4AI / LS-LLaMA

A Simple but Powerful SOTA NER Model | Official Code For Label Supervised LLaMA Finetuning
https://arxiv.org/abs/2310.01208
MIT License
131 stars 20 forks source link

Hi, what's the prompt for NER task? #18

Closed wmkai closed 4 months ago

SeanLee97 commented 4 months ago

@wmkai do you mean the prompt for the zero-shot NER setting?

wmkai commented 4 months ago

hmm, I want the prompt for the result of NER in CONLL2003. Does it have a prompt?

image
SeanLee97 commented 4 months ago

hi @wmkai, it is unnecessary to set a prompt for LS-unLLaMA since the generation ability will malfunction when transforming all layers into bidirectional.

BTW, it's recommended to use the BiLLM toolkit to train NER. Here is a training and inference example:

https://github.com/WhereIsAI/BiLLM?tab=readme-ov-file#ner

SeanLee97 commented 4 months ago

Here are two pretrained models for the conll2003 dataset:

wmkai commented 4 months ago

I see. Thank u very much :)