Summary of Distilling Text Style Transfer with Self-explanation From Llms, by Chiyu Zhang et al.
Distilling Text Style Transfer With Self-Explanation From LLMs
by Chiyu Zhang, Honglong Cai, Yuezhang, Yuexin Wu, Le Hou, Muhammad Abdul-Mageed
First submitted to arxiv on: 2 Mar 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
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 The proposed CoTeX framework leverages large language models and chain-of-thought prompting to facilitate Text Style Transfer (TST), which aims to alter text styles while retaining core content. By distilling the complex rewriting capabilities of LLMs into more streamlined models, CoTeX can work with both parallel and non-parallel data, outperforming traditional methods in low-resource settings. The framework is evaluated across four TST datasets, demonstrating its superiority over current unsupervised, supervised, and instruction-tuned techniques. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary CoTex is a new way to change the style of text while keeping its main ideas. Right now, it’s hard to find big datasets for this task, so CoTeX uses special prompts and large language models to make it work. This helps with both parallel and non-parallel data. It even beats other methods in situations where there isn’t much data. The researchers tested CoTex on four different sets of text and found that it’s the best approach. |
Keywords
» Artificial intelligence » Prompting » Style transfer » Supervised » Unsupervised
