parallel training, every machine has a different LightGBM_model.txt

[hlee13]

As title described, which I should use

config item

machine0

is_pre_partition = true tree_learner = data data = data.train.partition.0

output LightGBM_model.txt: f6c9af2b4026558d0931c95faba70a6e data_model.txt

machine1

is_pre_partition = true tree_learner = data data = data.train.partition.1 output LightGBM_model.txt: c2591e79db202b639b89e7ddfafd5fb2 data_model.txt

two model file only diff on Tree=0, other Trees are same:

machine0

Tree=0 num_leaves=2 split_feature=0 split_gain=-1 threshold=0 decision_type=0 default_value=0 left_child=-1 right_child=-2 leaf_parent=0 0 leaf_value=52.586637443099526 52.586637443099526 leaf_count=0 38749049 internal_value=0 internal_count=38749049 shrinkage=1 has_categorical=0

machine1

Tree=0 num_leaves=2 split_feature=0 split_gain=-1 threshold=0 decision_type=0 default_value=0 left_child=-1 right_child=-2 leaf_parent=0 0 leaf_value=52.142971758326794 52.142971758326794 leaf_count=0 41732951 internal_value=0 internal_count=41732951 shrinkage=1 has_categorical=0

[guolinke]

temporary solution: set boost_from_average=false . I will fix it soon.

[hlee13]

@guolinke , can directly give me a patch now .

[hlee13]

temporary solution: set boost_from_average=false .

@guolinke Is this a trick?

[guolinke]

@hlee13 it is a parameter.

[hlee13]

got it

[github-actions[bot]]

This issue has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.