Home | News | Documentation | Download

TMVA keras bug of LearningRateSchedule

It seems the method of below

TMVA.Types.kPyKeras

with option

LearningRateSchedule != ''

got the error as below

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/usr/local/lib/python3.7/site-packages/keras/engine/training.py", line 1239, in fit
    validation_freq=validation_freq)
  File "/usr/local/lib/python3.7/site-packages/keras/engine/training_arrays.py", line 146, in fit_loop
    callbacks.on_epoch_begin(epoch)
  File "/usr/local/lib/python3.7/site-packages/keras/callbacks/callbacks.py", line 136, in on_epoch_begin
    callback.on_epoch_begin(epoch, logs)
  File "/usr/local/lib/python3.7/site-packages/keras/callbacks/callbacks.py", line 930, in on_epoch_begin
    lr = self.schedule(epoch, lr)
  File "<string>", line 3, in schedule
AttributeError: 'float' object has no attribute 'optimizer'
<FATAL>                         : Failed to train model
***> abort program execution
Traceback (most recent call last):
  File "ClassificationKeras.py", line 69, in <module>
    factory.TrainAllMethods()
Exception: void TMVA::Factory::TrainAllMethods() =>
    FATAL error (C++ exception of type runtime_error)

the pyKeras need to be updated

May be a more complete reproducer will help to understand better the problem ?