Self-Created Tools to convert ONNX files (NCHW) to TensorFlow/TFLite/Keras format (NHWC). The purpose of this tool is to solve the massive Transpose extrapolation problem in onnx-tensorflow (onnx-tf). I don't need a Star, but give me a pull request.
MIT License
662
stars
65
forks
source link
InceptionV3 model convert result not as expected. #662
the part structure of full INT8 Quantization tflite is so different from the origin onnx model. Is it a bug? Can it be simplified as the structure of onnx model with quantized tflite op? Thanks!
Issue Type
Others
OS
Linux
onnx2tf version number
1.23.2
onnx version number
1.15.0
onnxruntime version number
1.17.1
onnxsim (onnx_simplifier) version number
0.4.33
tensorflow version number
2.16.1
Download URL for ONNX
Parameter Replacement JSON
Description
onnx model![image](https://github.com/PINTO0309/onnx2tf/assets/119873470/e064ea24-d75e-46a5-a926-3268bf2f078b)
Full INT8 Quantization tflite
.
command: onnx2tf.py -i inception_v3_Opset18.onnx -oiqt
the part structure of full INT8 Quantization tflite is so different from the origin onnx model. Is it a bug? Can it be simplified as the structure of onnx model with quantized tflite op? Thanks!