Discovering Interpretable Algorithms by Decompiling Transformers to RASP

Huang, Xinting; Bakalova, Aleksandra; Bhattamishra, Satwik; Merrill, William; Hahn, Michael

by Xinting Huang, Aleksandra Bakalova, Satwik Bhattamishra, William Merrill, Michael Hahn

Reference:

Discovering Interpretable Algorithms by Decompiling Transformers to RASPXinting Huang, Aleksandra Bakalova, Satwik Bhattamishra, William Merrill, Michael HahnThe Forty-Third International Conference on Machine Learning (ICML), 2026.

Bibtex Entry:

@inproceedings{huang2026discoveringinterpretablealgorithmsdecompiling,
      title={Discovering Interpretable Algorithms by Decompiling Transformers to RASP},
      author={Xinting Huang and Aleksandra Bakalova and Satwik Bhattamishra and William Merrill and Michael Hahn},
      year={2026},
      booktitle={The Forty-Third International Conference on Machine Learning (ICML)},
      url={https://arxiv.org/abs/2602.08857},
      month={May},
      blogpost={https://lacoco-lab.github.io/home/decompiling_transformers/}
}