ERROR: Didn't find op for builtin opcode 'TRANSPOSE_CONV' version '4'. An older version of this builtin might be supported.
CarlosNacher opened this issue · 2 comments
CarlosNacher commented
Fell free to reopen if issue still exist with compiler 16.0. Thanks!
Originally posted by @hjonnala in #147 (comment)
I am getting the same error:
Edge TPU Compiler version 16.0.384591198
ERROR: Didn't find op for builtin opcode 'TRANSPOSE_CONV' version '4'. An older version of this builtin might be supported. Are you using an old TFLite binary with a newer model?
ERROR: Registration failed.
Invalid model: model_full_integer_quant.tflite
Model could not be parsed
-
Edge TPU Compiler version: 16.0.384591198
-
Code which is being used to compile: https://colab.research.google.com/github/google-coral/tutorials/blob/master/compile_for_edgetpu.ipynb
-
The
.tflitealready quantized model (FULL INT 8) I am trying to compile: https://we.tl/t-VkFeyOEedI
vmanoj1996 commented
I am also facing this exact same problem. @CarlosNacher, have you figured out a solution/workaround?
vmanoj1996 commented
(update)
Fixed this issue by setting the experimental new converter to false!
converter.experimental_new_converter = False