TODO.md

Improve:

2. Architectural Improvements:

• Layer Normalization: Implement layer normalization or batch normalization between layers to stabilize learning and potentially improve convergence speed.

README.md

model noticed that blue side is objectively better, even with evened dataset model understood each champ becase of embedding layers model understood what champions to prioritize in draft because of attention layers difference between good draft and good pro play draft

Make utlity "search" functions for processed data

Make utility functions to search data in processed data(search teams like T1 or champs like Yone), calculate win-rate, etc..