stanfordnlp/CoreNLP · error · IOException

Couldn't load classifier from

Error message

Couldn't load classifier from {path}

What it means

loadClassifierFromPath tries to load a classifier from a filesystem path, first as a CRFClassifier then as a CMMClassifier. If both attempts fail it wraps the last exception in an IOException saying the classifier could not be loaded from that path.

Solutions

1. Verify the path exists and is readable: new File(path).canRead()
2. Re-serialize the classifier with the CoreNLP version on your classpath
3. Check the 'classifiers' property value for typos and correct resource prefixes
4. Inspect the wrapped cause to distinguish file-not-found from deserialization issues

Example fix

// before
props.setProperty("classifiers", "models/my-ner-model.ser.gz"); // file missing
// after
File f = new File("models/my-ner-model.ser.gz");
if (!f.canRead()) throw new IllegalStateException("classifier missing: " + f.getAbsolutePath());
props.setProperty("classifiers", f.getAbsolutePath());

for (String path : classifierPaths) {
  File f = new File(path);
  if (!f.canRead()) throw new IllegalStateException("unreadable classifier: " + f.getAbsolutePath());
  try (ObjectInputStream in = new ObjectInputStream(new BufferedInputStream(new FileInputStream(f)))) {
    // force read of stream header to catch corruption early
    in.available();
  }
}

try {
  nerPipeline = new NERClassifierCombiner(props);
} catch (IOException e) {
  throw new RuntimeException("Check 'classifiers' property paths and model versions", e);
}

When it happens

Trigger: Passing a nonexistent, unreadable, or non-classifier file path in the 'classifiers' property of NERClassifierCombiner/ClassifierCombiner, or a serialized model incompatible with the current library version.

Common situations: Typo'd classifier path in StanfordCoreNLP.properties; classifier file not shipped/deployed with the app; model trained/serialized with an older CoreNLP version; file permissions.

Understand the failure class

Background: 'Could not be found', 'does not exist', 'not found in database': the resource-not-found family when an ID, slug, key, or URI lookup comes back empty — this error's family across 20 libraries.

Related errors

• Couldn't load classifier!
• TokensRegexNERAnnotator
• Shouldn't happen:
• Error reading saved links
• RuntimeIOException wrapping IOException

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/10d8e6c9f7f7d655. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/ie/ClassifierCombiner.java:297

    }
  }


  public static <INN extends CoreMap & HasWord> AbstractSequenceClassifier<INN> loadClassifierFromPath(Properties props, String path)
      throws IOException {
    //try loading as a CRFClassifier
    try {
      return ErasureUtils.uncheckedCast(CRFClassifier.getClassifier(path, props));
    } catch (Exception e) {
      e.printStackTrace();
    }
    //try loading as a CMMClassifier
    try {
      return ErasureUtils.uncheckedCast(CMMClassifier.getClassifier(path));
    } catch (Exception e) {
      //fail
      //log.info("Couldn't load classifier from path :"+path);
      throw new IOException("Couldn't load classifier from " + path, e);
    }
  }

  @Override
  public Set<String> labels() {
    Set<String> labs = Generics.newHashSet();
    for(AbstractSequenceClassifier<? extends CoreMap> cls: baseClassifiers)
      labs.addAll(cls.labels());
    return labs;
  }


  /**
   * Reads the Answer annotations in the given labellings (produced by the base models)
   *   and combines them using a priority ordering, i.e., for a given baseDocument all
   *   labellings seen before in the baseDocuments list have higher priority.
   *   Writes the answer to AnswerAnnotation in the labeling at position 0
   *   (considered to be the main document).

View on GitHub (pinned to 1b7edd19c4)