Split size

Thanks for your great work! 

I want to run some experiments on your dataset, could you please tell me the split size of each dataset?  In README I found you set the size of train/dev to be 1,000/100, but in the text you say it is 10,000/100.