Closed cantonsir closed 3 months ago
I am currently using the hls4ml tool to convert a neural network, but the converted network is encountering some issues.
After converting to hls4ml files and then performing inference, the results are all zeros, which suggests there may be a bug somewhere.
The original neural network, which is based on Keras, works perfectly fine for inference.
My expectation is that, after converting to hls4ml files, the inference should yield approximate results to the original Keras-based model.
Code: https://colab.research.google.com/drive/1FNPvNIoaOCAdPsRIUdxp6TfVFZirgAPb?usp=sharing
Just training the model again, that could solve the problem
I am currently using the hls4ml tool to convert a neural network, but the converted network is encountering some issues.
After converting to hls4ml files and then performing inference, the results are all zeros, which suggests there may be a bug somewhere.
The original neural network, which is based on Keras, works perfectly fine for inference.
My expectation is that, after converting to hls4ml files, the inference should yield approximate results to the original Keras-based model.
Code: https://colab.research.google.com/drive/1FNPvNIoaOCAdPsRIUdxp6TfVFZirgAPb?usp=sharing