Error When Changing Number of Leaves with Callback function

[jturkewitz]

I get the following error when running on windows 10 with python and trying to use the callbacks to change the number of leaves:
Traceback (most recent call last):

File "", line 53, in
callbacks=[lgb.reset_parameter(num_leaves=lambda iter: np.random.choice([10,20,30]))])

File "C:\Users\User\Anaconda3\lib\site-packages\lightgbm-0.1-py3.6.egg\lightgbm\engine.py", line 180, in train
booster.update(fobj=fobj)

File "C:\Users\User\Anaconda3\lib\site-packages\lightgbm-0.1-py3.6.egg\lightgbm\basic.py", line 1355, in update
ctypes.byref(is_finished)))

OSError: exception: access violation writing 0x000002B697046EA8

Here is code to repro the bug:
import lightgbm as lgb
import pandas as pd
import numpy as np
from sklearn.metrics import mean_squared_error

try:
import cPickle as pickle
except:
import pickle

load or create your dataset

print('Load data...')
df_train = pd.read_csv('binary.train', header=None, sep='\t')
df_test = pd.read_csv('binary.test', header=None, sep='\t')

y_train = df_train[0].values
y_test = df_test[0].values
X_train = df_train.drop(0, axis=1).values
X_test = df_test.drop(0, axis=1).values

num_train, num_feature = X_train.shape

create dataset for lightgbm

if you want to re-use data, remember to set free_raw_data=False

lgb_train = lgb.Dataset(X_train, y_train,
free_raw_data=False)
lgb_eval = lgb.Dataset(X_test, y_test, reference=lgb_train,
free_raw_data=False)

specify your configurations as a dict

params = {
'boosting_type': 'gbdt',
'objective': 'binary',
'metric': 'binary_logloss',
'num_leaves': 31,
'learning_rate': 0.05,
'feature_fraction': 0.9,
'bagging_fraction': 0.8,
'bagging_freq': 5,
'verbose': 0
}

feature_name = ['feature_' + str(col) for col in range(num_feature)]
print('Start training...')

feature_name and categorical_feature

gbm = lgb.train(params,
lgb_train,
num_boost_round=10,
valid_sets=lgb_train, # eval training data
feature_name=feature_name,
categorical_feature=[21],
callbacks=[lgb.reset_parameter(num_leaves=lambda iter: np.random.choice([10,20,30]))])

[guolinke]

@jturkewitz can you check your re-produce code ?
it is very messy .

[jturkewitz]

Hi,
I copied and pasted most of the code from:
https://github.com/Microsoft/LightGBM/blob/master/examples/python-guide/advanced_example.py
I just moved the files binary.train and binary.test to the directory the code is run from and added the
callbacks=[lgb.reset_parameter(num_leaves=lambda iter: np.random.choice([10,20,30]))])
line at the end. So if you add this line to any dataset it should crash the kernel to repro the problem. I have reposted the code so markdown doesn't bold comments.

import lightgbm as lgb
import pandas as pd
import numpy as np
from sklearn.metrics import mean_squared_error

try:
    import cPickle as pickle
except:
    import pickle

# load or create your dataset
print('Load data...')
df_train = pd.read_csv('binary.train', header=None, sep='\t')
df_test = pd.read_csv('binary.test', header=None, sep='\t')

y_train = df_train[0].values
y_test = df_test[0].values
X_train = df_train.drop(0, axis=1).values
X_test = df_test.drop(0, axis=1).values

num_train, num_feature = X_train.shape

# create dataset for lightgbm
# if you want to re-use data, remember to set free_raw_data=False
lgb_train = lgb.Dataset(X_train, y_train,
                         free_raw_data=False)
lgb_eval = lgb.Dataset(X_test, y_test, reference=lgb_train,
                       free_raw_data=False)

# specify your configurations as a dict
params = {
    'boosting_type': 'gbdt',
    'objective': 'binary',
    'metric': 'binary_logloss',
    'num_leaves': 31,
    'learning_rate': 0.05,
    'feature_fraction': 0.9,
    'bagging_fraction': 0.8,
    'bagging_freq': 5,
    'verbose': 0
}

feature_name = ['feature_' + str(col) for col in range(num_feature)]
print('Start training...')
# feature_name and categorical_feature
gbm = lgb.train(params,
                lgb_train,
                num_boost_round=10,
                valid_sets=lgb_train,  # eval training data
                feature_name=feature_name,
                categorical_feature=[21],
                callbacks=[lgb.reset_parameter(num_leaves=lambda iter: np.random.choice([10,20,30]))])

[github-actions]

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.