ARIADNE: Agnostic Routing for Inference-time Adapter DyNamic sElection
Quick Answer
ARIADNE introduces a training-free, adapter-agnostic routing framework for dynamic adapter selection at inference time, achieving 97.44% of upper bound performance on Llama 3.2 1B Instruct across 23 NLP tasks.
Quick Take
It scales to 44 tasks with 89.7% average selection accuracy without requiring additional training or access to adapter internals.
Key Points
- ARIADNE uses centroids from training set embeddings for adapter selection.
- Compatible with any parameter-efficient fine-tuning methods without modifications.
- Achieves 97.44% performance on Llama 3.2 1B Instruct across 23 tasks.
- Scales to 44 tasks with an average selection accuracy of 89.7%.
- Eliminates the need for additional router training or internal adapter access.
Paper Resources
📖 Reader Mode
~2 min readAbstract:The increasing deployment of parameter-efficient fine-tuning (PEFT) has led to model ecosystems in which a single backbone is paired with many task-specialized adapters. In this setting, inference-time queries often arrive without task labels, requiring the system to automatically select the most appropriate adapter from a growing and heterogeneous adapter pool. Existing routing methods either depend on access to adapter internals, such as weight decompositions or gradient-based statistics, or require additional router training, which limits scalability and portability as new adapters are added. We introduce ARIADNE, a training-free, adapter-agnostic routing framework for dynamic adapter selection at inference time. ARIADNE represents each adapter through a set of centroids computed from embeddings of its training set, capturing the data distribution associated with that adapter. Given an unlabeled input, it selects an adapter by measuring proximity to these centroids in latent space. Because routing is performed entirely in the input embedding space, ARIADNE is compatible with arbitrary PEFT methods and requires no modification to the adapters or training procedures. Primarily evaluated with Llama 3.2 1B Instruct on 23 diverse NLP tasks, ARIADNE recovers 97.44% of the upper bound performance. Scaling to 44 tasks, it achieves 89.7% average selection accuracy, without additional training or access to adapter internals.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2606.19079 [cs.AI] |
| (or arXiv:2606.19079v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2606.19079 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Enrico Cassano [view email]
[v1]
Wed, 17 Jun 2026 13:50:51 UTC (257 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →HOBA: Hierarchical On-Policy Bidding Agents for Adaptive Online Advertising
HOBA (Hierarchical On-policy Bidding Agents) is a novel hierarchical reinforcement learning framework that enhances online advertising bidding systems by improving adaptability and reducing hyperparameter tuning costs. It utilizes a for hyperparameter inference, a SARSA agent for expert model selection, and a dynamic expert pool for bid execution, achieving a +3.6% increase in target cost during large-scale deployment and outperforming state-of-the-art baselines on AuctionNet.