Divesh Pandey

Are versions of keras and tensorflow matching with ```Keras==1.2.1 and tensorflow==0.12.1```?

You can try changing ```regul_alpha``` while keeping dropout 0. Also, if you have sufficient memory, try changing len of vectors to 300. It will boost up results. Regarding imbalanced dataset,...

Currently, I don't have any specific webpages(for NER) but if you like, you can go through these links. [how_to_choose_a_neural_network's_hyper-parameters](http://neuralnetworksanddeeplearning.com/chap3.html#how_to_choose_a_neural_network's_hyper-parameters) [distill.pub](https://distill.pub/)

definitely not an expert :smile:

Sorry for late replying. I had no specific logic while tuning model. The arguments to ```m.make_and_compile()``` played an important factor. The most important one being ```dropout```. In my case, setting...

You should see https://github.com/pandeydivesh15/NER-using-Deep-Learning/blob/master/Task%203:%20Hindi%20data/Convert%20SSF%20data%20to%20CoNLL%20form.ipynb The hindi dataset that I worked on was also in SSF. You will have to first convert your dataset into CoNLL form to be able to...

ok I can try. I will have to look into the data first.

Sorry for late reply. As I can see, you will have to design your own SSF->CoNLL converter. One ques, in line 21/22, the data shows, उत्तरांचल NNP Given this data,...

But still no idea about LOCATION named entity. Do you wish to train a new NER classifier (using this dataset) or you wish to just find/predict named entities for this...

For doing this, you would require already trained model for predicting NER tags. Currently my model isnt much powerful to predict better. You can see the results.