We built an automated crossword solver using state-of-the-art neural network models for open-domain question-answering and assorted techniques
for constraint satisfaction solving. Our system, known as the Berkeley Crossword Solver, is designed to solve American-style crossword puzzles.
These crosswords often involve challenging themes, puns, and world knowledge and typically range in grid size from 15x15 to 21x21.
The demo below illustrates the solving process on several example puzzles from the New York Times. These states are cached, and navigation can
be done with the buttons below. Some puzzles include local search steps at the end. Using your own puzzles is easy -- check the GitHub!