Examples with SubWordHashProvider zemberek.core.embeddings.SubWordHashProvider used on opensource projects

Example 1 with SubWordHashProvider

use of zemberek.core.embeddings.SubWordHashProvider in project zemberek-nlp by ahmetaa.

the class FastTextClassifierTrainer method train.

public FastTextClassifier train(Path corpus) {
    Args args = Args.forSupervised();
    args.loss = builder.type == LossType.SOFTMAX ? loss_name.softmax : loss_name.hierarchicalSoftmax;
    args.dim = builder.dimension;
    args.wordNgrams = builder.wordNgramOrder;
    args.thread = builder.threadCount;
    args.epoch = builder.epochCount;
    args.lr = builder.learningRate;
    args.ws = builder.contextWindowSize;
    SubWordHashProvider p = builder.subWordHashProvider;
    args.subWordHashProvider = p;
    args.minn = p.getMinN();
    args.maxn = p.getMaxN();
    args.minCount = builder.minWordCount;
    args.cutoff = builder.quantizationCutOff;
    FastTextTrainer trainer = new FastTextTrainer(args);
    // for catching and forwarding progress events.
    trainer.getEventBus().register(this);
    try {
        return new FastTextClassifier(trainer.train(corpus));
    } catch (Exception e) {
        e.printStackTrace();
        throw new RuntimeException(e);
    }
}