Closed wendysyn89 closed 6 years ago
Hi, If you want to train a BiLSTM-max on another dataset which for instance includes two input sentences and 5 output classes, you'll have to change the data loader here, propagate this change in the other files, and also change the number of classes. Also you can specify your own word embeddings with this parameter and this parameter in the train_nli.py file. For testing, you'll have to specify the correct word embeddings when setting W2V_PATH in demo.ipynb.
Thank you, Alexis
This project do not support training anymore. So do you know a better code for sentence embedding? Thank you!
Hi,
What should I do or is it possible if I would like to train an infersent model from my own dataset (instead of NLI) or change the vectors file(instead of Glove) for testing?
Your helps are much appreciated.
Thank you.