Quantifying the Plausibility of Context Reliance in Neural Machine Translation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sarti, Gabriele, Chrupała, Grzegorz, Nissim, Malvina, Bisazza, Arianna
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910365183377408
author Sarti, Gabriele
Chrupała, Grzegorz
Nissim, Malvina
Bisazza, Arianna
author_facet Sarti, Gabriele
Chrupała, Grzegorz
Nissim, Malvina
Bisazza, Arianna
contents Establishing whether language models can use contextual information in a human-plausible way is important to ensure their trustworthiness in real-world settings. However, the questions of when and which parts of the context affect model generations are typically tackled separately, with current plausibility evaluations being practically limited to a handful of artificial benchmarks. To address this, we introduce Plausibility Evaluation of Context Reliance (PECoRe), an end-to-end interpretability framework designed to quantify context usage in language models' generations. Our approach leverages model internals to (i) contrastively identify context-sensitive target tokens in generated texts and (ii) link them to contextual cues justifying their prediction. We use \pecore to quantify the plausibility of context-aware machine translation models, comparing model rationales with human annotations across several discourse-level phenomena. Finally, we apply our method to unannotated model translations to identify context-mediated predictions and highlight instances of (im)plausible context usage throughout generation.
format Preprint
id arxiv_https___arxiv_org_abs_2310_01188
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Quantifying the Plausibility of Context Reliance in Neural Machine Translation
Sarti, Gabriele
Chrupała, Grzegorz
Nissim, Malvina
Bisazza, Arianna
Computation and Language
Artificial Intelligence
Human-Computer Interaction
Machine Learning
I.2.7
Establishing whether language models can use contextual information in a human-plausible way is important to ensure their trustworthiness in real-world settings. However, the questions of when and which parts of the context affect model generations are typically tackled separately, with current plausibility evaluations being practically limited to a handful of artificial benchmarks. To address this, we introduce Plausibility Evaluation of Context Reliance (PECoRe), an end-to-end interpretability framework designed to quantify context usage in language models' generations. Our approach leverages model internals to (i) contrastively identify context-sensitive target tokens in generated texts and (ii) link them to contextual cues justifying their prediction. We use \pecore to quantify the plausibility of context-aware machine translation models, comparing model rationales with human annotations across several discourse-level phenomena. Finally, we apply our method to unannotated model translations to identify context-mediated predictions and highlight instances of (im)plausible context usage throughout generation.
title Quantifying the Plausibility of Context Reliance in Neural Machine Translation
topic Computation and Language
Artificial Intelligence
Human-Computer Interaction
Machine Learning
I.2.7
url https://arxiv.org/abs/2310.01188