asahi417's picture
model update
e212abd
metadata
license: cc-by-4.0
metrics:
  - bleu4
  - meteor
  - rouge-l
  - bertscore
  - moverscore
language: en
datasets:
  - lmqg/qag_tweetqa
pipeline_tag: text2text-generation
tags:
  - questions and answers generation
widget:
  - text: >-
      Beyonce further expanded her acting career, starring as blues singer Etta
      James in the 2008 musical biopic, Cadillac Records.
    example_title: Questions & Answers Generation Example 1
model-index:
  - name: research-backup/t5-small-tweetqa-qag-np
    results:
      - task:
          name: Text2text Generation
          type: text2text-generation
        dataset:
          name: lmqg/qag_tweetqa
          type: default
          args: default
        metrics:
          - name: BLEU4 (Question & Answer Generation)
            type: bleu4_question_answer_generation
            value: 10.71
          - name: ROUGE-L (Question & Answer Generation)
            type: rouge_l_question_answer_generation
            value: 34.77
          - name: METEOR (Question & Answer Generation)
            type: meteor_question_answer_generation
            value: 27.8
          - name: BERTScore (Question & Answer Generation)
            type: bertscore_question_answer_generation
            value: 89.48
          - name: MoverScore (Question & Answer Generation)
            type: moverscore_question_answer_generation
            value: 60.53
          - name: QAAlignedF1Score-BERTScore (Question & Answer Generation)
            type: qa_aligned_f1_score_bertscore_question_answer_generation
            value: 90.7
          - name: QAAlignedRecall-BERTScore (Question & Answer Generation)
            type: qa_aligned_recall_bertscore_question_answer_generation
            value: 90.23
          - name: QAAlignedPrecision-BERTScore (Question & Answer Generation)
            type: qa_aligned_precision_bertscore_question_answer_generation
            value: 91.19
          - name: QAAlignedF1Score-MoverScore (Question & Answer Generation)
            type: qa_aligned_f1_score_moverscore_question_answer_generation
            value: 62.94
          - name: QAAlignedRecall-MoverScore (Question & Answer Generation)
            type: qa_aligned_recall_moverscore_question_answer_generation
            value: 61.9
          - name: QAAlignedPrecision-MoverScore (Question & Answer Generation)
            type: qa_aligned_precision_moverscore_question_answer_generation
            value: 64.1

Model Card of research-backup/t5-small-tweetqa-qag-np

This model is fine-tuned version of t5-small for question & answer pair generation task on the lmqg/qag_tweetqa (dataset_name: default) via lmqg. This model is fine-tuned without a task prefix.

Overview

Usage

from lmqg import TransformersQG

# initialize model
model = TransformersQG(language="en", model="research-backup/t5-small-tweetqa-qag-np")

# 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", "research-backup/t5-small-tweetqa-qag-np")
output = pipe("Beyonce further expanded her acting career, starring as blues singer Etta James in the 2008 musical biopic, Cadillac Records.")

Evaluation

Score Type Dataset
BERTScore 89.48 default lmqg/qag_tweetqa
Bleu_1 35.61 default lmqg/qag_tweetqa
Bleu_2 23.38 default lmqg/qag_tweetqa
Bleu_3 15.73 default lmqg/qag_tweetqa
Bleu_4 10.71 default lmqg/qag_tweetqa
METEOR 27.8 default lmqg/qag_tweetqa
MoverScore 60.53 default lmqg/qag_tweetqa
QAAlignedF1Score (BERTScore) 90.7 default lmqg/qag_tweetqa
QAAlignedF1Score (MoverScore) 62.94 default lmqg/qag_tweetqa
QAAlignedPrecision (BERTScore) 91.19 default lmqg/qag_tweetqa
QAAlignedPrecision (MoverScore) 64.1 default lmqg/qag_tweetqa
QAAlignedRecall (BERTScore) 90.23 default lmqg/qag_tweetqa
QAAlignedRecall (MoverScore) 61.9 default lmqg/qag_tweetqa
ROUGE_L 34.77 default lmqg/qag_tweetqa

Training hyperparameters

The following hyperparameters were used during fine-tuning:

  • dataset_path: lmqg/qag_tweetqa
  • dataset_name: default
  • input_types: ['paragraph']
  • output_types: ['questions_answers']
  • prefix_types: None
  • model: t5-small
  • max_length: 256
  • max_length_output: 128
  • epoch: 16
  • batch: 64
  • lr: 0.0001
  • fp16: False
  • random_seed: 1
  • gradient_accumulation_steps: 1
  • 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",
}