英文

lmqg/mt5-small-esquad-qg的模型卡片

此模型是在 lmqg/qg_esquad 上通过 lmqg fine-tuning的 google/mt5-small 的问题生成任务的优化版本。

概述

用法

from lmqg import TransformersQG

# initialize model
model = TransformersQG(language="es", model="lmqg/mt5-small-esquad-qg")

# model prediction
questions = model.generate_q(list_context="a noviembre , que es también la estación lluviosa.", list_answer="noviembre")
  • 使用transformers
from transformers import pipeline

pipe = pipeline("text2text-generation", "lmqg/mt5-small-esquad-qg")
output = pipe("del <hl> Ministerio de Desarrollo Urbano <hl> , Gobierno de la India.")

评估

Score Type Dataset
BERTScore 84.07 default 12313321
Bleu_1 26.03 default 12313321
Bleu_2 17.75 default 12313321
Bleu_3 12.88 default 12313321
Bleu_4 9.61 default 12313321
METEOR 22.71 default 12313321
MoverScore 59.06 default 12313321
ROUGE_L 24.62 default 12313321
  • 指标(问题和答案生成,参考答案):每个问题都是从黄金答案生成的, raw metric file
Score Type Dataset
QAAlignedF1Score (BERTScore) 89.43 default 12313321
QAAlignedF1Score (MoverScore) 63.73 default 12313321
QAAlignedPrecision (BERTScore) 89.44 default 12313321
QAAlignedPrecision (MoverScore) 63.75 default 12313321
QAAlignedRecall (BERTScore) 89.41 default 12313321
QAAlignedRecall (MoverScore) 63.72 default 12313321
Score Type Dataset
QAAlignedF1Score (BERTScore) 79.89 default 12313321
QAAlignedF1Score (MoverScore) 54.82 default 12313321
QAAlignedPrecision (BERTScore) 77.46 default 12313321
QAAlignedPrecision (MoverScore) 53.31 default 12313321
QAAlignedRecall (BERTScore) 82.56 default 12313321
QAAlignedRecall (MoverScore) 56.52 default 12313321

训练超参数

在微调期间使用了以下超参数:

  • 数据集路径:lmqg/qg_esquad
  • 数据集名称:默认
  • 输入类型:['paragraph_answer']
  • 输出类型:['question']
  • 前缀类型:无
  • 模型:google/mt5-small
  • 最大长度:512
  • 输出最大长度:32
  • 时代:16
  • 批量:64
  • 学习率:0.0005
  • fp16:False
  • 随机种子:1
  • 梯度累积步数:1
  • 标签平滑:0.15

完整的配置可以在 fine-tuning config file 找到。

引用

@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",
}