search

Xtra-Computing / thundergbm

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

Check failed: [max_elem + n_columns*(max_elem + 1) < 0x7fffffff] Max_values is too large to be transformed.请问这个问题怎么解决？ #81

Open ninding opened 1 year ago

ninding commented 1 year ago 
2023-04-24 20:27:01,870 INFO [default] #instances = 349746, #features = 82
2023-04-24 20:27:01,874 INFO [default] convert csr to csc using gpu...
2023-04-24 20:27:02,171 INFO [default] Converting csr to csc using time: 0.297256 s
2023-04-24 20:27:02,171 INFO [default] Fast getting cut points...
2023-04-24 20:27:02,172 FATAL [default] Check failed: [max_elem + n_columns*(max_elem + 1) < 0x7fffffff] Max_values is too large to be transformed
2023-04-24 20:27:02,172 WARNING [default] Aborting application. Reason: Fatal log at [/home/zeyi/thundergbm/src/thundergbm/hist_cut.cu:99]
ninding commented 1 year ago

和TGBM相关的代码

model.classifier = TGBMClassifier()
model.classifier.fit(train_X, train_y)
hliuca commented 1 year ago

0x7fffffff, that is the maximal integer for int32. int64 should be used for index.