TMS320F28P550SJ: TMS320F28P550SJ: Deploy neural network model from my own ONNX file export by Pytorch.

Zehua Han

Part Number: TMS320F28P550SJ

Dear TI Community，

I have now generated mod.a and tvmgen_default.h in the artifacts_c28 folder, but the test_vector.c and user_input_config.h files have not been generated. I don't know what to do next. Additionally, could you provide guidance on how to use these two generated files (mod.a and tvmgen_default.h) in the artifacts_c28 folder?

Moreover, all the files compiled using the TI Edge AI Studio tool are present, but now there is an issue with converting the ONNX model.

Thanks!

6 天前

0 Taylor 5 天前

TI__Mastermind 38635 points

Hi,

We have received your post and the investigation will take some time. Thank you for your patience.

0 Lydia 1 天前

TI__Mastermind 33515 points

How to use tvmgen_default.h and mod.a are documented in the user's guide:

The tvmc compiler will not generate test_vector.c and user_input_config.h from an onnx model, as those information are not available in the model.

Is the customer using the EdgeAI Studio or ModelMaker?

C2000 微控制器

C2000™︎ 微控制器论坛

TMS320F28P550SJ: TMS320F28P550SJ: Deploy neural network model from my own ONNX file export by Pytorch.