Mixture of Languages: Improved Multilingual Encoders Through Language Grouping (2025.emnlp-main)
Copied to clipboard
João Maria Janeiro, Belen Alastruey, Francisco Massa, Maha Elbayad, Benjamin Piwowarski, Patrick Gallinari, Loic Barrault
| Challenge: | Recent work in this field relies on training transformer encoders on a large amount of multilingual data, with all parameters shared across all languages. |
| Approach: | They propose a mixture of languages strategy to pretrain largely multilingual encoders using masked language modeling. |
| Outcome: | The proposed architecture outperforms a dense counterpart, MoE models and public multilingual encoders on downstream tasks while minimizing interference. |
Similar Papers
GradSim: Gradient-Based Language Grouping for Effective Multilingual Training (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies show that not all languages positively influence each other . multilingual training can help in those cases by sharing knowledge across languages . |
| Approach: | They propose a gradient similarity-based language grouping method for multilingual training that is better correlated with cross-lingual model performance. |
| Outcome: | The proposed method leads to the largest performance gains on a multilingual dataset and is better correlated with cross-lingual model performance. |
Disentangling Pretrained Representation to Leverage Low-Resource Languages in Multilingual Machine Translation (2024.lrec-main)
Copied to clipboard
| Challenge: | Multilingual neural machine translation requires an enormous dataset, leaving the low-resource language (LRL) underdeveloped. |
| Approach: | They evaluated five languages using a parallel corpus of 1,000 instances each and found a zero-shot improvement of 7.4 from the baseline score of 7.1 to a score of 15.5 at best. |
| Outcome: | The proposed model improves performance in the linguistically diverse country of Indonesia by 7.4 from baseline score of 7.1 to 15.5 at best. |
Emerging Cross-lingual Structure in Pretrained Language Models (2020.acl-main)
Copied to clipboard
| Challenge: | Recent work has shown that multilingual pretraining works, but is unable to measure these effects. |
| Approach: | They propose to use multilingual masked language modeling to train a model on concatenated text from multiple languages to find universal latent symmetries in embedding spaces. |
| Outcome: | The proposed models can be trained on concatenated text from multiple languages without shared vocabulary or domain similarity. |
Learning Language-Specific Layers for Multilingual Machine Translation (2023.acl-long)
Copied to clipboard
| Challenge: | Multilingual Machine Translation (MNMT) is a promising new approach to improve translation quality between non-English languages. |
| Approach: | They propose a language-specific transformer layer to increase model capacity while keeping computation and parameters constant. |
| Outcome: | The proposed approach improves translation quality by 1.3 chrF (1.5 spBLEU) over not using LSLs on a separate decoder architecture. |
Mitigating Data Scarcity in Semantic Parsing across Languages with the Multilingual Semantic Layer and its Dataset (2024.findings-acl)
Copied to clipboard
Abelardo Carlos Martinez Lorenzo, Pere-Lluís Huguet Cabot, Karim Ghonim, Lu Xu, Hee-Soo Choi, Alberte Fernández-Castro, Roberto Navigli
| Challenge: | Large Language Models (LLMs) have advanced significantly in understanding human text, but semantic representations remain crucial for various applications. |
| Approach: | They introduce a multilingual semantic layer which decouples from disambiguation and external inventories and simplifies the task. |
| Outcome: | The proposed model reduces performance gap between languages and annotators by enabling them to understand semantic relations between concepts in any language. |
Breaking Down Multilingual Machine Translation (2022.findings-acl)
Copied to clipboard
| Challenge: | Multilingual training is an essential ingredient in machine translation systems . but it has different effects in different multilingual settings, such as many-to-one, one-tomany and many- to-many learning . |
| Approach: | They compare multilingual training settings with encoders and decoders initialized by multilingual learning . they find important attention heads for each language pair and compare their correlations during inference . |
| Outcome: | The proposed models outperform the best models for high-resource languages and one-to-many models for low-resourced languages. |
Multilingual Machine Translation: Closing the Gap between Shared and Language-specific Encoder-Decoders (2021.eacl-main)
Copied to clipboard
| Challenge: | State-of-the-art multilingual machine translation relies on a universal encoder-decoder, which requires retraining the entire system to add new languages. |
| Approach: | They propose an encoder-decoder approach that can be extended to new languages by learning their corresponding modules. |
| Outcome: | The proposed approach outperforms the universal encoder-decoder by 3.28 BLEU points on average while allowing to add new languages without retraining the rest of the modules. |
Rethinking Data Mixture for Large Language Models: A Comprehensive Survey and New Perspectives (2026.findings-eacl)
Copied to clipboard
| Challenge: | Existing methods for data mixture improve the generalization capability of large language models (LLMs) on downstream tasks. |
| Approach: | They propose a fine-grained categorization of existing methods and propose three subtypes of offline and online methods. |
| Outcome: | The proposed methods extend beyond offline and online classifications and highlight key challenges in the field of data mixture. |
Multilingual Denoising Pre-training for Neural Machine Translation (2020.tacl-1)
Copied to clipboard
Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer
| Challenge: | Existing approaches to pre-train models focus on only English corpora, but this is not common in machine translation. |
| Approach: | They propose a sequence-to-sequence denoising auto-encoder pre-trained on monolingual corpora . they show that it produces significant performance gains across MT tasks . |
| Outcome: | The proposed model can achieve significant performance gains across a wide variety of MT tasks. |
Multilingual Sequence-to-Sequence Models for Hebrew NLP (2023.findings-acl)
Copied to clipboard
| Challenge: | Recent work on pretrained language models for Hebrew is under-parameterized and under-trained . previous work on pretraining Hebrew LMs focused on encoder-only architectures . |
| Approach: | They propose to use sequence-to-sequence generative architectures to train large LMs in morphologically rich languages such as Hebrew. |
| Outcome: | The proposed model improves on all existing Hebrew NLP benchmarks. |