Large language models can be continually pre-trained or fine-tuned to improve performance in specific domains, languages, or skills, but this specialization often degrades other capabilities and may cause catastrophic forgetting. We investigate how abilities are distributed within LLM parameters by analyzing module activations under domain- and language-specific inputs for closely related models. Across layers and modules, we find that ability-related activations are highly concentrated in a small set of channels (typically <5\%), and these channels are largely disentangled with good sufficiency and stability. Building on these observations, we propose ACT (Activation-Guided Channel-wise Ability Transfer), which localizes ability-relevant channels via activation differences and selectively transfers only the corresponding parameters, followed by lightweight fine-tuning for compatibility. Experiments on multilingual mathematical and scientific reasoning show that ACT can recover forgotten abilities while preserving retained skills. It can also merge multiple specialized models to integrate several abilities into a single model with minimal interference. Our code and data will be publicly released.

翻译：大型语言模型可以通过持续预训练或微调来提升在特定领域、语言或技能上的性能，但这种专业化往往会削弱其他能力，并可能导致灾难性遗忘。我们通过分析相近模型在领域和语言特定输入下的模块激活情况，研究了能力在LLM参数中的分布方式。跨层和跨模块的分析表明，能力相关的激活高度集中在少量通道中（通常<5%），且这些通道在很大程度上是解耦的，具有良好的充分性和稳定性。基于这些观察，我们提出了ACT（基于激活引导的通道级能力迁移）方法，该方法通过激活差异定位能力相关通道，并选择性地仅迁移对应参数，随后进行轻量级微调以确保兼容性。在多语言数学与科学推理任务上的实验表明，ACT能够恢复被遗忘的能力，同时保持已掌握技能。该方法还可融合多个专业化模型，将多种能力整合至单一模型中，且干扰最小。我们的代码与数据将公开发布。