stanfordnlp/CoreNLP · error · IllegalArgumentException

Can only create a model using this method if…

Error message

Can only create a model using this method if combineClassification and simplifiedModel are turned on

What it means

SentimentModel.modelFromMatrices builds a model assuming the simplified single binary-transform/tensor layout keyed by "", "". The library throws this IllegalArgumentException because if RNNOptions.combineClassification or RNNOptions.simplifiedModel is off, the model would need per-label-category matrices that this factory method cannot derive from the given W/Wcat/Wt.

Solutions

Set combineClassification=true and simplifiedModel=true on the RNNOptions before calling modelFromMatrices
If you need non-simplified options, construct the SentimentModel via a constructor that reads real trees/options instead of modelFromMatrices

Example fix

// before
RNNOptions op = new RNNOptions();
op.simplifiedModel = false;
SentimentModel model = SentimentModel.modelFromMatrices(W, Wcat, Wt, wordVectors, op);
// after
RNNOptions op = new RNNOptions();
op.combineClassification = true;
op.simplifiedModel = true;
SentimentModel model = SentimentModel.modelFromMatrices(W, Wcat, Wt, wordVectors, op);

Defensive patterns

Strategy: validation

Validate before calling

if (!op.combineClassification || !op.simplifiedModel) {
  throw new IllegalArgumentException("modelFromMatrices requires combineClassification and simplifiedModel");
}
SentimentModel model = SentimentModel.modelFromMatrices(W, Wcat, Wt, wordVectors, op);

Try / catch

try {
  model = SentimentModel.modelFromMatrices(W, Wcat, Wt, wordVectors, op);
} catch (IllegalArgumentException e) {
  op.combineClassification = true;
  op.simplifiedModel = true;
  model = SentimentModel.modelFromMatrices(W, Wcat, Wt, wordVectors, op);
}

Prevention

Always build RNNOptions for modelFromMatrices with both flags explicitly set to true
Centralize option construction in one factory so flags are never forgotten
Never reuse options objects configured for full-grammar training with this method

When it happens

Trigger: Calling SentimentModel.modelFromMatrices(W, Wcat, Wt, wordVectors, op) where the passed RNNOptions has combineClassification=false or simplifiedModel=false.

Common situations: Converting a model trained with the original Matlab code into a Java SentimentModel while reusing options configured for full (non-simplified) training; copying op from a serialized/props-based pipeline that disabled simplifiedModel.

Understand the failure class

Background: "Invalid value" and "allowed values are" config errors: what your library rejected and how to fix it — this error's family across 41 libraries.

Related errors

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/16fd9e0d7b9068d3. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/sentiment/SentimentModel.java:216

    binaryTransform.replaceAll(x -> new SimpleMatrix(x));
    binaryTensors.replaceAll(x -> new SimpleTensor(x));
    binaryClassification.replaceAll(x -> new SimpleMatrix(x));
    unaryClassification.replaceAll((x, y) -> new SimpleMatrix(y));
    wordVectors.replaceAll((x, y) -> new SimpleMatrix(y));
  }
  */


  /**
   * Given single matrices and sets of options, create the
   * corresponding SentimentModel.  Useful for creating a Java version
   * of a model trained in some other manner, such as using the
   * original Matlab code.
   */
  static SentimentModel modelFromMatrices(SimpleMatrix W, SimpleMatrix Wcat, SimpleTensor Wt, Map<String, SimpleMatrix> wordVectors, RNNOptions op) {
    if (!op.combineClassification || !op.simplifiedModel) {
      throw new IllegalArgumentException("Can only create a model using this method if combineClassification and simplifiedModel are turned on");
    }
    TwoDimensionalMap<String, String, SimpleMatrix> binaryTransform = TwoDimensionalMap.treeMap();
    binaryTransform.put("", "", W);

    TwoDimensionalMap<String, String, SimpleTensor> binaryTensors = TwoDimensionalMap.treeMap();
    binaryTensors.put("", "", Wt);

    TwoDimensionalMap<String, String, SimpleMatrix> binaryClassification = TwoDimensionalMap.treeMap();

    Map<String, SimpleMatrix> unaryClassification = Generics.newTreeMap();
    unaryClassification.put("", Wcat);

    return new SentimentModel(binaryTransform, binaryTensors, binaryClassification, unaryClassification, wordVectors, op);
  }

  public SentimentModel(TwoDimensionalMap<String, String, SimpleMatrix> binaryTransform,
                        TwoDimensionalMap<String, String, SimpleTensor> binaryTensors,
                        TwoDimensionalMap<String, String, SimpleMatrix> binaryClassification,

View on GitHub (pinned to 1b7edd19c4)