Question about using the ML training dataset

[triplebeta]

Thanks for creating the dataset of the 110k solution for the Rubik cube and publishing it to Kaggle. When trying to use it I realize that I’m missing some information to do so.

1. The state of the cube is available in 1 line but I don’t know how to map those states to the sides of a cube. I can imagine multiple ways to map them to the 6 sides.

I assume the line contains 6 sequences of 9 colors. From the 5th color I can find out which side to apply it to, but how are they mapped?

1 2 3 1 4 7
4 5 6 or 2 5 8 or otherwise?
7 8 9 3 6 9

2. What’s the initial orientation of the cube when starting to apply the moves? Which color on the upside and which on the front?

It would be great if you could add that information to the dataset on Kaggle.

Thanks in advance,
Jeroen

PS: Sorry for creating an issue for this but I couldn't find another way to get in touch.