Simple and Effective Few-Shot Named Entity Recognition with Structured Nearest Neighbor Learning

Overview

structshot

Code and data for paper "Simple and Effective Few-Shot Named Entity Recognition with Structured Nearest Neighbor Learning", Yi Yang and Arzoo Katiyar, in EMNLP 2020.

Data

Due to license reason, we are only able to release the full CoNLL 2003 and WNUT 2017 dataset. We also release the support sets that we sampled from the CoNLL/WNUT/I2B2 dev sets to enable the reproducing of our evaluation results.

CoNLL 2003

The CoNLL 2003 NER train/dev/test datasets are data/train.txt, data/dev.txt, and data/test.txt respectively. The labels are available in data/labels.txt.

WNUT 2017

The WNUT 2017 NER dev/test datasets are data/dev-wnut.txt and data/test-wnut.txt respectively. The labels are available in data/labels-wnut.txt.

Support sets for CoNLL 2003, WNUT 2017, and I2B2 2014

The one-shot and five-shot support sets used in the paper are available in data/support-* folders.

Usage

Due to data license limitation, we will show how to do five-shot transfer learning from the CoNLL 2003 dataset to the WNUT 2017 dataset, instead of transfering from the OntoNotes 5 dataset, as presented in our paper.

The first step is to install the package and cd into the structshot directory:

pip install -e .
cd structshot

Pretrain BERT-NER model

The marjority of the code is copied from the HuggingFace transformers repo, which is used to pretrain a BERT-NER model:

# Pretrain a conventional BERT-NER model on CoNLL 2003 
bash run_pl.sh

In our paper, we actually merged B- and I- tags together for pretraining as well.

Few-shot NER with NNShot

Given the pretrained model located at output-model/checkpointepoch=2.ckpt, we now can perform five-shot NER transfer on the WNUT test set:

# Five-shot NER with NNShot
bash run_pred.sh output-model/checkpointepoch=2.ckpt NNShot

We use the IO tagging scheme rather than the BIO tagging scheme due to its simplicity and better performance. I obtained 22.8 F1 score.

Few-shot NER with StructShot

Given the same pretrained model, simply run:

# Five-shot NER with StructShot
bash run_pred.sh output-model/checkpointepoch=2.ckpt StructShot

I obtained 29.5 F1 score. You can tune the parameter tau in the run_pred.sh script based on dev set performance.

Notes

There are a few differences between this implementation and the one reported in the paper due to data license reason etc.:

  • This implementation pretrains the BERT-NER model with the BIO tagging scheme, while in our paper we uses the IO tagging scheme.
  • This implementation performs five-shot transfer learning from CoNLL 2003 to WNUT 2017, while in our paper we perform five-shot transfer learning from OntoNotes 5 to CoNLL'03/WNUT'17/I2B2'14.

If you can access OntoNotes 5 and I2B2'14, reproducing the results of the paper should be trivial.

Owner
ASAPP Research
AI for Enterprise
ASAPP Research
HyperLib: Deep learning in the Hyperbolic space

HyperLib: Deep learning in the Hyperbolic space Background This library implements common Neural Network components in the hypberbolic space (using th

105 Dec 25, 2022
FAST Aiming at the problems of cumbersome steps and slow download speed of GNSS data

FAST Aiming at the problems of cumbersome steps and slow download speed of GNSS data, a relatively complete set of integrated multi-source data download terminal software fast is developed. The softw

ChangChuntao 23 Dec 31, 2022
Library for time-series-forecasting-as-a-service.

TIMEX TIMEX (referred in code as timexseries) is a framework for time-series-forecasting-as-a-service. Its main goal is to provide a simple and generi

Alessandro Falcetta 8 Jan 06, 2023
A Keras implementation of YOLOv3 (Tensorflow backend)

keras-yolo3 Introduction A Keras implementation of YOLOv3 (Tensorflow backend) inspired by allanzelener/YAD2K. Quick Start Download YOLOv3 weights fro

7.1k Jan 03, 2023
Recreate CenternetV2 based on MMDET.

Introduction This project is trying to Recreate CenternetV2 based on MMDET, which is proposed in paper Probabilistic two-stage detection. This project

25 Dec 09, 2022
Vehicle Detection Using Deep Learning and YOLO Algorithm

VehicleDetection Vehicle Detection Using Deep Learning and YOLO Algorithm Dataset take or find vehicle images for create a special dataset for fine-tu

Maryam Boneh 96 Jan 05, 2023
Source code for Adaptively Calibrated Critic Estimates for Deep Reinforcement Learning

Adaptively Calibrated Critic Estimates for Deep Reinforcement Learning Official implementation of ACC, described in the paper "Adaptively Calibrated C

3 Sep 16, 2022
A heterogeneous entity-augmented academic language model based on Open Academic Graph (OAG)

Library | Paper | Slack We released two versions of OAG-BERT in CogDL package. OAG-BERT is a heterogeneous entity-augmented academic language model wh

THUDM 58 Dec 17, 2022
A privacy-focused, intelligent security camera system.

Self-Hosted Home Security Camera System A privacy-focused, intelligent security camera system. Features: Multi-camera support w/ minimal configuration

Scott Barnes 175 Jan 01, 2023
Ladder Variational Autoencoders (LVAE) in PyTorch

Ladder Variational Autoencoders (LVAE) PyTorch implementation of Ladder Variational Autoencoders (LVAE) [1]: where the variational distributions q at

Andrea Dittadi 63 Dec 22, 2022
Deep Inside Convolutional Networks - This is a caffe implementation to visualize the learnt model

Deep Inside Convolutional Networks This is a caffe implementation to visualize the learnt model. Part of a class project at Georgia Tech Problem State

Jigar 61 Apr 15, 2022
An updated version of virtual model making

Model-Swap-Face v2   这个项目是基于stylegan2 pSp制作的,比v1版本Model-Swap-Face在推理速度和图像质量上有一定提升。主要的功能是将虚拟模特进行环球不同区域的风格转换,目前转换器提供西欧模特、东亚模特和北非模特三种主流的风格样式,可帮我们实现生产资料零成

seeprettyface.com 62 Dec 09, 2022
SPLADE: Sparse Lexical and Expansion Model for First Stage Ranking

SPLADE 🍴 + 🥄 = 🔎 This repository contains the weights for four models as well as the code for running inference for our two papers: [v1]: SPLADE: S

NAVER 170 Dec 28, 2022
UMT is a unified and flexible framework which can handle different input modality combinations, and output video moment retrieval and/or highlight detection results.

Unified Multi-modal Transformers This repository maintains the official implementation of the paper UMT: Unified Multi-modal Transformers for Joint Vi

Applied Research Center (ARC), Tencent PCG 84 Jan 04, 2023
Repository of Vision Transformer with Deformable Attention

Vision Transformer with Deformable Attention This repository contains the code for the paper Vision Transformer with Deformable Attention [arXiv]. Int

410 Jan 03, 2023
PyTorch code for the ICCV'21 paper: "Always Be Dreaming: A New Approach for Class-Incremental Learning"

Always Be Dreaming: A New Approach for Data-Free Class-Incremental Learning PyTorch code for the ICCV 2021 paper: Always Be Dreaming: A New Approach f

49 Dec 21, 2022
Regulatory Instruments for Fair Personalized Pricing.

Fair pricing Source code for WWW 2022 paper Regulatory Instruments for Fair Personalized Pricing. Installation Requirements Linux with Python = 3.6 p

Renzhe Xu 6 Oct 26, 2022
Official code release for "Learned Spatial Representations for Few-shot Talking-Head Synthesis" ICCV 2021

Official code release for "Learned Spatial Representations for Few-shot Talking-Head Synthesis" ICCV 2021

Moustafa Meshry 16 Oct 05, 2022
Python Implementation of Chess Playing AI with variable difficulty

Chess AI with variable difficulty level implemented using the MiniMax AB-Pruning Algorithm

Ali Imran 7 Feb 20, 2022
PyTorch Implementation of AnimeGANv2

PyTorch implementation of AnimeGANv2

4k Jan 07, 2023