This is our thesis research, we trained an encoder decoder machine learning model to complete partially drawn shapes.

Constraint Satisfaction Approaches to Wordle: Novel Heuristics and Cross-Lexicon Validation