Work

LiST: Local-Simplex Test-Time LoRA Fusion

Year: 2026

Type: conference-paper

Abstract: Task-specific LoRA adapters offer a modular way to specialize large language and visionlanguage models.However, existing adapter composition methods are mostly static and cannot adapt to individual test inputs.To address these issues, we propose LiST, a label-free test-time LoRA fusion framework that converts an existing LoRA bank into a target-conditioned local simplex and searches sample-specific fusion weights at inference time.LiST builds joint task representations from LoRA parameter anchors and promptlevel behavior vectors, retrieves neighboring adapters as a local search space, and performs branch-preserving fusion without updating the backbone or adapters.Candidate weights are selected by a prompt-level energy with prior, geometric, and stochasticconsistency constraints, and are deployed only when they pass a safe acceptance rule.Otherwise, LiST falls back to a target-conditioned prior.Experiments on multimodal and language benchmarks show that LiST outperforms static LoRA merging and conventional testtime adaptation baselines, while preserving task-specific adapter utility and improving robustness on unseen tasks.

Source: Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP)

DOI: 10.18653/2026.findings-emnlp.906