Efficient Low-Resource Language Adaptation via Multi-Source Dynamic Logit Fusion
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917422593736704 |
|---|---|
| author | Zhang, Chen Lin, Jiuheng Liao, Zhiyuan Feng, Yansong |
| author_facet | Zhang, Chen Lin, Jiuheng Liao, Zhiyuan Feng, Yansong |
| contents | Adapting large language models (LLMs) to low-resource languages (LRLs) is constrained by the scarcity of task data and computational resources. Although Proxy Tuning offers a logit-level strategy for introducing scaling effects, it often fails in LRL settings because the large model's weak LRL competence might overwhelm the knowledge of specialized smaller models. We thus propose TriMix, a test-time logit fusion framework that dynamically balances capabilities from three different sources: LRL competence from a continually pretrained small model, task competence from high-resource language instruction tuning, and the scaling benefits of large models. It is data- and compute-efficient, requiring no LRL task annotations, and only continual pretraining on a small model. Experiments across four model families and eight LRLs show that TriMix consistently outperforms single-model baselines and Proxy Tuning. Our analysis reveals that prioritizing the small LRL-specialized model's logits is crucial for success, challenging the prevalent large-model-dominant assumption. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_18106 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Efficient Low-Resource Language Adaptation via Multi-Source Dynamic Logit Fusion Zhang, Chen Lin, Jiuheng Liao, Zhiyuan Feng, Yansong Computation and Language Adapting large language models (LLMs) to low-resource languages (LRLs) is constrained by the scarcity of task data and computational resources. Although Proxy Tuning offers a logit-level strategy for introducing scaling effects, it often fails in LRL settings because the large model's weak LRL competence might overwhelm the knowledge of specialized smaller models. We thus propose TriMix, a test-time logit fusion framework that dynamically balances capabilities from three different sources: LRL competence from a continually pretrained small model, task competence from high-resource language instruction tuning, and the scaling benefits of large models. It is data- and compute-efficient, requiring no LRL task annotations, and only continual pretraining on a small model. Experiments across four model families and eight LRLs show that TriMix consistently outperforms single-model baselines and Proxy Tuning. Our analysis reveals that prioritizing the small LRL-specialized model's logits is crucial for success, challenging the prevalent large-model-dominant assumption. |
| title | Efficient Low-Resource Language Adaptation via Multi-Source Dynamic Logit Fusion |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2604.18106 |