Summary of Lipsum-ft: Robust Fine-tuning Of Zero-shot Models Using Random Text Guidance, by Giung Nam et al.
Lipsum-FT: Robust Fine-Tuning of Zero-Shot Models Using Random Text Guidance
by Giung Nam, Byeongho Heo, Juho Lee
First submitted to arxiv on: 1 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV)
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 This paper presents a novel approach to fine-tuning large-scale contrastive vision-language pre-trained models for robustness against distribution shifts. The authors demonstrate that while zero-shot models can achieve competitive performance without additional training, they compromise on robustness. To address this issue, the study proposes Lipsum-FT, an algorithm that leverages the language modeling aspect of these pre-trained models to achieve better fine-tuning results. Experimental evaluations on DomainNet and ImageNet datasets show the superiority of Lipsum-FT over existing methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about making big advances in artificial intelligence by fine-tuning special computer programs called vision-language models. These programs can learn from lots of data, but they’re not always good at adapting to new situations. The researchers found a way to make these models more robust, which means they’ll work better even when the situation changes. They came up with a new approach called Lipsum-FT and tested it on some big datasets. It worked really well! |
Keywords
* Artificial intelligence * Fine tuning * Zero shot
