Tool/software:
Hi,
I am experimenting with the TVM model import to offload unsupported layers to the C71 with "c7x_codegen=1".
As far as I can see, currently only the timm example supports this.
I followed all instructions to build the model artifacts on my host and run the inference on the TDA4VM (release 9.2.9.0).
When running the dlr_inference_example.py script ont the target these are the two output images:
py_out_cl-dlr-timm_mobilenetv3_large_100_device_airshow.jpg
py_out_cl-dlr-timm_mobilenetv3_large_100_device_c7x_airshow.jpg
As you can see, the c7x version only outputs zeros.
Am I doing something wrong, or is this still a beta feature that is not expected to work at the moment?
