What causes "Model location not specified for true-case classifier!" in stanfordnlp/CoreNLP?

Set a valid 'truecase.model' property pointing at the classifier model file (e.g. the bundled English truecase model)

stanfordnlp/CoreNLP · error · RuntimeException

Model location not specified for true-case classifier!

Error message

Model location not specified for true-case classifier!

What it means

TrueCaseAnnotator's constructor resolves the true-case CRF classifier model from the 'truecase.model' property (defaulting to the bundled English model). If the resolved model location is null, it throws this RuntimeException while building the pipeline.

Solutions

Set a valid 'truecase.model' property pointing at the classifier model file (e.g. the bundled English truecase model)
Remove the empty/incorrect 'truecase.model' override so the built-in default model path is used
Verify the property key spelling — a typo key like 'truecase.model ' (trailing space) can yield a null/blank resolution
Ensure the model file exists at the given path (classpath or filesystem) before constructing the annotator

Example fix

// before
Properties props = new Properties();
props.setProperty("truecase.model", "");
// after
Properties props = new Properties();
props.setProperty("truecase.model", "edu/stanford/nlp/models/truecase/EnglishAll.untaased.caseless.ser.gz");

Defensive patterns

Strategy: validation

Validate before calling

String modelLoc = props.getProperty("truecase.model", DEFAULT_TRUECASE_MODEL);
if (modelLoc == null || modelLoc.trim().isEmpty())
  throw new IllegalStateException("truecase.model must point to a classifier model file");
if (!new java.io.File(modelLoc).exists() && getClass().getResource("/" + modelLoc) == null)
  throw new IllegalStateException("Truecase model not found: " + modelLoc);

Try / catch

try {
  annotators.add(new TrueCaseAnnotator(props));
} catch (RuntimeException e) {
  if (e.getMessage().contains("Model location not specified")) {
    log.error("Set 'truecase.model' to a valid model path or remove the empty override");
  }
  throw e;
}

Prevention

Never set truecase.model to an empty string; omit the key to use the default
Verify model file existence (filesystem or classpath) before constructing the pipeline
Centralize model path constants to avoid typos in keys
Test pipeline construction in CI so missing models fail early

When it happens

Trigger: Constructing a TrueCaseAnnotator or requesting annotator 'truecase' with a properties object where truecase.model is set to an empty string or otherwise resolves to null, so no model path exists to load.

Common situations: Overriding 'truecase.model' with an empty value in a properties file, copying a partial property subset that drops the default, or programmatic Properties construction that blanks the key.

Related errors

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/dfcd2ac560dc39f3. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/pipeline/TrueCaseAnnotator.java:76

  public TrueCaseAnnotator(String modelLoc,
                           String classBias,
                           String mixedCaseFileName,
                           boolean overwriteText,
                           boolean verbose) {
    this.overwriteText = overwriteText;
    this.verbose = verbose;

    Properties props = PropertiesUtils.asProperties(
            "loadClassifier", modelLoc,
            "mixedCaseMapFile", mixedCaseFileName,
            "classBias", classBias);
    trueCaser = new CRFBiasedClassifier<>(props);

    if (modelLoc != null) {
      trueCaser.loadClassifierNoExceptions(modelLoc, props);
    } else {
      throw new RuntimeException("Model location not specified for true-case classifier!");
    }

    if (classBias != null) {
      StringTokenizer biases = new java.util.StringTokenizer(classBias,",");
      while (biases.hasMoreTokens()) {
        StringTokenizer bias = new java.util.StringTokenizer(biases.nextToken(),":");
        String cname = bias.nextToken();
        double w = Double.parseDouble(bias.nextToken());
        trueCaser.setBiasWeight(cname,w);
        if (this.verbose) log.info("Setting bias for class " + cname + " to " + w);
      }
    }

    // Load map containing mixed-case words:
    // TODO:
    // this file has some weirdness in it
    // for example, 4 different interpretations of el-:
    //   EL-Dafla, el-Ela, El-Ein, EL-ALY

View on GitHub (pinned to 1b7edd19c4)