Closed lvpinrui closed 5 years ago
Dear, I have run several tests on similar dataset usps, and no error occurred. Could you please provide more details or zip your dataset and put it here if available? Make sure that your dataset is in a LibSVM format. Thanks.
Hi there, We found that your dataset is zero_based (the first index of feature is 0). The newest version is able to handle zero_based dataset. You can use the newest code or turn your dataset into one-based. Thanks.
hello,Running the original code after changing my dataset succeeded.but i use the newest code,This has a new problem. my thundergbm-train: my dataset is always 1.txt, When building thundergbm-train :
waiting for your advice,thank you very much.
Hi lvpinrui,
Thanks for imforming. You can delete the code in this line and rebuild the project. There is a bug here, we will fix it in no time.
Regards
i use the newest code,When my dataset is 1.txt running thundergbm-predict file succeeds, but when the dataset is 2.txt thundergbm-predict file fails ,this is a new problem my dataset is 2.txt,2.txt file size is 1.43GB,I don't know if it is because the file is too big.
It should work. You should check your dataset, or share it through the cloud (Baidu Yun?)
this my dataset(Baidu Yun): 链接:https://pan.baidu.com/s/15H6YW_mqZP6K4hrOVl9eOA 提取码:s9y7
thank you very much!
I have tested your dataset 2.txt on the newest code. No error has occurred. @lvpinrui
我已在最新代码上测试了您的数据集2.txt。没有发生错误。 @lvpinrui
hello ,thank you very much for your help.
but my Configuration is:
max_depth=6 num_round=40 n_gpus=1 verbosity=0 profiling=0 data=C:\Users\Administrator\Desktop\2\2.txt max_bin=255 colsample=1 bagging=1 num_parallel_tree=1 eta=1.0 objective=multi:softmax num_class=8 min_child_weight=1 lambda=1 gamma=1 tree_method=auto
when building thundergbm-train is ok,
when building thundergbm-predict is failed.
Yes, the same configuration as yours. I ran the test via command line, editing the machine.conf. It works fine for me. @lvpinrui
Yes, the same configuration as yours. I ran the test via command line, editing the machine.conf. It works fine for me. @lvpinrui
thank you very much for your help these days. I tried it a few times or not, I succeeded after changing my dataset.
Since the issue is solved, I will close it.
Cheers, Locke
My environment is: Windows 10 Visual Studio 2017 Community CMake3.14.0-rc3 CUDA10.1.105 Using test_dataset.txt in the library is successful,but Using my own .txt is failed. my thundergbm-predict:
When building thundergbm-predict :