Welcome to Huxli's Open Source AI Detector!

quick tour:

- Detector.py - where we run our model and predictions, calculate perplexity   

- gather_data.py - where we prepare our dataset for model training

- train_model.py - where we actually run our model

- runPrediction.py - where you can actually use this data

- model-5.pkl - our serialized Model, you can replace this with your own!

- results_1.csv - relevant data for training a model

- calculated.csv - contains our real and fake passages. values here are outdated and it is not recommended to use them with provided models(will be very inaccurate)

Getting Started:

1. Install dependences pip3 install -r requirements.txt
2. change text in runPredictions.py to text you want to test
3. run python3 runPrediction.py

System Requirements:

- At least 3gb of disk space
- at least 10gb of ram
- While this runs on CPU a GPU is recommended
- python
- torch was built using a CPU version for linux for python 3.10. if you use a different version of python or operating system you will have to change     requirements.txt
- A virtual environment is recommended for dependencies
- Installing dependencies may be better with Anaconda