# From Unaligned to Aligned: Scaling Multilingual LLMs with Multi-Way Parallel Corpora

arXiv: https://arxiv.org/abs/2505.14045 · HTML: https://mlsanity.com/paper/2505.14045/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 29 of its 47 references that are in the map.
- First posted on arXiv 20 May 2025.

## Builds on (most cited first, 29 of 29)

- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Similarity of Neural Network Representations Revisited](https://mlsanity.com/paper/1905.00414/) (2019)
- [LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models](https://mlsanity.com/paper/2403.13372/) (2024)
- [No Language Left Behind: Scaling Human-Centered Machine Translation](https://mlsanity.com/paper/2207.04672/) (2022)
- [COMET: A Neural Framework for MT Evaluation](https://mlsanity.com/paper/2009.09025/) (2020)
- [OLMo: Accelerating the Science of Language Models](https://mlsanity.com/paper/2402.00838/) (2024)
- [Making Monolingual Sentence Embeddings Multilingual using Knowledge Distillation](https://mlsanity.com/paper/2004.09813/) (2020)
- [The FLORES-101 Evaluation Benchmark for Low-Resource and Multilingual Machine Translation](https://mlsanity.com/paper/2106.03193/) (2021)
- [Word Translation Without Parallel Data](https://mlsanity.com/paper/1710.04087/) (2017)
- [Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond](https://mlsanity.com/paper/1812.10464/) (2018)
- [XCOPA: A Multilingual Dataset for Causal Commonsense Reasoning](https://mlsanity.com/paper/2005.00333/) (2020)
- [Aya Model: An Instruction Finetuned Open-Access Multilingual Language Model](https://mlsanity.com/paper/2402.07827/) (2024)
- [ChatGPT Beyond English: Towards a Comprehensive Evaluation of Large Language Models in Multilingual Learning](https://mlsanity.com/paper/2304.05613/) (2023)
- [Not All Languages Are Created Equal in LLMs: Improving Multilingual Capability by Cross-Lingual-Thought Prompting](https://mlsanity.com/paper/2305.07004/) (2023)
- [SIB-200: A Simple, Inclusive, and Big Evaluation Dataset for Topic Classification in 200+ Languages and Dialects](https://mlsanity.com/paper/2309.07445/) (2023)
- [Samanantar: The Largest Publicly Available Parallel Corpora Collection for 11 Indic Languages](https://mlsanity.com/paper/2104.05596/) (2021)
- [When and Why are Pre-trained Word Embeddings Useful for Neural Machine Translation?](https://mlsanity.com/paper/1804.06323/) (2018)
- [Reuse, Don't Retrain: A Recipe for Continued Pretraining of Language Models](https://mlsanity.com/paper/2407.07263/) (2024)
- [Cross-lingual Retrieval for Iterative Self-Supervised Training](https://mlsanity.com/paper/2006.09526/) (2020)
- [LLMs Beyond English: Scaling the Multilingual Capability of LLMs with Cross-Lingual Feedback](https://mlsanity.com/paper/2406.01771/) (2024)
- [Pivot-based Transfer Learning for Neural Machine Translation between Non-English Languages](https://mlsanity.com/paper/1909.09524/) (2019)
- [A Shocking Amount of the Web is Machine Translated: Insights from Multi-Way Parallelism](https://mlsanity.com/paper/2401.05749/) (2024)
- [AdaMergeX: Cross-Lingual Transfer with Large Language Models via Adaptive Adapter Merging](https://mlsanity.com/paper/2402.18913/) (2024)
- [Improving Both Domain Robustness and Domain Adaptability in Machine Translation](https://mlsanity.com/paper/2112.08288/) (2021)
- [Revealing the Parallel Multilingual Learning within Large Language Models](https://mlsanity.com/paper/2403.09073/) (2024)
- [Compositional Zero-Shot Domain Transfer with Text-to-Text Models](https://mlsanity.com/paper/2303.13386/) (2023)
- [MMCR4NLP: Multilingual Multiway Corpora Repository for Natural Language Processing](https://mlsanity.com/paper/1710.01025/) (2017)
- [Mitigating Data Imbalance and Representation Degeneration in Multilingual Machine Translation](https://mlsanity.com/paper/2305.12786/) (2023)
- [DCAD-2000: A Multilingual Dataset across 2000+ Languages with Data Cleaning as Anomaly Detection](https://mlsanity.com/paper/2502.11546/) (2025)

## Built on by (most cited first, 7 of 7)

- [When Meanings Meet: Investigating the Emergence and Quality of Shared Concept Spaces during Multilingual Language Model Training](https://mlsanity.com/paper/2601.22851/) (2026)
- [NiuTrans.LMT: Toward Inclusive and Scalable Multilingual Machine Translation with LLMs](https://mlsanity.com/paper/2511.07003/) (2025)
- [Parallel-SFT: Improving Zero-Shot Cross-Programming-Language Transfer for Code RL](https://mlsanity.com/paper/2604.20835/) (2026)
- [Inference-Time Steering for Cross-Lingual Factual Consistency in LLMs](https://mlsanity.com/paper/2607.19243/) (2026)
- [UPRPRC: Unified Pipeline for Reproducing Parallel Resources -- Corpus from the United Nations](https://mlsanity.com/paper/2509.15789/) (2025)
- [Cross-Lingual Alignment for Decoder-Only Models using MoE Routers](https://mlsanity.com/paper/2610.01921/) (2026)
- [Introducing Code-Switched Contexts to Cognitively-Inspired Bilingual Model Training](https://mlsanity.com/paper/2610.06161/) (2026)

## Abstract (arXiv)

Continued pretraining and instruction tuning on large-scale multilingual data have proven to be effective in scaling large language models (LLMs) to low-resource languages. However, the unaligned nature of such data limits its ability to effectively capture cross-lingual semantics. In contrast, multi-way parallel data, where identical content is aligned across multiple languages, provides stronger cross-lingual consistency and offers greater potential for improving multilingual performance. In this paper, we introduce a large-scale, high-quality multi-way parallel corpus, TED2025, based on TED Talks. The corpus spans 113 languages, with up to 50 languages aligned in parallel, ensuring extensive multilingual coverage. Using this dataset, we investigate best practices for leveraging multi-way parallel data to enhance LLMs, including strategies for continued pretraining, instruction tuning, and the analysis of key influencing factors. Experiments on six multilingual benchmarks show that models trained on multiway parallel data consistently outperform those trained on unaligned multilingual data.