@inproceedings{nehrdich-keutzer-2026-mitra,
title = "{MITRA}-{MT}: Continued Pretraining, Parallel Corpus Mining, and Multi-Directional Machine Translation for {S}anskrit, {T}ibetan, and Buddhist {C}hinese",
author = "Nehrdich, Sebastian and
Keutzer, Kurt",
editor = "Haddow, Barry and
Kocmi, Tom and
Koehn, Philipp and
Monz, Christof",
booktitle = "Proceedings of the Eleventh Conference on Machine Translation",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.wmt-main.26/",
pages = "492--506",
ISBN = "979-8-89176-474-3",
abstract = "Buddhist literature is preserved in a network of classical languages, above all Sanskrit, P{\={a}}li, Buddhist Chinese, and Tibetan, for which machine translation support remains limited, and largely restricted to translation into English. In order to address this shortcoming, we present three connected contributions. First, we release MITRA-parallel v2, a corpus of 1.69 million automatically aligned parallel records (2.34 million segment pairs) between Sanskrit, Tibetan, and Buddhist Chinese, mined with an embedding-based span-mining pipeline. Second, we present MITRA-MT, a domain-adapted large language model built by continued pretraining of Qwen3.5-9B on a 22.6-billion-token corpus of classical Asian languages and related modern material, followed by a lightweight translation fine-tune and pref- erence optimization. Third, we introduce a multi-directional evaluation suite covering 17 translation directions at sentence level (ten of them also at paragraph level), including cross-classical directions (Sanskrit{\ensuremath{\leftrightarrow}}Tibetan, Sanskrit{\ensuremath{\leftrightarrow}}Chinese, Tibetan{\ensuremath{\leftrightarrow}}Chinese) and from Tibetan into seven modern languages beyond English. On this suite, MITRA-MT outperforms all open baselines, including instruction-tuned LLMs up to 122B parameters and dedicated MT systems, on all directions, and matches commercial references on the large majority of directions when evaluated via chrF, with the largest margins for translation into Tibetan, Sanskrit, and Classical Chinese. We release the parallel corpus, the evaluation data, and the model weights."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="nehrdich-keutzer-2026-mitra">
<titleInfo>
<title>MITRA-MT: Continued Pretraining, Parallel Corpus Mining, and Multi-Directional Machine Translation for Sanskrit, Tibetan, and Buddhist Chinese</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sebastian</namePart>
<namePart type="family">Nehrdich</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kurt</namePart>
<namePart type="family">Keutzer</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Eleventh Conference on Machine Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Barry</namePart>
<namePart type="family">Haddow</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tom</namePart>
<namePart type="family">Kocmi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Philipp</namePart>
<namePart type="family">Koehn</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christof</namePart>
<namePart type="family">Monz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-474-3</identifier>
</relatedItem>
<abstract>Buddhist literature is preserved in a network of classical languages, above all Sanskrit, Pāli, Buddhist Chinese, and Tibetan, for which machine translation support remains limited, and largely restricted to translation into English. In order to address this shortcoming, we present three connected contributions. First, we release MITRA-parallel v2, a corpus of 1.69 million automatically aligned parallel records (2.34 million segment pairs) between Sanskrit, Tibetan, and Buddhist Chinese, mined with an embedding-based span-mining pipeline. Second, we present MITRA-MT, a domain-adapted large language model built by continued pretraining of Qwen3.5-9B on a 22.6-billion-token corpus of classical Asian languages and related modern material, followed by a lightweight translation fine-tune and pref- erence optimization. Third, we introduce a multi-directional evaluation suite covering 17 translation directions at sentence level (ten of them also at paragraph level), including cross-classical directions (Sanskrit\ensuremathłeftrightarrowTibetan, Sanskrit\ensuremathłeftrightarrowChinese, Tibetan\ensuremathłeftrightarrowChinese) and from Tibetan into seven modern languages beyond English. On this suite, MITRA-MT outperforms all open baselines, including instruction-tuned LLMs up to 122B parameters and dedicated MT systems, on all directions, and matches commercial references on the large majority of directions when evaluated via chrF, with the largest margins for translation into Tibetan, Sanskrit, and Classical Chinese. We release the parallel corpus, the evaluation data, and the model weights.</abstract>
<identifier type="citekey">nehrdich-keutzer-2026-mitra</identifier>
<location>
<url>https://aclanthology.org/2026.wmt-main.26/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>492</start>
<end>506</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T MITRA-MT: Continued Pretraining, Parallel Corpus Mining, and Multi-Directional Machine Translation for Sanskrit, Tibetan, and Buddhist Chinese
%A Nehrdich, Sebastian
%A Keutzer, Kurt
%Y Haddow, Barry
%Y Kocmi, Tom
%Y Koehn, Philipp
%Y Monz, Christof
%S Proceedings of the Eleventh Conference on Machine Translation
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-474-3
%F nehrdich-keutzer-2026-mitra
%X Buddhist literature is preserved in a network of classical languages, above all Sanskrit, Pāli, Buddhist Chinese, and Tibetan, for which machine translation support remains limited, and largely restricted to translation into English. In order to address this shortcoming, we present three connected contributions. First, we release MITRA-parallel v2, a corpus of 1.69 million automatically aligned parallel records (2.34 million segment pairs) between Sanskrit, Tibetan, and Buddhist Chinese, mined with an embedding-based span-mining pipeline. Second, we present MITRA-MT, a domain-adapted large language model built by continued pretraining of Qwen3.5-9B on a 22.6-billion-token corpus of classical Asian languages and related modern material, followed by a lightweight translation fine-tune and pref- erence optimization. Third, we introduce a multi-directional evaluation suite covering 17 translation directions at sentence level (ten of them also at paragraph level), including cross-classical directions (Sanskrit\ensuremathłeftrightarrowTibetan, Sanskrit\ensuremathłeftrightarrowChinese, Tibetan\ensuremathłeftrightarrowChinese) and from Tibetan into seven modern languages beyond English. On this suite, MITRA-MT outperforms all open baselines, including instruction-tuned LLMs up to 122B parameters and dedicated MT systems, on all directions, and matches commercial references on the large majority of directions when evaluated via chrF, with the largest margins for translation into Tibetan, Sanskrit, and Classical Chinese. We release the parallel corpus, the evaluation data, and the model weights.
%U https://aclanthology.org/2026.wmt-main.26/
%P 492-506
Markdown (Informal)
[MITRA-MT: Continued Pretraining, Parallel Corpus Mining, and Multi-Directional Machine Translation for Sanskrit, Tibetan, and Buddhist Chinese](https://aclanthology.org/2026.wmt-main.26/) (Nehrdich & Keutzer, WMT 2026)
ACL