模型:
lmqg/mt5-small-dequad-qg
这个模型是基于 google/mt5-small 在 lmqg/qg_dequad 数据集上通过 lmqg 进行的问句生成任务的微调版本(数据集名称:default)。
from lmqg import TransformersQG # initialize model model = TransformersQG(language="de", model="lmqg/mt5-small-dequad-qg") # model prediction questions = model.generate_q(list_context="das erste weltweit errichtete Hermann Brehmer 1855 im niederschlesischen ''Görbersdorf'' (heute Sokołowsko, Polen).", list_answer="1855")
from transformers import pipeline
pipe = pipeline("text2text-generation", "lmqg/mt5-small-dequad-qg")
output = pipe("Empfangs- und Sendeantenne sollen in ihrer Polarisation übereinstimmen, andernfalls <hl> wird die Signalübertragung stark gedämpft. <hl>")
| Score | Type | Dataset | |
|---|---|---|---|
| BERTScore | 79.9 | default | 12313321 |
| Bleu_1 | 10.18 | default | 12313321 |
| Bleu_2 | 4.02 | default | 12313321 |
| Bleu_3 | 1.6 | default | 12313321 |
| Bleu_4 | 0.43 | default | 12313321 |
| METEOR | 11.47 | default | 12313321 |
| MoverScore | 54.64 | default | 12313321 |
| ROUGE_L | 10.08 | default | 12313321 |
| Score | Type | Dataset | |
|---|---|---|---|
| QAAlignedF1Score (BERTScore) | 90.55 | default | 12313321 |
| QAAlignedF1Score (MoverScore) | 64.33 | default | 12313321 |
| QAAlignedPrecision (BERTScore) | 90.59 | default | 12313321 |
| QAAlignedPrecision (MoverScore) | 64.37 | default | 12313321 |
| QAAlignedRecall (BERTScore) | 90.51 | default | 12313321 |
| QAAlignedRecall (MoverScore) | 64.29 | default | 12313321 |
| Score | Type | Dataset | |
|---|---|---|---|
| QAAlignedF1Score (BERTScore) | 81.19 | default | 12313321 |
| QAAlignedF1Score (MoverScore) | 54.3 | default | 12313321 |
| QAAlignedPrecision (BERTScore) | 80 | default | 12313321 |
| QAAlignedPrecision (MoverScore) | 54.04 | default | 12313321 |
| QAAlignedRecall (BERTScore) | 82.46 | default | 12313321 |
| QAAlignedRecall (MoverScore) | 54.59 | default | 12313321 |
在微调过程中使用了以下超参数:
完整的配置可以在 fine-tuning config file 中找到。
@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",
}