[gustavotoyota]

Very curious how did you overcome the 2048 (or 4000) tokens limit? Did you have to retrain the model in some form?