CRF learn.

Share:

Description

CRF learn.

Usage

1
2
3
4
5
  crflearn(templet_file, training_file, model_file,
    sigma = 1, freq_thresh = 0, thread_num = 1,
    max_iter = 10000, eta = 1e-04,
    algorithm = c("CRFs", "AP", "PA", "L1CRFs"), depth = 5,
    isfast = TRUE)

Arguments

templet_file

Path of templet file.

training_file

Path of training file.

model_file

Path of model file.

sigma

Gaussian prior or L1 regularizer. Default value is 1.

freq_thresh

Frequency threshold. Default value is 0.

thread_num

Thread number. Default value is 1.

max_iter

Max iteration number. Default value is 10000.

eta

Controls training precision. Default value is 0.0001.

algorithm

Algorithm: CRFs means 'CRFs'; AP means 'Averaged Perceptron'; PA means 'Passive Aggressive'; L1CRFs means 'L1 CRFs'. Default value is 0.

depth

LBFGS depth. Default value is 5.

isfast

A logical value. TRUE means fast train crf, FALSE means slow but requires less memory. Default value is TRUE.

Value

A list with the input.

Author(s)

Jian Li <rweibo@sina.com>

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.