Language-Aware Token Boosting: LLM Language Confusion Reduction Without Tuning
About
Large language models (LLMs) sometimes exhibit language confusion when generating non-English text. Existing approaches typically rely on fine-tuning to mitigate this issue. In contrast, we propose a tuning-free paradigm for reducing language confusion. Within this paradigm, we introduce two methods: Language-Aware Token Boosting (LATB), which applies targeted perturbations to tokens associated with the desired language, and Adaptive Language-Aware Token Boosting (Adaptive-LATB), which dynamically adjusts these perturbations based on the model's confidence in the intended language. Experiments demonstrate that our methods effectively improve multilingual alignment by reducing language confusion, while maintain the summarization quality without requiring any additional fine-tuning. Our code is publicly available. https://github.com/scbdatax/genai-datax-language-aware-token-boosting.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Summarization | XLSum Korean (test) | ROUGE-27.03 | 14 | |
| Summarization | XL-Sum Arabic (test) | ROUGE-L12.45 | 12 | |
| Summarization | XLSum Japanese (test) | ROUGE-212.95 | 10 | |
| Multilingual Summarization | XL-SUM ru | Token-level Language Confusion0.28 | 5 | |
| Multilingual Summarization | XL-SUM zh | Token-level Language Confusion (%)4.78 | 5 | |
| Multilingual Summarization | XL-SUM ja | Token-level Language Confusion3.51 | 5 | |
| Multilingual Summarization | XL-SUM fr | Token-level Language Confusion (%)11 | 5 | |
| Multilingual Summarization | XL-SUM ko | Token-level Language Confusion3.45 | 5 | |
| Multilingual Summarization | XL-SUM th | Token-level Language Confusion0.38 | 5 | |
| Multilingual Summarization | XL-SUM hi | Token-level Language Confusion0.23 | 5 |