Summary of Selection-p: Self-supervised Task-agnostic Prompt Compression For Faithfulness and Transferability, by Tsz Ting Chung et al.
Selection-p: Self-Supervised Task-Agnostic Prompt Compression for Faithfulness and Transferability
by Tsz Ting Chung, Leyang Cui, Lemao Liu, Xinting Huang, Shuming Shi, Dit-Yan Yeung
First submitted to arxiv on: 15 Oct 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); 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 The proposed Selection-p method is a unified prompt compression technique for Large Language Models (LLMs) that leverages self-supervised pre-training to develop a probability distribution for each input token, indicating whether to preserve or discard it. This allows LLMs to achieve state-of-the-art performance across numerous classification tasks while compressing prompts up to 10 times without significant loss in performance. The method exhibits superior transferability to different models compared to prior work and maintains performance on in-context learning with long contexts. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Large Language Models have become very good at understanding and generating human language. To make them even better, researchers need to find ways to shrink the amount of information they use without losing their abilities. This paper shows a new way to do this by giving models a special set of rules that help decide what information is important to keep or throw away. The results are impressive – the model can understand and generate text just as well with much less data, making it more efficient and useful for real-world applications. |
Keywords
» Artificial intelligence » Classification » Probability » Prompt » Self supervised » Token » Transferability
