Jason's toy model for predicting the halo mass in next snapshot from current data. The data I use is the merger tree data in TNG100. Simple 2-layer LSTM network using keras. I only train 1k samples and better results can be achieved using more samples. Future editions will focus on other halo properties.