Cross-lingual Transfer of Reward Models in Multilingual Alignment
This work addresses the challenge of aligning AI systems with human preferences in multilingual contexts, which is incremental as it builds on existing RLHF methods by extending them to non-English languages.
The paper tackles the problem of limited multilingual applicability in reinforcement learning with human feedback (RLHF) by investigating cross-lingual transfer of reward models (RMs) from English to other languages, resulting in a 3-4% average increase in Multilingual RewardBench compared to target language RMs.
Reinforcement learning with human feedback (RLHF) is shown to largely benefit from precise reward models (RMs). However, recent studies in reward modeling schemes are skewed towards English, limiting the applicability of RLHF in multilingual alignments. In this work, we investigate the cross-lingual transfer of RMs trained in diverse languages, primarily from English. Our experimental results demonstrate the strong cross-lingual transfer of English RMs, exceeding target language RMs by 3~4% average increase in Multilingual RewardBench. Furthermore, we analyze the cross-lingual transfer of RMs through the representation shifts. Finally, we perform multilingual alignment to exemplify how cross-lingual transfer in RM propagates to enhanced multilingual instruction-following capability, along with extensive analyses on off-the-shelf RMs. We release the code, model, and data.