# AdapterDistillation: Non-Destructive Task Composition with Knowledge Distillation

arXiv: https://arxiv.org/abs/2312.16261 · HTML: https://mlsanity.com/paper/2312.16261/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Knowledge distillation for LLMs, one of 12 directions in Knowledge distillation.
- Builds on 9 of its 19 references that are in the map.
- First posted on arXiv 26 Dec 2023.

## Placed in

- Knowledge distillation / [Knowledge distillation for LLMs](https://mlsanity.com/distill/knowledge-distillation-for-llms/)

## Builds on (most cited first, 9 of 9)

- [Parameter-Efficient Transfer Learning for NLP](https://mlsanity.com/paper/1902.00751/) (2019)
- [Prefix-Tuning: Optimizing Continuous Prompts for Generation](https://mlsanity.com/paper/2101.00190/) (2021)
- [The Power of Scale for Parameter-Efficient Prompt Tuning](https://mlsanity.com/paper/2104.08691/) (2021)
- [Towards a Unified View of Parameter-Efficient Transfer Learning](https://mlsanity.com/paper/2110.04366/) (2021)
- [AdapterFusion: Non-Destructive Task Composition for Transfer Learning](https://mlsanity.com/paper/2005.00247/) (2020)
- [MAD-X: An Adapter-Based Framework for Multi-Task Cross-Lingual Transfer](https://mlsanity.com/paper/2005.00052/) (2020)
- [AdapterHub: A Framework for Adapting Transformers](https://mlsanity.com/paper/2007.07779/) (2020)
- [Distributed Learning With Sparsified Gradient Differences](https://mlsanity.com/paper/2202.02491/) (2022)
- [Communication Efficient Federated Learning via Ordered ADMM in a Fully Decentralized Setting](https://mlsanity.com/paper/2202.02580/) (2022)

## Built on by (most cited first, 1 of 1)

- [No Train but Gain: Language Arithmetic for training-free Language Adapters enhancement](https://mlsanity.com/paper/2404.15737/) (2024)

## Abstract (arXiv)

Leveraging knowledge from multiple tasks through introducing a small number of task specific parameters into each transformer layer, also known as adapters, receives much attention recently. However, adding an extra fusion layer to implement knowledge composition not only increases the inference time but also is non-scalable for some applications. To avoid these issues, we propose a two-stage knowledge distillation algorithm called AdapterDistillation. In the first stage, we extract task specific knowledge by using local data to train a student adapter. In the second stage, we distill the knowledge from the existing teacher adapters into the student adapter to help its inference. Extensive experiments on frequently asked question retrieval in task-oriented dialog systems validate the efficiency of AdapterDistillation. We show that AdapterDistillation outperforms existing algorithms in terms of accuracy, resource consumption and inference time.