i have a trained neural network with 4 input neurons, 7 hidden neurons and 3 output neurons , 49 weights and 10 biases . Now i have to implement it on an FPGA . How do i start verilog code for this ?
i'm very confused . There are so many MAC operations involved . how to make them run in parallel . can u just tell me step by step ? input has 4 neurons.
This could be similar to an electronic circuit simulator. Many branches, interacting with one another. In reality the branches run simultaneously. However your simulation needs to examine them one by one. Then it sums them up to reach a tentative solution. The algorithm checks whether the tentative solution cooperates with conditions in every branch. If not then it goes through several iterations, until it converges on a solution.