ReasonAny: Incorporating Reasoning Capability to Any Model via Simple and Effective Model Merging
This addresses the problem of performance collapse when merging reasoning and domain models for practitioners needing specialized AI with reasoning, though it appears incremental as an improvement over existing model merging methods.
The paper tackles the challenge of equipping domain-specialized models with reasoning capabilities without training, identifying that reasoning ability resides in parameter regions with low gradient sensitivity, and proposes ReasonAny, a merging framework that outperforms state-of-the-art baselines across safety, biomedicine, and finance domains.
Large Reasoning Models (LRMs) with long chain-of-thought reasoning have recently achieved remarkable success. Yet, equipping domain-specialized models with such reasoning capabilities, referred to as "Reasoning + X", remains a significant challenge. While model merging offers a promising training-free solution, existing methods often suffer from a destructive performance collapse: existing methods tend to both weaken reasoning depth and compromise domain-specific utility. Interestingly, we identify a counter-intuitive phenomenon underlying this failure: reasoning ability predominantly resides in parameter regions with low gradient sensitivity, contrary to the common assumption that domain capabilities correspond to high-magnitude parameters. Motivated by this insight, we propose ReasonAny, a novel merging framework that resolves the reasoning-domain performance collapse through Contrastive Gradient Identification. Experiments across safety, biomedicine, and finance domains show that ReasonAny effectively synthesizes "Reasoning + X" capabilities, significantly outperforming state-of-the-art baselines while retaining robust reasoning performance.