zhitinghu/logicnn

logicnn's Issues

preprocess_stsa.py throws MemoryError

This is what I run : python preprocess_stsa.py ./raw/ ./w2v/GoogleNews-vectors-negative300.bin

And this is what I get:

loading data...data loaded!
number of status: 79653
vocab size: 33797
max sentence length: 54
loading word2vec vectors...
Traceback (most recent call last):
File "C:\Users\Merom\OneDrive\Desktop\logicnn-master\logicnn-master\data\preprocess_data_updated.py", line 190, in
w2v = load_bin_vec(w2v_file, vocab)
File "C:\Users\Merom\OneDrive\Desktop\logicnn-master\logicnn-master\data\preprocess_data_updated.py", line 113, in load_bin_vec
word.append(ch)
MemoryError

Can you please help?

Some question about the implemention

Hi!
Thanks for your release of the code!
I have some question here. It seems that the implemention is different from the formula(5). Besides, I don't know what's the definition of the 'r' function in formula(4). Could you tell me what it is?
What's more, I am also curiosty about the implemention about NER and it is difficult to understand it if just read your paper. Would you mind release your code about NER?
Thanks very mush!

Question for calc_rule_constraints in logicnn_classes.py

Hi, thanks for sharing the code. I have some questions following.
The details in function calc_rule_constraints of logicnn_classes.py make me puzzled.

1. What's the aim to add the last distr again out of the loop?
2. Why to minus the distr_y0 from distr_all?

question on log_distribution function for BUT-rule.

Hello Zhiting,

It seems your implementation is quite different from the formula(5) in the paper. It is very confused. Could you explain more details on the log_distribution function. Thanks a lot.

The code for RNN?

Hi,

Do you have plan to release the code for RNN (NER problem) in the paper?

Thanks.