Error: java.io.FileNotFoundException: File /lib/linux/x86_64/libxgboost4j.so was not found inside JAR

#1

I have cloned the java repo for XGBoost4j and I would like to use the functions. Every time I try to run the predict function (from a python model that I trained) I get this issue in the title. Excuse my ignorance but what exactly should I do to get this running?

#3

I want to not have to use a jar file. I have made my own classes for DMatrix, XGBooost, etc. But something is not allowing me to run them without this jar file. Is there any bypass? Like I removed the static class at the begining of the XGBoostJNI:
// static {
// try {
// NativeLibLoader.initXGBoost();
// } catch (Exception ex) {
// logger.error(“Failed to load native library”, ex);
// throw new RuntimeException(ex);
// }
// }

And now I have the following error:
Exception in thread “main” java.lang.UnsatisfiedLinkError: xgboost.XGBoostJNI.XGBoosterCreate([J[J)I

#4

XGBoost4J requires the native library to run, and XGBoostJNI forms the interface to the native library. You’ll need to load in the native library before any of those functions will work. Is there a reason you can’t use the binary on Maven Central (e.g. https://repo1.maven.org/maven2/ml/dmlc/xgboost4j_2.12/1.6.2/xgboost4j_2.12-1.6.2.jar)?