GenerationPrograms: Fine-grained Attribution with Executable Programs

TL;DR

GenerationPrograms introduces a modular, executable framework for text generation that improves attribution accuracy and interpretability by explicitly decomposing the process into distinct, programmable steps.

Contribution

It proposes a novel two-stage generation approach using executable programs, enhancing attribution quality and interpretability over traditional methods.

Findings

Significantly improves attribution quality in long-form QA and summarization tasks.

Outperforms traditional attribution techniques in recovering accurate attributions.

Enables localized refinement through modular program adjustments.

Abstract

Recent large language models (LLMs) achieve impressive performance in source-conditioned text generation but often fail to correctly provide fine-grained attributions for their outputs, undermining verifiability and trust. Moreover, existing attribution methods do not explain how and why models leverage the provided source documents to generate their final responses, limiting interpretability. To overcome these challenges, we introduce a modular generation framework, GenerationPrograms, inspired by recent advancements in executable "code agent" architectures. Unlike conventional generation methods that simultaneously generate outputs and attributions or rely on post-hoc attribution, GenerationPrograms decomposes the process into two distinct stages: first, creating an executable program plan composed of modular text operations (such as paraphrasing, compression, and fusion) explicitly tailored to the query, and second, executing these operations following the program's specified instructions to produce the final response. Empirical evaluations demonstrate that GenerationPrograms significantly improves attribution quality at both the document level and sentence level across two long-form question-answering tasks and a multi-document summarization task. We further demonstrate that GenerationPrograms can effectively function as a post-hoc attribution method, outperforming traditional techniques in recovering accurate attributions. In addition, the interpretable programs generated by GenerationPrograms enable localized refinement through modular-level improvements that further enhance overall attribution quality.