TMVA Multiclass example crash when running with argument "DNN"

Run
{{
cd $ROOTSYS/tutorials/tmva
root -l
.x TMVAMulticlass.C("DNN")
}}

Output
{{
... snip ...
: -----------------------------------------------------------
TFHandler_DNN : Variable Mean RMS [ Min Max ]
: -----------------------------------------------------------
: var1: 0.077564 0.26797 [ -0.89529 0.91993 ]
: var2: 0.071274 0.28374 [ -1.0016 1.0000 ]
: var3: -0.087585 0.27454 [ -1.2342 1.0000 ]
: var4: -0.035473 0.25923 [ -0.98958 0.83435 ]
: -----------------------------------------------------------
: Using Standard Implementation.Training with learning rate = 0.1, momentum = 0.3, repetitions = 1
:
Error in <TRint::HandleTermInput()>: std::out_of_range caught: vector::_M_range_check
}}

Backtrace
{{
(lldb) bt

• thread #1: tid = 0x2c56e2, 0x00007fff89022f06 libsystem_kernel.dylib`__pthread_kill + 10, queue = 'com.apple.main-thread', stop reason = signal SIGABRT
• frame #0: 0x00007fff89022f06 libsystem_kernel.dylib`__pthread_kill + 10
  frame #1: 0x00007fff95f4a4ec libsystem_pthread.dylib`pthread_kill + 90
  frame #2: 0x00007fff8e0c56e7 libsystem_c.dylib`abort + 129
  frame #3: 0x00007fff93feff81 libc++abi.dylib`abort_message + 257
  frame #4: 0x00007fff94015a2f libc++abi.dylib`default_terminate_handler() + 243
  frame #5: 0x00007fff8aa956c3 libobjc.A.dylib`_objc_terminate() + 124
  frame #6: 0x00007fff9401319e libc++abi.dylib`std::__terminate(void ()) + 8
  frame #7: 0x00007fff94012c12 libc++abi.dylib`__cxa_throw + 121
  frame #8: 0x00007fff95424781 libc++.1.dylib`std::_1::vector_base_common<true>::_throw_out_of_range() const + 71
  frame #9: 0x0000000101869146 libTMVA.so`double TMVA::DNN::Net::train<TMVA::DNN::Steepest>(std::_1::vector<double, std::1::allocator<double> >&, std::1::vector<Pattern, std::1::allocator<Pattern> >&, std::1::vector<Pattern, std::1::allocator<Pattern> > const&, TMVA::DNN::Steepest&, TMVA::DNN::Settings&) [inlined] std::1::vector<double, std::1::allocator<double> >::at(_n=0) const + 25 at vector:1526 [opt]
  frame #10: 0x000000010186912d libTMVA.so`double TMVA::DNN::Net::train<TMVA::DNN::Steepest>(this=<unavailable>, weights=<unavailable>, trainPattern=size=4000, testPattern=size=4000, minimizer=<unavailable>, settings=<unavailable>) + 3341 at NeuralNet.icc:829 [opt]
  frame #11: 0x00000001018675d4 libTMVA.so`TMVA::MethodDNN::Train(this=<unavailable>) + 6868 at MethodDNN.cxx:717 [opt]
  frame #12: 0x00000001018257f0 libTMVA.so`TMVA::MethodBase::TrainMethod(this=0x0000000115861600) + 512 at MethodBase.cxx:675 [opt]
  frame #13: 0x00000001017dc405 libTMVA.so`TMVA::Factory::TrainAllMethods(this=<unavailable>) + 1301 at Factory.cxx:886 [opt]
  frame #14: 0x0000000100004360 a.out`TMVAMulticlass(myMethodList=TString @ 0x00007fff5fbff5f8) + 12256 at TMVAMulticlass.C:150
  frame #15: 0x000000010000575b a.out`main(argc=2, argv=0x00007fff5fbff6a0) + 491 at TMVAMulticlass.C:185
  frame #16: 0x00007fff9cda45ad libdyld.dylib`start + 1
  }}

Expected Output
Not crashing on training.