You may need to set up an ssh certificate for git.

https://docs.github.com/en/authentication/connecting-to-github-with-ssh/generating-a-new-ssh-key-and-adding-it-to-the-ssh-agent

1. Make the file using make
2. Get the output using ./output
3. If needed, clear the object files (*.o) using make clean

There are three different examples for you to choose:

1. Very simple integer BP: example_fc_int_bp_very_simple()
2. Fully connected (fc) mnist integer BP: example_fc_int_bp_mnist()
3. Fully connected (fc) mnist integer DFA: example_fc_int_dfa_mnist()
4. Fully connected (fc) fashion mnist integer DFA: example_fc_int_dfa_fashion_mnist()

You can comment out the current function and call the new function in the main.cpp file

1. For mnist dataset training,use classification loss function batchCrossEntropyLoss() and test the accuracy

1. Error: output: pktnn_mat.cpp:566: pktnn::pktmat& pktnn::pktmat::matElemDivMat(pktnn::pktmat&, pktnn::pktmat&): Assertion `mat1.dimsEqual(mat2)' failed

   Cause: backward() --> computeDeltas() --> matElemDivMat()

2. Training Accuracy low, for example 9.8% Cause: useDfa(false)

3. Training Accuracy 100%, maybe not correct either Try to modify the layer setting to match with BP_simple_training

   //Modified the fc layers to do BP training
   fc1.useDfa(false).initWeightBias().setActv(a).setNextLayer(fc2);         //Use BP for training, set useDfa(false)
   fc2.useDfa(false).initWeightBias().setActv(a).setPrevLayer(fc1).setNextLayer(fcLast);
   fcLast.useDfa(false).setActv(b).setPrevLayer(fc2);
   

1. Neural Networks and Deep Learning - Michael Nielsen http://neuralnetworksanddeeplearning.com/
2. Different activation functions and initialization functions https://shahaab-co.com/mag/en-articles/weight-initialization-in-deep-learning/