mt5-base-zhquad-ae / README.md
asahi417's picture
commit files to HF hub
cb578da
metadata
license: cc-by-4.0
metrics:
  - bleu4
  - meteor
  - rouge-l
  - bertscore
  - moverscore
language: zh
datasets:
  - lmqg/qg_zhquad
pipeline_tag: text2text-generation
tags:
  - answer extraction
widget:
  - text: >-
      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。
    example_title: Answering Extraction Example 1
model-index:
  - name: lmqg/mt5-base-zhquad-ae
    results:
      - task:
          name: Text2text Generation
          type: text2text-generation
        dataset:
          name: lmqg/qg_zhquad
          type: default
          args: default
        metrics:
          - name: BLEU4 (Answer Extraction)
            type: bleu4_answer_extraction
            value: 79.86
          - name: ROUGE-L (Answer Extraction)
            type: rouge_l_answer_extraction
            value: 94.53
          - name: METEOR (Answer Extraction)
            type: meteor_answer_extraction
            value: 68.41
          - name: BERTScore (Answer Extraction)
            type: bertscore_answer_extraction
            value: 99.48
          - name: MoverScore (Answer Extraction)
            type: moverscore_answer_extraction
            value: 97.97
          - name: AnswerF1Score (Answer Extraction)
            type: answer_f1_score__answer_extraction
            value: 92.68
          - name: AnswerExactMatch (Answer Extraction)
            type: answer_exact_match_answer_extraction
            value: 92.62

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

This model is fine-tuned version of google/mt5-base for answer extraction 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-ae")

# model prediction
answers = model.generate_a("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-ae")
output = pipe("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
AnswerExactMatch 92.62 default lmqg/qg_zhquad
AnswerF1Score 92.68 default lmqg/qg_zhquad
BERTScore 99.48 default lmqg/qg_zhquad
Bleu_1 90.95 default lmqg/qg_zhquad
Bleu_2 87.44 default lmqg/qg_zhquad
Bleu_3 83.75 default lmqg/qg_zhquad
Bleu_4 79.86 default lmqg/qg_zhquad
METEOR 68.41 default lmqg/qg_zhquad
MoverScore 97.97 default lmqg/qg_zhquad
ROUGE_L 94.53 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_sentence']
  • output_types: ['answer']
  • prefix_types: None
  • model: google/mt5-base
  • max_length: 512
  • max_length_output: 32
  • epoch: 18
  • batch: 8
  • lr: 0.0001
  • fp16: False
  • random_seed: 1
  • gradient_accumulation_steps: 8
  • 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",
}