What causes "Error loading word counts" in stanfordnlp/CoreNLP?

Ensure the CoreNLP models jar/file distribution is on the classpath so the default word-counts resource is found

stanfordnlp/CoreNLP · error · RuntimeException

Error loading word counts

Error message

Error loading word counts

What it means

FeatureExtractor.loadVocabulary loads a word-count file to build a frequency-thresholded vocabulary and wraps any failure in RuntimeException("Error loading word counts"). The word counts resource is required input for statistical/neural coref feature extraction.

Solutions

Ensure the CoreNLP models jar/file distribution is on the classpath so the default word-counts resource is found
If you set a custom word-counts path, verify the file exists and is a tab/space-separated word-count file
Check file permissions
Inspect e.getCause() to distinguish missing-file from parse errors

Example fix

// before
// models jar missing; default edu/stanford/nlp/models/dcoref/... uncountable
// after
classpath += ":/path/to/stanford-corenlp-<ver>-models.jar"   // supplies word counts
// or set explicitly:
props.setProperty("coref.word.counts", "/abs/path/word_counts.txt");

Defensive patterns

Strategy: try-catch

Validate before calling

String countsPath = props.getProperty("coref.word.counts",
    "edu/stanford/nlp/models/dcoref/... (default in models jar)");
boolean resolvable = Thread.currentThread().getContextClassLoader()
    .getResource(countsPath) != null || new java.io.File(countsPath).canRead();
if (!resolvable) throw new IllegalStateException("Word counts resource not on classpath: " + countsPath);

Try / catch

try {
  FeatureExtractor fx = new FeatureExtractor(props, dictionaries, compressor);
} catch (RuntimeException e) {
  if ("Error loading word counts".equals(e.getMessage()))
    throw new IllegalStateException("Word-counts resource missing — install CoreNLP models jar", e);
  throw e;
}

Prevention

Always include the stanford-corenlp models jar in the classpath
Validate custom coref.word.counts paths at startup
Keep counts file format consistent (word + whitespace + count per line)
Check e.getCause() to distinguish missing file vs parse error

When it happens

Trigger: Constructing a FeatureExtractor when the word-counts resource (dictionaries.getCounts (coref.word.counts file)) is missing, unreadable, or has an unexpected format so reading/parsing throws.

Common situations: CoreNLP models jar not on the classpath (the default word counts live inside the model distribution); custom word-counts file path wrong; corrupted or wrong-format counts file.

Understand the failure class

Background: "failed to read file", EACCES, ENOENT and "could not read <path>" errors: when a program can't read a file from disk — this error's family across 49 libraries.

Related errors

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/5dc6d1091adbc044. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/coref/statistical/FeatureExtractor.java:97

      Compressor<String> compressor, Set<String> vocabulary) {
    this.dictionaries = dictionaries;
    this.compressor = compressor;
    this.vocabulary = vocabulary;
    this.useDocSource = CorefProperties.conll(props);
    this.useConstituencyParse = CorefProperties.useConstituencyParse(props);
  }

  private static Set<String> loadVocabulary(String wordCountsPath) {
    Set<String> vocabulary = new HashSet<>();
    try {
      Counter<String> counts = IOUtils.readObjectFromURLOrClasspathOrFileSystem(wordCountsPath);
      for (Map.Entry<String, Double> e : counts.entrySet()) {
        if (e.getValue() > MIN_WORD_COUNT) {
          vocabulary.add(e.getKey());
        }
      }
    } catch (Exception e) {
      throw new RuntimeException("Error loading word counts", e);
    }
    return vocabulary;
  }

  public DocumentExamples extract(int id, Document document,
      Map<Pair<Integer, Integer>, Boolean> labeledPairs) {
    return extract(id, document, labeledPairs, compressor);
  }

  public DocumentExamples extract(int id, Document document,
      Map<Pair<Integer, Integer>, Boolean> labeledPairs, Compressor<String> compressor) {
    List<Mention> mentionsList = CorefUtils.getSortedMentions(document);
    Map<Integer, List<Mention>> mentionsByHeadIndex = new HashMap<>();
    for (Mention m : mentionsList) {
      List<Mention> withIndex = mentionsByHeadIndex.get(m.headIndex);
      if (withIndex == null) {
        withIndex = new ArrayList<>();
        mentionsByHeadIndex.put(m.headIndex, withIndex);

View on GitHub (pinned to 1b7edd19c4)