Daniel Ke (Front end)
Michael Xu (Front end)
Zhiyuan Li (Back end, AI)
Aiden Lundberg (Pitch, idea)
git clone git@github.com:FakeZhiyuanLi/GPTSense.git
cd GPTSense
python -m venv env
./env/bin/activate
pip install requirements.txt
Created with python version 3.9.12
Tensorflow text classification model trained on 32,000 examples of OpenAI GPT-3.5-Turbo API and 10,000 examples of human responses scraped from blogs
flask run
Enter text into input area (works best between 5-30 lines)
Click submit
Result shows between 0-100% chance of chatGPT response
Percentages above 60% are likely chatGPT, can be false positive at times Percentages above 90% are most likely chatGPT, can also be false positive
Scroll down for a history of past checks
Click on the box to copy the response
Questions database: https://github.com/xinyadu/nqg
This is likely the fault of tensorflow-macos and tensorflow-metal, these are macos specific versions of Tensorflow and do not exist on windows
Solution: delete these modules from requirements.txt and run pip install tensorflow