Language Model Planners do not Scale, but do Formalizers?
Owen Jiang Cassie Huang Ashish Sabharwal Li Zhang · 2026
Read Language Model Planners do not Scale, but do Formalizers? by Owen Jiang Cassie Huang Ashish Sabharwal Li Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-23844.txt.
Description
Jiang and co-authors argue that language models should write programs, not raw planning files, for hard tasks. Models often turn problem text into PDDL for a solver, but scores drop as problems grow. The authors define unraveling problems, where a short text hides a huge PDDL file. Standard formalizers fail to scale on these. Their higher-order approach has the model write a small program that prints the large PDDL. This separates output length from problem size and improves results.
Catalog record
| CREATOR | Owen Jiang Cassie Huang Ashish Sabharwal Li Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50537 |
| LICENSE | http://creativecommons.org/licenses/by-sa/4.0/ |
| RIGHTS | CC BY SA |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.23844