Loading Now

Summary of Towards Infinite-long Prefix in Transformer, by Yingyu Liang et al.


Towards Infinite-Long Prefix in Transformer

by Yingyu Liang, Zhenmei Shi, Zhao Song, Chiwun Yang

First submitted to arxiv on: 20 Jun 2024

Categories

  • Main: Machine Learning (cs.LG)
  • Secondary: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)

     Abstract of paper      PDF of paper


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 proposes a new approach to fine-tuning language models called Prefix Learning, which aims to enhance their performance on various downstream tasks. The authors investigate the effectiveness of prompting and context-based methods in a stylized setting using the Neural Tangent Kernel (NTK) framework. They provide a convergence guarantee for training an ultra-long prefix and design an algorithm that only requires a few extra trainable parameters. Experimental results show that this method achieves superior or competitive performance compared to existing approaches like full parameter fine-tuning, P-Tuning V2, and LoRA.
Low GrooveSquid.com (original content) Low Difficulty Summary
Prefix Learning is a new way to make language models better at understanding specific tasks. The authors wanted to see if their approach could work well without needing to change the whole model. They tested it on vision, natural language, and math data and found that it did really well compared to other methods!

Keywords

» Artificial intelligence  » Fine tuning  » Lora  » Prompting