Closed Meier-Matthias closed 3 years ago
This is strange. It should have to do with the xgboost core and its compilation. I found two recent posts reporting the same issue (https://discuss.xgboost.ai/t/segmentation-fault-whenever-i-try-to-train-or-predict/2254 , https://stackoverflow.com/questions/67899117/segmentation-fault-when-fitting-train-data-into-xgboost) but without responses/solutions.
I assume that you are using the actual version (1.4.1.1 R package version)? If so, try an older version, or upgrade to the latest version. Which gcc version are you using?
gcc: 9.3.0 R: 4.1.0 xgboost: 1.4.1.1
It seems that xgboost has this issues recurrently. I tried to update xgboost but nothing changed.
Did you try an older version?
(you can install older versions via devtools::install_version(...))
What version do you prefer?
Maybe we can wait a few weeks with this to see if they correct the bug (if this is a bug), because it would be a bit annoying if the code runs only with a specific old version.
Yes, I think it is only a temporary issue and I suggested to use an older version so you @Meier-Matthias can continue compiling the book. Doesn't matter which version works at the moment.
Seems to work for the moment.
Can't run examples under "### Boosted Regression Trees", because xgboost causes segfaults.