It was but that shouldn’t matter, the coolest part about these models is they should eventually be able to figure out a game like wordle themselves. If the user explains how to play the game, it shouldn’t matter when the model was trained