save_model on regressor results in crash (Windows 10)

Xtra-Computing / thundergbm

ThunderGBM: Fast GBDTs and Random Forests on GPUs

Apache License 2.0

689 stars 85 forks source link

save_model on regressor results in crash (Windows 10) #35

Open counterpoint1 opened 4 years ago

counterpoint1 commented 4 years ago

I get this error, causing python to crash, when I invoke the save_model method after a model run. Trees are about 10+ deep, and 200 trees to save. Any ideas?

zeyiwen commented 4 years ago

We couldn't reproduce your issue. Would you please try to run ThunderGBM using command-line? Our command-line version also saves the trained model. This should help you identify whether the error is from ThunderGBM or other parts of your code.

counterpoint1 commented 4 years ago

How to best get the arrays outside python for command line execution?

also, we are attempting to migrate from scikit learn’s random forest, which has an option ‘ min_samples_leaf’ ...

Is it possible to have this option or guidance on how to replicate that parameter in thundergbm? Also can thundergbm apply a weighting vector to put more weight on recent samples?

counterpoint1 commented 4 years ago

Also, where to download the package for command line execution (for Windows 64 environments)? I just installed the python wheel for Windows...

zeyiwen commented 4 years ago

You need to build the executable first. Please refer to the document for more details.