Hello everybody,
After having successfully converted my model from Onnx to DLC, I'm getting an error at runtime (Android) when trying to build the model.
Error is the following:
Caused by: com.qualcomm.qti.snpe.SnpeError$NativeException: Unable to create network! Cause: error_code=1002; error_message=Layer parameter value is invalid. error_code=1002; error_message=Layer parameter value is invalid. No backend could validate Op=/core/core/layers.0/blocks.0/Transpose Type=Transpose error code=3110; error_component=Model Validation; line_no=1073; thread_id=527913553072; error_component=Model Validation; line_no=277; thread_id=543938721016
The Transpose operation from model info is the following:
| 9 | /core/core/layers.0/blocks.0/Transpose | Transpose | /core/core/layers.0/blocks.0/Reshape_1_output_0 (data type: Float_32; tensor dimension: [1, 4, 16, 4, 16, 96]; tensor type: NATIVE) | /core/core/layers.0/blocks.0/Transpose_output_0 (data type: Float_32; tensor dimension: [1, 4, 4, 16, 16, 96]; tensor type: NATIVE) | 1x4x4x16x16x96 | A D G C | packageName: qti.aisw |
Is there a way to know why this error occurs and how to solve it?
Environment:
Ubuntu 20.04
SNPE: 2.16.0.23.10.29
Python: 3.8
ONNX: 1.11.0
Thanks a lot for your help
Eric
Helo Eric.descheres
According to snpe documentation, at Network models/Supported network layers. ElementWiseSelect (Where) has support for gpu, cpu. However I can only use it with cpu but not gpu.
Environment:
Ubuntu 20.04
SNPE: 2.21.0.240401120655_85612
Python: 3.8
ONNX: 1.11.0