Paper
EstLLM: Enhancing Estonian Capabilities in Multilingual LLMs via Continued Pretraining and Post-Training
Authors
Aleksei Dorkin, Taido Purason, Emil Kalbaliyev, Hele-Andra Kuulmets, Marii Ojastu, Mark Fišel, Tanel Alumäe, Eleri Aedmaa, Krister Kruusmaa, Kairit Sirts
Abstract
Large language models (LLMs) are predominantly trained on English-centric data, resulting in uneven performance for smaller languages. We study whether continued pretraining (CPT) can substantially improve Estonian capabilities in a pretrained multilingual LLM while preserving its English and general reasoning performance. Using Llama 3.1 8B as the main base model, we perform CPT on a mixture that increases Estonian exposure while approximating the original training distribution through English replay and the inclusion of code, mathematics, and instruction-like data. We subsequently apply supervised fine-tuning, preference optimization, and chat vector merging to introduce robust instruction-following behavior. Evaluation on a comprehensive suite of Estonian benchmarks shows consistent gains in linguistic competence, knowledge, reasoning, translation quality, and instruction-following compared to the original base model and its instruction-tuned variant, while maintaining competitive performance on English benchmarks. These findings indicate that CPT, with an appropriately balanced data mixture, together with post-training alignment, can substantially improve single-language capabilities in pretrained multilingual LLMs.
Metadata
Related papers
Vibe Coding XR: Accelerating AI + XR Prototyping with XR Blocks and Gemini
Ruofei Du, Benjamin Hersh, David Li, Nels Numan, Xun Qian, Yanhe Chen, Zhongy... • 2026-03-25
Comparing Developer and LLM Biases in Code Evaluation
Aditya Mittal, Ryan Shar, Zichu Wu, Shyam Agarwal, Tongshuang Wu, Chris Donah... • 2026-03-25
The Stochastic Gap: A Markovian Framework for Pre-Deployment Reliability and Oversight-Cost Auditing in Agentic Artificial Intelligence
Biplab Pal, Santanu Bhattacharya • 2026-03-25
Retrieval Improvements Do Not Guarantee Better Answers: A Study of RAG for AI Policy QA
Saahil Mathur, Ryan David Rittner, Vedant Ajit Thakur, Daniel Stuart Schiff, ... • 2026-03-25
MARCH: Multi-Agent Reinforced Self-Check for LLM Hallucination
Zhuo Li, Yupeng Zhang, Pengyu Cheng, Jiajun Song, Mengyu Zhou, Hao Li, Shujie... • 2026-03-25
Raw Data (Debug)
{
"raw_xml": "<entry>\n <id>http://arxiv.org/abs/2603.02041v1</id>\n <title>EstLLM: Enhancing Estonian Capabilities in Multilingual LLMs via Continued Pretraining and Post-Training</title>\n <updated>2026-03-02T16:24:36Z</updated>\n <link href='https://arxiv.org/abs/2603.02041v1' rel='alternate' type='text/html'/>\n <link href='https://arxiv.org/pdf/2603.02041v1' rel='related' title='pdf' type='application/pdf'/>\n <summary>Large language models (LLMs) are predominantly trained on English-centric data, resulting in uneven performance for smaller languages. We study whether continued pretraining (CPT) can substantially improve Estonian capabilities in a pretrained multilingual LLM while preserving its English and general reasoning performance. Using Llama 3.1 8B as the main base model, we perform CPT on a mixture that increases Estonian exposure while approximating the original training distribution through English replay and the inclusion of code, mathematics, and instruction-like data. We subsequently apply supervised fine-tuning, preference optimization, and chat vector merging to introduce robust instruction-following behavior. Evaluation on a comprehensive suite of Estonian benchmarks shows consistent gains in linguistic competence, knowledge, reasoning, translation quality, and instruction-following compared to the original base model and its instruction-tuned variant, while maintaining competitive performance on English benchmarks. These findings indicate that CPT, with an appropriately balanced data mixture, together with post-training alignment, can substantially improve single-language capabilities in pretrained multilingual LLMs.</summary>\n <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n <published>2026-03-02T16:24:36Z</published>\n <arxiv:primary_category term='cs.CL'/>\n <author>\n <name>Aleksei Dorkin</name>\n </author>\n <author>\n <name>Taido Purason</name>\n </author>\n <author>\n <name>Emil Kalbaliyev</name>\n </author>\n <author>\n <name>Hele-Andra Kuulmets</name>\n </author>\n <author>\n <name>Marii Ojastu</name>\n </author>\n <author>\n <name>Mark Fišel</name>\n </author>\n <author>\n <name>Tanel Alumäe</name>\n </author>\n <author>\n <name>Eleri Aedmaa</name>\n </author>\n <author>\n <name>Krister Kruusmaa</name>\n </author>\n <author>\n <name>Kairit Sirts</name>\n </author>\n </entry>"
}