sld-columbia / esp

Embedded Scalable Platforms: Heterogeneous SoC architecture and IP integration made easy
Other
344 stars 107 forks source link

Regarding input and output prediction data for hls4ml integration #153

Closed mayank-kabra2001 closed 2 years ago

mayank-kabra2001 commented 2 years ago

Hi, As mentioned in the guide for integrating accelerators in keras(hls4ml) - https://www.esp.cs.columbia.edu/docs/hls4ml/hls4ml-guide/ , we are using KERAS_3layer_input_features.dat and KERAS_3layer_predictions.dat.

Now if I want to include MNIST image classifier, as mentioned in hls4ml framework that in master branch doesn’t provide sets of inputs and expected outputs, which however are important in our case to be able to test the accelerator after the integration in ESP. So how should I create the input_features and ouput_predictions for MNIST classifier ?