bart-large-squad-qg Open-source Question Generation Model - Freely Generate Relevant Questions from Text

Bart Large Squad Qg

Developed by lmqg

This is a question generation model based on the BART-large architecture, specifically fine-tuned on the SQuAD dataset, capable of generating relevant questions from given text.

Question Answering System

Transformers

English#Chinese question generation #High-precision question answering #Text summary conversion

Downloads 84

Release Time : 3/2/2022

Model Overview

This model is mainly used to generate relevant questions from text paragraphs and is particularly suitable for scenarios such as education and question-answering systems. The model is based on the BART-large architecture and has been fine-tuned on the SQuAD dataset.

Model Features

High-quality question generation

Performs well on the SQuAD dataset and can generate high-quality questions relevant to the context

Multi-language support

Although mainly targeted at Chinese, the characteristics of the BART architecture give it the potential to handle multi-language texts

Context understanding

Can understand the text context and generate relevant questions

Model Capabilities

Text generation

Question generation

Context understanding

Use Cases

Education

Automatically generate reading comprehension questions

Automatically generate reading comprehension questions from textbook texts

Achieved a BLEU4 score of 26.17 on the SQuAD dataset

Question-answering system

Data augmentation for question-answering systems

Generate training data for question-answering systems

BERTScore reached 91.0

🚀 Model Card of `lmqg/bart-large-squad-qg`

This model is a fine - tuned version of facebook/bart-large for the question generation task on the lmqg/qg_squad (dataset_name: default) via lmqg. It aims to generate high - quality questions from given text, which is useful in various natural language processing applications.

🚀 Quick Start

Prerequisites

Make sure you have the necessary libraries installed. You can use lmqg or transformers to interact with the model.

Installation

If you haven't installed lmqg or transformers yet, you can install them using pip:

pip install lmqg transformers

✨ Features

High - quality Question Generation: Trained on the lmqg/qg_squad dataset, the model can generate relevant and accurate questions.
Multiple Evaluation Metrics: Evaluated using a variety of metrics such as BLEU4, METEOR, ROUGE - L, BERTScore, and MoverScore, ensuring its performance from different perspectives.
Out - of - Domain Adaptability: Shows decent performance on out - of - domain datasets like lmqg/qg_squadshifts and lmqg/qg_subjqa.

💻 Usage Examples

Basic Usage with `lmqg`

from lmqg import TransformersQG

# initialize model
model = TransformersQG(language="en", model="lmqg/bart-large-squad-qg")

# model prediction
questions = model.generate_q(list_context="William Turner was an English painter who specialised in watercolour landscapes", list_answer="William Turner")

Basic Usage with `transformers`

from transformers import pipeline

pipe = pipeline("text2text-generation", "lmqg/bart-large-squad-qg")
output = pipe("<hl> Beyonce <hl> further expanded her acting career, starring as blues singer Etta James in the 2008 musical biopic, Cadillac Records.")

📚 Documentation

Overview

Property	Details
Model Type	facebook/bart-large
Language	en
Training Data	lmqg/qg_squad (default)
Online Demo	https://autoqg.net/
Repository	https://github.com/asahi417/lm-question-generation
Paper	https://arxiv.org/abs/2210.03992

Evaluation

Question Generation

Metric	Score	Type	Dataset	Link
BERTScore	91	default	lmqg/qg_squad	raw metric file
Bleu_1	58.79	default	lmqg/qg_squad	raw metric file
Bleu_2	42.79	default	lmqg/qg_squad	raw metric file
Bleu_3	33.11	default	lmqg/qg_squad	raw metric file
Bleu_4	26.17	default	lmqg/qg_squad	raw metric file
METEOR	27.07	default	lmqg/qg_squad	raw metric file
MoverScore	64.99	default	lmqg/qg_squad	raw metric file
ROUGE_L	53.85	default	lmqg/qg_squad	raw metric file

Question & Answer Generation (Reference Answer)

Metric	Score	Type	Dataset	Link
QAAlignedF1Score (BERTScore)	95.54	default	lmqg/qg_squad	raw metric file
QAAlignedF1Score (MoverScore)	70.82	default	lmqg/qg_squad	raw metric file
QAAlignedPrecision (BERTScore)	95.59	default	lmqg/qg_squad	raw metric file
QAAlignedPrecision (MoverScore)	71.13	default	lmqg/qg_squad	raw metric file
QAAlignedRecall (BERTScore)	95.49	default	lmqg/qg_squad	raw metric file
QAAlignedRecall (MoverScore)	70.54	default	lmqg/qg_squad	raw metric file

Question & Answer Generation (Pipeline Approach)

Metric	Score	Type	Dataset	Link
QAAlignedF1Score (BERTScore)	93.23	default	lmqg/qg_squad	raw metric file
QAAlignedF1Score (MoverScore)	64.76	default	lmqg/qg_squad	raw metric file
QAAlignedPrecision (BERTScore)	93.13	default	lmqg/qg_squad	raw metric file
QAAlignedPrecision (MoverScore)	64.98	default	lmqg/qg_squad	raw metric file
QAAlignedRecall (BERTScore)	93.35	default	lmqg/qg_squad	raw metric file
QAAlignedRecall (MoverScore)	64.63	default	lmqg/qg_squad	raw metric file

Question Generation (Out - of - Domain)

Dataset	Type	BERTScore	Bleu_4	METEOR	MoverScore	ROUGE_L	Link
lmqg/qg_squadshifts	amazon	90.93	6.53	22.3	60.87	25.03	link
lmqg/qg_squadshifts	new_wiki	93.23	11.12	27.32	66.23	29.68	link
lmqg/qg_squadshifts	nyt	92.49	8.12	25.25	64.06	25.29	link
lmqg/qg_squadshifts	reddit	90.95	5.95	21.5	60.59	22.37	link
lmqg/qg_subjqa	books	88.07	0.63	11.58	55.56	12.37	link
lmqg/qg_subjqa	electronics	87.83	0.87	15.35	56.35	16.02	link
lmqg/qg_subjqa	grocery	87.79	0.53	15.13	57.02	12.34	link
lmqg/qg_subjqa	movies	87.49	0.0	11.86	55.29	12.51	link
lmqg/qg_subjqa	restaurants	87.98	0.0	12.42	55.43	13.08	link
lmqg/qg_subjqa	tripadvisor	88.91	0.0	13.72	56.05	14.03	link

🔧 Technical Details

Training Hyperparameters

The following hyperparameters were used during fine - tuning:

dataset_path: lmqg/qg_squad
dataset_name: default
input_types: ['paragraph_answer']

📄 License

This model is released under the CC - BY - 4.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご