Skip to content
forked from ayyyq/T-LSTM

Event factuality prediction.Trigger state LSTM

License

Notifications You must be signed in to change notification settings

zhnzhang/T-LSTM

 
 

Repository files navigation

2020年毕业设计

基于语义理解的事件真实性预测

Datasets/数据集

FACTBANK UW UDS_IH2 MEANTIME

Unified Datasets Tool/统一数据处理工具

工具 论文 需要:python2.7

Baseline/基线模型

https://arxiv.org/pdf/1804.02472.pdf
https://arxiv.org/pdf/1907.03227.pdf

T-LSTM/触发状态长短期记忆网络

通过拓展lstm视野域,加入通过GCN提取的trigger-state。
事件真实性预测回归架构

Experiment/实验

预处理

torchtext glove-42b-300d
transformers bert-large-base

环境

linux 16.04
4*titan xp
python 3.6
pytorch 1.4 cuda10.1

Hyperparameter optimization/超参数优化

工具:NNI

Tuner/调优器

Tree-structured Parzen Estimator

Assessor/评估器

Curve Fitting Assessor

模型优点

  • [✓] 线性迭代+树形结构
  • [✓] 长依赖关系+语义融合

实验结果

  • [✓] MAE+R 指标表现优良

About

Event factuality prediction.Trigger state LSTM

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%