KinGPT variant trained on 13,341,057 unique puzzle positions (FEN + best move pairs).
KinGPT variant trained on 54,681 unique positions generated from 1050 Stockfish 18 self-play games.
KinGPT variant trained on combined dataset of 13,395,738 Woodpecker and Beaver variant positions.
1@misc{tang2026generalizationmemorizationbrittlenesstesting,
2 title={Generalization or Memorization? Brittleness Testing for Chess-Trained Language Models},
3 author={Ethan Tang},
4 year={2026},
5 eprint={2605.17565},
6 archivePrefix={arXiv},
7 primaryClass={cs.AI},
8 url={https://arxiv.org/abs/2605.17565},
9}
