Atomic Inference for NLI with Generated Facts as Atoms

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Stacey, Joe, Minervini, Pasquale, Dubossarsky, Haim, Camburu, Oana-Maria, Rei, Marek
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913525466660864
author Stacey, Joe
Minervini, Pasquale
Dubossarsky, Haim
Camburu, Oana-Maria
Rei, Marek
author_facet Stacey, Joe
Minervini, Pasquale
Dubossarsky, Haim
Camburu, Oana-Maria
Rei, Marek
contents With recent advances, neural models can achieve human-level performance on various natural language tasks. However, there are no guarantees that any explanations from these models are faithful, i.e. that they reflect the inner workings of the model. Atomic inference overcomes this issue, providing interpretable and faithful model decisions. This approach involves making predictions for different components (or atoms) of an instance, before using interpretable and deterministic rules to derive the overall prediction based on the individual atom-level predictions. We investigate the effectiveness of using LLM-generated facts as atoms, decomposing Natural Language Inference premises into lists of facts. While directly using generated facts in atomic inference systems can result in worse performance, with 1) a multi-stage fact generation process, and 2) a training regime that incorporates the facts, our fact-based method outperforms other approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2305_13214
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Atomic Inference for NLI with Generated Facts as Atoms
Stacey, Joe
Minervini, Pasquale
Dubossarsky, Haim
Camburu, Oana-Maria
Rei, Marek
Computation and Language
I.2.7
With recent advances, neural models can achieve human-level performance on various natural language tasks. However, there are no guarantees that any explanations from these models are faithful, i.e. that they reflect the inner workings of the model. Atomic inference overcomes this issue, providing interpretable and faithful model decisions. This approach involves making predictions for different components (or atoms) of an instance, before using interpretable and deterministic rules to derive the overall prediction based on the individual atom-level predictions. We investigate the effectiveness of using LLM-generated facts as atoms, decomposing Natural Language Inference premises into lists of facts. While directly using generated facts in atomic inference systems can result in worse performance, with 1) a multi-stage fact generation process, and 2) a training regime that incorporates the facts, our fact-based method outperforms other approaches.
title Atomic Inference for NLI with Generated Facts as Atoms
topic Computation and Language
I.2.7
url https://arxiv.org/abs/2305.13214