prueba fastext.txt

datatext <- data.frame(cbind(csvData[, "text"], csvData[, "gender"]))
datatext[datatext$gender == "male", "gender"] <- "human"
datatext[datatext$gender == "female", "gender"] <- "human"
rowsTrain <- sample(nrow(datatext), round(0.8 * nrow(datatext)), replace = FALSE)
train_sentences <- datatext[rowsTrain,]
test_sentences <- datatext[-rowsTrain,]

tmp_file_model <- tempfile()
train_labels <- paste0("__label__", train_sentences[,"gender"])
train_texts <- tolower(train_sentences[,"text"])
train_to_write <- paste(train_labels, train_texts)
train_tmp_file_txt <- tempfile()
writeLines(text = train_to_write, con = train_tmp_file_txt)


test_labels <- paste0("__label__", test_sentences[,"gender"])
test_labels_without_prefix <- test_sentences[,"gender"]
test_texts <- tolower(test_sentences[,"text"])
test_to_write <- paste(test_labels, test_texts)

execute(commands = c("supervised", "-input", train_tmp_file_txt, "-output", tmp_file_model, "-dim", 20, "-lr", 1, "-epoch", 20, "-wordNgrams", 2, "-verbose", 2))
model <- load_model(tmp_file_model)
predictions <- predict(model, sentences = test_to_write)
print(head(predictions, 5))

mean(names(unlist(predictions)) == test_labels_without_prefix)
get_hamming_loss(as.list(test_labels_without_prefix), predictions)
predictions <- predict(model, sentences = test_to_write)
print(head(predictions, 5))
print(head(predict(model, sentences = test_to_write, simplify = TRUE)))
unlink(train_tmp_file_txt)
unlink(tmp_file_model)
rm(model)
gc()