Bike Xie
Bike Xie
About
- Username
- Bike Xie
- Joined
- Visits
- 46
- Last Active
- Roles
- Member, ModeratorsAIModel
Comments
-
Hi Hyun, The toolchain will process the quantization based on the given model and the typical data set. Even if the model weights are UINT8, toolchain will still convert all weights into INT8 and the quantized model will be different from the origin…
-
Hi Ellen, Could you please provide the original keras model (dummy weight is fine)? Thanks Bike
-
Hi Max, Is this onnx model before onnx2onnx.py? I think we needs to check the issue of pytorch2onnx.py first as this is a necessary step.
-
Hi Max, Could you please share the generated onnx with us? We just need to check the onnx structure and you can set the weight to be dummy weight. If this is a pytorch model, please also share the pth file, dummy weight is fine.
-
Thanks for supporting us and helping us to improve our tool chain.
-
Hi Phidias, We found and fixed some simulation bug. It will be released soon in the next tool chain release. For now, you can ignore Csim output and try port to hardware and check performance directly.
-
Hi Phidias, Could you please find "simulator" and "csim" folder, pack them and attach here? We can take a look. Thanks, Bike
-
Hi Phidias, Here is what you need to do to resolve the issue. Follow the instructions in http://doc.kneron.com/docs/#manual_520/#316-onnx-to-onnx-onnx-optimization to process onnx2onnx optimization after tflite2onnx.py and before fpAnalyserCompilerI…
-
This warning is typically when the weight size does not match the expectation based on the onnx node. For example, the conv weight size is smaller than what the conv layer needs. Could you please attach the tensorflow file and onnx file (dummy weigh…