mlsanityData to 14 Oct 2026

HexaBee-4B: Six-Language Translation with Multilingual Answering and Bidirectional Supervision

arXiv 2610.16361 ↗First posted 12 Oct 2026cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

Multilingual translation requires a model to preserve meaning while producing text in a requested language. We present HexaBee-4B, a translator adapted from Gemma 3 4B IT through three stages: Wikipedia continued pretraining, joint multilingual answering and answer-text translation, and bidirectional translation specialization. The intermediate stage connects two tasks through shared multilingual answers: answering a question in the requested language and translating an answer into that language. It comprises 19.46 million answering records, including 5.76 million cross-language question-answer pairs, and 10.70 million answer-text translations. The final stage uses 658,274 forward and reverse synthetic translation examples. On 36,572 synthetic test examples spanning all 30 directions among Arabic, Chinese, English, Hebrew, Persian, and Spanish, HexaBee-4B achieves direction-macro chrF++ of 73.75, compared with 57.11 for Gemma 3 4B IT and 66.83 for Gemma 3 12B IT. BLEU, TER, COMET, and a matched model-judged subset support the same overall ranking. These results demonstrate improved translation across the six-language set, with chrF++ gains in every evaluated direction over both baselines.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 16 of its 35 references that are in the map. First posted on arXiv 12 Oct 2026.

Builds on

16 of 35 references.

References · 16

  1. 2018A Call for Clarity in Reporting BLEU Scorescited by 1,799
  2. 2020Don't Stop Pretraining: Adapt Language Models to Domains and Taskscited by 1,565
  3. 2024LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Modelscited by 1,464
  4. 2020COMET: A Neural Framework for MT Evaluationcited by 1,033
  5. 2016Sequence-Level Knowledge Distillationcited by 801
  6. 2021The FLORES-101 Evaluation Benchmark for Low-Resource and Multilingual Machine Translationcited by 560
  7. 2020TyDi QA: A Benchmark for Information-Seeking Question Answering in Typologically Diverse Languagescited by 488
  8. 2016Google's Multilingual Neural Machine Translation System: Enabling Zero-Shot Translationcited by 469
  9. 2024Aya Model: An Instruction Finetuned Open-Access Multilingual Language Modelcited by 295
  10. 2024Aya Dataset: An Open-Access Collection for Multilingual Instruction Tuningcited by 177
  11. 2020MKQA: A Linguistically Diverse Benchmark for Multilingual Open Domain Question Answeringcited by 153
  12. 2023GEMBA-MQM: Detecting Translation Quality Error Spans with GPT-4cited by 146
4 more
  1. 2024Multilingual Instruction Tuning With Just a Pinch of Multilingualitycited by 73
  2. 2023Eliciting the Translation Ability of Large Language Models via Multilingual Finetuning with Translation Instructionscited by 56
  3. 2023Empowering Cross-lingual Abilities of Instruction-tuned Large Language Models by Translation-following demonstrationscited by 35
  4. 2024CrossIn: An Efficient Instruction Tuning Approach for Cross-Lingual Knowledge Alignmentcited by 11

Built on by

Not cited yet.