NeurIPS2023

Data Preprocessing, LSTM Model, and Analysis for something.

TODO:

1. Unify single passes and training functions if they're the same (seems to be at least in case of LSTM and NFO, I don't see why we can't get a general training function for all models)
2. Separate dataset and data loader for the CustomSequence (will eff up training code but we'll fix it)
3. Fix the transformer model (needs proper testing)