TU Darmstadt / ULB / TUbiblio

MAD-G: Multilingual Adapter Generation for Efficient Cross-Lingual Transfer

Ansell, Alan ; Edoardo, Maria Ponti ; Pfeiffer, Jonas ; Ruder, Sebastian ; Glavaš, Goran ; Vulić, Ivan ; Korhonen, Anna (2021)
MAD-G: Multilingual Adapter Generation for Efficient Cross-Lingual Transfer.
Conference on Empirical Methods in Natural Language Processing (EMNLP 2021). Punta Cana, Dominican Republic (07.-11.11.2021)
Conference or Workshop Item, Bibliographie

Abstract

Adapter modules have emerged as a general parameter-efficient means to specialize a pretrained encoder to new domains. Massively multilingual transformers (MMTs) have particularly benefited from additional training of language-specific adapters. However, this approach is not viable for the vast majority of languages, due to limitations in their corpus size or compute budgets. In this work, we propose MAD-G (Multilingual ADapter Generation), which contextually generates language adapters from language representations based on typological features. In contrast to prior work, our time- and space-efficient MAD-G approach enables (1) sharing of linguistic knowledge across languages and (2) zero-shot inference by generating language adapters for unseen languages. We thoroughly evaluate MAD-G in zero-shot cross-lingual transfer on part-of-speech tagging, dependency parsing, and named entity recognition. While offering (1) improved fine-tuning efficiency (by a factor of around 50 in our experiments), (2) a smaller parameter budget, and (3) increased language coverage, MAD-G remains competitive with more expensive methods for language-specific adapter training across the board. Moreover, it offers substantial benefits for low-resource languages, particularly on the NER task in low-resource African languages. Finally, we demonstrate that MAD-G’s transfer performance can be further improved via: (i) multi-source training, i.e., by generating and combining adapters of multiple languages with available task-specific training data; and (ii) by further fine-tuning generated MAD-G adapters for languages with monolingual data.

Item Type: Conference or Workshop Item
Erschienen: 2021
Creators: Ansell, Alan ; Edoardo, Maria Ponti ; Pfeiffer, Jonas ; Ruder, Sebastian ; Glavaš, Goran ; Vulić, Ivan ; Korhonen, Anna
Type of entry: Bibliographie
Title: MAD-G: Multilingual Adapter Generation for Efficient Cross-Lingual Transfer
Language: English
Date: 7 November 2021
Publisher: ACL
Book Title: Findings of the Association for Computational Linguistics: EMNLP 2021
Event Title: Conference on Empirical Methods in Natural Language Processing (EMNLP 2021)
Event Location: Punta Cana, Dominican Republic
Event Dates: 07.-11.11.2021
URL / URN: https://aclanthology.org/2021.findings-emnlp.410/
Abstract:

Adapter modules have emerged as a general parameter-efficient means to specialize a pretrained encoder to new domains. Massively multilingual transformers (MMTs) have particularly benefited from additional training of language-specific adapters. However, this approach is not viable for the vast majority of languages, due to limitations in their corpus size or compute budgets. In this work, we propose MAD-G (Multilingual ADapter Generation), which contextually generates language adapters from language representations based on typological features. In contrast to prior work, our time- and space-efficient MAD-G approach enables (1) sharing of linguistic knowledge across languages and (2) zero-shot inference by generating language adapters for unseen languages. We thoroughly evaluate MAD-G in zero-shot cross-lingual transfer on part-of-speech tagging, dependency parsing, and named entity recognition. While offering (1) improved fine-tuning efficiency (by a factor of around 50 in our experiments), (2) a smaller parameter budget, and (3) increased language coverage, MAD-G remains competitive with more expensive methods for language-specific adapter training across the board. Moreover, it offers substantial benefits for low-resource languages, particularly on the NER task in low-resource African languages. Finally, we demonstrate that MAD-G’s transfer performance can be further improved via: (i) multi-source training, i.e., by generating and combining adapters of multiple languages with available task-specific training data; and (ii) by further fine-tuning generated MAD-G adapters for languages with monolingual data.

Uncontrolled Keywords: UKP_p_emergencity, emergenCITY_INF
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Ubiquitous Knowledge Processing
LOEWE
LOEWE > LOEWE-Zentren
LOEWE > LOEWE-Zentren > emergenCITY
Date Deposited: 22 Dec 2021 11:13
Last Modified: 28 Feb 2023 13:53
PPN:
Export:
Suche nach Titel in: TUfind oder in Google
Send an inquiry Send an inquiry

Options (only for editors)
Show editorial Details Show editorial Details