max-rl/maze_17x17_diverse_1.3m download history

max-rl/maze_17x17_diverse_1.3m is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 80 times (15 in the last 7 days), and 310 times in total. It ranks #134,522 among datasets by monthly downloads.

17x17 Maze with Diverse Paths (1.3M, 16 paths/maze) This is v2 of the 17x17 maze SFT corpus used in the MaxRL paper. It is designed to make judge_maze_continuous non-trivially different from the binary reward by ensuring path lengths span a wide range per maze, instead of always being sho

Open max-rl/maze_17x17_diverse_1.3m on Hugging Face

Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.