Model Card of lmqg/mbart-large-cc25-frquad-qa
This model is fine-tuned version of facebook/mbart-large-cc25 for question answering task on the lmqg/qg_frquad (dataset_name: default) via lmqg
.
Overview
- Language model: facebook/mbart-large-cc25
- Language: fr
- Training data: lmqg/qg_frquad (default)
- Online Demo: https://autoqg.net/
- Repository: https://github.com/asahi417/lm-question-generation
- Paper: https://arxiv.org/abs/2210.03992
Usage
- With
lmqg
from lmqg import TransformersQG
# initialize model
model = TransformersQG(language="fr", model="lmqg/mbart-large-cc25-frquad-qa")
# model prediction
answers = model.answer_q(list_question="En quelle année a-t-on trouvé trace d'un haut fourneau similaire?", list_context=" Cette technologie ne disparaît qu'au début du XXe siècle. On retrouve vers 1900 un haut fourneau similaire dans le Bulacan, aux Philippines. Plus tard encore, le « haut fourneau dans la cour » prôné par Mao Zedong pendant le Grand Bond en avant est de ce type. L'expérience n'est un échec technique que dans les régions où le savoir-faire n'existe pas, ou a disparu.")
- With
transformers
from transformers import pipeline
pipe = pipeline("text2text-generation", "lmqg/mbart-large-cc25-frquad-qa")
output = pipe("question: En quelle année a-t-on trouvé trace d'un haut fourneau similaire?, context: Cette technologie ne disparaît qu'au début du XXe siècle. On retrouve vers 1900 un haut fourneau similaire dans le Bulacan, aux Philippines. Plus tard encore, le « haut fourneau dans la cour » prôné par Mao Zedong pendant le Grand Bond en avant est de ce type. L'expérience n'est un échec technique que dans les régions où le savoir-faire n'existe pas, ou a disparu.")
Evaluation
- Metric (Question Answering): raw metric file
Score | Type | Dataset | |
---|---|---|---|
AnswerExactMatch | 39.34 | default | lmqg/qg_frquad |
AnswerF1Score | 60.48 | default | lmqg/qg_frquad |
BERTScore | 92.2 | default | lmqg/qg_frquad |
Bleu_1 | 37.27 | default | lmqg/qg_frquad |
Bleu_2 | 32.61 | default | lmqg/qg_frquad |
Bleu_3 | 29.23 | default | lmqg/qg_frquad |
Bleu_4 | 26.33 | default | lmqg/qg_frquad |
METEOR | 31.8 | default | lmqg/qg_frquad |
MoverScore | 77.16 | default | lmqg/qg_frquad |
ROUGE_L | 38.14 | default | lmqg/qg_frquad |
Training hyperparameters
The following hyperparameters were used during fine-tuning:
- dataset_path: lmqg/qg_frquad
- dataset_name: default
- input_types: ['paragraph_question']
- output_types: ['answer']
- prefix_types: None
- model: facebook/mbart-large-cc25
- max_length: 512
- max_length_output: 32
- epoch: 15
- batch: 32
- lr: 0.0002
- 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
- 9
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.
Evaluation results
- BLEU4 (Question Answering) on lmqg/qg_frquadself-reported26.330
- ROUGE-L (Question Answering) on lmqg/qg_frquadself-reported38.140
- METEOR (Question Answering) on lmqg/qg_frquadself-reported31.800
- BERTScore (Question Answering) on lmqg/qg_frquadself-reported92.200
- MoverScore (Question Answering) on lmqg/qg_frquadself-reported77.160
- AnswerF1Score (Question Answering) on lmqg/qg_frquadself-reported60.480
- AnswerExactMatch (Question Answering) on lmqg/qg_frquadself-reported39.340