LeonardoBizzoni [Mon, 27 May 2024 12:32:51 +0000 (14:32 +0200)]
Fixed creation of weight matrix
A weight was supposed to be a DMatrix<f64> with:
- nrows = node_count in the next layer
- ncols = node_count in the current layer
so that the resulting matrix could be multiplied with the input matrix (column vector)