StreamingFCLayer_Batch weight order

[MohamedA95]

Hello,
I am trying to implement a network using Finn-hlslib directly. I am a little bit confused about the correct weight order for StreamingFCLayer_Batch. I understand that all layers read the input in a depth-wise raster mode. But when it comes to FCLayer, the input is supposed to be 1D. In my network, I have a QuantMaxPool2d which generates an output of size [6466] followed by a QuantLinear layer where the num of input features is 6466=2304 & num of output features is 120. Hence the QuantLinear weight's size is [120,2304]. When I try to extract the weights to a .hpp how should I order the QuantLinear weight? For conv layers, I followed a depth-wise raster order, but for conv layers, the weight tensor is a 3D tensor. But for QuantLinear weight it's a 2D object!