The personalized fine-tuning of large language models (LLMs) on edge devices is severely constrained by limited computation resources. Although split federated learning alleviates on-device burdens, its effectiveness diminishes in few-shot reasoning scenarios due to the low data efficiency of conventional supervised fine-tuning, which leads to excessive communication overhead. To address this, we propose Language-Empowered Split Fine-Tuning (LESFT), a framework that integrates split architectures with a contrastive-inspired fine-tuning paradigm. LESFT simultaneously learns from multiple logically equivalent but linguistically diverse reasoning chains, providing richer supervisory signals and improving data efficiency. This process-oriented training allows more effective reasoning adaptation with fewer samples. Extensive experiments demonstrate that LESFT consistently outperforms strong baselines such as SplitLoRA in task accuracy. LESFT consistently outperforms strong baselines on GSM8K, CommonsenseQA, and AQUA_RAT, with the largest gains observed on Qwen2.5-3B. These results indicate that LESFT can effectively adapt large language models for reasoning tasks under the computational and communication constraints of edge environments.

With the development of ubiquitous computing and multimodal sensing, intelligent terminals are evolving towards a “human−machine symbiosis” paradigm, increasing demands for intelligent, autonomous, and self-adaptive terminal systems. However, existing systems still face fundamental challenges in real-world deployments: (1) Trade-off between user comfort, data fidelity, and privacy; (2) absence of context-aware scheduling mechanisms; and (3) limited capabilities in semantic generalization. To address these challenges, we propose Meta Fiberverse, a fabric-based computational platform designed for human−machine−environment symbiosis. The system integrates high-density fabric sensing, high-fidelity scheduling mechanisms, and plugin-enhanced semantic coordination framework powered by Large Language Models (LLMs). Experimental results demonstrate the system’s performance in communication latency and task accuracy, offering a feasible path and technical reference for next-generation human-centered intelligent terminals with human−machine symbiosis, resource self-consistency, and semantic autonomy.