Summary of Owled: Outlier-weighed Layerwise Pruning For Efficient Autonomous Driving Framework, by Jiaxi Li et al.
OWLed: Outlier-weighed Layerwise Pruning for Efficient Autonomous Driving Framework
by Jiaxi Li, Lu Yin, Xilu Wang
First submitted to arxiv on: 12 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Robotics (cs.RO)
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 Medium Difficulty Summary: The paper presents OWLed, a novel framework for efficiently deploying Large Language Models (LLMs) in autonomous driving systems. Current approaches suffer from computational demands that render them unfeasible for real-world applications. To address this challenge, the authors introduce Outlier-Weighed Layerwise Pruning (OWLed), which leverages outlier-weighted layerwise sparsity to compress models while preserving their functionality. The method assigns non-uniform sparsity ratios to different layers based on the distribution of outlier features, reducing model size without fine-tuning. To ensure adaptation to autonomous driving tasks, the authors incorporate driving environment data into both calibration and pruning processes. Experimental results demonstrate that OWLed outperforms existing methods in perception, action prediction, and language understanding while significantly lowering computational requirements. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Low Difficulty Summary: Researchers have been working on a way to make self-driving cars smarter by using Large Language Models (LLMs). The problem is that these models are too big and require too much power to work well on a car. To solve this, the authors created a new method called OWLed that makes the model smaller while keeping it just as good at recognizing things and making decisions. They used special data about driving environments to help the model work better in real-world scenarios. The results show that OWLed is better than other methods at doing tasks like recognizing objects, predicting actions, and understanding language, all while using much less power. |
Keywords
» Artificial intelligence » Fine tuning » Language understanding » Pruning
