Searching Latent Program Spaces
Matthew V Macfarlane, Clement Bonnet

TL;DR
The paper introduces the Latent Program Network (LPN), a neural architecture that learns a latent space of implicit programs, enabling efficient test-time search and adaptation for programming tasks, combining strengths of symbolic and neural methods.
Contribution
LPN is a novel neural architecture that learns a latent program space allowing test-time search without predefined DSLs, improving adaptability and scalability.
Findings
LPN outperforms or matches existing methods on programming-by-examples tasks.
LPN doubles performance on out-of-distribution tasks with test-time search.
LPN effectively learns a compact program space and adapts to new tasks.
Abstract
General intelligence requires systems that acquire new skills efficiently and generalize beyond their training distributions. Although program synthesis approaches have strong generalization power, they face scaling issues due to the large combinatorial spaces that quickly render them impractical, requiring human-generated DSLs or pre-trained priors to narrow this search space. On the other hand, deep learning methods have had high successes, but they lack structured test-time adaptation and rely on heavy stochastic sampling or expensive gradient updates for fine-tuning. In this work, we propose the Latent Program Network (LPN), a novel architecture that builds in test-time search directly into neural models. LPN learns a latent space of implicit programs -- neurally mapping inputs to outputs -- through which it can search using gradients at test time. LPN combines the adaptability of…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
Taxonomy
TopicsSoftware Engineering Research · Software Testing and Debugging Techniques · Model-Driven Software Engineering Techniques
