stanfordnlp/CoreNLP · error · RuntimeException

Could not read from float initial weight file

Error message

Could not read from float initial weight file 

What it means

CRFClassifierFloat.trainWeights loads initial weights from flags.initialWeights as a gzip-compressed float array. If opening/reading the stream or decoding the array throws IOException, it throws RuntimeException('Could not read from float initial weight file <path>') — notably WITHOUT the cause chained, so the original IOException detail is lost.

Solutions

1. Check that flags.initialWeights points to an existing, readable gzip-compressed float-array file.
2. Verify the file is valid gzip ('file initialweights.gz') and was written with ConvertByteArray.writeFloatArr.
3. Regenerate the initial weights file as a float array with the matching utility/serialization.
4. If you only have a double array, convert it to float[] before writing/using it for the float trainer.
5. Wrap the call yourself or patch to chain the cause (new RuntimeException(msg, e)) to see the true IOException.

Example fix

// before
props.setProperty("initialWeights", "init.txt"); // plain text, not gzip float array
// after
// write with ConvertByteArray.writeFloatArr(new GZIPOutputStream(new FileOutputStream("init.f32.gz")), floatWeights);
props.setProperty("initialWeights", "init.f32.gz");

File f = new File(flags.initialWeights);
if (!f.isFile() || !f.canRead()) throw new IllegalArgumentException("initialWeights missing/unreadable: " + flags.initialWeights);
try (InputStream in = new GZIPInputStream(new FileInputStream(f))) {
  in.read(); // throws IOException('Not in GZIP format') if not gzip
}

try {
  initialWeights = ConvertByteArray.readFloatArr(
      new DataInputStream(new BufferedInputStream(new GZIPInputStream(new FileInputStream(flags.initialWeights)))));
} catch (IOException e) {
  throw new RuntimeException("Could not read float initial weights from " + flags.initialWeights, e); // chain the cause!
}

When it happens

Trigger: Setting flags.initialWeights to a nonexistent file, unreadable file, or a file that is not a gzip stream of a float array (wrong format, plain text, double array) during training with float weights.

Common situations: Typo in initialWeights path; file compressed differently (plain vs gzip); passing a double-array dump instead of a float dump; permissions; file created by another tool with a different binary layout.

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

• Could not read from double initial weight file
• Could not read from double initial LOP scales file
• edu.stanford.nlp.io.RuntimeIOException
• Error reading saved links
• Error setting up training

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/81b3a001dda02b20. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/ie/crf/CRFClassifierFloat.java:92

    if (pruneFeatureItr == 0) {
      minimizer.setM(flags.QNsize);
    } else {
      minimizer.setM(flags.QNsize2);
    }

    float[] initialWeights;
    if (flags.initialWeights == null) {
      initialWeights = func.initial();
    } else {
      try {
        log.info("Reading initial weights from file " + flags.initialWeights);
        try (DataInputStream dis = new DataInputStream(new BufferedInputStream(new GZIPInputStream(new FileInputStream(
            flags.initialWeights))))) {
          initialWeights = ConvertByteArray.readFloatArr(dis);
        }
      } catch (IOException e) {
        throw new RuntimeException("Could not read from float initial weight file " + flags.initialWeights);
      }
    }
    log.info("numWeights: " + initialWeights.length);
    float[] weightsArray = minimizer.minimize(func, (float) flags.tolerance, initialWeights);
    return ArrayMath.floatArrayToDoubleArray(weightsArray);
  }

} // end class CRFClassifierFloat

View on GitHub (pinned to 1b7edd19c4)