Fasttext: Multiple labels in training data

[duyvuleo]

Hi,

Is it possible for fastText to run on the training data in which multiple labels for each instance are employed, e.g.,

__label__A __label__C particularly in already existing electrical buildings installations , it is a simple matter to integrate the inventive circuit .

Thanks!

[spate141]

You can use multiple labels like this:
__label__A, __label__C particularly in already existing electrical buildings installations , it is a simple matter to integrate the inventive circuit .

Related information here : #72

[duyvuleo]

@spate141 : Thanks for pointing to the related issue. Still a bit confused. Which one of the followings is correct:

1) __label__A, __label__C particularly in already existing electrical buildings installations , it is a simple matter to integrate the inventive circuit .

2) __label__A , __label__C particularly in already existing electrical buildings installations , it is a simple matter to integrate the inventive circuit .

Thanks!

[spate141]

It doesn't make any difference! You can use either of any format,
__label__1,__label__2 or __label__1 , __label__2 or __label__1, __label__2

[duyvuleo]

Cool. Thanks!

[nguyentheanhhnvntd]

Hello, if I use __label__1,__label__2 and it will consider that just one label, right @spate141 ??