Hi,
I am running Karel's script to train CNN in Kaldi. However, I am getting Kaldi assert failure error in initialization step. Here is the log file (iter00_initial.log):
kaldi::nnet1::Nnet::Read(std::istream&, bool) kaldi::nnet1::Nnet::Read(std::string const&) nnet-train-frmshuff(main+0xe44) [0x525068] /lib64/libc.so.6(__libc_start_main+0xfd) [0x373c41ed5d] nnet-train-frmshuff() [0x524169] KALDI_ASSERT: at nnet-train-frmshuff:ReadData:nnet/nnet-max-pooling-component.h:81, failed: (num_patches - pool_size_) % pool_step_ == 0 Stack trace is: kaldi::KaldiGetStackTrace() kaldi::KaldiAssertFailure_(char const, char const, int, char const*) kaldi::nnet1::MaxPoolingComponent::ReadData(std::istream&, bool) kaldi::nnet1::Component::Read(std::istream&, bool) kaldi::nnet1::Nnet::Read(std::istream&, bool)
Btw, I am using CUDA 7.0.
Hi, most likely the input-feature dimension is wrong, i.e. different from the RM example. K.
Hi,
I am running Karel's script to train CNN in Kaldi. However, I am getting Kaldi assert failure error in initialization step. Here is the log file (iter00_initial.log):
kaldi::nnet1::Nnet::Read(std::istream&, bool)
kaldi::nnet1::Nnet::Read(std::string const&)
nnet-train-frmshuff(main+0xe44) [0x525068]
/lib64/libc.so.6(__libc_start_main+0xfd) [0x373c41ed5d]
nnet-train-frmshuff() [0x524169]
KALDI_ASSERT: at nnet-train-frmshuff:ReadData:nnet/nnet-max-pooling-component.h:81, failed: (num_patches - pool_size_) % pool_step_ == 0
Stack trace is:
kaldi::KaldiGetStackTrace()
kaldi::KaldiAssertFailure_(char const, char const, int, char const*)
kaldi::nnet1::MaxPoolingComponent::ReadData(std::istream&, bool)
kaldi::nnet1::Component::Read(std::istream&, bool)
kaldi::nnet1::Nnet::Read(std::istream&, bool)
Btw, I am using CUDA 7.0.
Hi, most likely the input-feature dimension is wrong, i.e. different from the RM example.
K.