From d2d5b7400d2bdb413a2ce9a5b9a8b0717bfbedae Mon Sep 17 00:00:00 2001 From: Hammouda Elbez <hammouda.elbez@univ-lille.fr> Date: Tue, 22 Nov 2022 09:55:40 +0100 Subject: [PATCH] ReadMe updated --- README.md | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index bce089e..16d4993 100644 --- a/README.md +++ b/README.md @@ -37,9 +37,17 @@ For example: to run MNIST simulation without compression and reinforcement: In the `apps` folder, you will find the source code for each simulation, where you can change the architecture and the network parameters or activate the [layerwise compression](https://gitlab.univ-lille.fr/hammouda.elbez/progressive-layer-based-compression-for-convolutional-spiking-neural-network/-/blob/main/CSNN-Simulator/apps/Mnist.cpp#L21). ## Going from CSNN to SpiNNaker +First, run the simulation of the two dense layers using CSNN: + + ./MnistForSpiNNaker 0 0 + +then + + ./MnistForSpiNNaker 1 1 + To transfer the learned weights from CSNN to SpiNNaker, we use the following command: - ./Weight_extractor [binary file geenrated from a simulation] [name_layer] + ./Weight_extractor [binary file generated from a simulation] [name_layer] For example: -- GitLab