English

A Pseudo-Semantic Loss for Autoregressive Models with Logical Constraints

Machine Learning 2024-01-30 v2 Artificial Intelligence Computation and Language

Abstract

Neuro-symbolic AI bridges the gap between purely symbolic and neural approaches to learning. This often requires maximizing the likelihood of a symbolic constraint w.r.t the neural network's output distribution. Such output distributions are typically assumed to be fully-factorized. This limits the applicability of neuro-symbolic learning to the more expressive autoregressive distributions, e.g., transformers. Under such distributions, computing the likelihood of even simple constraints is #P-hard. Instead of attempting to enforce the constraint on the entire output distribution, we propose to do so on a random, local approximation thereof. More precisely, we optimize the likelihood of the constraint under a pseudolikelihood-based approximation centered around a model sample. Our approximation is factorized, allowing the reuse of solutions to sub-problems, a main tenet for efficiently computing neuro-symbolic losses. Moreover, it is a local, high-fidelity approximation of the likelihood, exhibiting low entropy and KL-divergence around the model sample. We evaluate our approach on Sudoku and shortest-path prediction cast as autoregressive generation, and observe that we greatly improve upon the base model's ability to predict logically-consistent outputs. We also evaluate on the task of detoxifying large language models. Using a simple constraint disallowing a list of toxic words, we are able to steer the model's outputs away from toxic generations, achieving SoTA detoxification compared to previous approaches.

Keywords

Cite

@article{arxiv.2312.03905,
  title  = {A Pseudo-Semantic Loss for Autoregressive Models with Logical Constraints},
  author = {Kareem Ahmed and Kai-Wei Chang and Guy Van den Broeck},
  journal= {arXiv preprint arXiv:2312.03905},
  year   = {2024}
}

Comments

Updated detoxification experiments; moved example toxic generations to Github and added link

R2 v1 2026-06-28T13:43:25.241Z