xingdi-eric-yuan/matchlstm-pytorch

a pytorch implementation of match lstm question answering model

matchlstm-pytorch's Issues

Squad json to txt code?

Hi! Thanks for your last reply! I wonder if you still have the code used to produce the txt files in tokenized_squad_v1.1.2 folder? We are trying to run some expanded Squad dataset that (in original json format) on the implementation so it would be really helpful if you can upload that to your directory!

epoch count to get the result ?

Hello, how much epoch needed to get the result like in README ?
I try running first epoch and it only get:
INFO - epoch=0, valid nll loss=6.04093, valid f1=0.23862, valid em=0.14327, lr=0.002000

Am i in a right path ?

How to run the code for the first time using CPU only?

Hi! Thank you for the implementation! I'm trying to run the train.py file on my laptop, and found an error about using NVIDIA GPU, my computer doesn't have a NVIDIA GPU so I wonder if there is a flag we can use to use CPU only? Many thanks!