A custom micropython firmware integrating tensorflow lite for microcontrollers and ulab to implement the tensorflow micro examples.
179
stars
86
forks
source link
Expose tensor type to validate model is as expected #4
Closed
mocleiri closed 3 years ago
I got caught out using the wrong model because I didn't validate the type of the input and output tensors.
We can either expose this on the tensor or add an an option when building the interpreter to validate the types there.
In both cases we need to make some constants that can be used to represent the different tensor types.