Papers by Jonas Pfeiffer

26 papers
Fine-Tuned Neural Models for Propaganda Detection at the Sentence and Fragment levels (D19-50)

Copied to clipboard

Challenge: The system was evaluated on a unified development set without distributing the gold labels.
Approach: They propose to use fine-grained propaganda detection to build models that can explain why an article is propagandistic.
Outcome: The proposed model performed on all eighteen propaganda techniques in the corpus of the shared task.
How Good is Your Tokenizer? On the Monolingual Performance of Multilingual Language Models (2021.acl-long)

Copied to clipboard

Challenge: Using pretraining data, we find that a designated monolingual tokenizer plays an equally important role in the downstream performance of the model.
Approach: They propose to compare pretrained multilingual models with their monolingual counterparts on a set of five diverse monolingual downstream tasks.
Outcome: The proposed models offer previously unmatched performance in all NLP tasks.
Adapters: A Unified Library for Parameter-Efficient and Modular Transfer Learning (2023.emnlp-demo)

Copied to clipboard

Challenge: Adapters is an open-source library that unifies parameter-efficient and modular transfer learning in large language models.
Approach: They propose to integrate 10 different methods into a unified interface for parameter-efficient and modular transfer learning in large language models.
Outcome: The proposed library is able to perform on multiple NLP tasks and is open-source.
AdapterHub: A Framework for Adapting Transformers (2020.emnlp-demos)

Copied to clipboard

Challenge: AdapterHub framework enables dynamic “stiching-in” of pre-trained adapters for different tasks and languages.
Approach: They propose a framework that allows dynamic "stiching-in" of pre-trained adapters for different tasks and languages.
Outcome: The proposed framework allows dynamic “stiching-in” of pre-trained adapters for different tasks and languages.
M2QA: Multi-domain Multilingual Question Answering (2024.findings-emnlp)

Copied to clipboard

Challenge: Language varies along several axes, most importantly, language instance and domain . lack of evaluation datasets prevents transfer of NLP systems to non-dominant languages .
Approach: They propose a multi-domain multilingual question answering benchmark to explore cross-lingual cross-domain performance of fine-tuned models and state-of-the-art LLMs.
Outcome: The proposed benchmark compared 13,500 SQuAD 2.0-style question-answer instances in German, Turkish, and Chinese for the domains of product reviews, news, and creative writing.
What to Pre-Train on? Efficient Intermediate Task Selection (2021.emnlp-main)

Copied to clipboard

Challenge: Existing methods for fine-tuning intermediate tasks are inefficient and expensive.
Approach: They propose to use a set of 42 intermediate and 11 target English classification, multiple choice, question answering, and sequence tagging tasks to identify the best settings for intermediate transfer learning.
Outcome: The proposed methods achieve an average Regret@3 of 1% across all target tasks.
MAD-X: An Adapter-Based Framework for Multi-Task Cross-Lingual Transfer (2020.emnlp-main)

Copied to clipboard

Challenge: Current deep pretrained models lack capacity to represent all languages . limited capacity is an issue even for high-resource languages where models are not included in training data at all.
Approach: They propose an adapter-based framework that enables high portability and parameter-efficient transfer to arbitrary tasks and languages by learning modular language and task representations.
Outcome: The proposed framework outperforms state-of-the-art models on cross-lingual transfer across languages and typologically diverse models.
AdapterDrop: On the Efficiency of Adapters in Transformers (2021.emnlp-main)

Copied to clipboard

Challenge: Recent approaches to transformer models are expensive to fine-tune, slow for inference, and have large storage requirements.
Approach: They propose a method to remove adapters from transformer layers during training and inference . they show that AdapterDrop can dynamically reduce computational overhead .
Outcome: The proposed approach reduces computational overhead while maintaining performance over multiple tasks with minimal loss of performance.
Romanization-based Large-scale Adaptation of Multilingual Language Models (2023.findings-emnlp)

Copied to clipboard

Challenge: Large multilingual pretrained language models are limited by their vocabulary size and parameter budget.
Approach: They explore the potential of leveraging transliteration on a massive scale to improve performance for multilingual pretrained language models.
Outcome: The proposed transliteration tool outperforms other methods on low-resource languages.
Delving Deeper into Cross-lingual Visual Question Answering (2023.findings-eacl)

Copied to clipboard

Challenge: Existing studies on cross-lingual VQA have reported poor zero-shot transfer performance of current multilingual multimodal Transformers . lack of multilingual resources has hindered development and evaluation of VQA methods beyond the English language .
Approach: They analyze cross-lingual VQA across different question types of varying complexity . they show that simple modifications to the standard training setup can substantially reduce the transfer gap to monolingual English performance.
Outcome: The proposed model significantly reduces the transfer gap to monolingual English performance . the proposed model also improves on question types and languages .
Lifting the Curse of Multilinguality by Pre-training Modular Transformers (2022.naacl-main)

Copied to clipboard

Challenge: Recent work on multilingual pre-trained models has focused on pre-training transformers on concatenated corpora of a large number of languages.
Approach: They propose a language-specific module approach that allows for more languages to be trained post-hoc.
Outcome: The proposed model can be pre-trained on multiple languages with no drop in performance .
AdapterHub Playground: Simple and Flexible Few-Shot Learning with Adapters (2022.acl-demo)

Copied to clipboard

Challenge: AdapterHub Playground is an open-access tool for researchers to use pretrained language models without writing a single line of code.
Approach: They propose a tool which allows researchers to leverage pretrained models without writing a single line of code for a variety of NLP tasks.
Outcome: The proposed model can be used for prediction, training and analysis of textual data without writing a single line of code.
MAD-G: Multilingual Adapter Generation for Efficient Cross-Lingual Transfer (2021.findings-emnlp)

Copied to clipboard

Challenge: Massively multilingual transformers (MMTs) have benefited from additional training of language-specific adapters, but this approach is not viable for the vast majority of languages due to limitations in their corpus size or compute budgets.
Approach: They propose a multilingual ADapter generation approach which contextually generates language adapters from language representations based on typological features.
Outcome: The proposed method improves cross-lingual transfer performance on part-of-speech tagging, dependency parsing, and named entity recognition tasks while remaining cost-effective.
mmT5: Modular Multilingual Pre-Training Solves Source Language Hallucinations (2023.findings-emnlp)

Copied to clipboard

Challenge: Recent large language models display surprising multilingual capabilities despite being pre-trained on English data.
Approach: They propose a multilingual sequence-to-sequence model that disentangles language-specific information from language-agnostic information.
Outcome: The proposed model outperforms existing models on representative natural language understanding and generation tasks in 40+ languages.
Where’s the Point? Self-Supervised Multilingual Punctuation-Agnostic Sentence Segmentation (2023.acl-long)

Copied to clipboard

Challenge: Prior sentence segmentation tools rely on punctuation or require a large amount of training data . a new method for multilingual sentence segmenting is proposed to replace the best prior tools by using only sentence-segmented examples.
Approach: They propose a punctuation-agnostic sentence segmentation method that uses newline characters which implicitly perform segmentation into paragraphs.
Outcome: The proposed method outperforms all prior best sentence segmentation tools by 6.1% F1 points.
MultiCQA: Zero-Shot Transfer of Self-Supervised Text Matching Models on a Massive Scale (2020.emnlp-main)

Copied to clipboard

Challenge: a new study examines the zero-shot transfer capabilities of text matching models on a massive scale.
Approach: They propose to integrate self-supervised with supervised multi-task learning on all available source domains to study the zero-shot transfer capabilities of text matching models on a massive scale.
Outcome: The proposed model outperforms in-domain BERT and the previous state of the art on six benchmarks.
FAMULUS: Interactive Annotation and Feedback Generation for Teaching Diagnostic Reasoning (D19-3)

Copied to clipboard

Challenge: Existing systems for technologyenhanced learning address skills on recalling, explaining, and applying knowledge, e.g., in automatically generated language learning exercises and math word problems.
Approach: They propose to leverage a NLP model to support experts in their further data annotation with automatic suggestions and provide automatic feedback for students.
Outcome: The proposed system improves on two user studies on diagnostic reasoning in medicine and teacher education and can be extended to further use cases.
AdapterFusion: Non-Destructive Task Composition for Transfer Learning (2021.eacl-main)

Copied to clipboard

Challenge: Existing methods for incorporating knowledge from multiple tasks suffer from catastrophic forgetting and difficulties in dataset balancing.
Approach: They propose an algorithm that extracts and combine adapters in a knowledge composition step.
Outcome: The proposed class outperforms traditional methods such as full fine-tuning and multi-task learning on 16 diverse NLU tasks.
xGQA: Cross-Lingual Visual Question Answering (2022.findings-acl)

Copied to clipboard

Challenge: a lack of multilingual multimodal datasets has hindered multimodal vision and language modeling efforts.
Approach: They propose a multilingual evaluation benchmark for the visual question answering task . they extend the established English GQA dataset to 7 typologically diverse languages .
Outcome: The proposed methods outperform current state-of-the-art models in zero-shot cross-lingual settings, but the accuracy remains low across languages.
Retrieve Fast, Rerank Smart: Cooperative and Joint Approaches for Improved Cross-Modal Retrieval (2022.tacl-1)

Copied to clipboard

Challenge: Current approaches to cross-modal retrieval process text and visual input jointly . current approaches are pretrained from scratch and suffer from huge retrieval latency and inefficiency issues .
Approach: They propose a cooperative retrieve-and-rerank framework that turns pretrained text-image multi-modal models into efficient retrieval models.
Outcome: The proposed framework improves retrieval performance over current approaches . it uses twin networks to encode all items of a corpus and a cross-encoder component for a more nuanced ranking .
UKP-SQUARE: An Online Platform for Question Answering Research (2022.acl-demo)

Copied to clipboard

Challenge: Recent advances in NLP and information retrieval have given rise to a diverse set of question answering tasks that are of different formats and require different model architectures and setups.
Approach: They propose an extensible online QA platform that allows users to query and analyze a large collection of modern Skills via a user-friendly web interface and integrated behavioural tests.
Outcome: The proposed tool allows users to query and analyze a large collection of modern Skills via a user-friendly web interface and integrated behavioural tests.
CompoundPiece: Evaluating and Improving Decompounding Performance of Language Models (2023.emnlp-main)

Copied to clipboard

Challenge: Currently, there is no dataset containing compound and non-compound words across languages . however, current LLMs perform poorly on words tokenized unfavorably by subword tokenization.
Approach: They propose to use a Wiktionary dataset to evaluate large language models on decompounding . they find that current LLMs perform poorly on words tokenized unfavorably .
Outcome: The proposed model outperforms the best unsupervised models by 13.9% accuracy on average.
Modular and Parameter-Efficient Fine-Tuning for NLP Models (2022.emnlp-tutorials)

Copied to clipboard

Challenge: State-of-the-art language models in NLP perform best when fine-tuned even on small datasets.
Approach: They provide an overview of parameter-efficient fine-tuning methods and highlight similarities and differences . they highlight benefits and usage scenarios of a neglected property of parameter efficient models .
Outcome: This paper provides an overview of parameter-efficient fine-tuning methods . it highlights similarities and differences by presenting them in a unified view .
UNKs Everywhere: Adapting Multilingual Language Models to New Scripts (2021.emnlp-main)

Copied to clipboard

Challenge: Massively multilingual language models offer state-of-the-art cross-lingual transfer performance on a range of NLP tasks, but there is a profound performance gap between resource-rich and resource-poor target languages.
Approach: They propose a series of data-efficient methods that enable quick and effective adaptation of pretrained multilingual models to low-resource languages and unseen scripts.
Outcome: The proposed methods improve learning of the new dedicated embedding matrix in the target language and for low-resource languages written in unseen scripts.
FUN with Fisher: Improving Generalization of Adapter-Based Cross-lingual Transfer with Scheduled Unfreezing (2024.naacl-long)

Copied to clipboard

Challenge: Standard fine-tuning of language models suffers with generalization to distribution shifts.
Approach: They propose to use Fisher Information to investigate scheduled unfreezing algorithms for adapter-based cross-lingual task transfer to improve generalization to distribution shifts.
Outcome: The proposed method achieves an average of 2 points improvement over four datasets compared to standard fine-tuning and provides empirical evidence for a theory-based justification of the proposed method.
Smelting Gold and Silver for Improved Multilingual AMR-to-Text Generation (2021.emnlp-main)

Copied to clipboard

Challenge: Recent work on multilingual AMR-to-text generation has focused on data augmentation strategies that utilize generated silver AMRs, but this assumes a high quality of generated AMR.
Approach: They propose to combine gold AMR with silver AMRs to generate multilingual AMR annotations.
Outcome: The proposed models outperform the current state of the art for German, Italian, Spanish, and Chinese by a large margin.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations