Wordle Solver

This repository contains my submission, attempts and utilities for an assignment* to build a guesser for (a close cousin of) Wordle. The solver must guess a hidden five-letter word in up to six tries, receiving feedback on each guess in the form of correct letters, misplaced letters, and letters not in the solution.

*First assignment for the exam 20879 Language Technology, from the MSc in AI at Bocconi University, Spring 2025.

DISCLAIMER

The guesser_general.py file has been so far tested only for the current configuration.

Repository Contents

• Assignment_1_Wordle.ipynb
  A Jupyter notebook describing the assignment requirements and some development notes.

• game.py
  A script to run the Wordle game. You can play manually or run multiple rounds in automated mode.

• wordle.py
  The Wordle game logic itself. It checks each guess, provides feedback, and tracks the number of attempts.

• guesser_original.py
  The skeleton Guesser class initially provided by the professor. Contains only minimal logic.

• guesser_submitted.py
  The final version I submitted for evaluation, optimized for speed on smaller test sets.

• guesser_general.py
  A more feature-complete version with additional toggles (dummy guesses, second-guess distinct letters, frequency-based heuristics, etc.). Useful for experimenting with larger wordlists or advanced strategies. Closely inspired by a version kindly shared by fellow student Giacomo Cirò.

• multiple.py
  A script to run multiple tests on random subsets of the dev word list. For each run, it picks a subset of words, writes them to data/r_wordlist.yaml, then invokes game.py for a specified number of rounds. Finally, it collects and prints aggregate statistics (accuracy, average guess length, and run time).

• attempts/
  A folder containing intermediate or alternative attempts and experiments.

• data/
  Contains supporting files such as word lists (wordlist.yaml) or frequency tables (wordlist.tsv).

How to Run

1. Install dependencies (ensure you have yaml, rich, and numpy).

2. From the command line, navigate to this repository’s folder.

3. Run:

   python game.py

   This starts an interactive game where you can guess words manually.

4. To run automated rounds (e.g., 500 rounds), use:

   python game.py --r 500 --p

   This will run 500 games automatically using the selected Guesser class. "--p" will print the results of each round.

Feel free to explore the different guesser implementations and switch them in game.py to compare performance and strategies.