InfiFusion: A Unified Framework for Enhanced Cross-Model Reasoning via LLM Fusion
This addresses the need for flexible and efficient cross-model reasoning in AI, though it appears incremental as an enhancement to existing fusion methods.
The paper tackles the problem of integrating multiple domain-specialized LLMs into a single model by introducing InfiFusion, an efficient training pipeline that outperforms state-of-the-art models like Qwen-2.5-14B-Instruct and Phi-4 across 11 benchmarks while reducing computational costs to 160 H800 GPU hours.
We introduce InfiFusion, an efficient training pipeline designed to integrate multiple domain-specialized Large Language Models (LLMs) into a single pivot model, effectively harnessing the strengths of each source model. Traditional fusion methods either merge model parameters directly or rely on knowledge distillation with rigid assumptions, limiting their flexibility and efficiency. InfiFusion overcomes these limitations by enhancing Universal Logit Distillation (ULD) with Top-K selection and Logits Standardization. We propose two fusion strategies: Pairwise Fusion (InfiFusion$_p$), where each source model knowledge is distilled individually into the pivot model followed by merging and Unified Fusion (InfiFusion$_u$), where knowledge from all source models is distilled simultaneously into the pivot model. InfiFusion outperforms the state-of-the-art models, such as Qwen-2.5-14B-Instruct and Phi-4, across 11 widely applied benchmarks covering reasoning, coding, mathematics, and instruction-following tasks. Notably, InfiFusion achieves this superior performance while significantly reduces computational costs, completing full training with only 160 H800 GPU hours compared to the millions typically required for traditional LLM training.