User Tools

Site Tools


it:fasttext

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
it:fasttext [2019/01/07 12:22]
pmay [Links]
it:fasttext [2019/01/07 12:25] (current)
pmay [Help: fastText Commnand Line Tool]
Line 23: Line 23:
 </​code>​ </​code>​
  
 +<​code>​
 +The following arguments are mandatory:
 +  -input ​             training file path
 +  -output ​            ​output file path
 +
 +The following arguments are optional:
 +  -verbose ​           verbosity level [2]
 +
 +The following arguments for the dictionary are optional:
 +  -minCount ​          ​minimal number of word occurences [5]
 +  -minCountLabel ​     minimal number of label occurences [0]
 +  -wordNgrams ​        max length of word ngram [1]
 +  -bucket ​            ​number of buckets [2000000]
 +  -minn               min length of char ngram [3]
 +  -maxn               max length of char ngram [6]
 +  -t                  sampling threshold [0.0001]
 +  -label ​             labels prefix [__label__]
 +
 +The following arguments for training are optional:
 +  -lr                 ​learning rate [0.05]
 +  -lrUpdateRate ​      ​change the rate of updates for the learning rate [100]
 +  -dim                size of word vectors [100]
 +  -ws                 size of the context window [5]
 +  -epoch ​             number of epochs [5]
 +  -neg                number of negatives sampled [5]
 +  -loss               loss function {ns, hs, softmax, one-vs-all} [ns]
 +  -thread ​            ​number of threads [12]
 +  -pretrainedVectors ​ pretrained word vectors for supervised learning []
 +  -saveOutput ​        ​whether output params should be saved [false]
 +
 +The following arguments for quantization are optional:
 +  -cutoff ​            ​number of words and ngrams to retain [0]
 +  -retrain ​           whether embeddings are finetuned if a cutoff is applied [false]
 +  -qnorm ​             whether the norm is quantized separately [false]
 +  -qout               ​whether the classifier is quantized [false]
 +  -dsub               size of each sub-vector [2]
 +</​code>​
 ===== Help: Python Usage ===== ===== Help: Python Usage =====
 To load a model use ''​load_model''​ of ''​fastText.FastText''​. On how to use it see code of ''​class _FastText''​ in ''​FastText''​ module. ​ To load a model use ''​load_model''​ of ''​fastText.FastText''​. On how to use it see code of ''​class _FastText''​ in ''​FastText''​ module. ​
it/fasttext.txt · Last modified: 2019/01/07 12:25 by pmay