LM-Debugger: An Interactive Tool for Inspection and Intervention in Transformer-Based Language Models

Mor Geva, Avi Caciularu, Guy Dar, Paul Roit, Shoval Sadde, Micah Shlain, Bar Tamir, Yoav Goldberg

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

10 Scopus citations

Abstract

The opaque nature and unexplained behavior of transformer-based language models (LMs) have spurred a wide interest in interpreting their predictions. However, current interpretation methods mostly focus on probing models from outside, executing behavioral tests, and analyzing salience input features, while the internal prediction construction process is largely not understood. In this work, we introduce LM-Debugger, an interactive debugger tool for transformer-based LMs, which provides a fine-grained interpretation of the model's internal prediction process, as well as a powerful framework for intervening in LM behavior. For its backbone, LM-Debugger relies on a recent method that interprets the inner token representations and their updates by the feed-forward layers in the vocabulary space. We demonstrate the utility of LM-Debugger for single-prediction debugging, by inspecting the internal disambiguation process done by GPT2. Moreover, we show how easily LM-Debugger allows to shift model behavior in a direction of the user's choice, by identifying a few vectors in the network and inducing effective interventions to the prediction process. We release LM-Debugger as an open-source tool and a demo over GPT2 models.

Original languageEnglish
Title of host publicationEMNLP 2022 - 2022 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Demonstrations Session
EditorsWanxiang Che, Ekaterina Shutova
PublisherAssociation for Computational Linguistics (ACL)
Pages12-21
Number of pages10
ISBN (Electronic)9781959429418
StatePublished - 2022
Externally publishedYes
Event2022 Conference on Empirical Methods in Natural Language Processing, EMNLP 2022 - Abu Dhabi, United Arab Emirates
Duration: 7 Dec 202211 Dec 2022

Publication series

NameEMNLP 2022 - 2022 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Demonstrations Session

Conference

Conference2022 Conference on Empirical Methods in Natural Language Processing, EMNLP 2022
Country/TerritoryUnited Arab Emirates
CityAbu Dhabi
Period7/12/2211/12/22

Funding

FundersFunder number
Séphora Berrebi Foundation
Horizon 2020 Framework Programme
European Research Council
Horizon 2020802774
Planning and Budgeting Committee of the Council for Higher Education of Israel

    Fingerprint

    Dive into the research topics of 'LM-Debugger: An Interactive Tool for Inspection and Intervention in Transformer-Based Language Models'. Together they form a unique fingerprint.

    Cite this