stanfordnlp/CoreNLP · error · IllegalArgumentException

KBestSequenceFinder only works with rightWindow == 0 not

Error message

KBestSequenceFinder only works with rightWindow == 0 not 

What it means

KBestSequenceFinder.kBestSequences enumerates k-best label sequences with a left-to-right algorithm that requires the model to have no right context window (rightWindow == 0). Models with a non-zero right window throw IllegalArgumentException('KBestSequenceFinder only works with rightWindow == 0 not ' + rightWindow).

Solutions

1. Rebuild/wrap the SequenceModel so it has rightWindow == 0 (move right-context features into left context or drop them).
2. Use ExactBestSequenceFinder (or another finder supporting right windows) if you only need the single best sequence.
3. If you need k-best with right windows, implement a search that handles them, since this finder explicitly does not.

Example fix

// before
new KBestSequenceFinder().kBestSequences(modelWithRightWindow, k); // model.rightWindow() == 1
// after
new ExactBestSequenceFinder().bestSequence(modelWithRightWindow);
// or construct the model with rightWindow = 0 and then:
new KBestSequenceFinder().kBestSequences(modelRightWindowZero, k);

// check the finder's precondition before calling
if (ts.rightWindow() != 0)
  throw new IllegalArgumentException("use ExactBestSequenceFinder for rightWindow=" + ts.rightWindow());

boolean supportsKBest(SequenceModel ts) { return ts.rightWindow() == 0; }

When it happens

Trigger: Calling kBestSequences (or bestSequences/bestSequence/bestLabelsCounter, which delegate to it) on a SequenceModel whose rightWindow() is non-zero, at KBestSequenceFinder.java:46.

Common situations: Requesting n-best output from taggers/CRFs whose SequenceModel carries right-side context (e.g. models built with a right window for lookahead features); switching a pipeline from ExactBestSequenceFinder to KBestSequenceFinder without adjusting the model.

Understand the failure class

Background: UnsupportedOperationException and "is not supported" errors: when a library deliberately refuses a call — this error's family across 30 libraries.

Related errors

• BiLexPCFGParser doesn't support k best parses
• linearConstraints.length (
• Two models must have the same number of classes
• Two models must have the same sequence length
• LogPrior.getSigmaSquaredM is undefined for any prior but…

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/93adb8d392042d48. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/sequences/KBestSequenceFinder.java:46

  /**
   * Runs the Viterbi algorithm on the sequence model, and then proceeds to efficiently
   * backwards decode the best k label sequence assignments.
   * This sequence finder only works on SequenceModel's with rightWindow == 0.
   *
   * @param ts The SequenceModel to find the best k label sequence assignments of
   * @param k The number of top-scoring assignments to find.
   * @return A Counter with k entries that map from a sequence assignment (int array) to a double score
   */
  @SuppressWarnings("MethodMayBeStatic")
  public Counter<int[]> kBestSequences(SequenceModel ts, int k) {

    // Set up tag options
    int length = ts.length();
    int leftWindow = ts.leftWindow();
    int rightWindow = ts.rightWindow();

    if (rightWindow != 0) {
      throw new IllegalArgumentException("KBestSequenceFinder only works with rightWindow == 0 not " + rightWindow);
    }

    int padLength = length + leftWindow + rightWindow;

    int[][] tags = new int[padLength][];
    int[] tagNum = new int[padLength];
    for (int pos = 0; pos < padLength; pos++) {
      tags[pos] = ts.getPossibleValues(pos);
      tagNum[pos] = tags[pos].length;
    }

    int[] tempTags = new int[padLength];

    // Set up product space sizes
    int[] productSizes = new int[padLength];

    int curProduct = 1;
    for (int i = 0; i < leftWindow; i++) {

View on GitHub (pinned to 1b7edd19c4)