0

I am new to kur and just trying it out.

with kur -v train speech.yml, it runs fine but with some annoying messages: 1. The tensorflow is not compiling SSE instructions; 2. system library 'libmagic' can't be found.

Should I worry about these? should I do something about these?

Thanks a lot! @AdamSypniewski where is the best place to ask questions about kur?

W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.
[INFO 2017-02-26 14:33:20,651 kur.backend.keras_backend:654] Waiting for model to finish compiling...
[WARNING 2017-02-26 14:33:20,657 kur.utils.audiotools:98] Python package "magic" could not be loaded, possibly because system library "libmagic" could not be found. We are falling back on our own heuristics.

Daniel
  • 1,428
  • 3
  • 16
  • 35

1 Answers1

1

The "TensorFlow library wasn't compiled" warning is just a warning (c.f., this answer). Every thing will still work fine, but it is possible that you'll get a boost in performance if you compile/install TensorFlow from source, rather than from pip/conda. That's all.

The "libmagic" warning can (probably) be ignored as well. "libmagic" is the name of a library which uses magic numbers to determine the file type of an arbitrary file. Current versions of macOS don't have it installed by default, so Kur falls back onto its own file-format heuristic. If you aren't seeing errors and if your model output seems reasonable, you're probably fine. If you want to be super-careful, you can use a macOS package manager to install it; for example, using Homebrew, the command is brew install libmagic.

And since you asked, the best place to ask questions about Kur is on Gitter.

Community
  • 1
  • 1
Adam Sypniewski
  • 236
  • 1
  • 3