Is the whole data supposed to be in an single file without any \n character?

[deshwalmahesh]

Yout example_text_file is long and consists no newline character. I can get the idea about MLM as you could split the data in equal length sentences (if this is what you are doing) but what do you do for NSP? How do you verify that a new Paper / Article has started? It means that Sent_N and Sent_N+1 could be from different articles? Could you please shed some light on how MLM and NSP training have been done?