KL720 compiling error. Need assistance for conversion of onnx model to nef model.
The opset version of onnx used is 11 on VGG-16, simple object detection model. I need to know about all the steps that must be followed for model conversion.
The code and the error is attached above. I am using the docker env provided by kneron.
The discussion has been closed due to inactivity. To continue with the topic, please feel free to post a new discussion.
Comments
Hi Haresh,
There might be a problem with the onnx file. Could you provide your onnx file so we could test it out and look into it? Thanks!
You could also check if your onnx model has operators supported by KL720: Hardware Supported Operators - Document Center (kneron.com)
If it has operators not supported, you might need to use onnx editor to delete them: 3. Floating-Point Model Preparation - Document Center (kneron.com)
For model conversion, we would normally go through these steps:
Model conversion and optimization and config: 3. Floating-Point Model Preparation - Document Center (kneron.com)
Quantization to BIE: 4. Fixed-Point Model Generation - Document Center (kneron.com)
Batch compile to NEF: 5. Compilation - Document Center (kneron.com)
Example: 1. Toolchain Manual Overview - Document Center (kneron.com) (from 1.4-1.6)