langcaptcha's Introduction

LangCaptcha

How to run the backend and DB server

In order to use the COMET model in the backend, you first need to download the model. Use the function
```
cd utils && python convert_to_onnx.py
```
The downloading process is going to take a while. When it's done, copy or move the CONVERTED_COMET folder to $PROJECT_ROOT/backend/app.

Spin up a Postgres database and the backend in containers. From the project root directory, run:

cd backend && docker compose up  # use the --build flag if this is your first time spinning up the project

Note: make sure that you don't have any other Postgres running in the same port. You can check it by

lsof -i:5432  # or any other port where you are running the containerized Postgres DB

Make sure that the containers are running using
```
docker ps
```
If this is the first time you set up the environment, (in a new terminal) save the authentication data into the database by running
```
cd ../utils
python preprocess_first_text.py
```
This will create the data table first in the database and populate it with English-Korean parallel corpus data. The table will look like

id en ko

1 text in text in

2 english korean

To set up the second part of the verification, run
```
python setup_second.py
```
This will create the prompt pool data prompt and populate it with three sample prompts.

id text

1 I cannot find my umbrella.

2 Where is the organizer of this event

3 There should be a policy that limits the number of visitors in the school per day.

It will also create a table to store the translations submitted in the second part, second.

id prompt_id translation

In order to check if the tables are created, you can enter the container using the command
```
docker exec -it <postgres-container-id> psql -U postgres -d bytes
```
In order to run the frontend, in another terminal, type in the command
```
npm run dev  # make sure that node is installed
```
and open the website using the address mentioned in the terminal.