'fastText' Wrapper for Text Classification and Word Representation

Learning text representations and text classifiers may rely on the same simple and efficient approach. 'fastText' is an open-source, free, lightweight library that allows users to perform both tasks. It transforms text into continuous vectors that can later be used on many language related task. It works on standard, generic hardware (no 'GPU' required). It also includes model size reduction feature. 'fastText' original source code is available at < https://github.com/facebookresearch/fastText>.


News

0.2.5

  • add get word id function
  • add tokenizer function
  • change the way sentence vector are computed (use fastText internal code to improve speed)
  • remove RcppThread due to change in FB source code (no more print from any threadr)
  • add possibility to interrupte learning

0.2.4

  • major refactoring
  • update to last version of fastText sourcecode
  • sentence representation function
  • add tags function
  • fix compilation on Windows R Dev
  • better Makevars (related to strippedLib task)

0.2.3 (9.11.17)

  • fix a cran note related to the DESCRIPTION file
  • remove documentation not anymore useful because of previous update
  • add some asserts to avoid the case where some sentences have no prediction because all their words are unknown (not seen during training)
  • fix compilation on Mac OS

0.2.2 (07.11.17)

  • make possible to interrupt long computation (not for model training part)
  • add simplify option to predict (to get flat vector as a result)
  • remove prefix label in predict result
  • update fastText source code
  • fix crash when learning and setting verbose to 2 (calling Rcout from multiple threads crash the application)

0.2.1 (18.09.17)

  • fix small bugs in compilation (mostly for mac os)
  • remove all notes (Cran)

0.2.0 (15.09.17)

  • first Cran release
  • covers all basic features of fastText

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.

install.packages("fastrtext")

0.2.5 by Michaël Benesty, 2 months ago


https://github.com/pommedeterresautee/fastrtext, https://pommedeterresautee.github.io/fastrtext/


Report a bug at https://github.com/pommedeterresautee/fastrtext/issues


Browse source code at https://github.com/cran/fastrtext


Authors: Michaël Benesty [aut, cre, cph], Facebook, Inc [cph]


Documentation:   PDF Manual  


MIT + file LICENSE license


Imports methods, Rcpp, assertthat

Suggests knitr, testthat

Linking to Rcpp


Suggested by projector.


See at CRAN