A Machine Learning Approach That Beats Large Rubik's Cubes. A subset of DeepCubeA's dataset hardest scrambles.

This dataset contains a subset of 16 hardest scrambles from the DeepCubeA dataset that were not solved optimally with our approach (see research paper "A Machine Learning Approach That Beats Large Rubik's Cubes").

Files:
- data.pt – A 2D Torch tensor (Int8) where each row represents a scrambled 3x3x3 Rubik’s Cube state. Values in [0, 6) correspond to face colors.
- generators.json – Defines permutation actions and their symbolic names used in searcher.py.

Optimal solution lengths for the scrambles in this dataset:  
[20, 20, 20, 21, 20, 20, 20, 20, 19, 20, 20, 19, 21, 20, 20, 20].