![]() % Compute first-order Taylor scores and accumulate the score across % previous mini-batches of data. = sgdmupdate(prunableNet, netGradients, velocity, learnRate, momentum) % Update the network parameters using the SGDM optimizer. % Evaluate the pruning activations, gradients of the pruning % activations, model gradients, state, and loss using the dlfeval and % modelLossPruning functions. % Reset the velocity parameter for the SGDM solver in every pruning % iteration.įineTuningIteration = fineTuningIteration + 1 The following steps are used:įor pruningIteration = 1:maxPruningIterations Prune the network by repeatedly fine-tuning the network and removing the low scoring filters.įor each pruning iteration. Title( "Number of Prunable Convolution Filters After Pruning") ![]() LineNumPrunables = animatedline(Color=,LineWidth=2,Marker= "^") Īddpoints(lineNumPrunables,0,double(maxPrunableFilters)) Title( "Validation Accuracy After Pruning") LineAccuracyPruning = animatedline(Color=,LineWidth=2,Marker= "o") Īddpoints(lineAccuracyPruning,0,accuracyOfTrainedNet) ![]()
0 Comments
Leave a Reply. |