| fasttext_interface {fastText} | R Documentation |
Interface for the fasttext library
Description
Interface for the fasttext library
Usage
fasttext_interface(
list_params,
path_output = "",
MilliSecs = 100,
path_input = "",
remove_previous_file = TRUE,
print_process_time = FALSE
)
Arguments
list_params |
a list of valid parameters |
path_output |
a character string specifying the file path where the process-logs (or output in generally) should be saved |
MilliSecs |
an integer specifying the delay in milliseconds when printing the results to the specified path_output |
path_input |
a character string specifying the path to the input data file |
remove_previous_file |
a boolean. If TRUE, in case that the path_output is not an empty string (""), then an existing file with the same output name will be removed |
print_process_time |
a boolean. If TRUE then the processing time of the function will be printed out in the R session |
Details
This function allows the user to run the various methods included in the fasttext library from within R
The "output" parameter which exists in the named list (see examples section) and is passed to the "list_params" parameter of the "fasttext_interface()" function, is a file path and not a directory name and will actually return two files (a *.vec* and a *.bin*) to the output directory.
Value
a vector of class character that includes the parameters and file paths used as input to the function
References
https://github.com/facebookresearch/fastText
https://github.com/facebookresearch/fastText/blob/master/docs/supervised-tutorial.md
Examples
## Not run:
library(fastText)
####################################################################################
# If the user intends to run the following examples then he / she must replace #
# the 'input', 'output', 'path_input', 'path_output', 'model' and 'test_data' file #
# paths depending on where the data are located or should be saved! #
# ( 'tempdir()' is used here as an example folder ) #
####################################################################################
# ------------------------------------------------
# print information for the Usage of each function [ parameters ]
# ------------------------------------------------
fastText::printUsage()
fastText::printTestUsage()
fastText::printTestLabelUsage()
fastText::printQuantizeUsage()
fastText::printPrintWordVectorsUsage()
fastText::printPrintSentenceVectorsUsage()
fastText::printPrintNgramsUsage()
fastText::printPredictUsage()
fastText::printNNUsage()
fastText::printDumpUsage()
fastText::printAnalogiesUsage()
fastText::print_parameters(command = "supervised")
# -----------------------------------------------------------------------
# In case that the 'command' is one of 'cbow', 'skipgram' or 'supervised'
# -----------------------------------------------------------------------
list_params = list(command = 'cbow',
lr = 0.1,
dim = 200,
input = file.path(tempdir(), "doc.txt"),
output = tempdir(),
verbose = 2,
thread = 1)
res = fasttext_interface(list_params,
path_output = file.path(tempdir(),"model_logs.txt"),
MilliSecs = 100)
# ---------------------
# 'supervised' training
# ---------------------
list_params = list(command = 'supervised',
lr = 0.1,
dim = 200,
input = file.path(tempdir(), "cooking.train"),
output = file.path(tempdir(), "model_cooking"),
verbose = 2,
thread = 1)
res = fasttext_interface(list_params,
path_output = file.path(tempdir(), 'logs_supervise.txt'),
MilliSecs = 5)
# ---------------------------------------
# In case that the 'command' is 'predict'
# ---------------------------------------
list_params = list(command = 'predict',
model = file.path(tempdir(), 'model_cooking.bin'),
test_data = file.path(tempdir(), 'cooking.valid'),
k = 1,
th = 0.0)
res = fasttext_interface(list_params,
path_output = file.path(tempdir(), 'predict_valid.txt'))
# ------------------------------------
# In case that the 'command' is 'test' [ k = 5 , means that precision and recall are at 5 ]
# ------------------------------------
list_params = list(command = 'test',
model = file.path(tempdir(), 'model_cooking.bin'),
test_data = file.path(tempdir(), 'cooking.valid'),
k = 5,
th = 0.0)
res = fasttext_interface(list_params) # It only prints 'Precision', 'Recall' to the R session
# ------------------------------------------
# In case that the 'command' is 'test-label' [ k = 5 , means that precision and recall are at 5 ]
# ------------------------------------------
list_params = list(command = 'test-label',
model = file.path(tempdir(), 'model_cooking.bin'),
test_data = file.path(tempdir(), 'cooking.valid'),
k = 5,
th = 0.0)
res = fasttext_interface(list_params, # prints also 'Precision', 'Recall' to R session
path_output = file.path(tempdir(), "test_valid.txt"))
# -----------------
# quantize function [ it will take a .bin file and return an .ftz file ]
# -----------------
# the quantize function is currenlty (01/02/2019) single-threaded
# https://github.com/facebookresearch/fastText/issues/353#issuecomment-342501742
list_params = list(command = 'quantize',
input = file.path(tempdir(), 'model_cooking.bin'),
output = file.path(tempdir(), gsub('.bin', '.ftz', 'model_cooking.bin')))
res = fasttext_interface(list_params)
# -----------------
# quantize function [ by using the optional parameters 'qnorm' and 'qout' ]
# -----------------
list_params = list(command = 'quantize',
input = file.path(tempdir(), 'model_cooking.bin'),
output = file.path(tempdir(), gsub('.bin', '.ftz', 'model_cooking.bin')),
qnorm = TRUE,
qout = TRUE)
res = fasttext_interface(list_params)
# ------------------
# print-word-vectors [ each line of the 'queries.txt' must be a single word ]
# ------------------
list_params = list(command = 'print-word-vectors',
model = file.path(tempdir(), 'model_cooking.bin'))
res = fasttext_interface(list_params,
path_input = file.path(tempdir(), 'queries.txt'),
path_output = file.path(tempdir(), 'print_vecs_file.txt'))
# ----------------------
# print-sentence-vectors [ See also the comments in the main.cc file about the input-file ]
# ----------------------
list_params = list(command = 'print-sentence-vectors',
model = file.path(tempdir(), 'model_cooking.bin'))
res = fasttext_interface(list_params,
path_input = file.path(tempdir(), 'text.txt'),
path_output = file.path(tempdir(), 'SENTENCE_VECs.txt'))
# ------------
# print-ngrams [ print to console or to output-file ]
# ------------
list_params = list(command = 'skipgram', lr = 0.1, dim = 200,
input = file.path(tempdir(), "doc.txt"),
output = tempdir(), verbose = 2, thread = 1,
minn = 2, maxn = 2)
res = fasttext_interface(list_params,
path_output = file.path(tempdir(), "ngram_out.txt"),
MilliSecs = 5)
list_params = list(command = 'print-ngrams',
model = file.path(tempdir(), 'ngram_out.bin'),
word = 'word') # print n-grams for specific word
res = fasttext_interface(list_params, path_output = "") # print output to console
res = fasttext_interface(list_params,
path_output = file.path(tempdir(), "NGRAMS.txt")) # output to file
# -------------
# 'nn' function
# -------------
list_params = list(command = 'nn',
model = file.path(tempdir(), 'model_cooking.bin'),
k = 20,
query_word = 'word') # a 'query_word' is required
res = fasttext_interface(list_params,
path_output = file.path(tempdir(), "nn_output.txt"))
# ---------
# analogies [ in the output file each analogy-triplet-result is separated with a newline ]
# ---------
list_params = list(command = 'analogies',
model = file.path(tempdir(), 'model_cooking.bin'),
k = 5)
res = fasttext_interface(list_params,
path_input = file.path(tempdir(), 'analogy_queries.txt'),
path_output = file.path(tempdir(), 'analogies_output.txt'))
# -------------
# dump function [ the 'option' param should be one of 'args', 'dict', 'input' or 'output' ]
# -------------
list_params = list(command = 'dump',
model = file.path(tempdir(), 'model_cooking.bin'),
option = 'args')
res = fasttext_interface(list_params,
path_output = file.path(tempdir(), "DUMP.txt"))
## End(Not run)
[Package fastText version 1.0.4 Index]