Di Wu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Di Wu has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Christof Monz
Language Technology Lab University of Amsterdam
14 shared publications - Shaomu Tan
University of Amsterdam
7 shared publications - David Stap
Language Technology Lab University of Amsterdam
5 shared publications - Andrew Yates
Johns Hopkins University, Baltimore, Maryland, USA
3 shared publications - Seth Aycock
Language Technology Lab University of Amsterdam
3 shared publications - Yan Meng
Language Technology Lab University of Amsterdam
3 shared publications
Latest Publications
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 688-694
UvA-MT’s Participation in the WMT25 General Translation Shared Task
Di Wu, Yan Meng, Maya Konstantinovna Nachesa, Seth Aycock, Christof Monz. Proceedings of the Tenth Conference on Machine Translation. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 688-694
Open AccessUvA-MT’s Participation in the WMT25 General Translation Shared Task
Di Wu, Yan Meng, Maya Konstantinovna Nachesa, Seth Aycock, Christof Monz. Proceedings of the Tenth Conference on Machine Translation. 2025.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 20435-20451
Please Translate Again: Two Simple Experiments on Whether Human-Like Reasoning Helps Translation
Di Wu, Seth Aycock, Christof Monz. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 20435-20451
Open AccessPlease Translate Again: Two Simple Experiments on Whether Human-Like Reasoning Helps Translation
Di Wu, Seth Aycock, Christof Monz. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 974-983
UvA-MT at WMT25 Evaluation Task: LLM Uncertainty as a Proxy for Translation Quality
Di Wu, Christof Monz. Proceedings of the Tenth Conference on Machine Translation. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 974-983
Open AccessUvA-MT at WMT25 Evaluation Task: LLM Uncertainty as a Proxy for Translation Quality
Di Wu, Christof Monz. Proceedings of the Tenth Conference on Machine Translation. 2025.
- Open Access
Neuron Specialization: Leveraging intrinsic task modularity for multilingual machine translation
Training a unified multilingual model promotes knowledge transfer but inevitably introduces negative interference. Language-specific modeling methods show promise in reducing interference. However, they often rely on heuristics to distribute capacity and struggle to foster cross-lingual transfer via isolated modules. In this paper, we explore intrinsic task modularity within multilingual networks and leverage these observations to circumvent interference under multilingual translation. We show that neurons in the feed-forward layers tend to be activated in a language-specific manner. Meanwhile, these specialized neurons exhibit structural overlaps that reflect language proximity, which progress across layers. Based on these findings, we propose Neuron Specialization, an approach that identifies specialized neurons to modularize feed-forward layers and then continuously updates them through sparse networks. Extensive experiments show that our approach achieves consistent performance gains over strong baselines with additional analyses demonstrating reduced interference and increased knowledge transfer.
Open AccessNeuron Specialization: Leveraging intrinsic task modularity for multilingual machine translation
Training a unified multilingual model promotes knowledge transfer but inevitably introduces negative interference. Language-specific modeling methods show promise in reducing interference. However, they often rely on heuristics to distribute capacity and struggle to foster cross-lingual transfer via isolated modules. In this paper, we explore intrinsic task modularity within multilingual networks and leverage these observations to circumvent interference under multilingual translation. We show that neurons in the feed-forward layers tend to be activated in a language-specific manner. Meanwhile, these specialized neurons exhibit structural overlaps that reflect language proximity, which progress across layers. Based on these findings, we propose Neuron Specialization, an approach that identifies specialized neurons to modularize feed-forward layers and then continuously updates them through sparse networks. Extensive experiments show that our approach achieves consistent performance gains over strong baselines with additional analyses demonstrating reduced interference and increased knowledge transfer.
- Open Access
How Far Can 100 Samples Go? Unlocking Overall Zero-Shot Multilingual Translation via Tiny Multi-Parallel Data
Zero-shot translation aims to translate between language pairs not seen during training in Multilingual Machine Translation (MMT) and is largely considered an open problem. A common, albeit resource-consuming, solution is to add as many related translation directions as possible to the training corpus. In this paper, we show that for an English-centric model, surprisingly large zero-shot improvements can be achieved by simply fine-tuning with a very small amount of multi-parallel data. For example, on the EC30 dataset, we obtain up to +21.7 ChrF non-English overall improvements (870 directions) by using only 100 multi-parallel samples while preserving English-centric translation quality. When investigating the size effect of fine-tuning data and its transfer capabilities, we found that already a small, randomly sampled set of fine-tuning directions is sufficient to achieve comparable improvements. The resulting non-English performance is close to the complete translation upper bound. Even in a minimal setting -- fine-tuning with only one single sample -- the well-known off-target issue is almost completely resolved, explaining parts -- but not all -- of the observed improvements in translation quality.
Open AccessHow Far Can 100 Samples Go? Unlocking Overall Zero-Shot Multilingual Translation via Tiny Multi-Parallel Data
Zero-shot translation aims to translate between language pairs not seen during training in Multilingual Machine Translation (MMT) and is largely considered an open problem. A common, albeit resource-consuming, solution is to add as many related translation directions as possible to the training corpus. In this paper, we show that for an English-centric model, surprisingly large zero-shot improvements can be achieved by simply fine-tuning with a very small amount of multi-parallel data. For example, on the EC30 dataset, we obtain up to +21.7 ChrF non-English overall improvements (870 directions) by using only 100 multi-parallel samples while preserving English-centric translation quality. When investigating the size effect of fine-tuning data and its transfer capabilities, we found that already a small, randomly sampled set of fine-tuning directions is sufficient to achieve comparable improvements. The resulting non-English performance is close to the complete translation upper bound. Even in a minimal setting -- fine-tuning with only one single sample -- the well-known off-target issue is almost completely resolved, explaining parts -- but not all -- of the observed improvements in translation quality.
- Open Access
Beyond Shared Vocabulary: Increasing Representational Word Similarities across Languages for Multilingual Machine Translation
Using a vocabulary that is shared across languages is common practice in Multilingual Neural Machine Translation (MNMT). In addition to its simple design, shared tokens play an important role in positive knowledge transfer, assuming that shared tokens refer to similar meanings across languages. However, when word overlap is small, especially due to different writing systems, transfer is inhibited. In this paper, we define word-level information transfer pathways via word equivalence classes and rely on graph networks to fuse word embeddings across languages. Our experiments demonstrate the advantages of our approach: 1) embeddings of words with similar meanings are better aligned across languages, 2) our method achieves consistent BLEU improvements of up to 2.3 points for high- and low-resource MNMT, and 3) less than 1.0\% additional trainable parameters are required with a limited increase in computational costs, while inference time remains identical to the baseline. We release the codebase to the community.
Open AccessBeyond Shared Vocabulary: Increasing Representational Word Similarities across Languages for Multilingual Machine Translation
Using a vocabulary that is shared across languages is common practice in Multilingual Neural Machine Translation (MNMT). In addition to its simple design, shared tokens play an important role in positive knowledge transfer, assuming that shared tokens refer to similar meanings across languages. However, when word overlap is small, especially due to different writing systems, transfer is inhibited. In this paper, we define word-level information transfer pathways via word equivalence classes and rely on graph networks to fuse word embeddings across languages. Our experiments demonstrate the advantages of our approach: 1) embeddings of words with similar meanings are better aligned across languages, 2) our method achieves consistent BLEU improvements of up to 2.3 points for high- and low-resource MNMT, and 3) less than 1.0\% additional trainable parameters are required with a limited increase in computational costs, while inference time remains identical to the baseline. We release the codebase to the community.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 6506-6527
Neuron Specialization: Leveraging Intrinsic Task Modularity for Multilingual Machine Translation
Shaomu Tan, Di Wu, Christof Monz. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 6506-6527
Open AccessNeuron Specialization: Leveraging Intrinsic Task Modularity for Multilingual Machine Translation
Shaomu Tan, Di Wu, Christof Monz. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 176-184
UvA-MT’s Participation in the WMT24 General Translation Shared Task
Shaomu Tan, David Stap, Seth Aycock, Christof Monz, Di Wu. Proceedings of the Ninth Conference on Machine Translation. 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 176-184
Open AccessUvA-MT’s Participation in the WMT24 General Translation Shared Task
Shaomu Tan, David Stap, Seth Aycock, Christof Monz, Di Wu. Proceedings of the Ninth Conference on Machine Translation. 2024.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 293-297
Representational Isomorphism and Alignment of Multilingual Large Language Models
Di Wu, Yibin Lei, Andrew Yates, Christof Monz. Proceedings of the Fourth Workshop on Multilingual Representation Learning (MRL 2024). 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 293-297
Open AccessRepresentational Isomorphism and Alignment of Multilingual Large Language Models
Di Wu, Yibin Lei, Andrew Yates, Christof Monz. Proceedings of the Fourth Workshop on Multilingual Representation Learning (MRL 2024). 2024.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 14074-14085
Representational Isomorphism and Alignment of Multilingual Large Language Models
Di Wu, Yibin Lei, Andrew Yates, Christof Monz. Findings of the Association for Computational Linguistics: EMNLP 2024. 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 14074-14085
Open AccessRepresentational Isomorphism and Alignment of Multilingual Large Language Models
Di Wu, Yibin Lei, Andrew Yates, Christof Monz. Findings of the Association for Computational Linguistics: EMNLP 2024. 2024.