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

This model is fine-tuned version of google/flan-t5-base for question generation and answer extraction jointly on the lmqg/qg_squad (dataset_name: default) via lmqg.

Overview

• Language model: google/flan-t5-base
• 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

Usage

• With lmqg

from lmqg import TransformersQG

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

# model prediction
question_answer_pairs = model.generate_qa("William Turner was an English painter who specialised in watercolour landscapes")

• With transformers

from transformers import pipeline

pipe = pipeline("text2text-generation", "lmqg/flan-t5-base-squad-qg-ae")

# answer extraction
answer = pipe("generate question: <hl> Beyonce <hl> further expanded her acting career, starring as blues singer Etta James in the 2008 musical biopic, Cadillac Records.")

# question generation
question = pipe("extract answers: <hl> Beyonce further expanded her acting career, starring as blues singer Etta James in the 2008 musical biopic, Cadillac Records. <hl> Her performance in the film received praise from critics, and she garnered several nominations for her portrayal of James, including a Satellite Award nomination for Best Supporting Actress, and a NAACP Image Award nomination for Outstanding Supporting Actress.")

Evaluation

• Metric (Question Generation): raw metric file

	Score	Type	Dataset
BERTScore	90.61	default	lmqg/qg_squad
Bleu_1	58.99	default	lmqg/qg_squad
Bleu_2	42.92	default	lmqg/qg_squad
Bleu_3	33.3	default	lmqg/qg_squad
Bleu_4	26.43	default	lmqg/qg_squad
METEOR	26.99	default	lmqg/qg_squad
MoverScore	64.75	default	lmqg/qg_squad
ROUGE_L	53.37	default	lmqg/qg_squad

• Metric (Question & Answer Generation): raw metric file

	Score	Type	Dataset
QAAlignedF1Score (BERTScore)	93.31	default	lmqg/qg_squad
QAAlignedF1Score (MoverScore)	64.59	default	lmqg/qg_squad
QAAlignedPrecision (BERTScore)	92.65	default	lmqg/qg_squad
QAAlignedPrecision (MoverScore)	63.7	default	lmqg/qg_squad
QAAlignedRecall (BERTScore)	93.99	default	lmqg/qg_squad
QAAlignedRecall (MoverScore)	65.57	default	lmqg/qg_squad

• Metric (Answer Extraction): raw metric file

	Score	Type	Dataset
AnswerExactMatch	57.39	default	lmqg/qg_squad
AnswerF1Score	68.88	default	lmqg/qg_squad
BERTScore	91.28	default	lmqg/qg_squad
Bleu_1	49.4	default	lmqg/qg_squad
Bleu_2	44.53	default	lmqg/qg_squad
Bleu_3	39.73	default	lmqg/qg_squad
Bleu_4	35.6	default	lmqg/qg_squad
METEOR	42.76	default	lmqg/qg_squad
MoverScore	81.31	default	lmqg/qg_squad
ROUGE_L	68.47	default	lmqg/qg_squad

Training hyperparameters

The following hyperparameters were used during fine-tuning:

• dataset_path: lmqg/qg_squad
• dataset_name: default
• input_types: ['paragraph_answer', 'paragraph_sentence']
• output_types: ['question', 'answer']
• prefix_types: ['qg', 'ae']
• model: google/flan-t5-base
• max_length: 512
• max_length_output: 32
• epoch: 7
• batch: 8
• lr: 0.0001
• fp16: False
• random_seed: 1
• gradient_accumulation_steps: 16
• 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",
}