stanfordnlp/CoreNLP · error · RuntimeException

after blockInitialize, param Index ( ) not equal to…

Error message

after blockInitialize, param Index ( ) not equal to beforeOutputWeights ( )

What it means

During initial() parameter initialization, the blockInitialize path fills the first beforeOutputWeights parameters; the count afterwards must exactly equal beforeOutputWeights. A mismatch means the block initialization loops wrote a different number of values than the output-layer offset, desynchronizing the initialization vector layout.

Solutions

Verify the flags (sparseOutputLayer, tieOutputLayer, softmaxOutputLayer) match the weight-array shapes implied by inputLayerSize and numClasses.
Check blockInitialize loops iterate the full U matrix so exactly beforeOutputWeights values are written.
Keep constructor arguments consistent with the training data's label indices and feature map.
Fall back to the non-block branch (flags disabling block init) to see if dimensions line up without custom initialization.

Example fix

// before (block loop truncated by an early break)
for (int i = 0; i < U.length - 1; i++) initial[count++] = ...;
// after
for (int i = 0; i < U.length; i++) initial[count++] = ...;

Defensive patterns

Strategy: validation

Validate before calling

// after building the objective function
if (fn.domainDimension() <= 0)
  throw new IllegalStateException("degenerate parameter space; check dimension arguments");

Try / catch

try {
  minimizer.minimize(fn, tol, fn.initial());
} catch (RuntimeException e) {
  if (e.getMessage().contains("not equal to beforeOutputWeights")) {
    // retry with default (non-block) initialization flags
  } else throw e;
}

Prevention

Avoid patching blockInitialize without updating its count assertion targets
Keep sparse/tied layer flags consistent across construction and initialization
Unit-test initialization length on small synthetic data

When it happens

Trigger: initial() invoked (first call by the optimizer, e.g. QNMinimizer) with sparseOutputLayer / tieOutputLayer / softmaxOutputLayer block initialization active while array dimensions (inputLayerSize, numClasses, eU shapes) do not match the beforeOutputWeights computation.

Common situations: Custom hidden-layer sizes or modified blockInitialize code during non-linear second-order CRF experiments; inconsistent constructor arguments (numNodeFeatures, numEdgeFeatures, window).

Understand the failure class

Background: "This is a bug, please report it": internal invariant violations, unreachable panics, and SNH errors explained — this error's family across 47 libraries.

Related errors

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/a94208359a8e2155. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/ie/crf/CRFNonLinearSecondOrderLogConditionalObjectiveFunction.java:202

        int interval = numNodeFeatures / numHiddenUnits;
        for (int i = 0; i < numHiddenUnits; i++) {
          int lower = i * interval;
          int upper = (i + 1) * interval;
          if (i == numHiddenUnits - 1)
            upper = numNodeFeatures;
          for (int j = 0; j < outputLayerSize; j++) {
            for (int k = 0; k < numNodeFeatures; k++) {
              val = 0;
              if (k >= lower && k < upper) {
                val = random.nextDouble() * twoEpsilon - epsilon;
              }
              initial[count++] = val;
            }
          }
        }
        if (count != beforeOutputWeights) {
          throw new RuntimeException("after blockInitialize, param Index (" + count + ") not equal to beforeOutputWeights (" + beforeOutputWeights + ")");
        }
      } else {
        for (int i = 0; i < beforeOutputWeights; i++) {
          val = random.nextDouble() * twoEpsilon - epsilon;
          initial[count++] = val;
        }
      }

      if (flags.sparseOutputLayer) {
        for (int i = 0; i < outputLayerSize4Edge; i++) {
          double total = 1;
          for (int j = 0; j < numHiddenUnits-1; j++) {
            val = random.nextDouble() * total;
            initial[count++] = val;
            total -= val;
          }
          initial[count++] = total;
        }

View on GitHub (pinned to 1b7edd19c4)