novastar111/sokoban_eval_3box_certified_deadlock_100 download history

novastar111/sokoban_eval_3box_certified_deadlock_100 is a reinforcement learning dataset on the Hugging Face Hub. In the last 30 days it was downloaded 19 times (5 in the last 7 days), and 59 times in total. It ranks #387,631 among datasets by monthly downloads.

3-box Deadlock This is the exact 100-episode shard used by the finalized UWM Sokoban evaluations. Episode selection is model-independent; no checkpoint output or evaluation score was used to select an episode. Evaluation contract suite: sokoban_paper_final_v2_max40 box

Open novastar111/sokoban_eval_3box_certified_deadlock_100 on Hugging Face

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