TimeSeries API PyTorch Engine support

[keklol5050]

The documentation states (https://djl.ai/extensions/timeseries/docs/forecast_with_M5_data.html) that PyTorch also supports the TimeSeries API, but when trying to train a model with the PyTorch engine, an error occurs:
Exception in thread "main" java.lang.UnsupportedOperationException: Not implemented yet.
at ai.djl.pytorch.engine.PtNDArray.gammaln(PtNDArray.java:845)
at ai.djl.timeseries.distribution.NegativeBinomial.logProb(NegativeBinomial.java:47)
at ai.djl.timeseries.distribution.DistributionLoss.evaluate(DistributionLoss.java:54)
at ai.djl.training.EasyTrain.trainSplit(EasyTrain.java:124)
at ai.djl.training.EasyTrain.trainBatch(EasyTrain.java:110)
at ai.djl.training.EasyTrain.fit(EasyTrain.java:58)
at com.crypto.analysis.main.ex.NNew.train(NNew.java:134)
at com.crypto.analysis.main.ex.NNew.main(NNew.java:63)

`public static void train(Path outputDir) {
try (Model model = Model.newInstance("DeepAR");
NDManager manager = NDManager.newBaseManager()) {

        DistributionOutput distributionOutput = new NegativeBinomialOutput();
        DefaultTrainingConfig config = setupTrainingConfig(distributionOutput);
        DeepARNetwork trainingNetwork = getDeepARModel(freq, predictionLength, distributionOutput, true);

        model.setBlock(trainingNetwork);

        List<TimeSeriesTransform> trainingTransformation = trainingNetwork.createTrainingTransformation(manager);
        int contextLength = trainingNetwork.getContextLength();

        NDataset trainSet = getDataset(trainingTransformation, contextLength);

        try (Trainer trainer = model.newTrainer(config)) {
            trainer.setMetrics(new Metrics());

            System.out.println("+++++" + trainSet.availableSize());
            int historyLength = trainingNetwork.getHistoryLength();
            Shape[] inputShapes = new Shape[4];
            // (N, num_cardinality)
            inputShapes[0] = new Shape(1, 1);
            // (N, num_real) if use_feat_stat_real else (N, 1)
            inputShapes[1] = new Shape(1, 1);

            inputShapes[2] =
                    new Shape(
                            1,
                            historyLength,
                            TimeFeature.timeFeaturesFromFreqStr("M").size() + 1);
            inputShapes[3] = new Shape(1, historyLength);
            trainer.initialize(inputShapes);

            EasyTrain.fit(trainer, 1500, trainSet, null);
            logger.info(String.valueOf(trainer.getTrainingResult()));

            model.setProperty("Epochs", "1500");
            model.save(outputDir, "DeepAR");
        }

    } catch (TranslateException | IOException e) {
        throw new RuntimeException(e);
    }
}

private static DeepARNetwork getDeepARModel(String freq, int predictionLength,
                                            DistributionOutput distributionOutput, boolean training) {

    List<Integer> cardinality = new ArrayList<>();
    cardinality.add(allSize-testSize);

    DeepARNetwork.Builder builder =
            DeepARNetwork.builder()
                    .setFreq(freq)
                    .setPredictionLength(predictionLength)
                    .setCardinality(cardinality)
                    .optDistrOutput(distributionOutput)
                    .optUseFeatDynamicReal(false)
                    .optUseFeatStaticCat(false)
                    .optUseFeatStaticReal(false);
    return training ? builder.buildTrainingNetwork() : builder.buildPredictionNetwork();
}

private static DefaultTrainingConfig setupTrainingConfig(DistributionOutput distributionOutput) {
    return new DefaultTrainingConfig(new DistributionLoss("Loss", distributionOutput))
            .addEvaluator(new Rmsse(distributionOutput))
            .optInitializer(new XavierInitializer(), Parameter.Type.WEIGHT)
            .addTrainingListeners(TrainingListener.Defaults.logging());
}`