Newest papers

From Abstraction to Action: DSLs for Natural Language Abstraction Interpretation and Execution

Amit Lisha, Reut Tsarfaty · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28311 · cited by 0 works (OpenAlex) · open-access PDF

French

L'interprétation des instructions abstraites exige que les modèles identifient à la fois l'abstraction envisagée par le locuteur, et le dérollent dans de nombreuses actions concrètes qui l'instantanent. Les modèles de langue luttent ici: le travail précédent a montré que les noms de la communauté documentés dégradent la performance à mesure que le niveau d'abstraction dans la partie d'instruction augmente. Pour briser l'écart entre l'abstraction du langage naturel et les actions concrètes, nous proposons d'emprunter des environnements de codage la notion d'un langage spécifique du domaine (DSL), une couche de représentation explicite sur laquelle la représentation repose, dont les noms primitifs documentés rec

Machine translation (Meta M2M100); the English original follows.

Abstract

Natural-language instructions often rely on high-level abstractions, such as asking an agent to ``stack blocks in a pyramid.'' Interpreting abstract instructions requires models to both identify the abstraction intended by the speaker, and unroll it into the many concrete actions that instantiate it. Language models struggle here: prior work has shown that performance degrades as the level of abstraction in the instruction rises. To bridge the gap between natural language abstraction and concrete actions, we propose to borrow from coding environments the notion of a {\em domain-specific language} (DSL), an explicit representation layer whose documented primitives name recurring abstractions that expand into concrete actions. We demonstrate the efficacy of DSLs for NL interpretation on \textsc{Hexagons}, a visually-grounded instruction-following task graded for abstraction. We further ask whether language models can induce such DSLs themselves from domain data. Our results show that while code generation consistently improves over direct inference, code generation over a human-curated DSL yields dramatic gains, up to 10 absolute F1 points, reaching the human skyline, with gains concentrated at {\em higher} abstraction levels. We confirm this pattern on the \textsc{Text2CAD} benchmark. At the same time, model-induced DSLs recover only part of this gain, remaining inconsistent across runs and short of their human-curated counterpart. Overall, the right representation closes much of the gap caused by high abstraction --- but inducing that layer of representation automatically is where models still fall short, and where we see a strong opportunity for the community to make progress.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Abstraction, Computer science, Programming language, Interpretation (philosophy), Representation (politics), Abstract interpretation

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange