Edit model card

Model Card of lmqg/mt5-base-zhquad-qg-ae

This model is fine-tuned version of google/mt5-base for question generation and answer extraction jointly on the lmqg/qg_zhquad (dataset_name: default) via lmqg.

Overview

Usage

from lmqg import TransformersQG

# initialize model
model = TransformersQG(language="zh", model="lmqg/mt5-base-zhquad-qg-ae")

# model prediction
question_answer_pairs = model.generate_qa("南安普敦的警察服务由汉普郡警察提供。南安普敦行动的主要基地是一座新的八层专用建筑,造价3000万英镑。该建筑位于南路,2011年启用,靠近南安普敦中央火车站。此前,南安普顿市中心的行动位于市民中心西翼,但由于设施老化,加上计划在旧警察局和地方法院建造一座新博物馆,因此必须搬迁。在Portswood、Banister Park、Hille和Shirley还有其他警察局,在南安普顿中央火车站还有一个英国交通警察局。")
  • With transformers
from transformers import pipeline

pipe = pipeline("text2text-generation", "lmqg/mt5-base-zhquad-qg-ae")

# answer extraction
answer = pipe("generate question: 南安普敦的警察服务由汉普郡警察提供。南安普敦行动的主要基地是一座新的八层专用建筑,造价3000万英镑。该建筑位于南路,2011年启用,靠近<hl> 南安普敦中央 <hl>火车站。此前,南安普顿市中心的行动位于市民中心西翼,但由于设施老化,加上计划在旧警察局和地方法院建造一座新博物馆,因此必须搬迁。在Portswood、Banister Park、Hille和Shirley还有其他警察局,在南安普顿中央火车站还有一个英国交通警察局。")

# question generation
question = pipe("extract answers: 南安普敦的警察服务由汉普郡警察提供。 南安普敦行动的主要基地是一座新的八层专用建筑,造价3000万英镑。 <hl> 该建筑位于南路,2011年启用,靠近 南安普敦中央 火车站。 <hl> 此前,南安普顿市中心的行动位于市民中心西翼,但由于设施老化,加上计划在旧警察局和地方法院建造一座新博物馆,因此必须搬迁。 在Portswood、Banister Park、Hille和Shirley还有其他警察局,在南安普顿中央火车站还有一个英国交通警察局。")

Evaluation

Score Type Dataset
BERTScore 76.82 default lmqg/qg_zhquad
Bleu_1 36.9 default lmqg/qg_zhquad
Bleu_2 25.74 default lmqg/qg_zhquad
Bleu_3 19.13 default lmqg/qg_zhquad
Bleu_4 14.63 default lmqg/qg_zhquad
METEOR 23.69 default lmqg/qg_zhquad
MoverScore 57.24 default lmqg/qg_zhquad
ROUGE_L 34.07 default lmqg/qg_zhquad
Score Type Dataset
QAAlignedF1Score (BERTScore) 78.4 default lmqg/qg_zhquad
QAAlignedF1Score (MoverScore) 53.55 default lmqg/qg_zhquad
QAAlignedPrecision (BERTScore) 75.27 default lmqg/qg_zhquad
QAAlignedPrecision (MoverScore) 51.56 default lmqg/qg_zhquad
QAAlignedRecall (BERTScore) 81.92 default lmqg/qg_zhquad
QAAlignedRecall (MoverScore) 55.82 default lmqg/qg_zhquad
Score Type Dataset
AnswerExactMatch 95.07 default lmqg/qg_zhquad
AnswerF1Score 95.15 default lmqg/qg_zhquad
BERTScore 99.76 default lmqg/qg_zhquad
Bleu_1 92.37 default lmqg/qg_zhquad
Bleu_2 89.37 default lmqg/qg_zhquad
Bleu_3 86.14 default lmqg/qg_zhquad
Bleu_4 82.63 default lmqg/qg_zhquad
METEOR 71.18 default lmqg/qg_zhquad
MoverScore 98.8 default lmqg/qg_zhquad
ROUGE_L 95.72 default lmqg/qg_zhquad

Training hyperparameters

The following hyperparameters were used during fine-tuning:

  • dataset_path: lmqg/qg_zhquad
  • dataset_name: default
  • input_types: ['paragraph_answer', 'paragraph_sentence']
  • output_types: ['question', 'answer']
  • prefix_types: ['qg', 'ae']
  • model: google/mt5-base
  • max_length: 512
  • max_length_output: 32
  • epoch: 5
  • batch: 32
  • lr: 0.0005
  • fp16: False
  • random_seed: 1
  • gradient_accumulation_steps: 2
  • 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",
}
Downloads last month
3
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Dataset used to train lmqg/mt5-base-zhquad-qg-ae

Evaluation results

  • BLEU4 (Question Generation) on lmqg/qg_zhquad
    self-reported
    14.630
  • ROUGE-L (Question Generation) on lmqg/qg_zhquad
    self-reported
    34.070
  • METEOR (Question Generation) on lmqg/qg_zhquad
    self-reported
    23.690
  • BERTScore (Question Generation) on lmqg/qg_zhquad
    self-reported
    76.820
  • MoverScore (Question Generation) on lmqg/qg_zhquad
    self-reported
    57.240
  • QAAlignedF1Score-BERTScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_zhquad
    self-reported
    78.400
  • QAAlignedRecall-BERTScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_zhquad
    self-reported
    81.920
  • QAAlignedPrecision-BERTScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_zhquad
    self-reported
    75.270
  • QAAlignedF1Score-MoverScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_zhquad
    self-reported
    53.550
  • QAAlignedRecall-MoverScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_zhquad
    self-reported
    55.820