Enhancing Multilingual Reasoning via Steerable Model Merging
For practitioners of multilingual AI, this work provides a method to improve reasoning performance by dynamically balancing contributions from multilingual and reasoning models.
The authors propose a Steerable Model Merging (ST-Merge) framework that uses a gated cross-attention mechanism to adaptively weight source models, addressing conflicts in multilingual reasoning. The method consistently outperforms baselines on four benchmarks across 21 languages.
Model merging is an effective technique for composing the capabilities of a multilingual model and a reasoning model. It has achieved promising generalization in multilingual reasoning tasks by aligning feature spaces of different models. However, the merged single model often fails to address the conflicts between source models, leading to suboptimal performance. In other words, the one-size-fits-all merging strategy may not align with the characteristics of different inputs which may require prioritizing certain models over others. To this end, we propose a Steerable Model Merging (ST-Merge) framework to modulate the contribution of each source model. To realize this idea, we introduce a gated cross-attention mechanism to weight or filter the two attended source models in an adaptive manner. Extensive experiments demonstrate that ST-Merge consistently outperforms multiple strong baselines on four multilingual reasoning benchmarks across 21 different languages.