outputs = net3(Variable(images))
_, predicted = torch.max(outputs.data, 1)
correct3 += (predicted == labels).sum()
total += labels.size(0)
print('Accuracy of the network on the %d test images:%%%f %%%f %%%f' %
(total, 100 * correct1 / total, 100 * correct2 / total,
100 * correct3 / total))
if save_model:
torch.save(
{
'model_state_dict': net1.state_dict(),
'epoch': epoch,
'optimizer_state_dict': optimizer1.state_dict(),
'running_loss': running_loss1,
}, './s1.model')
torch.save(
{
'model_state_dict': net2.state_dict(),
'epoch': epoch,
'optimizer_state_dict': optimizer2.state_dict(),
'running_loss': running_loss2,
}, './s2.model')
torch.save(
{
'model_state_dict': net3.state_dict(),
'epoch': epoch,
'optimizer_state_dict': optimizer3.state_dict(),
'running_loss': running_loss3,
torch.norm(x3 - z).item()) / (3 * N)
print('layer=%d(%d,%f) ADMM=%d primal=%e dual=%e' %
(ci, N, rho, nadmm, primal_residual, dual_residual))
if check_results:
verification_error_check(net1, net2, net3)
print('Finished Training')
if save_model:
torch.save(
{
'model_state_dict': net1.state_dict(),
'epoch': epoch,
'optimizer_state_dict': opt1.state_dict(),
'running_loss': running_loss1,
}, './s1.model')
torch.save(
{
'model_state_dict': net2.state_dict(),
'epoch': epoch,
'optimizer_state_dict': opt2.state_dict(),
'running_loss': running_loss2,
}, './s2.model')
torch.save(
{
'model_state_dict': net3.state_dict(),
'epoch': epoch,
'optimizer_state_dict': opt3.state_dict(),
'running_loss': running_loss3,
# print statistics
if i%(batches_for_report) == (batches_for_report-1): # after every 'batches_for_report'
print('%f: [%d, %5d] loss: %.5f accuracy: %.3f'%
(time.time()-start_time,epoch+1,i+1,running_loss/batches_for_report,
verification_error_check(net)))
running_loss=0.0
print('Finished Training')
# save model (and other extra items)
torch.save({
'model_state_dict':net.state_dict(),
'epoch':epoch,
'optimizer_state_dict':optimizer.state_dict(),
'running_loss':running_loss,
},'./res.model')
# whole dataset
correct=0
total=0
for data in trainloader:
images,labels=data
outputs=net(Variable(images).to(mydevice)).cpu()
_,predicted=torch.max(outputs.data,1)
total += labels.size(0)
correct += (predicted==labels).sum()
print('Accuracy of the network on the %d train images: %d %%'%