Awesome

Insurance-QA deeplearning model

This is a repo for Q&A Mathing, includes some deep learning models, such as CNN、RNN.

CNN. Basic CNN model from 《Applying Deep Learning To Answer Selection: A Study And An Open Task》
RNN. RNN seems the best model on Insurance-QA dataset.
SWEM. SWEM is the fastest, and has good effect on other datasets, such as WikiQA ..., but is seems not so good on Insurance-QA dataset. I think that, SWEM is more suitable for Q&Q matching, not Q&A matching.

It's hard to say which model is the best in other datasets, you have to choose the most suitable model for you. More models are on the way, pay attention to the updates.

Requirements

tensorflow 1.4.0
python3.5

Performance

margin loss version

Model/Score	Ins_qa_top1_precision	quora_best_prec
CNN	62%	None
LSTM+CNN	68%	None
SWEM	<55%	None

logloss version

Model/Score	Insqa_top1_precision	quora_best_prec
CNN	None	79.60%
LSTM+CNN	None	None
SWEM	<40%	82.69%

Running

Change configuration to your own environment, just like data pathes

vim config.py

Data processing

python3 gen.py

Run CNN model

cd ./cnn/tensorflow && python3 insqa_train.py

It will take few hours(thousands of epoches) to train this model on a single GPU.

Downloads