navyas321 / c3n2

Clever and Complete Convolutional Neural Network (C3N2) is a complete implementation of Lenet, including both the training and inference layer, on the Xilinx Zynqboard FPGA. The layers are implemented in High Level Synthesis (HLS)

C3N2

Clever and Complete Convolutional Neural Network (C3N2) is a complete implementation of Lenet, including both the training and inference layer, on the Xilinx Zynqboard FPGA. The layers are implemented in High Level Synthesis (HLS)

c3n2's Issues

Training on FPGA zcu102

Hi, this is a very good project! It really helps me. I have compiled the code on SDSoC, and run the program on the zcu102 platform. I just have two questions:

1. Which functions I should move to the HW(PL), all functions in lenet.cpp (like convolution1), or all functions in main.cpp (like testing)?

2. I find that training on the FPGA is very slow, so can I train the model on a host server and just run the testing part on the FPGA board? And would you mind providing me the trained model file (model.dat)?

Thank you.

Best,
Yifan