Korean stereoypte detector with TUNiB-Electra and K-StereoSet

Last update: Feb 18, 2022

Related tags

Text Data & NLP Stereotype-Detector

Overview

Korean Stereotype Detector

Korean stereotype sentence classifier using K-StereoSet with TUNiB-Electra

Web demo

you can test this model easily in demo page
LINK: https://share.streamlit.io/newfull5/stereotype-detector/demo.py

Dataset

There are 12,688 human-labeled comments in total.
They ar splitted into 10,688 train set, 1000 test set and 1000 valid set

stereotype, anti-stereotype, unrelated, profession, race, gender, religion, comment
1.0,    0.0,    0.0,    1.0,    0.0,    0.0,    0.0,    축구선수의 오랜 경력의 전형적인 결과는 부상을 입은 발목과 두개골이다.
1.0,    0.0,    0.0,    0.0,    1.0,    0.0,    0.0,    콜롬비아 사람들은 모두 마약 카르텔에 소속되어 있다.
0.0,    1.0,    0.0,    0.0,    1.0,    0.0,    0.0,    그는 무기를 소지한 적이 없다.
0.0,    1.0,    0.0,    0.0,    0.0,    1.0,    0.0,    그는 주말에만 지루할 수 있는 일을 한다.

Detail

	stereotype	anti-stereotype	unrelated	profession	race	gender	religion	Total
Train	3,550	3,556	3,581	4,140	4,896	1,268	383	10,688
Valid	341	347	312	410	435	110	45	1,000
Test	334	324	336	361	483	113	43	1,000

Score

	precision	recall	F1
stereotype	0.814	0.601	0.691
anti-stereotype	0.894	0.509	0.648
unrelated	0.872	0.870	0.871
profession	0.943	0.711	0.811
race	0.787	0.907	0.843
gender	0.639	0.836	0.724
religion	0.724	1.0	0.840
total (macro score)	0.810	0.776	0.775

Usage

training

python3 train.py --model_name tunib/electra-ko-base \
                 --data_dir YOUR_PATH \
                 --batch_size BATCH_SIZE \

threshold optimizing

python3 threshold.py --model_name tunib/electra-ko-base \
                     --data_dir YOUR_CKPT_DIR_PATH \
                     --file_path YOUR_CKPT_FILE_NAME \
                     --batch_size BATCH_SIZE \
                     --data_path TEST_DATA_PATH

test

python3 score.py --model_name tunib/electra-ko-base \
                 --data_dir YOUR_CKPT_DIR_PATH \
                 --file_path YOUR_CKPT_FILE_NAME \
                 --batch_size BATCH_SIZE \
                 --data_path TEST_DATA_PATH

Korean stereoypte detector with TUNiB-Electra and K-StereoSet

Related tags

Overview

Korean Stereotype Detector

Web demo

Dataset

Detail

Score

Usage

Owner

Sae_Chan_Oh

OCR을 이용하여 인원수를 인식 후 줌을 Kill 해줍니다

spaCy plugin for Transformers , Udify, ELmo, etc.

Client library to download and publish models and other files on the huggingface.co hub

A PyTorch implementation of VIOLET

Continuously update some NLP practice based on different tasks.

An example project using OpenPrompt under pytorch-lightning for prompt-based SST2 sentiment analysis model

vits chinese, tts chinese, tts mandarin

Source code of the "Graph-Bert: Only Attention is Needed for Learning Graph Representations" paper

Under the hood working of transformers, fine-tuning GPT-3 models, DeBERTa, vision models, and the start of Metaverse, using a variety of NLP platforms: Hugging Face, OpenAI API, Trax, and AllenNLP

L3Cube-MahaCorpus a Marathi monolingual data set scraped from different internet sources.

Code for the paper "Language Models are Unsupervised Multitask Learners"

中文問句產生器；使用台達電閱讀理解資料集(DRCD)

LewusBot - Twitch ChatBot built in python with twitchio library

Simple Speech to Text, Text to Speech

💛 Code and Dataset for our EMNLP 2021 paper: "Perspective-taking and Pragmatics for Generating Empathetic Responses Focused on Emotion Causes"

Official code for "Parser-Free Virtual Try-on via Distilling Appearance Flows", CVPR 2021

Visual Automata is a Python 3 library built as a wrapper for Caleb Evans' Automata library to add more visualization features.

test

Bidirectional LSTM-CRF and ELMo for Named-Entity Recognition, Part-of-Speech Tagging and so on.

Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities