K BERT
1.0.0
Código Sorce y conjuntos de datos para "K-Bert: habilitando la representación del lenguaje con el gráfico de conocimiento", que se implementa en función del marco UER.
Noticias
Software:
Python3
Pytorch >= 1.0
argparse == 1.1
google_model.bin desde aquí y guárdelo en los models/ directorio.CnDbpedia.spo desde aquí y guárdelo en el brain/kgs/ directorio.datasets/ directorio.El árbol de directorio de K-Bert:
K-BERT
├── brain
│ ├── config.py
│ ├── __init__.py
│ ├── kgs
│ │ ├── CnDbpedia.spo
│ │ ├── HowNet.spo
│ │ └── Medical.spo
│ └── knowgraph.py
├── datasets
│ ├── book_review
│ │ ├── dev.tsv
│ │ ├── test.tsv
│ │ └── train.tsv
│ ├── chnsenticorp
│ │ ├── dev.tsv
│ │ ├── test.tsv
│ │ └── train.tsv
│ ...
│
├── models
│ ├── google_config.json
│ ├── google_model.bin
│ └── google_vocab.txt
├── outputs
├── uer
├── README.md
├── requirements.txt
├── run_kbert_cls.py
└── run_kbert_ner.py
Ejemplo de ejemplo en la revisión del libro con CNDBPEDIA:
CUDA_VISIBLE_DEVICES= ' 0 ' nohup python3 -u run_kbert_cls.py
--pretrained_model_path ./models/google_model.bin
--config_path ./models/google_config.json
--vocab_path ./models/google_vocab.txt
--train_path ./datasets/book_review/train.tsv
--dev_path ./datasets/book_review/dev.tsv
--test_path ./datasets/book_review/test.tsv
--epochs_num 5 --batch_size 32 --kg_name CnDbpedia
--output_model_path ./outputs/kbert_bookreview_CnDbpedia.bin
> ./outputs/kbert_bookreview_CnDbpedia.log &Resultados:
Best accuracy in dev : 88.80%
Best accuracy in test: 87.69%
Opciones de run_kbert_cls.py :
useage: [--pretrained_model_path] - Path to the pre-trained model parameters.
[--config_path] - Path to the model configuration file.
[--vocab_path] - Path to the vocabulary file.
--train_path - Path to the training dataset.
--dev_path - Path to the validating dataset.
--test_path - Path to the testing dataset.
[--epochs_num] - The number of training epoches.
[--batch_size] - Batch size of the training process.
[--kg_name] - The name of knowledge graph, "HowNet", "CnDbpedia" or "Medical".
[--output_model_path] - Path to the output model.
Precisión (dev/test %) en diferentes conjuntos de datos:
| Conjunto de datos | Tocador | Cndbpedia |
|---|---|---|
| Reseña de libros | 88.75/87.75 | 88.80/87.69 |
| Chnsenticorp | 95.00/95.50 | 94.42/95.25 |
| Compras | 97.01/96.92 | 96.94/96.73 |
| 98.22/98.33 | 98.29/98.33 | |
| LCQMC | 88.97/87.14 | 88.91/87.20 |
| Xnli | 77.11/77.07 | 76.99/77.43 |
Ejecute un ejemplo en el conjunto de datos MSRA_NER con CNDBPEDIA:
CUDA_VISIBLE_DEVICES='0' nohup python3 -u run_kbert_ner.py
--pretrained_model_path ./models/google_model.bin
--config_path ./models/google_config.json
--vocab_path ./models/google_vocab.txt
--train_path ./datasets/msra_ner/train.tsv
--dev_path ./datasets/msra_ner/dev.tsv
--test_path ./datasets/msra_ner/test.tsv
--epochs_num 5 --batch_size 16 --kg_name CnDbpedia
--output_model_path ./outputs/kbert_msraner_CnDbpedia.bin
> ./outputs/kbert_msraner_CnDbpedia.log &
Resultados:
The best in dev : precision=0.957, recall=0.962, f1=0.960
The best in test: precision=0.953, recall=0.959, f1=0.956
Opciones de run_kbert_ner.py :
useage: [--pretrained_model_path] - Path to the pre-trained model parameters.
[--config_path] - Path to the model configuration file.
[--vocab_path] - Path to the vocabulary file.
--train_path - Path to the training dataset.
--dev_path - Path to the validating dataset.
--test_path - Path to the testing dataset.
[--epochs_num] - The number of training epoches.
[--batch_size] - Batch size of the training process.
[--kg_name] - The name of knowledge graph.
[--output_model_path] - Path to the output model.
Resultados experimentales en tareas específicas del dominio (precisión/recuperación/F1 %):
| Kg | Finanzas_qa | Law_qa | Finanza_ner | Medicina_ner |
|---|---|---|---|---|
| Tocador | 0.805/0.888/0.845 | 0.842/0.903/0.871 | 0.860/0.888/0.874 | 0.935/0.939/0.937 |
| Cn-dbpedia | 0.814/0.881/0.846 | 0.814/0.942/0.874 | 0.860/0.887/0.873 | 0.935/0.937/0.936 |
| Médico | - | - | - | 0.944/0.943/0.944 |
Este trabajo es un estudio conjunto con el apoyo de la Universidad de Pekín y Tencent Inc.
Si usa este código, cite este documento:
@inproceedings{weijie2019kbert,
title={{K-BERT}: Enabling Language Representation with Knowledge Graph},
author={Weijie Liu, Peng Zhou, Zhe Zhao, Zhiruo Wang, Qi Ju, Haotang Deng, Ping Wang},
booktitle={Proceedings of AAAI 2020},
year={2020}
}