Smoothie-Qwen: Post-Hoc Smoothing to Reduce Language Bias in Multilingual LLMs
This provides a practical solution for improving language controllability in LLMs, making them more reliable for global applications, though it is incremental as it builds on existing models.
The paper tackled language confusion in multilingual LLMs, where models generate responses in a dominant language regardless of the prompt, and proposed Smoothie-Qwen, a post-hoc smoothing method that reduced unintended Chinese output by over 95% while maintaining task accuracy.
Multilingual large language models (LLMs) often exhibit language confusion, a tendency to generate responses in a dominant language irrespective of the prompt's language. To address this, we propose Smoothie-Qwen, a lightweight, post-hoc method that mitigates language bias without retraining. This technique selectively adjusts token-level output probabilities to effectively suppress undesired language generation. Applied to the Qwen model, our method reduces unintended Chinese output by over 95% while preserving task accuracy on multilingual benchmarks. This work provides a practical and efficient solution for enhancing the language controllability of LLMs, making them more reliable for global applications.