Enhancing Multilingual Capabilities of Large Language Models through Self-Distillation from Resource-Rich Languages
This work addresses the issue of multilingual performance gaps in LLMs for users needing cross-lingual applications, representing an incremental improvement over existing translation-based methods.
The paper tackles the problem of large language models underperforming in most languages compared to resource-rich ones by proposing SDRRL, a self-distillation method that leverages internal capabilities from resource-rich languages, resulting in significant enhancement of multilingual performance across various tasks while minimizing impact on original performance.
While large language models (LLMs) have been pre-trained on multilingual corpora, their performance still lags behind in most languages compared to a few resource-rich languages. One common approach to mitigate this issue is to translate training data from resource-rich languages into other languages and then continue training. However, using the data obtained solely relying on translation while ignoring the original capabilities of LLMs across languages is not always effective, which we show will limit the performance of cross-lingual knowledge transfer. In this work, we propose SDRRL, a method based on Self-Distillation from Resource-Rich Languages that effectively improve multilingual performance by leveraging the internal capabilities of LLMs on resource-rich languages. We evaluate on different LLMs (LLaMA-2 and SeaLLM) and source languages across various comprehension and generation tasks, experimental results demonstrate that SDRRL can significantly enhance multilingual capabilities while minimizing the impact on original performance in resource-rich languages.