Xtra-Computing / thundergbm

ThunderGBM: Fast GBDTs and Random Forests on GPUs
Apache License 2.0
689 stars 84 forks source link

issue with saving and reloading model? #75

Closed AlanSpencer2 closed 1 year ago

AlanSpencer2 commented 2 years ago

I am trying to save a trained model using: clf.save_model(path)... But when I reload the model, it doesn´t produce the same results as the saved model, despite using the same exact test data.

I reloaded with the command clf.load_model(path).

What is the cause of this problem? How can I save and reload models successfully?

Kurt-Liuhf commented 2 years ago

Hi @AlanSpencer2, I tried to reproduce your issue on dataset covtypewhich is a dataset for classification. And I used the same methods as you mentioned, clf.save_model(path) andclf.load_model(path), but the prediction results are exactly the same before and after model saving and reloading. Please provide more information (e.g., dataset you used, type of task, model parameters, etc.) to help us reproduce the issue.

zeyiwen commented 1 year ago

@AlanSpencer2 I'm closing this issue, as the codebase works fine in the case and we are unable to reproduce the issue reported.