Exploring Fpga Implementation For Binarized Neural Network Inference