Dynamic Collaboration of Multi-Language Models based on Minimal Complete Semantic Units
This work addresses the challenge of improving reasoning in language models for AI applications, though it appears incremental as it builds on existing multi-model collaboration techniques.
The paper tackles the problem of enhancing reasoning capabilities in language models by proposing a token-level multi-model collaboration method with a dynamic selection strategy and minimal complete semantic units to address vocabulary misalignment, achieving superior results across various benchmarks.
This paper investigates the enhancement of reasoning capabilities in language models through token-level multi-model collaboration. Our approach selects the optimal tokens from the next token distributions provided by multiple models to perform autoregressive reasoning. Contrary to the assumption that more models yield better results, we introduce a distribution distance-based dynamic selection strategy (DDS) to optimize the multi-model collaboration process. To address the critical challenge of vocabulary misalignment in multi-model collaboration, we propose the concept of minimal complete semantic units (MCSU), which is simple yet enables multiple language models to achieve natural alignment within the linguistic space. Experimental results across various benchmarks demonstrate the superiority of our method. The code will be available at https://github.com/Fanye12/DDS.