The main task of this competition is to detect fake texts in a given dataset. Each data sample contains two texts - one real and one fake. The data used in this Challenge come from [The Messenger] (https://messenger.eso.org) journal. Importantly, *both* texts in each sample - real (optimal for the recipient, as close as possible to the hidden original text) and fake (more or much more distant from the hidden original text) - have been significantly modified using LLMs.

You have access to:
1) a specially curated dataset designed for this competition,
2) Jupyter notebooks with a baseline solution and a sample submission.

The Impostor Hunt starts now! 🚀