neural-question-generation icon indicating copy to clipboard operation
neural-question-generation copied to clipboard

Published 20 hours ago • verified publisher icon seanie12

After training, how to generate a question for a custom text?

Open rajasekar-venkatesan opened this issue 5 years ago • 9 comments

Hi, After I train the model, how do I generate questions for a custom text?

rajasekar-venkatesan avatar Jul 12 '19 06:07 rajasekar-venkatesan

Same question here... I'm looking for a solution to generate a specific domain SQuAD like dataset using a question generation (QG) tool. In the present case, it seems we have to process the input (text) in order to convert it into the expect format (CoNLL). Might be a too heavy process for my need. Seanie did a great job to evaluate QG on SQuAD 1.1 but the solution has not been designed to easily generate questions from arbitrary text.

nleclercq avatar Jul 16 '19 15:07 nleclercq

@rajasekar-venkatesan, @nleclercq TL;DR: I will update the repo so you can generate question for a custom text. Thanks for having interest on my repo. The reason why it is necessary to prerocess text into coNLL format is because the model requires answer position in the passage to generate answer specific question.

seanie12 avatar Jul 17 '19 00:07 seanie12

Hi, Thanks for your repo, I have a question ,after I downloaded the dataset and processed the data should I train the model or do you have trained weights available? and after training the model or loading the trained weights, how can I use it for getting questions from custom paragraph using your new inference.py code? I'd really appreciate your help.

armheb avatar Aug 27 '19 01:08 armheb

Can you please show an example of how to generate questions based on a given paragraph?

armheb avatar Sep 04 '19 23:09 armheb

Was there any update on this request? If not can you please provide the ideal source and target file.

sarthak221995 avatar Nov 06 '19 04:11 sarthak221995

Hi, Thanks for your repo, I have a question ,after I downloaded the dataset and processed the data should I train the model or do you have trained weights available? and after training the model or loading the trained weights, how can I use it for getting questions from custom paragraph using your new inference.py code? I'd really appreciate your help.

hello， do you know how to install torch_scatter?

ECNUHP avatar Feb 12 '20 12:02 ECNUHP

Does anyone able to generate questions from test data? @seanie12 Thanks for providing the code. can you please provide a solution to generate questions from test dataset.

Thanks

bhavikapanara avatar Mar 22 '20 12:03 bhavikapanara

Sorry, I am busy with other projects. I will update the repo as soon as possible. @bhavikapanara

seanie12 avatar Apr 01 '20 00:04 seanie12

Sorry, I am busy with other projects. I will update the repo as soon as possible. @bhavikapanara

Thanks, @seanie12 for your reply.

Hoping that you will update your repo in coming days.

bhavikapanara avatar Apr 01 '20 04:04 bhavikapanara