Soft Alignment Objectives for Robust Adaptation of Language Generation

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216224%3A14330%2F23%3A00130912" target="_blank" >RIV/00216224:14330/23:00130912 - isvavai.cz</a>

Result on the web

<a href="https://aclanthology.org/2023.acl-long.492" target="_blank" >https://aclanthology.org/2023.acl-long.492</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.18653/v1/2023.acl-long.492" target="_blank" >10.18653/v1/2023.acl-long.492</a>

Result language

angličtina

Original language name

Soft Alignment Objectives for Robust Adaptation of Language Generation

Original language description

Domain adaptation allows generative language models to address specific flaws caused by the domain shift of their application. However, the traditional adaptation by further training on in-domain data rapidly weakens the model's ability to generalize to other domains, making the open-ended deployments of the adapted models prone to errors. This work introduces novel training objectives built upon a semantic similarity of the predicted tokens to the reference. Our results show that (1) avoiding the common assumption of a single correct prediction by constructing the training target from tokens' semantic similarity can largely mitigate catastrophic forgetting of adaptation, while (2) preserving the adaptation in-domain quality, (3) with negligible additions to compute costs. In the broader context, the objectives grounded in a continuous token similarity pioneer the exploration of the middle ground between the efficient but na"{i}ve exact-match token-level objectives and expressive but computationally- and resource-intensive sequential objectives.

Czech name

—

Czech description

—

Type

D - Article in proceedings

CEP classification

—

OECD FORD branch

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Project

—

Continuities

S - Specificky vyzkum na vysokych skolach

Publication year

2023

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

ISBN

9781959429722

ISSN

0736-587X

e-ISSN

—

Number of pages

17

Pages from-to

8837-8853

Publisher name

Association for Computational Linguistics

Place of publication

Toronto, Canada

Event location

Toronto, Canada

Event date

Jul 9, 2023

Type of event by nationality

WRD - Celosvětová akce

UT code for WoS article

—