Correctly formatted data

[kstats]

Hello!

I'm trying to run this repo, and am running into problems with the data. Can you please share how to obtain and format the data? What was the source?

Thank you!

[joemzhao]

Hi,

I am not the author of this repo but I guess the author used the pickle file in:
https://github.com/luchi007/RNN_Text_Classify

I am also implementing this topic and I found this from the author's blog (unfortunately in Chinese) where RNN_Text_Classify is referenced. Hope this helps.

j

[kstats]

I'm more interested in if this format aligns with the dialogue data produced with the original paper's code (available here: https://github.com/jiweil/Neural-Dialogue-Generation).

That repository takes the past two utterances in as input to the gen/disc models, and I'm wondering if this repository is set up to do the same.

[joemzhao]

I guess the data formats are not aligned. Hence we have to make some modifications accordingly.

[brightmart]

1.only need following data under train_data. each line is a a answer(or query)
'chitchat.train.answer','chitchat.train.query','chitchat.dev.answer','chitchat.dev.query','chitchat.test.answer'.'chitchat.test.query'

2.data/subj0.pkl:https:
find it http://github.com/luchi007/RNN_Text_Classify/tree/master/data

[liuyuemaicha]

sorry, the Correctly formatted data of every train_files like this -- one row , one sentence:

you ' r e a bad girl and a false sister and i have done with you
i ' m not talking about the furs or the silverware
let ' s see you carry her down the hill my boy
perhaps it was just an accident as he claimed at the time
oh what ' s the use porter ?
my father demanded that i say it today in front of everyone
oh you can tell them he is better than good
i ' m usually a very calm and collected person
you ' r e partly right but only partly
all my black shit on black leather big ass medallion and shit on like this
man you ain ' t got no credentials
whatever it is you wish to tell me tell me without any hesitation
then why did you go to work for her ?

[liuyuemaicha]

I have added the training data (only for examples)