paper-with-me

Papers

Bitext Mining for Low-Resource Languages via Contrastive Learning

2022-08-23 · Weiting Tan, Philipp Koehn

Mining high-quality bitexts for low-resource languages is challenging. This paper shows that sentence representation of language models fine-tuned with multiple negatives ranking loss, a contrastive objective, helps retrieve clean bitexts. Experiments show that parallel data mined from our approach substantially outperform the previous state-of-the-art method on low resource languages Khmer and Pashto.

📄 PDF Abstract BibTeX arXiv:2208.11194

Code (1)

steventan0110/align-filter 공식 구현 pytorch

Tasks

Contrastive LearningSentence

Similar Papers 제목 키워드 기반

xSIM++: An Improved Proxy to Bitext Mining Performance for Low-Resource Languages

2023-06-22 · Mingda Chen, Kevin Heffernan, Onur Çelebi, Alex Mourachko 외

We introduce a new proxy score for evaluating bitext mining based on similarity in a multilingual embedding space: xSIM++. In comparison to xSIM, this improved proxy leverages rule-based approaches to extend English sent…

NMT

Bitext Mining Using Distilled Sentence Representations for Low-Resource Languages

2022-05-25 · Kevin Heffernan, Onur Çelebi, Holger Schwenk

Scaling multilingual representation learning beyond the hundred most frequent languages is challenging, in particular to cover the long tail of low-resource languages. A promising approach has been to train one-for-all m…

Cross-Lingual TransferNMTRepresentation LearningSentence

Cross-Lingual Representation Alignment Through Contrastive Image-Caption Tuning

2025-05-19 · Nathaniel Krasner, Nicholas Lanuzo, Antonios Anastasopoulos

Multilingual alignment of sentence representations has mostly required bitexts to bridge the gap between languages. We investigate whether visual information can bridge this gap instead. Image caption datasets are very e…

Natural Language UnderstandingRetrievalSentence

Enhancing Multilingual Embeddings via Multi-Way Parallel Text Alignment

2026-02-25 · Barah Fazili, Koustava Goswami arxiv

Multilingual pretraining typically lacks explicit alignment signals, leading to suboptimal cross-lingual alignment in the representation space. In this work, we show that training standard pretrained models for cross-lin…

Contrastive LearningSemantic Similarity

WikiMatrix: Mining 135M Parallel Sentences in 1620 Language Pairs from Wikipedia

2019-07-10 · EACL 2021 2 · Holger Schwenk, Vishrav Chaudhary, Shuo Sun, Hongyu Gong 외

We present an approach based on multilingual sentence embeddings to automatically extract parallel sentences from the content of Wikipedia articles in 85 languages, including several dialects or low-resource languages. W…

ArticlesSentenceSentence Embeddings