
sudo curl -L "https://github.com/docker/compose/releases/latest/download/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose
sudo chmod +x /usr/local/bin/docker-compose
docker-compose --version

If there is a given docker-compose version it should be fine. Then, in the frontend folder, we have to download npm/node and run npm i To do this, run the following command:

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.5/install.sh | bash

Then, after restarting terminal window, run the following command:

cd /frontend
nvm install node
npm i 

The .env file should already be in the vm filesystem, but in case it isn't, you will need to get that as well from somebody. After doing this and getting your .env file, run

docker-compose up

and the application should be available at http://localhost:8000