Summary of Rethinking Low-rank Adaptation in Vision: Exploring Head-level Responsiveness Across Diverse Tasks, by Yibo Zhong et al.
Rethinking Low-Rank Adaptation in Vision: Exploring Head-Level Responsiveness across Diverse Tasks
by Yibo Zhong, Yao Zhou
First submitted to arxiv on: 13 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary In this paper, researchers propose an efficient approach for adapting pre-trained Vision Transformers (ViT) called Head-level responsiveness tuning for low-rank adaptation (Heart-LoRA). The method selectively activates task-responsive heads and dynamically adjusts the activation of approximated heads tailored to the current task. This leads to improved performance on visual adaptation benchmark datasets compared to state-of-the-art PETL approaches. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper proposes a way to adapt pre-trained Vision Transformers, called Heart-LoRA, which makes it more efficient by only updating some important parts. It also finds that different parts of the model are good at different tasks and adjusts them accordingly. |
Keywords
* Artificial intelligence * Lora * Low rank adaptation * Vit
