Naga Kiran

Similar error , i too faced. Please change the pretrained weights with default Null string, for the sake first training. Next after that you can use trained model for predictions...

Error is from line 1345 line in run_squad.py, please check the encoding format of FLAGS.context file by opening in python, if possible change the encoding format relevant to that with...

Its Gensim library installation problem bro. you please take care ( https://stackoverflow.com/questions/8115280/importerror-no-module-named-bz2-for-python-2-7-2 ) Regards, Naga.

Hi Vincenzo, Thanks for your response. Please have a look at BERT release from Google-research (https://github.com/google-research/bert). BERT actually trained to understand only 100-200 words paragraphs( SQuAD dataset) only for Question-Answering....

Yes Vincenzo, Thats correct. Just to distinguish feeding context paragraphs to limited words, we have followed that approach. And make sure that each delimited paragraphs have different context(not exactly or...

Hi Vinny, Sry for late response. Could you show the line number, where it came from. mostly we should pass values as class variables than tensorflow.flags, please check. Regards, Naga.

Hi Emanuel, Sorry for the late reply. I appreciate your efforts in using the repository with concise information. Actually, i thought of implementing this approach with BERT model to get...

Manual effort is needed at there in labelling each character, I have labelled some characters which are extracted for Lumos font characters. I just edited readme in providing link for...

Hi Thai Hoc, Actually OpenCV contours are well in segmenting the characters, if not tesseract trained well in doing that so. If both would not work in your case. I...

Hi @Pranav-Goel is your network trained well in answering Custom questions? i have replicated the data with my custom answers with some questions. Its not working well in answering question...