| # 💡Model Description | |
| Official model repository for our **ACL 2026 Main Conference** paper "*Language on Demand, Knowledge at Core*: Composing LLMs with Encoder-Decoder Translation Models for Extensible Multilinguality". | |
| ## ✨XBridge-base | |
| [`XBridge-base`](https://huggingface.co/ICTNLP/XBridge-base) is trained with stage 1 (cross-model alignment) using trilingual translation data, composing [`LLaMA3-8B`](https://huggingface.co/meta-llama/Meta-Llama-3-8B) with [`NLLB-200-1.3B`](https://huggingface.co/facebook/nllb-200-1.3B). Training is conducted on 10 languages: | |
| > Bn, De, En, Es, Fr, Ja, Ru, Sw, Th, Zh | |
| Despite being trained on a limited set of languages, we observe in our analysis that **stage 1 learns a language-agnostic cross-model alignment**, which generalizes well beyond the seen languages. | |
| ## ✨XBridge-SFT | |
| [`XBridge-SFT`](https://huggingface.co/ICTNLP/XBridge-SFT) further extends `XBridge-base` by training stage 2 (encoder-side adaptation) and stage 3 (decoder-side adaptation) for instruction-following tasks. Notably, we directly scale to 50 languages in these stages. This design is motivated by our finding of cross-model generalization. We train on the multilingual instruction-following dataset [`Bactrian-X`](https://huggingface.co/datasets/MBZUAI/Bactrian-X), and expand to the following additional languages: | |
| > Af, Ar, Az, Cs, El, Et, Fa, Fi, Gl, Gu, He, Hi, Hr, Id, It, Ka, Kk, Km, Lt, Lv, Mk, Ml, Mn, Mr, My, Ne, Nl, Pl, Ps, Pt, Ro, Sl, Sv, Ta, Te, Tr, Uk, Ur, Vi, Xh | |
| Empirically, we find that this direct scaling strategy achieves strong performance, demonstrating the robustness and generalization ability of the stage 1 alignment. | |
| See our [paper](https://arxiv.org/abs/2603.17512) for more details, and try our Gradio demo in the [github repository](https://github.com/ictnlp/XBridge)! | |
| # 📚Citation | |
| If you find this model or our work useful, please cite: | |
| ```tex | |
| @misc{bu2026languagedemandknowledgecore, | |
| title={Language on Demand, Knowledge at Core: Composing LLMs with Encoder-Decoder Translation Models for Extensible Multilinguality}, | |
| author={Mengyu Bu and Yang Feng}, | |
| year={2026}, | |
| eprint={2603.17512}, | |
| archivePrefix={arXiv}, | |
| primaryClass={cs.CL}, | |
| url={https://arxiv.org/abs/2603.17512}, | |
| } | |
| ``` | |
| # 📮Contact | |
| For questions, please contact: `bumengyu23z@ict.ac.cn` |