InsuranceQA_zh

Question and Answering Language Model

Welcome

Convolutional Neural Network for Chinese InsuranceQA Corpus with TensorFlow, implementation of paper "Applying Deep Learning to Answer Selection: A Study and An Open Task".

Deps

• py2
• TensorFlow v1.0+

pip install -r Requirements.txt

Train

python2 train.py

Customize the hyper parameters: python2 train.py --help

Metrics

scripts/start_tensorboard.sh
open http://localhost:6006

在默认参数下，运行57,123 steps: loss 0.591968, acc 0.8

Data

insuranceQA Chinese Corpus

Documentation

https://github.com/l11x0m7/InsuranceQA_zh/wiki

• 网络设计

• 数据设计

• 调优过程

• 结果

• 参考文献

Others

insuranceQA English Corpus

License

Apache 2.0