Hi there,
I am trying to run a container, which is converted by a tensorflow version yolo weight file, on DSP with example demo snpe-net-run-udl, but it reports error log below:
error_code=902; error_message=Layer parameter value is invalid in DSP. Layer MobileNet/conv_ds_3/pointwise_conv/convolution: Convolution parameters not supported for DSP runtime. Parameters: 208x208x64 input dims, 104x104x128 output dims, 0x0 padding, 1x1 stride, 1x1 kernel, 1x1 dilation, 128 filters, 1 group.; error_component=DSP Runtime; line_no=223; thread_id=-396294856
The container file runs perfectly on CPU and GPU.
The error message says the layer parameter value is invalid in DSP, but I don't know what's wrong with the parameters, and I have checked the limitations in SDK documentations, nothing is related to layer parameters.
Could you please provide the details about layer parameter specification?
Thank you very much.
Hi DamonZhou,
Greetings.
Did you solved this problem ?
I am also facing this issue with my model, it works fine with CPU and GPU but with DLC throws this error.
error_code=902; error_message=Layer parameter value is invalid in DSP. Layer conv2d_5/convolution: Convolution parameters not supported for DSP runtime. Parameters: 2x2x128 input dims, 1x1x2 output dims, 0x0 padding, 1x1 stride, 1x1 kernel, 1x1 dilation, 2 filters, 1 group.; error_component=DSP Runtime; line_no=244; thread_id=-286317400
Can you please guide me with this?
Thanks in Advance,
Kshama Vete