0

I have implemented LSTM-CTC model in android device and comes to this code block.

SparseToDense sparseDense = ctc_decode(out, input_length, true, beam_width, top_paths);

I am getting the sparseDense object perfectly with expected shape but i don't know how to get retrieve float array from this.

If anyone know, please reply...

Thanks.

Mayur Kanojiya
  • 158
  • 1
  • 8
  • What are you asking? You get the dense array from the sparse then you're trying to convert the tensor into floating values to be used? – IanQ Feb 21 '19 at 16:01
  • ** session.runner().fetch(sparseToDense.asOutput()).run() ** were throwing INT64 Sparse OPKernel error, but after converting all the parameters of the SparseToDense to the INT32, it get resolved. However now i worry about the accuracy mismatch in android devices due to this casting. – Mayur Kanojiya Feb 22 '19 at 06:26
  • Please update the question with more relevant information and the actual error messages – IanQ Feb 22 '19 at 14:21

1 Answers1

0

** session.runner().fetch(sparseToDense.asOutput()).run() ** were throwing INT64 Sparse OPKernel error, but after converting all the parameters of the SparseToDense to the INT32, it get resolved.

Mayur Kanojiya
  • 158
  • 1
  • 8