Thank you for pointing this out. The Project Details document has been corrected to say that your answers should be output to the screen only. No output file should be generated.

The story input files are in the one-sentence-per-line format. The answer files for the training & test sets are in the multiple-sentences-per-line format. This was done out of laziness (one of the virtues of a good engineer) — we need to make less modifications to the grading script this way. The instructions given on the website for input and output should be followed, regardless of the format of the answer files.

Unix and Windows handle end-of-lines differently. In Windows, lines end with both a line feed character and a carriage return character. In Unix, lines only end with a line feed character. As a result, when you open up the story files in Notepad (or most other Windows text editors), it will appear that all the sentences are on the same line. This shouldn't effect you at all though. Most programming environments should interpet the line feed character as an end-of-line so everything should work properly.

A common problem in this situation is that the case of the username/password is important. The case specified on the NDA has been used. There may be instances where the intended case is different from the interpreted case. If you are having problems with accessing the training/test files, you may want to send an email to verify the case of your username/password.