After I set the backend of Keras to CNTK on Google Colab, getting the following warning and error despite I do not get any issues when I use TensorFlow or Theano as the backend of Keras:
/usr/local/lib/python3.6/dist-packages/cntk/cntk_py_init.py:56: UserWarning: Unsupported Linux distribution (ubuntu-18.04). CNTK supports Ubuntu 16.04 and above, only.
warnings.warn('Unsupported Linux distribution (%s-%s). CNTK supports Ubuntu 16.04 and above, only.' % (__my_distro__, __my_distro_ver__))
/usr/local/lib/python3.6/dist-packages/keras/backend/cntk_backend.py:26: UserWarning: CNTK backend warning: GPU is not detected. CNTK's CPU version is not fully optimized,please run with GPU to get better performance.
'CNTK backend warning: GPU is not detected. '
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-1-3c177e7557ce> in <module>()
85
86 if __name__ == '__main__':
---> 87 main()
8 frames
/usr/local/lib/python3.6/dist-packages/cntk/ops/__init__.py in convolution(convolution_map, operand, strides, sharing, auto_padding, sequential, dilation, reduction_rank, groups, max_temp_mem_size_in_samples, name)
419 dilation = sanitize_shape(dilation)
420 return convolution(convolution_map, operand, strides, sharing, auto_padding, dilation,
--> 421 reduction_rank, groups, max_temp_mem_size_in_samples, sequential, name)
422
423 @typemap
ValueError: Convolution operation requires that kernel dim 3 <= input dim 2.
[CALL STACK]
[0x7efc8f367d99] + 0x67bd99
[0x7efc8f508c65] Microsoft::MSR::CNTK::ConvolveGeometry:: ComputeOutputShape (Microsoft::MSR::CNTK::TensorShape const&, Microsoft::MSR::CNTK::TensorShape const&, Microsoft::MSR::CNTK::TensorShape const&, Microsoft::MSR::CNTK::TensorShape const&, std::vector<bool,std::allocator<bool>> const&, std::vector<bool,std::allocator<bool>> const&, Microsoft::MSR::CNTK::TensorShape const&, Microsoft::MSR::CNTK::TensorShape const&, Microsoft::MSR::CNTK::TensorShape const&, unsigned long, bool, bool, bool) + 0x795
[0x7efc8f4f7ad0] CNTK::PrimitiveFunction:: ConvolutionOpOutputShape (CNTK::PrimitiveOpType, CNTK::NDShape const&, CNTK::NDShape&, CNTK::NDShape&, CNTK::NDShape&, std::vector<bool,std::allocator<bool>>&, std::vector<bool,std::allocator<bool>>&, CNTK::NDShape&, CNTK::NDShape&, bool, bool, CNTK::NDShape&, unsigned long, bool) + 0x1d0
[0x7efc8f5002bd] CNTK::PrimitiveFunction:: InferOutputs (std::vector<CNTK::Variable,std::allocator<CNTK::Variable>>&) + 0x22bd
[0x7efc8f4aaca4] + 0x7beca4
[0x7efcf5c62827] + 0xf827
[0x7efc8f4a70be] CNTK::Function:: InitOutputs () + 0x6e
[0x7efc8f54b63d] CNTK::CompositeFunction:: InferOutputs (std::vector<CNTK::Variable,std::allocator<CNTK::Variable>>&) + 0xd
[0x7efc8f4aaca4] + 0x7beca4
[0x7efcf5c62827] + 0xf827
[0x7efc8f4a70be] CNTK::Function:: InitOutputs () + 0x6e
[0x7efc8f4aef0f] CNTK:: AsComposite (std::shared_ptr<CNTK::Function> const&, std::__cxx11::basic_string<wchar_t,std::char_traits<wchar_t>,std::allocator<wchar_t>> const&) + 0x58f
[0x7efc8f4be732] CNTK:: BinaryOp (CNTK::PrimitiveOpType, CNTK::Variable const&, CNTK::Variable const&, CNTK::Dictionary&&, std::__cxx11::basic_string<wchar_t,std::char_traits<wchar_t>,std::allocator<wchar_t>> const&, bool) + 0xf2
[0x7efc8f4d1396] CNTK::Internal:: Convolution (CNTK::Variable const&, CNTK::Variable const&, CNTK::NDShape const&, std::vector<bool,std::allocator<bool>> const&, std::vector<bool,std::allocator<bool>> const&, CNTK::NDShape const&, bool, CNTK::NDShape const&, unsigned long, unsigned long, std::__cxx11::basic_string<wchar_t,std::char_traits<wchar_t>,std::allocator<wchar_t>> const&) + 0xd6
[0x7efc8f4d2e0e] CNTK:: Convolution (CNTK::Variable const&, CNTK::Variable const&, CNTK::NDShape const&, std::vector<bool,std::allocator<bool>> const&, std::vector<bool,std::allocator<bool>> const&, CNTK::NDShape const&, unsigned long, unsigned long, unsigned long, bool, std::__cxx11::basic_string<wchar_t,std::char_traits<wchar_t>,std::allocator<wchar_t>> const&) + 0x23e
[0x7efc9046fb68] + 0x2bab68
[0x50a2bf]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x588e5c]
[0x59fc4e] PyObject_Call + 0x3e
[0x50d356] _PyEval_EvalFrameDefault + 0x17e6
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50cd96] _PyEval_EvalFrameDefault + 0x1226
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50cd96] _PyEval_EvalFrameDefault + 0x1226
[0x508e55] _PyFunction_FastCallDict + 0xf5
[0x594931]
[0x59fc4e] PyObject_Call + 0x3e
[0x50d356] _PyEval_EvalFrameDefault + 0x17e6
[0x507d64]
[0x509042] _PyFunction_FastCallDict + 0x2e2
[0x594931]
[0x54a941]
[0x5a9cbc] _PyObject_FastCallKeywords + 0x19c
[0x50a5c3]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x588e5c]
[0x59fc4e] PyObject_Call + 0x3e
[0x50d356] _PyEval_EvalFrameDefault + 0x17e6
[0x507d64]
[0x588e5c]
[0x59fc4e] PyObject_Call + 0x3e
[0x50d356] _PyEval_EvalFrameDefault + 0x17e6
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50cd96] _PyEval_EvalFrameDefault + 0x1226
[0x509758]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x516345]
[0x50a2bf]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50cd96] _PyEval_EvalFrameDefault + 0x1226
[0x507d64]
[0x509042] _PyFunction_FastCallDict + 0x2e2
[0x594931]
[0x59fc4e] PyObject_Call + 0x3e
[0x50d356] _PyEval_EvalFrameDefault + 0x17e6
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50cd96] _PyEval_EvalFrameDefault + 0x1226
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x509758]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x509758]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x588d41]
[0x59fc4e] PyObject_Call + 0x3e
[0x50d356] _PyEval_EvalFrameDefault + 0x17e6
[0x507d64]
[0x588d41]
[0x59fc4e] PyObject_Call + 0x3e
[0x50d356] _PyEval_EvalFrameDefault + 0x17e6
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x509758]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x508e55] _PyFunction_FastCallDict + 0xf5
[0x594931]
[0x59fc4e] PyObject_Call + 0x3e
[0x50d356] _PyEval_EvalFrameDefault + 0x17e6
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x509758]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x509758]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x516345]
[0x50a2bf]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x509a90]
[0x50a48d]
[0x50bfb4] _PyEval_EvalFrameDefault + 0x444
[0x507d64]
[0x588dcd]
[0x59fc4e] PyObject_Call + 0x3e
[0x63828b]
[0x638f5f] Py_Main + 0x45f
[0x4b0d00] main + 0xe0
[0x7efcf5e93b97] __libc_start_main + 0xe7
[0x5b250a] _start + 0x2a
p.s. Software Stack:
Kerasversion:2.3.1.Pythonversion:3.6.9.
p.s. I'm able to use CNTK package (e.g., import cntk) as I've followed the answer provided in a similar question.
