pred = output.data.max(1)[1]
correct += pred.eq(target.data).sum()
# test loss average
loss_avg += loss.data[0]
state['test_loss'] = loss_avg / len(test_loader)
state['test_accuracy'] = correct / len(test_loader.dataset)
# Main loop
best_accuracy = 0.0
for epoch in range(args.epochs):
if epoch in args.schedule:
state['learning_rate'] *= args.gamma
for param_group in optimizer.param_groups:
param_group['lr'] = state['learning_rate']
state['epoch'] = epoch
train()
test()
if state['test_accuracy'] > best_accuracy:
best_accuracy = state['test_accuracy']
torch.save(net.state_dict(),
os.path.join(args.save, 'model.pytorch'))
log.write('%s\n' % json.dumps(state))
log.flush()
print(state)
print("Best accuracy: %f" % best_accuracy)
log.close()
best_accuracy = 0.0
for epoch in range(args.epochs):
epochstarttime = datetime.datetime.now()
if epoch in args.schedule:
state['learning_rate'] *= args.gamma
for param_group in optimizer.param_groups:
param_group['lr'] = state['learning_rate']
state['epoch'] = epoch
train()
test()
if state['test_accuracy'] > best_accuracy:
best_accuracy = state['test_accuracy']
torch.save(net.state_dict(), os.path.join(args.save, modeloutput))
log.write('%s\n' % json.dumps(state))
# print(state)
log.write("Best accuracy: %f\n" % best_accuracy)
epochendtime = datetime.datetime.now()
log.write(
f'end: {epochendtime}; len: {epochendtime - epochstarttime}{nextline}'
)
log.flush()
endtime = datetime.datetime.now()
log.write(f'end: {endtime}; len: {endtime - starttime}{nextline}')
log.flush()
log.close()