Summary of Padellm-ner: Parallel Decoding in Large Language Models For Named Entity Recognition, by Jinghui Lu et al.
PaDeLLM-NER: Parallel Decoding in Large Language Models for Named Entity Recognition
by Jinghui Lu, Ziwei Yang, Yanjie Wang, Xuejing Liu, Brian Mac Namee, Can Huang
First submitted to arxiv on: 7 Feb 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 This study aims to improve the efficiency of Named Entity Recognition (NER) using Large Language Models (LLMs). The main challenge is the sequential decoding process, which increases the sequence length and latency. To address this, the authors introduce Parallel Decoding in LLM for NE} (PaDeLLM-NER), a method that integrates seamlessly with existing generative models without requiring additional modules or modifications. PaDeLLM-NER allows for simultaneous decoding of all mentions, reducing generation latency by 1.76 to 10.22 times compared to the autoregressive approach on both English and Chinese datasets. The results show that PaDeLLM-NER maintains state-of-the-art performance across various datasets while significantly improving inference speed. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study tries to make computers better at finding important information in text, like names and places. Right now, these computers are slow because they have to do things one step at a time. The authors came up with a new way for the computer to do all of this processing at once, making it much faster. They tested their idea on English and Chinese texts and found that it’s 1-10 times faster than before! And it still works just as well. |
Keywords
» Artificial intelligence » Autoregressive » Inference » Named entity recognition » Ner
