The maze-30x30-hard-1k used in the Tiny Recursion Models (TRM) paper hosted here for convenience.

[!NOTE] NOTE: Value encoding. The original TRM data uses 1..5. We shift everything down by 1, so final values are 0..4 The mapping is:

space/open cell -> 1 S start -> 2 G goal -> 3 o solution path -> 4

Use seq_len=900 and vocab_size=5 with this dataset.

Generated with the official code by running: python dataset/build_maze_dataset.py --aug # 1000 examples, 8 augmentsand then from… See the full description on the dataset page: https://huggingface.co/datasets/emiliocantuc/maze-30x30-hard-1k.