def test_img_local(net_g, dataset_test, dict_users, args):
net_g.eval()
# testing
test_loss = 0
correct = 0
acc_locals, local_data_sizes, loss_locals = [], [], []
idxs_users = range(args.num_users)
for idx in idxs_users:
local = LocalUpdate(args=args,
dataset=dataset_test,
idxs=dict_users[idx],
train=False)
(acc, local_data_size), loss = local.test(net=net_g.to(args.device))
acc_locals.append(acc)
local_data_sizes.append(local_data_size)
loss_locals.append(loss)
test_loss = sum([i * j for i, j in zip(local_data_sizes, loss_locals)
]) / sum(local_data_sizes)
correct = sum([i * j for i, j in zip(local_data_sizes, acc_locals)])
accuracy = correct / sum(local_data_sizes)
if args.verbose:
print('\nTest set: Average loss: {:.4f} \nAccuracy: {}/{} ({:.2f}%)\n'.
format(test_loss, correct, sum(local_data_sizes), accuracy))
return acc_locals, accuracy, test_loss
# print loss
loss_avg = sum(loss_locals) / len(loss_locals)
print('Round {:3d}, Average loss {:.3f}'.format(iter, loss_avg))
loss_train.append(loss_avg)
# plot loss curve
plt.figure()
plt.plot(range(len(loss_train)), loss_train)
plt.ylabel('train_loss')
plt.savefig('./save/fed_{}_{}_{}_C{}_iid{}.png'.format(
args.dataset, args.model, args.epochs, args.frac, args.iid))
# testing
list_acc, list_loss = [], []
net_glob.eval()
for c in range(args.num_users):
net_local = LocalUpdate(args=args,
dataset=dataset_train,
idxs=dict_users[c],
tb=summary)
acc, loss = net_local.test(net=net_glob)
list_acc.append(acc)
list_loss.append(loss)
print("Average testing acc.: {:.2f}%".format(100. * sum(list_acc) /
len(list_acc)))
torch.save(
net_glob,
'./save/model_{}_{}_{}_C{}_iid{}.pt'.format(args.dataset, args.model,
args.epochs, args.frac,
args.iid))
