I'm using FastText pre-trained-embedding for tackling a classification task, but I saw it supports also online training (incremental training) for adding domain-specific corpus.

How does it work?

As far as I know, starting from the "model.bin" file it retrains the model only on the new corpus updating the old word-vectors, is it right?

  • $\begingroup$ Are you referring to github.com/facebookresearch/fastText/pull/423 ? It seems the official FastText does not support incremental learning yet. $\endgroup$ Jun 25, 2019 at 13:01
  • $\begingroup$ yeah exactly i saw that link. Probably i misunderstood. Anyway i opted for the pretrained one with Wikipedia corpus. Thanks for your answer, hoping and waiting for the official "incremental" features in Fast Textt as soon as possible :). $\endgroup$
    – Alfonso
    Jun 25, 2019 at 20:05


You must log in to answer this question.