Edit model card

Model Card of lmqg/mbart-large-cc25-esquad-qg-ae

This model is fine-tuned version of facebook/mbart-large-cc25 for question generation and answer extraction jointly on the lmqg/qg_esquad (dataset_name: default) via lmqg.

Overview

Usage

from lmqg import TransformersQG

# initialize model
model = TransformersQG(language="es", model="lmqg/mbart-large-cc25-esquad-qg-ae")

# model prediction
question_answer_pairs = model.generate_qa("a noviembre , que es también la estación lluviosa.")
  • With transformers
from transformers import pipeline

pipe = pipeline("text2text-generation", "lmqg/mbart-large-cc25-esquad-qg-ae")

# answer extraction
answer = pipe("generate question: del <hl> Ministerio de Desarrollo Urbano <hl> , Gobierno de la India.")

# question generation
question = pipe("extract answers: <hl> En la diáspora somalí, múltiples eventos islámicos de recaudación de fondos se llevan a cabo cada año en ciudades como Birmingham, Londres, Toronto y Minneapolis, donde los académicos y profesionales somalíes dan conferencias y responden preguntas de la audiencia. <hl> El propósito de estos eventos es recaudar dinero para nuevas escuelas o universidades en Somalia, para ayudar a los somalíes que han sufrido como consecuencia de inundaciones y / o sequías, o para reunir fondos para la creación de nuevas mezquitas como.")

Evaluation

Score Type Dataset
BERTScore 79.36 default lmqg/qg_esquad
Bleu_1 22.05 default lmqg/qg_esquad
Bleu_2 14.55 default lmqg/qg_esquad
Bleu_3 10.34 default lmqg/qg_esquad
Bleu_4 7.61 default lmqg/qg_esquad
METEOR 19.58 default lmqg/qg_esquad
MoverScore 56.05 default lmqg/qg_esquad
ROUGE_L 20.95 default lmqg/qg_esquad
Score Type Dataset
QAAlignedF1Score (BERTScore) 81.13 default lmqg/qg_esquad
QAAlignedF1Score (MoverScore) 54.86 default lmqg/qg_esquad
QAAlignedPrecision (BERTScore) 77.75 default lmqg/qg_esquad
QAAlignedPrecision (MoverScore) 52.82 default lmqg/qg_esquad
QAAlignedRecall (BERTScore) 84.91 default lmqg/qg_esquad
QAAlignedRecall (MoverScore) 57.16 default lmqg/qg_esquad
Score Type Dataset
AnswerExactMatch 52.81 default lmqg/qg_esquad
AnswerF1Score 70.95 default lmqg/qg_esquad
BERTScore 86.7 default lmqg/qg_esquad
Bleu_1 32.77 default lmqg/qg_esquad
Bleu_2 28.12 default lmqg/qg_esquad
Bleu_3 24.52 default lmqg/qg_esquad
Bleu_4 21.5 default lmqg/qg_esquad
METEOR 40.42 default lmqg/qg_esquad
MoverScore 77.96 default lmqg/qg_esquad
ROUGE_L 46.66 default lmqg/qg_esquad

Training hyperparameters

The following hyperparameters were used during fine-tuning:

  • dataset_path: lmqg/qg_esquad
  • dataset_name: default
  • input_types: ['paragraph_answer', 'paragraph_sentence']
  • output_types: ['question', 'answer']
  • prefix_types: ['qg', 'ae']
  • model: facebook/mbart-large-cc25
  • max_length: 512
  • max_length_output: 32
  • epoch: 5
  • batch: 2
  • lr: 0.0001
  • fp16: False
  • random_seed: 1
  • gradient_accumulation_steps: 32
  • 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
2

Dataset used to train research-backup/mbart-large-cc25-esquad-qg-ae

Evaluation results

  • BLEU4 (Question Generation) on lmqg/qg_esquad
    self-reported
    7.610
  • ROUGE-L (Question Generation) on lmqg/qg_esquad
    self-reported
    20.950
  • METEOR (Question Generation) on lmqg/qg_esquad
    self-reported
    19.580
  • BERTScore (Question Generation) on lmqg/qg_esquad
    self-reported
    79.360
  • MoverScore (Question Generation) on lmqg/qg_esquad
    self-reported
    56.050
  • QAAlignedF1Score-BERTScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_esquad
    self-reported
    81.130
  • QAAlignedRecall-BERTScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_esquad
    self-reported
    84.910
  • QAAlignedPrecision-BERTScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_esquad
    self-reported
    77.750
  • QAAlignedF1Score-MoverScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_esquad
    self-reported
    54.860
  • QAAlignedRecall-MoverScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_esquad
    self-reported
    57.160
  • QAAlignedPrecision-MoverScore (Question & Answer Generation (with Gold Answer)) on lmqg/qg_esquad
    self-reported
    52.820
  • BLEU4 (Answer Extraction) on lmqg/qg_esquad
    self-reported
    21.500
  • ROUGE-L (Answer Extraction) on lmqg/qg_esquad
    self-reported
    46.660
  • METEOR (Answer Extraction) on lmqg/qg_esquad
    self-reported
    40.420
  • BERTScore (Answer Extraction) on lmqg/qg_esquad
    self-reported
    86.700
  • MoverScore (Answer Extraction) on lmqg/qg_esquad
    self-reported
    77.960
  • AnswerF1Score (Answer Extraction) on lmqg/qg_esquad
    self-reported
    70.950
  • AnswerExactMatch (Answer Extraction) on lmqg/qg_esquad
    self-reported
    52.810