Hi,
We used ensemble to perform regression task, and found after fitting, the StackingRegressor model was not able to be saved by using pickle or other ways.
So could you give me some suggestions on how to saving models when using AutoML ensemble?
The code of FLAML AutoML models:
am = automl()
am.fit(x_train,
y_train,
task='regression',
metric='rmse',
split_ratio=0.2,
ensemble={
'final_estimator': LGBMRegressor(),
'passthrough': True,
},
time_budget=3600)
with open('am_model_test.pickle', 'wb') as am_output:
pickle.dump(am, am_output, pickle.HIGHEST_PROTOCOL)
am_output.close()
sonichi
commented
1 year ago
Hi, We used ensemble to perform regression task, and found after fitting, the StackingRegressor model was not able to be saved by using pickle or other ways. So could you give me some suggestions on how to saving models when using AutoML ensemble?
The code of FLAML AutoML models: am = automl() am.fit(x_train, y_train, task='regression', metric='rmse', split_ratio=0.2, ensemble={ 'final_estimator': LGBMRegressor(), 'passthrough': True, }, time_budget=3600) with open('am_model_test.pickle', 'wb') as am_output: pickle.dump(am, am_output, pickle.HIGHEST_PROTOCOL) am_output.close()