CLSep 26, 2025

Context Parametrization with Compositional Adapters

arXiv:2509.22158v11 citationsh-index: 6
Originality Incremental advance
AI Analysis

This work addresses efficiency and flexibility issues in adapting LLMs to new tasks, offering a practical solution for scaling deployment, though it appears incremental as it builds on prior adapter generation methods.

The paper tackles the limitations of in-context learning and supervised fine-tuning for large language models by introducing CompAs, a meta-learning framework that translates context into compositional adapter parameters, which outperforms existing methods on multiple-choice and extractive question answering tasks when scaling to more inputs.

Large language models (LLMs) often seamlessly adapt to new tasks through in-context learning (ICL) or supervised fine-tuning (SFT). However, both of these approaches face key limitations: ICL is inefficient when handling many demonstrations, and SFT incurs training overhead while sacrificing flexibility. Mapping instructions or demonstrations from context directly into adapter parameters offers an appealing alternative. While prior work explored generating adapters based on a single input context, it has overlooked the need to integrate multiple chunks of information. To address this gap, we introduce CompAs, a meta-learning framework that translates context into adapter parameters with a compositional structure. Adapters generated this way can be merged algebraically, enabling instructions, demonstrations, or retrieved passages to be seamlessly combined without reprocessing long prompts. Critically, this approach yields three benefits: lower inference cost, robustness to long-context instability, and establishes a principled solution when input exceeds the model's context window. Furthermore, CompAs encodes information into adapter parameters in a reversible manner, enabling recovery of input context through a decoder, facilitating safety and security. Empirical results on diverse multiple-choice and extractive question answering tasks show that CompAs outperforms ICL and prior generator-based methods, especially when scaling to more inputs. Our work establishes composable adapter generation as a practical and efficient alternative for scaling LLM deployment.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes