tensorflowlite on Redboard artemis Nano

Development Boards Artemis
1

I can not get the Tensorflow Lite examples to compile for the Artemis nano. I get an error saying that the macro F is not happy about getting more than one argument:

error: macro “F” passed 2 arguments, but takes just 1

output_data = F(input1_data, input2_data);
I am currently on an Ubuntu machine running Arduino IDE 1.8.12, Tensorflow Lite 2.4.0-ALPHA, and Sparkfun Apollo3 Boards 2.0.5.
Tried running this same combination on windows and had the same error.
Any help would be appreciated!

2

Thanks for reaching out to us on this.

The portion of the code

output_data = F(input1_data, input2_data);

is definitely passing 2 arguments (input1_data,input2_data)

Perhaps try removing the input2_data portion? If that doesn’t work, following what the user did in this post might help: https://forum.sparkfun.com/viewtopic.php?t=52065

Finally, it might be helpful to start-over altogether, following this guide: https://www.hackster.io/javagoza/artemi … -example-8 with a micro speech example.

Hope this helps, and happy sparking!

3

hugoleon00:
I can not get the Tensorflow Lite examples to compile for the Artemis nano. I get an error saying that the macro F is not happy about getting more than one argument:

error: macro “F” passed 2 arguments, but takes just 1

output_data = F(input1_data, input2_data);
I am currently on an Ubuntu machine running Arduino IDE 1.8.12, Tensorflow Lite 2.4.0-ALPHA, and Sparkfun Apollo3 Boards 2.0.5.
Tried running this same combination on windows and had the same error.
Any help would be appreciated!
[/quote]
Did you ever mange to get it working I’ve been having the same problems

4

Something has already defined F which is used in the template as the comparison function parameter name. I didn’t have time to look for where, I just renamed the function parameter F to __F in the template functions to avoid the name collision. I also changed ComparisonFn to __ComparisonFn . The parameter F appears in 13 places in the file. Be sure and changed them all to __F.

Line 53:

template

using __ComparisonFn = bool (*)(T, T);

template <typename T, __ComparisonFn __F>

inline void ComparisonImpl(

const ComparisonParams& op_params, const RuntimeShape& input1_shape,

const T* input1_data, const RuntimeShape& input2_shape,

const T* input2_data, const RuntimeShape& output_shape, bool* output_data) {

const int64_t flatsize =

MatchingFlatSize(input1_shape, input2_shape, output_shape);

for (int64_t i = 0; i < flatsize; ++i) {

output_data = __F(input1_data, input2_data);
}
}