---
license: apache-2.0
pipeline_tag: image-segmentation
---

Semantic Library Adaptation (SemLA) is a training-free, test-time domain adaptation framework that dynamically retrieves and merges the most relevant LoRA adapters from a library based on semantic similarity to the target domain. SemLA constructs tailored models for each input without additional training, offering scalability, explainability, and privacy preservation.  See [the paper](https://arxiv.org/abs/2503.21780) for more details.

Code: https://github.com/rezaqorbani/SemLA

Project page: https://thegoodailab.org/semla