search

ttpro1995 / TreeLSTMSentiment

Pytorch implementation of Sentiment Classification in Improved Semantic Representations From Tree-Structured Long Short-Term Memory Networks
MIT License
113 stars 40 forks source link

Problem with fine-grain classification #8

Closed lihy-1224 closed 5 years ago

lihy-1224 commented 5 years ago

In fine-grained mode, dev percentage is more than 80% in my test, but the result in paper is about 50%. I just wonder how to test the fine-grained classification correctly, thx~

ttpro1995 commented 5 years ago

well, it should like in paper ... :| but i haven't focus on fine grained for a while

ttpro1995 commented 5 years ago

I think you need to fix this https://github.com/ttpro1995/TreeLSTMSentiment/issues/3 and https://github.com/ttpro1995/TreeLSTMSentiment/issues/2 (delete all temp data, because you might use binary preprocess data for fine grain)

please let me know if it work

by the way, please confirm if binary classification work with result as in paper

lihy-1224 commented 5 years ago

thanks, I solve it by remove those preprocess data. And binary/fine-grain classification work with result as in original paper