from lmqg import TransformersQG

# initialize model
model = TransformersQG(language="en", model="lmqg/t5-base-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")

With transformers

python

from transformers import pipeline

pipe = pipeline("text2text-generation", "lmqg/t5-base-squad-qg")
output = pipe("generate question: <hl> Beyonce <hl> further expanded her acting career, starring as blues singer Etta James in the 2008 musical biopic, Cadillac Records.")

Evaluation

Metric (Question Generation): raw metric file

	Score	Type	Dataset
BERTScore	90.6	default	lmqg/qg_squad
Bleu_1	58.69	default	lmqg/qg_squad
Bleu_2	42.66	default	lmqg/qg_squad
Bleu_3	32.99	default	lmqg/qg_squad
Bleu_4	26.13	default	lmqg/qg_squad
METEOR	26.97	default	lmqg/qg_squad
MoverScore	64.74	default	lmqg/qg_squad
ROUGE_L	53.33	default	lmqg/qg_squad

Metric (Question & Answer Generation, Reference Answer): Each question is generated from the gold answer. raw metric file

	Score	Type	Dataset
QAAlignedF1Score (BERTScore)	95.42	default	lmqg/qg_squad
QAAlignedF1Score (MoverScore)	70.63	default	lmqg/qg_squad
QAAlignedPrecision (BERTScore)	95.48	default	lmqg/qg_squad
QAAlignedPrecision (MoverScore)	70.92	default	lmqg/qg_squad
QAAlignedRecall (BERTScore)	95.37	default	lmqg/qg_squad
QAAlignedRecall (MoverScore)	70.34	default	lmqg/qg_squad

Metric (Question & Answer Generation, Pipeline Approach): Each question is generated on the answer generated by lmqg/t5-base-squad-ae. raw metric file

	Score	Type	Dataset
QAAlignedF1Score (BERTScore)	92.75	default	lmqg/qg_squad
QAAlignedF1Score (MoverScore)	64.36	default	lmqg/qg_squad
QAAlignedPrecision (BERTScore)	92.59	default	lmqg/qg_squad
QAAlignedPrecision (MoverScore)	64.45	default	lmqg/qg_squad
QAAlignedRecall (BERTScore)	92.93	default	lmqg/qg_squad
QAAlignedRecall (MoverScore)	64.35	default	lmqg/qg_squad

Metrics (Question Generation, Out-of-Domain)

Dataset	Type	BERTScore	Bleu_4	METEOR	MoverScore	ROUGE_L	Link
lmqg/qg_squadshifts	amazon	90.75	6.57	22.37	60.8	24.81	link
lmqg/qg_squadshifts	new_wiki	93.02	11.09	27.23	65.97	29.59	link
lmqg/qg_squadshifts	nyt	92.2	7.77	25.16	63.83	24.56	link
lmqg/qg_squadshifts	reddit	90.59	5.68	21.3	60.23	21.96	link
lmqg/qg_subjqa	books	88.14	0.49	13.51	55.65	9.44	link
lmqg/qg_subjqa	electronics	87.71	0.0	16.53	55.77	13.48	link
lmqg/qg_subjqa	grocery	87.46	0.0	16.24	56.59	10.26	link
lmqg/qg_subjqa	movies	87.66	0.72	13.06	55.45	11.89	link
lmqg/qg_subjqa	restaurants	87.83	0.0	13.3	55.45	10.7	link
lmqg/qg_subjqa	tripadvisor	89.23	0.93	16.51	56.67	13.51	link

Training hyperparameters

The following hyperparameters were used during fine-tuning:

dataset_path: lmqg/qg_squad
dataset_name: default
input_types: ['paragraph_answer']
output_types: ['question']
prefix_types: ['qg']
model: t5-base
max_length: 512
max_length_output: 32
epoch: 5
batch: 16
lr: 0.0001
fp16: False
random_seed: 1
gradient_accumulation_steps: 4
label_smoothing: 0.15

The full configuration can be found at fine-tuning config file.

Citation

@inproceedings{ushio-etal-2022-generative,
    title = "{G}enerative {L}anguage {M}odels for {P}aragraph-{L}evel {Q}uestion {G}eneration",
    author = "Ushio, Asahi  and
        Alva-Manchego, Fernando  and
        Camacho-Collados, Jose",
    booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing",
    month = dec,
    year = "2022",
    address = "Abu Dhabi, U.A.E.",
    publisher = "Association for Computational Linguistics",
}

lmqg/t5-base-squad-qg

Model Card of lmqg/t5-base-squad-qg

Overview

Usage

Evaluation

Training hyperparameters

Citation

lmqg/
t5-base-squad-qg

Model Card of `lmqg/t5-base-squad-qg`