stanfordnlp/CoreNLP · error · java.lang.IllegalArgumentException

This parser does not contain a DVModel reranker

Error message

This parser does not contain a DVModel reranker

What it means

DVParser.getModelFromLexicalizedParser extracts the DVModel from a loaded parser by casting parser.reranker to DVModelReranker; if the reranker is not an instance of DVModelReranker it throws IllegalArgumentException. Only parsers trained with an embedded DV model expose their neural parameters through this method.

Solutions

1. Ensure the parser was loaded from a model trained by DVParser with an embedded DVModelReranker
2. Check parser.reranker instanceof DVModelReranker before calling the extraction method
3. Re-train or re-save the parser with DVParser so the reranker is embedded

Example fix

// before
LexicalizedParser p = LexicalizedParser.loadModel("englishPCFG.ser.gz");
DVModel m = DVParser.getModelFromLexicalizedParser(p); // throws
// after
LexicalizedParser p = LexicalizedParser.loadModel("dvparser.ser.gz");
if (p.reranker instanceof DVModelReranker) {
  DVModel m = DVParser.getModelFromLexicalizedParser(p);
}

if (parser == null || !(parser.reranker instanceof DVModelReranker)) {
    throw new IllegalArgumentException("Parser must be a DVParser model with embedded DVModelReranker");
}

boolean hasDvModelReranker(LexicalizedParser p) {
    return p != null && p.reranker instanceof DVModelReranker;
}

try {
    DVModel m = DVParser.getModelFromLexicalizedParser(parser);
} catch (IllegalArgumentException e) {
    if (e.getMessage().contains("DVModel reranker")) {
        System.err.println("Load a DVParser-trained model, not a plain parser model");
    }
}

When it happens

Trigger: Calling DVParser.getModelFromLexicalizedParser(parser) with a parser loaded from a plain PCFG/factored model file, or a DVParser model whose reranker was replaced/absent.

Common situations: Passing a standard Stanford parser model (englishPCFG.ser.gz) instead of a DVParser-trained model; programmatic pipelines that reassemble LexicalizedParser objects and drop the reranker; old model files predating the reranker field.

Understand the failure class

Background: Type mismatch errors: IllegalArgumentException, TypeError and type guards across 150 open-source libraries — this error's family across 150 libraries.

Related errors

• Expected parsers with DVModel embedded
• format error in embeddings
• format error unexpected featureFactory line:
• Unknown word vector not specified in the word vector file
• Unable to annotate attribute " + key

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/9a2e27290e95c459. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/parser/dvparser/DVParser.java:395

  public static DVParser loadModel(String filename, String[] args) {
    log.info("Loading serialized model from " + filename);
    DVParser dvparser;
    try {
      dvparser = IOUtils.readObjectFromURLOrClasspathOrFileSystem(filename);
      dvparser.op.setOptions(args);
    } catch (IOException e) {
      throw new RuntimeIOException(e);
    } catch (ClassNotFoundException e) {
      throw new RuntimeIOException(e);
    }
    log.info("... done");
    return dvparser;
  }

  public static DVModel getModelFromLexicalizedParser(LexicalizedParser parser) {
    if (!(parser.reranker instanceof DVModelReranker)) {
      throw new IllegalArgumentException("This parser does not contain a DVModel reranker");
    }
    DVModelReranker reranker = (DVModelReranker) parser.reranker;
    return reranker.getModel();
  }

  public static void help() {
    log.info("Options supplied by this file:");
    log.info("  -model <name>: When training, the name of the model to save.  Otherwise, the name of the model to load.");
    log.info("  -parser <name>: When training, the LexicalizedParser to use as the base model.");
    log.info("  -cachedTrees <name>: The name of the file containing a treebank with cached parses.  See CacheParseHypotheses.java");
    log.info("  -treebank <name> [filter]: A treebank to use instead of cachedTrees.  Trees will be reparsed.  Slow.");
    log.info("  -testTreebank <name> [filter]: A treebank for testing the model.");
    log.info("  -train: Run training over the treebank, testing on the testTreebank.");
    log.info("  -continueTraining <name>: The name of a file to continue training.");
    log.info("  -nofilter: Rules for the parser will not be filtered based on the training treebank.");
    log.info("  -runGradientCheck: Run a gradient check.");
    log.info("  -resultsRecord: A file for recording info on intermediate results");
    log.info();

View on GitHub (pinned to 1b7edd19c4)