train_loader = DataLoader(train_dataset, cfg['batch_size'], shuffle=True, num_workers=1)
if not cfg['use_both']:
val_loader = DataLoader(val_dataset, cfg['batch_size'], shuffle=True, num_workers=1)
print('Creating Model...')
net = models[cfg['model_name']](cfg).to(device)
net = nn.DataParallel(net)
n_params = count_parameters(net)
print("model: {:,} M parameters".format(n_params / 1024 / 1024))
criterion = bce_with_logits#nn.CrossEntropyLoss()
optimizer = optim.Adamax(net.parameters(), lr=cfg['init_lr'])
sched = LambdaLR(optimizer, lr_lambda=lr_schedule_func_builder())
checkpoint_path = 'checkpoint/{}'.format(cfg_name)
if os.path.exists(checkpoint_path) is False:
os.mkdir(checkpoint_path)
if cfg['train_check_point'] is not None:
net_checkpoint = torch.load(cfg['train_check_point'])
net.load_state_dict(net_checkpoint)
optim_checkpoint = torch.load('optim_{}'.format(cfg['train_check_point']))
optimizer.load_state_dict(optim_checkpoint)
logger = Logger(os.path.join(checkpoint_path, "log.txt"))
for k, v in cfg.items():
logger.write(k+': {}'.format(v))
if args.optim == 'adam':
optimizer = optim.Adam(model.parameters(),
lr=args.lr,
eps=1e-9,
weight_decay=args.l2_norm,
betas=[0.9, 0.98])
elif args.optim == 'sparseadam':
optimizer = optim.SparseAdam(model.parameters(),
lr=args.lr,
eps=1e-9,
weight_decay=args.l2_norm,
betas=[0.9, 0.98])
elif args.optim == 'adamax':
optimizer = optim.Adamax(model.parameters(),
lr=args.lr,
eps=1e-9,
weight_decay=args.l2_norm,
betas=[0.9, 0.98])
elif args.optim == 'rmsprop':
optimizer = optim.RMSprop(model.parameters(),
lr=args.lr,
eps=1e-10,
weight_decay=args.l2_norm,
momentum=0.9)
elif args.optim == 'sgd':
optimizer = optim.SGD(model.parameters(),
lr=args.lr,
weight_decay=args.l2_norm,
momentum=0.9) # 0.01
elif args.optim == 'adagrad':
optimizer = optim.Adagrad(model.parameters(),
def __init__(self, opt, embedding=None, state_dict=None):
self.opt = opt
self.updates = state_dict[
'updates'] if state_dict and 'updates' in state_dict else 0
self.eval_embed_transfer = True
self.train_loss = AverageMeter()
self.network = DNetwork(opt, embedding)
if state_dict:
new_state = set(self.network.state_dict().keys())
for k in list(state_dict['network'].keys()):
if k not in new_state:
del state_dict['network'][k]
for k, v in list(self.network.state_dict().items()):
if k not in state_dict['network']:
state_dict['network'][k] = v
self.network.load_state_dict(state_dict['network'])
parameters = [p for p in self.network.parameters() if p.requires_grad]
if opt['optimizer'] == 'sgd':
self.optimizer = optim.SGD(parameters,
opt['learning_rate'],
momentum=opt['momentum'],
weight_decay=opt['weight_decay'])
elif opt['optimizer'] == 'adamax':
self.optimizer = optim.Adamax(parameters,
opt['learning_rate'],
weight_decay=opt['weight_decay'])
elif opt['optimizer'] == 'adam':
self.optimizer = optim.Adam(parameters,
opt['learning_rate'],
weight_decay=opt['weight_decay'])
elif opt['optimizer'] == 'adadelta':
self.optimizer = optim.Adadelta(parameters,
opt['learning_rate'],
rho=0.95)
else:
raise RuntimeError('Unsupported optimizer: %s' % opt['optimizer'])
if state_dict and 'optimizer' in state_dict:
self.optimizer.load_state_dict(state_dict['optimizer'])
if opt['fix_embeddings']:
wvec_size = 0
else:
wvec_size = (opt['vocab_size'] -
opt['tune_partial']) * opt['embedding_dim']
if opt.get('have_lr_scheduler', False):
if opt.get('scheduler_type', 'rop') == 'rop':
self.scheduler = ReduceLROnPlateau(self.optimizer,
mode='max',
factor=opt['lr_gamma'],
patience=3)
elif opt.get('scheduler_type', 'rop') == 'exp':
self.scheduler = ExponentioalLR(self.optimizer,
gamma=opt.get('lr_gamma', 0.5))
else:
milestones = [
int(step)
for step in opt.get('multi_step_lr', '10,20,30').split(',')
]
self.scheduler = MultiStepLR(self.optimizer,
milestones=milestones,
gamma=opt.get('lr_gamma'))
else:
self.scheduler = None
self.total_param = sum([p.nelement() for p in parameters]) - wvec_size
Nbatches = int(math.ceil(Ntrain / batch_size)) #batch_size is defined above
Nepochs = 500
Nrep = 1
#model = conv3DNet(grid_size, Noutputs, batch_size)
#model = conv3DNet(grid_size, Noutputs, batch_size)
#model = conv3DNet(grid_size, Noutputs, batch_size)
#model = conv3DNet(grid_size, Noutputs, batch_size)
#model = UnetGenerator_3d(in_dim=1, out_dim=Noutputs, num_filter=4)
#model = UnetGenerator_3d_softmax(in_dim=1, out_dim=Noutputs, num_filter=8)
model = UnetGenerator_3d_log_softmax(in_dim=1, out_dim=Noutputs, num_filter=4)
#optimizer = optim.SGD(model.parameters(), lr=1e-2, momentum=0.90)
#optimizer = optim.Adam(model.parameters())
#optimizer = optim.Adagrad(model.parameters())
optimizer = optim.Adamax(model.parameters())
#optimizer = optim.ASGD(model.parameters())
#optimizer = optim.RMSprop(model.parameters())
#optimizer = optim.Rprop(model.parameters())
scheduler = optim.lr_scheduler.ReduceLROnPlateau(
optimizer, 'min', patience=10, verbose=True
) #Reduces the learning rate if it did not decreased by more than 10^-4 in 10 steps
train_errors = torch.Tensor(Nepochs).zero_()
validation_errors = torch.Tensor(Nepochs).zero_()
ep_loss = torch.Tensor(Nepochs).zero_()
for i_ep in range(Nepochs):
for b_start in range(0, Ntrain, batch_size):
def _opti(self, parameters):
return optim.Adamax(parameters)
torch.manual_seed(args.seed)
if torch.cuda.is_available():
if not args.cuda:
print("WARNING: You have a CUDA device, so you should probably run with --cuda")
else:
torch.cuda.manual_seed(args.seed)
corpus = Corpus(args.task)
model = eval(args.model)(corpus, args)
model.train()
criterion = nn.NLLLoss()
parameters = filter(lambda p: p.requires_grad, model.parameters())
optimizer = optim.Adamax(parameters, lr=args.lr)
if args.cuda:
model.cuda()
criterion.cuda()
start_time = time.time()
total_loss = 0
interval = args.interval
save_interval = len(corpus.data_all['train']) // args.batch_size
best_dev_score = -99999
iterations = args.epochs*len(corpus.data_all['train']) // args.batch_size
print('max iterations: '+str(iterations))
count = 0
def run(args, kwargs):
print('\nMODEL SETTINGS: \n', args, '\n')
print("Random Seed: ", args.manual_seed)
# ==================================================================================================================
# SNAPSHOTS
# ==================================================================================================================
args.model_signature = str(datetime.datetime.now())[0:19].replace(' ', '_')
args.model_signature = args.model_signature.replace(':', '_')
snapshots_path = os.path.join(args.out_dir, 'vae_' + args.dataset + '_')
snap_dir = snapshots_path + args.flow + '_' + str(args.gpu_num)
if args.flow != 'no_flow':
snap_dir += '_' + 'num_flows_' + str(args.num_flows)
if args.flow == 'orthogonal':
snap_dir = snap_dir + '_num_vectors_' + str(args.num_ortho_vecs)
elif args.flow == 'orthogonalH':
snap_dir = snap_dir + '_num_householder_' + str(args.num_householder)
elif args.flow == 'iaf':
snap_dir = snap_dir + '_madehsize_' + str(args.made_h_size)
elif args.flow == 'permutation':
snap_dir = snap_dir + '_' + 'kernelsize_' + str(args.kernel_size)
elif args.flow == 'mixed':
snap_dir = snap_dir + '_' + 'num_householder_' + str(args.num_householder)
snap_dir = snap_dir + '__' + args.model_signature + '/'
args.snap_dir = snap_dir
if not os.path.exists(snap_dir):
os.makedirs(snap_dir)
# SAVING
torch.save(args, snap_dir + args.flow + '.config')
# ==================================================================================================================
# LOAD DATA
# ==================================================================================================================
train_loader, val_loader, test_loader, args = load_dataset(args, **kwargs)
# ==================================================================================================================
# SELECT MODEL
# ==================================================================================================================
# flow parameters and architecture choice are passed on to model through args
if args.flow == 'no_flow':
model = VAE.VAE(args)
elif args.flow == 'planar':
model = VAE.PlanarVAE(args)
elif args.flow == 'iaf':
model = VAE.IAFVAE(args)
elif args.flow == 'orthogonal':
model = VAE.OrthogonalSylvesterVAE(args)
elif args.flow == 'householder':
model = VAE.HouseholderSylvesterVAE(args)
elif args.flow == 'triangular':
model = VAE.TriangularSylvesterVAE(args)
else:
raise ValueError('Invalid flow choice')
if args.cuda:
print("Model on GPU")
model.cuda()
print(model)
optimizer = optim.Adamax(model.parameters(), lr=args.learning_rate, eps=1.e-7)
# ==================================================================================================================
# TRAINING
# ==================================================================================================================
train_loss = []
val_loss = []
# for early stopping
best_loss = np.inf
best_bpd = np.inf
e = 0
epoch = 0
train_times = []
for epoch in range(1, args.epochs + 1):
t_start = time.time()
tr_loss = train(epoch, train_loader, model, optimizer, args)
train_loss.append(tr_loss)
train_times.append(time.time()-t_start)
print('One training epoch took %.2f seconds' % (time.time()-t_start))
v_loss, v_bpd = evaluate(val_loader, model, args, epoch=epoch)
val_loss.append(v_loss)
# early-stopping
if v_loss < best_loss:
e = 0
best_loss = v_loss
if args.input_type != 'binary':
best_bpd = v_bpd
print('->model saved<-')
torch.save(model, snap_dir + args.flow + '.model')
# torch.save(model, snap_dir + args.flow + '_' + args.architecture + '.model')
elif (args.early_stopping_epochs > 0) and (epoch >= args.warmup):
e += 1
if e > args.early_stopping_epochs:
break
if args.input_type == 'binary':
print('--> Early stopping: {}/{} (BEST: loss {:.4f})\n'.format(e, args.early_stopping_epochs, best_loss))
else:
print('--> Early stopping: {}/{} (BEST: loss {:.4f}, bpd {:.4f})\n'.format(e, args.early_stopping_epochs,
best_loss, best_bpd))
if math.isnan(v_loss):
raise ValueError('NaN encountered!')
train_loss = np.hstack(train_loss)
val_loss = np.array(val_loss)
plot_training_curve(train_loss, val_loss, fname=snap_dir + '/training_curve_%s.pdf' % args.flow)
# training time per epoch
train_times = np.array(train_times)
mean_train_time = np.mean(train_times)
std_train_time = np.std(train_times, ddof=1)
print('Average train time per epoch: %.2f +/- %.2f' % (mean_train_time, std_train_time))
# ==================================================================================================================
# EVALUATION
# ==================================================================================================================
test_score_file = snap_dir + 'test_scores.txt'
with open('experiment_log.txt', 'a') as ff:
print(args, file=ff)
print('Stopped after %d epochs' % epoch, file=ff)
print('Average train time per epoch: %.2f +/- %.2f' % (mean_train_time, std_train_time), file=ff)
final_model = torch.load(snap_dir + args.flow + '.model')
if args.testing:
validation_loss, validation_bpd = evaluate(val_loader, final_model, args)
test_loss, test_bpd = evaluate(test_loader, final_model, args, testing=True)
with open('experiment_log.txt', 'a') as ff:
print('FINAL EVALUATION ON VALIDATION SET\n'
'ELBO (VAL): {:.4f}\n'.format(validation_loss), file=ff)
print('FINAL EVALUATION ON TEST SET\n'
'NLL (TEST): {:.4f}\n'.format(test_loss), file=ff)
if args.input_type != 'binary':
print('FINAL EVALUATION ON VALIDATION SET\n'
'ELBO (VAL) BPD : {:.4f}\n'.format(validation_bpd), file=ff)
print('FINAL EVALUATION ON TEST SET\n'
'NLL (TEST) BPD: {:.4f}\n'.format(test_bpd), file=ff)
else:
validation_loss, validation_bpd = evaluate(val_loader, final_model, args)
# save the test score in case you want to look it up later.
_, _ = evaluate(test_loader, final_model, args, testing=True, file=test_score_file)
with open('experiment_log.txt', 'a') as ff:
print('FINAL EVALUATION ON VALIDATION SET\n'
'ELBO (VALIDATION): {:.4f}\n'.format(validation_loss), file=ff)
if args.input_type != 'binary':
print('FINAL EVALUATION ON VALIDATION SET\n'
'ELBO (VAL) BPD : {:.4f}\n'.format(validation_bpd), file=ff)
def train(self):
"""
Train the model and print out training data.
"""
# Define Loss function and Optimizer
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adamax(self._net.parameters(), lr=0.001)
# Use CUDA device if available
device = self._set_device()
start_time = time.perf_counter()
train_loss_history = []
train_acc_history = []
# Train the network
for epoch in range(10):
running_loss = 0.0
train_loss = 0.0
correct = 0
total = 0
for i, data in enumerate(self._data_loader, 0):
# data is a list of [inputs, labels]
inputs, labels = data[0].to(device), data[1].to(device)
# clear the parameter gradients
optimizer.zero_grad()
# forward + backward + optimize
outputs = self._net(inputs)
loss = criterion(outputs, labels)
loss.backward()
optimizer.step()
# calculate training accuracy and loss
_, predictions = torch.max(outputs, 1)
correct += (predictions == labels).sum().item()
total += labels.size(0)
train_loss += loss.item()
# print loss and accuracy every 500 mini-batches
running_loss += loss.item()
if i % 500 == 499:
print(
'Epoch %d/10, %5d mini-batches, Loss: %.3f, Accuracy: %.3f'
% (epoch + 1, i + 1, running_loss / 500,
correct / total))
running_loss = 0.0
train_loss_history.append(train_loss / len(self._data_loader))
train_acc_history.append(correct / total)
# print training time
end_time = time.perf_counter()
print(
f'Finished training in {(end_time - start_time)/60:.2f} minutes.')
# plot training accuracy and loss curve
plt.plot(np.array(train_loss_history), 'b', label='Training Loss')
plt.plot(np.array(train_acc_history), 'y', label='Training Accuracy')
plt.legend()
plt.show()
self.save_network()
def main_train(epoch_start, epoch_end, train, arg):
#%%
net_name = arg['net_name']
loss_name = arg['loss_name']
filename = get_filename(net_name, loss_name)
print('train model: ' + filename)
if epoch_start == epoch_end:
print('epoch_end is epoch_start, exist main_train')
return
#---------------------------------------
device = arg['device']
lr = arg['lr']
norm_type = arg['norm_type']
rand_pad = arg['rand_pad']
#%%
num_classes = 9
if norm_type == np.inf:
noise_norm = 0.1
max_iter = 1
step = 1.0
elif norm_type == 2:
noise_norm = 5.0
max_iter = 1
step = 1.0
#%%
loader_train, loader_val = get_dataloader(rand_pad=rand_pad)
#%%
loss_train_list = []
acc_train_list = []
acc_val_list = []
acc_test_list = []
epoch_save = epoch_start - 1
#%%
model = Net(net_name)
if epoch_start > 0:
print('load', filename + '_epoch' + str(epoch_save) + '.pt')
checkpoint = torch.load(filename + '_epoch' + str(epoch_save) + '.pt',
map_location=torch.device('cpu'))
model.load_state_dict(checkpoint['model_state_dict'])
#------------------------
loss_train_list = checkpoint['result']['loss_train_list']
acc_train_list = checkpoint['result']['acc_train_list']
acc_val_list = checkpoint['result']['acc_val_list']
acc_test_list = checkpoint['result']['acc_test_list']
if 'E' in arg.keys():
if arg['E'] is None:
arg['E'] = checkpoint['result']['arg']['E']
print('load E')
#------------------------
model.to(device)
#------------------------
if arg['optimizer'] == 'Adam':
optimizer = optim.Adam(model.parameters(), lr=lr)
elif arg['optimizer'] == 'AdamW':
optimizer = optim.AdamW(model.parameters(), lr=lr)
elif arg['optimizer'] == 'Adamax':
optimizer = optim.Adamax(model.parameters(), lr=lr)
elif arg['optimizer'] == 'SGD':
optimizer = optim.SGD(model.parameters(),
lr=lr,
momentum=0.9,
weight_decay=0.001,
nesterov=True)
else:
raise NotImplementedError('unknown optimizer')
#%%
for epoch in range(epoch_save + 1, epoch_end):
#-------- training --------------------------------
start = time.time()
loss_train, acc_train = train(model, device, optimizer, loader_train,
epoch, arg)
loss_train_list.append(loss_train)
acc_train_list.append(acc_train)
print('epoch', epoch, 'training loss:', loss_train, 'acc:', acc_train)
end = time.time()
print('time cost:', end - start)
#-------- validation --------------------------------
result_val = test(model,
device,
loader_val,
num_classes=num_classes,
class_balanced_acc=True)
acc_val_list.append(result_val['acc'])
#-------- test --------------------------------
#result_test = test(model, device, loader_test, num_classes=num_classes, class_balanced_acc=True)
#acc_test_list.append(result_test['acc'])
#--------save model-------------------------
result = {}
result['arg'] = arg
result['loss_train_list'] = loss_train_list
result['acc_train_list'] = acc_train_list
result['acc_val_list'] = acc_val_list
result['acc_test_list'] = acc_test_list
if (epoch + 1) % 10 == 0:
save_checkpoint(filename + '_epoch' + str(epoch) + '.pt', model,
result, epoch)
epoch_save = epoch
#------- show result ----------------------
#plt.close('all')
display.clear_output(wait=False)
fig, ax = plot_result(loss_train_list, acc_train_list, acc_val_list,
acc_test_list)
display.display(fig)
fig.savefig(filename + '_epoch' + str(epoch) + '.png')
plt.close(fig)
model = Net()
model.to(device)
if config.LOSS == "l1":
loss_function = nn.L1Loss()
elif config.LOSS == "vgg":
loss_function = loss.VggLoss()
elif config.LOSS == "ssim":
loss_function = loss.SsimLoss()
elif config.LOSS == "l1+vgg":
loss_function = loss.CombinedLoss()
else:
raise ValueError(f"Unknown loss: {config.LOSS}")
optimizer = optim.Adamax(model.parameters(), lr=0.001)
board_writer = SummaryWriter()
# ----------------------------------------------------------------------
def train(epoch):
print("===> Training...")
before_pass = [p.data.clone() for p in model.parameters()]
epoch_loss = 0
for iteration, batch in enumerate(training_data_loader, 1):
input, target = batch[0].to(device), batch[1].to(device)
optimizer.zero_grad()
if __name__ == '__main__':
model = init_model(config.START_FROM_EXISTING_MODEL)
if config.LOSS == "l1":
loss_function = nn.L1Loss()
elif config.LOSS == "vgg":
loss_function = loss.VggLoss()
elif config.LOSS == "ssim":
loss_function = loss.SsimLoss()
elif config.LOSS == "l1+vgg":
loss_function = loss.CombinedLoss()
else:
raise ValueError(f"Unknown loss: {config.LOSS}")
optimizer = optim.Adamax(model.parameters(), lr=config.LEARNING_RATE[0])
board_writer = SummaryWriter()
# ----------------------------------------------------------------------
def train(epoch):
print("===> Training...")
before_pass = [p.data.clone() for p in model.parameters()]
epoch_loss = 0
target_crop = _make_target_crop(config.PATCH_SIZE[0], config.PATCH_SIZE[1],
config.CROP_SIZE, config.CROP_SIZE)
epoch_lr = config.LEARNING_RATE[-1]
def forward(self, x):
x = self.hidden(x)
x = self.hidden2(x)
x = self.hidden3(x)
x = self.sig(x)
x = self.output(x)
return x
in_d, out_d, in_test, out_test = load_data()
net = Net()
crit = nn.MSELoss()
# opt = optim.SGD(params=net.parameters(),lr= 0.01)
opt = optim.Adamax(params=net.parameters(), lr=0.002, betas=(0.9, 0.999))
for epoch in range(100):
loss_value = 0.0
i = 0
for values in in_d:
opt.zero_grad()
temp = np.matrix(values)
tensorIn = torch.from_numpy(temp).float()
outs = net(tensorIn)
tensorOut = torch.from_numpy(out_d[i]).float()
loss = crit(outs, tensorOut)
def optim_selection(self):
if self.config.optim == "Nesterov":
return optim.SGD(
self.model.parameters(),
lr=self.config.lr,
momentum=0.9,
nesterov=True,
weight_decay=0.0001,
)
elif self.config.optim == "SGD": # weight_decay = l2 regularization
return optim.SGD(
self.model.parameters(),
lr=self.config.lr,
momentum=0.9,
nesterov=False,
weight_decay=0.0001,
)
elif self.config.optim == "Adadelta": # default lr = 1.0
return optim.Adadelta(
self.model.parameters(),
lr=self.config.lr,
rho=0.9,
eps=1e-06,
weight_decay=1e-6,
)
elif self.config.optim == "Adagrad": # default lr = 0.01
return optim.Adagrad(
self.model.parameters(),
lr=self.config.lr,
lr_decay=0,
weight_decay=1e-6,
initial_accumulator_value=0,
eps=1e-10,
)
elif self.config.optim == "Adam": # default lr=0.001
return optim.Adam(self.model.parameters(),
lr=self.config.lr,
weight_decay=1e-6)
elif self.config.optim == "AdamW": # default lr=0.001
return optim.AdamW(
self.model.parameters(),
lr=self.config.lr,
betas=(0.9, 0.999),
eps=1e-08,
weight_decay=0.01,
amsgrad=False,
)
elif self.config.optim == "SparseAdam": # default lr = 0.001
return optim.SparseAdam(
self.model.parameters(),
lr=self.config.lr,
betas=(0.9, 0.999),
eps=1e-08,
)
elif self.config.optim == "Adamax": # default lr=0.002
return optim.Adamax(
self.model.parameters(),
lr=self.config.lr,
betas=(0.9, 0.999),
eps=1e-08,
weight_decay=1e-6,
)
elif self.config.optim == "ASGD":
return optim.ASGD(
self.model.parameters(),
lr=self.config.lr,
lambd=0.0001,
alpha=0.75,
t0=1000000.0,
weight_decay=1e-6,
)
elif self.config.optim == "RMSprop": # default lr=0.01
return optim.RMSprop(
self.model.parameters(),
lr=self.config.lr,
alpha=0.99,
eps=1e-08,
weight_decay=0,
momentum=0,
centered=False,
)
elif self.config.optim == "Rprop": # default lr=0.01
return optim.Rprop(
self.model.parameters(),
lr=self.config.lr,
etas=(0.5, 1.2),
step_sizes=(1e-06, 50),
)
def main(argv):
config = Config()
config.load_user_config()
config.log.info("finish loading user config")
train_file = config.args["train_file"]
dev_file = config.args["dev_file"]
old_glove_file = config.args["glove_file"]
new_glove_file = config.args["glove_file"] + ".subset"
# TODO(demi): switch "overwrite" to False
train_data_raw, dev_data_raw, i2w, w2i, i2c, c2i, new_glove_file, glove_dim, vocab_size, char_vocab_size\
= squad_read_data(config, train_file, dev_file, old_glove_file, new_glove_file, overwrite=True)
config.log.info("finish reading squad data in raw formats")
config.update_batch([("glove_file", new_glove_file),
("glove_dim", glove_dim),
("vocab_size", vocab_size),
("char_vocab_size", char_vocab_size)])
config.log.warning("reminder: now we only support train/fake mode")
assert config.args["mode"] in ["train", "fake"], "mode (%s) not found" % config.args["mode"]
train_id_conversion, train_data = make_dataset(config, train_data_raw, w2i, c2i)
dev_id_conversion, dev_data = make_dataset(config, dev_data_raw, w2i, c2i)
config.log.info("finish making datasets: reformatting raw data")
train_data = QnADataset(train_data, config)
dev_data = QnADataset(dev_data, config)
config.log.info("finish generating datasets")
train_loader = torch.utils.data.DataLoader(train_data, batch_size=1, shuffle=True, **config.kwargs)
dev_loader = torch.utils.data.DataLoader(dev_data, batch_size=1, **config.kwargs)
config.log.info("finish generating data loader")
model = BiDAF(config, i2w)
config.log.info("finish creating model")
if config.args["use_cuda"]:
model.cuda()
# log config and model
config.log.info(config.format_string())
config.log.info("model:{}".format(model))
if config.args['optimizer'] == "Adam":
optimizer = optim.Adam(model.get_train_parameters(), lr=config.args['lr'], weight_decay=config.args['weight_decay'])
if config.args['optimizer'] == "Adamax":
optimizer = optim.Adamax(model.get_train_parameters(), lr=config.args['lr'], weight_decay=config.args['weight_decay'])
if config.args['optimizer'] == "SGD":
optimizer = torch.optim.SGD(model.get_train_parameters(), lr=config.args['lr'], momentum=0.9, weight_decay=config.args['weight_decay'])
if config.args['optimizer'] == "Adadelta":
optimizer = torch.optim.Adadelta(model.get_train_parameters(), lr=config.args["lr"])
#if config.args['optimizer'] == "Adagrad":
config.log.info("model = %s" % model)
config.log.info("config = %s" % config.format_string())
trainer = Trainer(config)
evaluator = Evaluator(config)
""" save model checkpoint """
def save_checkpoint(epoch):
checkpoint = {"model_state_dict": model.state_dict(),
"config_args" : config.args}
if config.args["optimizer"] != "YF": # YF can't save state dict right now
checkpoint["optimizer_state_dict"] = optimizer.state_dict()
checkpoint_file = config.args["model_dir"] + config.args["model_name"] + "-EPOCH%d" % epoch
torch.save(checkpoint, checkpoint_file)
config.log.info("saving checkpoint: {}".format(checkpoint_file))
for epoch in range(1, config.args["max_epoch"] + 1):
config.log.info("training: epoch %d" % epoch)
# QS(demi): do i need to return model & optimizer?
model, optimizer, train_avg_loss, train_answer_dict = trainer.run(model, train_id_conversion[0], train_loader, optimizer, mode="train")
model, optimizer, dev_avg_loss, dev_answer_dict = trainer.run(model, dev_id_conversion[0], dev_loader, optimizer, mode="dev")
# loss is a float tensor with size 1
config.log.info("[EPOCH %d] LOSS = (train)%.5lf | (dev)%.5lf" % (epoch, train_avg_loss[0], dev_avg_loss[0]))
answer_filename = "{}/{}-EPOCH{}".format(config.args["model_dir"], config.args["model_name"], epoch)
config.log.info("[EVAUATION] TRAIN EVAL")
evaluator.eval("official", train_file, train_answer_dict, "{}/answer.train".format(config.args["model_dir"], answer_filename))
config.log.info("[EVAUATION] DEV EVAL")
evaluator.eval("official", dev_file, dev_answer_dict, "{}/answer.dev".format(config.args["model_dir"], answer_filename))
save_checkpoint(epoch)
def main(args):
device = torch.device("cuda:0")
# model hyperparameters
dataset = args.dataset
batch_size = args.batch_size
hps = Hyperparameters(base_dim=args.base_dim,
res_blocks=args.res_blocks,
bottleneck=args.bottleneck,
skip=args.skip,
weight_norm=args.weight_norm,
coupling_bn=args.coupling_bn,
affine=args.affine)
scale_reg = 5e-5 # L2 regularization strength
# optimization hyperparameters
lr = args.lr
momentum = args.momentum
decay = args.decay
# prefix for images and checkpoints
filename = 'bs%d_' % batch_size \
+ 'normal_' \
+ 'bd%d_' % hps.base_dim \
+ 'rb%d_' % hps.res_blocks \
+ 'bn%d_' % hps.bottleneck \
+ 'sk%d_' % hps.skip \
+ 'wn%d_' % hps.weight_norm \
+ 'cb%d_' % hps.coupling_bn \
+ 'af%d' % hps.affine \
# load dataset
train_split, val_split, data_info = data_utils.load(dataset)
train_loader = torch.utils.data.DataLoader(train_split,
batch_size=batch_size,
shuffle=True,
num_workers=2)
val_loader = torch.utils.data.DataLoader(val_split,
batch_size=batch_size,
shuffle=False,
num_workers=2)
prior = distributions.Normal( # isotropic standard normal distribution
torch.tensor(0.).to(device),
torch.tensor(1.).to(device))
flow = realnvp.RealNVP(datainfo=data_info, prior=prior, hps=hps).to(device)
optimizer = optim.Adamax(flow.parameters(),
lr=lr,
betas=(momentum, decay),
eps=1e-7)
epoch = 0
running_loss = 0.
running_log_ll = 0.
optimal_log_ll = float('-inf')
early_stop = 0
image_size = data_info.channel * data_info.size**2 # full image dimension
while epoch < args.max_epoch:
epoch += 1
print('Epoch %d:' % epoch)
flow.train()
for batch_idx, data in enumerate(train_loader, 1):
optimizer.zero_grad()
x, _ = data
# log-determinant of Jacobian from the logit transform
x, log_det = data_utils.logit_transform(x)
x = x.to(device)
log_det = log_det.to(device)
# log-likelihood of input minibatch
log_ll, weight_scale = flow(x)
log_ll = (log_ll + log_det).mean()
# add L2 regularization on scaling factors
loss = -log_ll + scale_reg * weight_scale
running_loss += loss.item()
running_log_ll += log_ll.item()
loss.backward()
optimizer.step()
if batch_idx % 10 == 0:
bit_per_dim = (-log_ll.item() + np.log(256.) * image_size) \
/ (image_size * np.log(2.))
print('[%d/%d]\tloss: %.3f\tlog-ll: %.3f\tbits/dim: %.3f' % \
(batch_idx*batch_size, len(train_loader.dataset),
loss.item(), log_ll.item(), bit_per_dim))
mean_loss = running_loss / batch_idx
mean_log_ll = running_log_ll / batch_idx
mean_bit_per_dim = (-mean_log_ll + np.log(256.) * image_size) \
/ (image_size * np.log(2.))
print('===> Average train loss: %.3f' % mean_loss)
print('===> Average train log-likelihood: %.3f' % mean_log_ll)
print('===> Average train bit_per_dim: %.3f' % mean_bit_per_dim)
running_loss = 0.
running_log_ll = 0.
flow.eval()
with torch.no_grad():
for batch_idx, data in enumerate(val_loader, 1):
x, _ = data
x, log_det = data_utils.logit_transform(x)
x = x.to(device)
log_det = log_det.to(device)
# log-likelihood of input minibatch
log_ll, weight_scale = flow(x)
log_ll = (log_ll + log_det).mean()
# add L2 regularization on scaling factors
loss = -log_ll + scale_reg * weight_scale
running_loss += loss.item()
running_log_ll += log_ll.item()
mean_loss = running_loss / batch_idx
mean_log_ll = running_log_ll / batch_idx
mean_bit_per_dim = (-mean_log_ll + np.log(256.) * image_size) \
/ (image_size * np.log(2.))
print('===> Average validation loss: %.3f' % mean_loss)
print('===> Average validation log-likelihood: %.3f' % mean_log_ll)
print('===> Average validation bits/dim: %.3f' % mean_bit_per_dim)
running_loss = 0.
running_log_ll = 0.
samples = flow.sample(args.sample_size)
samples, _ = data_utils.logit_transform(samples, reverse=True)
utils.save_image(
utils.make_grid(samples),
'./samples/' + dataset + '/' + filename + '_ep%d.png' % epoch)
if mean_log_ll > optimal_log_ll:
early_stop = 0
optimal_log_ll = mean_log_ll
torch.save(flow, './models/' + dataset + '/' + filename + '.model')
print('[MODEL SAVED]')
else:
early_stop += 1
if early_stop >= 100:
break
print('--> Early stopping %d/100 (BEST validation log-likelihood: %.3f)' \
% (early_stop, optimal_log_ll))
print('Training finished at epoch %d.' % epoch)
def train():
parser = argparse.ArgumentParser()
# 配置文件
parser.add_argument(
"--config-yml",
default="exp_fvqa/exp2.yml",
help=
"Path to a config file listing reader, model and solver parameters.")
parser.add_argument("--cpu-workers",
type=int,
default=8,
help="Number of CPU workers for dataloader.")
parser.add_argument(
"--save-dirpath",
default="fvqa/exp_data/checkpoints",
help=
"Path of directory to create checkpoint directory and save checkpoints."
)
parser.add_argument(
"--load-pthpath",
default="",
help="To continue training, path to .pth file of saved checkpoint.")
parser.add_argument("--gpus", default="", help="gpus")
parser.add_argument(
"--overfit",
action="store_true",
help="Whether to validate on val split after every epoch.")
parser.add_argument(
"--validate",
action="store_true",
help="Whether to validate on val split after every epoch.")
args = parser.parse_args()
# set mannual seed
torch.manual_seed(10)
torch.cuda.manual_seed(10)
cudnn.benchmark = True
cudnn.deterministic = True
config = yaml.load(open(args.config_yml))
device = torch.device("cuda:0") if args.gpus != "cpu" else torch.device(
"cpu")
# Print config and args.
print(yaml.dump(config, default_flow_style=False))
for arg in vars(args):
print("{:<20}: {}".format(arg, getattr(args, arg)))
print('Loading TrainDataset...')
train_dataset = FvqaTrainDataset(config, overfit=args.overfit)
train_dataloader = DataLoader(train_dataset,
batch_size=config['solver']['batch_size'],
num_workers=args.cpu_workers,
shuffle=True,
collate_fn=collate_fn)
if args.validate:
print('Loading TestDataset...')
val_dataset = FvqaTestDataset(config, overfit=args.overfit)
val_dataloader = DataLoader(val_dataset,
batch_size=config['solver']['batch_size'],
num_workers=args.cpu_workers,
shuffle=True,
collate_fn=collate_fn)
print('Loading glove...')
que_vocab = Vocabulary(config['dataset']['word2id_path'])
glove = np.load(config['dataset']['glove_vec_path'])
glove = torch.Tensor(glove)
print('Building Model...')
model = CMGCNnet(config,
que_vocabulary=que_vocab,
glove=glove,
device=device)
if torch.cuda.device_count() > 1 and args.gpus != "cpu":
print("Let's use", torch.cuda.device_count(), "GPUs!")
model = nn.DataParallel(model)
model = model.to(device)
print(model)
iterations = len(train_dataset) // config["solver"]["batch_size"] + 1
def lr_lambda_fun(current_iteration: int) -> float:
current_epoch = float(current_iteration) / iterations
if current_epoch <= config["solver"]["warmup_epochs"]:
alpha = current_epoch / float(config["solver"]["warmup_epochs"])
return config["solver"]["warmup_factor"] * (1. - alpha) + alpha
else:
idx = bisect(config["solver"]["lr_milestones"], current_epoch)
return pow(config["solver"]["lr_gamma"], idx)
optimizer = optim.Adamax(model.parameters(),
lr=config["solver"]["initial_lr"])
scheduler = lr_scheduler.LambdaLR(optimizer, lr_lambda=lr_lambda_fun)
T = iterations * (config["solver"]["num_epochs"] -
config["solver"]["warmup_epochs"] + 1)
scheduler2 = lr_scheduler.CosineAnnealingLR(
optimizer, int(T), eta_min=config["solver"]["eta_min"], last_epoch=-1)
summary_writer = SummaryWriter(log_dir=args.save_dirpath)
checkpoint_manager = CheckpointManager(model,
optimizer,
args.save_dirpath,
config=config)
if args.load_pthpath == "":
start_epoch = 0
else:
start_epoch = int(args.load_pthpath.split("_")[-1][:-4])
model_state_dict, optimizer_state_dict = load_checkpoint(
args.load_pthpath)
if isinstance(model, nn.DataParallel):
model.module.load_state_dict(model_state_dict)
else:
model.load_state_dict(model_state_dict)
optimizer.load_state_dict(optimizer_state_dict)
print("Loading resume model from {}...".format(args.load_pthpath))
global_iteration_step = start_epoch * iterations
for epoch in range(start_epoch, config['solver']['num_epochs']):
print(f"\nTraining for epoch {epoch}:")
train_answers = []
train_preds = []
for i, batch in enumerate(tqdm(train_dataloader)):
optimizer.zero_grad()
fact_batch_graph = model(batch)
batch_loss = cal_batch_loss(fact_batch_graph,
batch,
device,
neg_weight=0.1,
pos_weight=0.9)
batch_loss.backward()
optimizer.step()
fact_graphs = dgl.unbatch(fact_batch_graph)
for i, fact_graph in enumerate(fact_graphs):
train_pred = fact_graph.ndata['h'].squeeze() # (num_nodes,1)
train_preds.append(train_pred) # [(num_nodes,)]
train_answers.append(batch['facts_answer_id_list'][i])
summary_writer.add_scalar('train/loss', batch_loss,
global_iteration_step)
summary_writer.add_scalar("train/lr",
optimizer.param_groups[0]["lr"],
global_iteration_step)
summary_writer.add_text('train/loss', str(batch_loss.item()),
global_iteration_step)
summary_writer.add_text('train/lr',
str(optimizer.param_groups[0]["lr"]),
global_iteration_step)
if global_iteration_step <= iterations * config["solver"][
"warmup_epochs"]:
scheduler.step(global_iteration_step)
else:
global_iteration_step_in_2 = iterations * config["solver"][
"warmup_epochs"] + 1 - global_iteration_step
scheduler2.step(int(global_iteration_step_in_2))
global_iteration_step = global_iteration_step + 1
torch.cuda.empty_cache()
checkpoint_manager.step()
train_acc_1, train_acc_3 = cal_acc(
train_answers, train_preds)
print(
"trainacc@1={:.2%} & trainacc@3={:.2%} "
.format(train_acc_1, train_acc_3))
summary_writer.add_scalars(
'train/acc', {
'acc@1': train_acc_1,
'acc@3': train_acc_3
}, epoch)
if args.validate:
model.eval()
answers = [] # [batch_answers,...]
preds = [] # [batch_preds,...]
print(f"\nValidation after epoch {epoch}:")
for i, batch in enumerate(tqdm(val_dataloader)):
with torch.no_grad():
fact_batch_graph = model(batch)
batch_loss = cal_batch_loss(fact_batch_graph,
batch,
device,
neg_weight=0.1,
pos_weight=0.9)
summary_writer.add_scalar('test/loss', batch_loss, epoch)
fact_graphs = dgl.unbatch(fact_batch_graph)
for i, fact_graph in enumerate(fact_graphs):
pred = fact_graph.ndata['h'].squeeze() # (num_nodes,1)
preds.append(pred) # [(num_nodes,)]
answers.append(batch['facts_answer_id_list'][i])
acc_1, acc_3 = cal_acc(answers, preds)
print("acc@1={:.2%} & acc@3={:.2%} ".
format(acc_1, acc_3))
summary_writer.add_scalars('test/acc', {
'acc@1': acc_1,
'acc@3': acc_3
}, epoch)
model.train()
torch.cuda.empty_cache()
print('Train finished !!!')
summary_writer.close()
def train(config_path):
logger.info('------------MODEL TRAIN--------------')
logger.info('loading config file...')
global_config = read_config(config_path)
# set random seed
seed = global_config['global']['random_seed']
torch.manual_seed(seed)
#set default gpu
os.environ["CUDA_VISIBLE_DEVICES"] = str(global_config['train']["gpu_id"])
enable_cuda = global_config['train']['enable_cuda']
device = torch.device("cuda" if enable_cuda else "cpu")
if torch.cuda.is_available() and not enable_cuda:
logger.warning("CUDA is avaliable, you can enable CUDA in config file")
elif not torch.cuda.is_available() and enable_cuda:
raise ValueError(
"CUDA is not abaliable, please unable CUDA in config file")
logger.info('reading squad dataset...')
dataset = SquadDataset(global_config)
logger.info('constructing model...')
model_choose = global_config['global']['model']
dataset_h5_path = global_config['data']['dataset_h5']
if model_choose == 'base':
model_config = read_config('config/base_model.yaml')
model = BaseModel(dataset_h5_path, model_config)
elif model_choose == 'match-lstm':
model = MatchLSTM(dataset_h5_path)
elif model_choose == 'match-lstm+':
model = MatchLSTMPlus(dataset_h5_path,
global_config['preprocess']['use_domain_tag'])
elif model_choose == 'r-net':
model = RNet(dataset_h5_path)
elif model_choose == 'm-reader':
model = MReader(dataset_h5_path)
else:
raise ValueError('model "%s" in config file not recoginized' %
model_choose)
model = model.to(device)
criterion = MyNLLLoss()
# optimizer
optimizer_choose = global_config['train']['optimizer']
optimizer_lr = global_config['train']['learning_rate']
optimizer_param = filter(lambda p: p.requires_grad, model.parameters())
if optimizer_choose == 'adamax':
optimizer = optim.Adamax(optimizer_param)
elif optimizer_choose == 'adadelta':
optimizer = optim.Adadelta(optimizer_param)
elif optimizer_choose == 'adam':
optimizer = optim.Adam(optimizer_param)
elif optimizer_choose == 'sgd':
optimizer = optim.SGD(optimizer_param, lr=optimizer_lr)
else:
raise ValueError('optimizer "%s" in config file not recoginized' %
optimizer_choose)
# check if exist model weight
weight_path = global_config['data']['model_path']
if os.path.exists(weight_path):
logger.info('loading existing weight...')
weight = torch.load(weight_path,
map_location=lambda storage, loc: storage)
if enable_cuda:
weight = torch.load(
weight_path, map_location=lambda storage, loc: storage.cuda())
# weight = pop_dict_keys(weight, ['pointer', 'init_ptr_hidden']) # partial initial weight
model.load_state_dict(weight, strict=False)
# training arguments
logger.info('start training...')
train_batch_size = global_config['train']['batch_size']
valid_batch_size = global_config['train']['valid_batch_size']
num_workers = global_config['global']['num_data_workers']
batch_train_data = dataset.get_dataloader_train(train_batch_size,
num_workers)
batch_dev_data = dataset.get_dataloader_dev(valid_batch_size, num_workers)
clip_grad_max = global_config['train']['clip_grad_norm']
best_avg = 0.
# every epoch
for epoch in range(global_config['train']['epoch']):
# train
model.train() # set training = True, make sure right dropout
sum_loss = train_on_model(model=model,
criterion=criterion,
optimizer=optimizer,
batch_data=batch_train_data,
epoch=epoch,
clip_grad_max=clip_grad_max,
device=device)
logger.info('epoch=%d, sum_loss=%.5f' % (epoch, sum_loss))
# evaluate
with torch.no_grad():
model.eval() # let training = False, make sure right dropout
valid_score_em, valid_score_f1, valid_loss = eval_on_model(
model=model,
criterion=criterion,
batch_data=batch_dev_data,
epoch=epoch,
device=device)
valid_avg = (valid_score_em + valid_score_f1) / 2
logger.info(
"epoch=%d, ave_score_em=%.2f, ave_score_f1=%.2f, sum_loss=%.5f" %
(epoch, valid_score_em, valid_score_f1, valid_loss))
# save model when best avg score
if valid_avg > best_avg:
save_model(
model,
epoch=epoch,
model_weight_path=global_config['data']['model_path'],
checkpoint_path=global_config['data']['checkpoint_path'])
logger.info("saving model weight on epoch=%d" % epoch)
best_avg = valid_avg
logger.info('pretraining finished.')
if global_config['global']['finetune']:
batch_train_data = dataset.get_dataloader_train2(
train_batch_size, num_workers)
batch_dev_data = dataset.get_dataloader_dev2(valid_batch_size,
num_workers)
for epoch in range(global_config['train']['finetune_epoch']):
# train
model.train() # set training = True, make sure right dropout
sum_loss = train_on_model(model=model,
criterion=criterion,
optimizer=optimizer,
batch_data=batch_train_data,
epoch=epoch,
clip_grad_max=clip_grad_max,
device=device)
logger.info('finetune epoch=%d, sum_loss=%.5f' % (epoch, sum_loss))
# evaluate
with torch.no_grad():
model.eval() # let training = False, make sure right dropout
valid_score_em, valid_score_f1, valid_loss = eval_on_model(
model=model,
criterion=criterion,
batch_data=batch_dev_data,
epoch=epoch,
device=device)
valid_avg = (valid_score_em + valid_score_f1) / 2
logger.info(
"finetune epoch=%d, ave_score_em=%.2f, ave_score_f1=%.2f, sum_loss=%.5f"
% (epoch, valid_score_em, valid_score_f1, valid_loss))
# save model when best avg score
if valid_avg > best_avg:
save_model(
model,
epoch=epoch,
model_weight_path=global_config['data']['model_path'],
checkpoint_path=global_config['data']['checkpoint_path'])
logger.info("saving model weight on epoch=%d" % epoch)
best_avg = valid_avg
if global_config['global']['finetune2']:
batch_train_data = dataset.get_dataloader_train3(
train_batch_size, num_workers)
batch_dev_data = dataset.get_dataloader_dev3(valid_batch_size,
num_workers)
for epoch in range(global_config['train']['finetune_epoch2']):
# train
model.train() # set training = True, make sure right dropout
sum_loss = train_on_model(model=model,
criterion=criterion,
optimizer=optimizer,
batch_data=batch_train_data,
epoch=epoch,
clip_grad_max=clip_grad_max,
device=device)
logger.info('finetune2 epoch=%d, sum_loss=%.5f' %
(epoch, sum_loss))
# evaluate
with torch.no_grad():
model.eval() # let training = False, make sure right dropout
valid_score_em, valid_score_f1, valid_loss = eval_on_model(
model=model,
criterion=criterion,
batch_data=batch_dev_data,
epoch=epoch,
device=device)
valid_avg = (valid_score_em + valid_score_f1) / 2
logger.info(
"finetune2 epoch=%d, ave_score_em=%.2f, ave_score_f1=%.2f, sum_loss=%.5f"
% (epoch, valid_score_em, valid_score_f1, valid_loss))
# save model when best avg score
if valid_avg > best_avg:
save_model(
model,
epoch=epoch,
model_weight_path=global_config['data']['model_path'],
checkpoint_path=global_config['data']['checkpoint_path'])
logger.info("saving model weight on epoch=%d" % epoch)
best_avg = valid_avg
logger.info('finished.')
{"params": model.parameters()},
optim.Adagrad(lr=0.1, params=model.parameters()),
id="AdagradConf",
),
pytest.param(
"Adam",
{"lr": 0.1},
{"params": model.parameters()},
optim.Adam(lr=0.1, params=model.parameters()),
id="AdamConf",
),
pytest.param(
"Adamax",
{"lr": 0.1},
{"params": model.parameters()},
optim.Adamax(lr=0.1, params=model.parameters()),
id="AdamaxConf",
),
pytest.param(
"AdamW",
{"lr": 0.1},
{"params": model.parameters()},
optim.AdamW(lr=0.1, params=model.parameters()),
id="AdamWConf",
),
pytest.param(
"ASGD",
{"lr": 0.1},
{"params": model.parameters()},
optim.ASGD(lr=0.1, params=model.parameters()),
id="ASGDConf",
def optimization_algorithms(SCI_optimizer, cnn, LR, SCI_SGD_MOMENTUM,
REGULARIZATION):
if type(SCI_optimizer) is str:
if (SCI_optimizer == 'Adam'):
optimizer = optim.Adam(cnn.parameters(),
lr=LR,
betas=(0.01, 0.999),
weight_decay=REGULARIZATION)
if (SCI_optimizer == 'AMSGrad'):
optimizer = optim.Adam(cnn.parameters(),
lr=LR,
betas=(0.01, 0.999),
weight_decay=REGULARIZATION,
amsgrad=True)
if (SCI_optimizer == 'AdamW'):
optimizer = AdamW(cnn.parameters(),
lr=LR,
betas=(0.01, 0.999),
weight_decay=REGULARIZATION)
if (SCI_optimizer == 'RMSprop'):
optimizer = optim.RMSprop(cnn.parameters(), lr=LR)
#if (SCI_optimizer == 'SparseAdam') or (int(SCI_optimizer) == 4) :
#optimizer = optim.SparseAdam(cnn.parameters(), lr=LR)
if (SCI_optimizer == 'SGD'):
optimizer = optim.SGD(cnn.parameters(),
lr=LR,
momentum=SCI_SGD_MOMENTUM,
weight_decay=REGULARIZATION)
if (SCI_optimizer == 'Adadelta'):
optimizer = optim.Adadelta(cnn.parameters(),
lr=LR,
weight_decay=REGULARIZATION)
if (SCI_optimizer == 'Rprop'):
optimizer = optim.Rprop(cnn.parameters(), lr=LR)
#if (SCI_optimizer == 'Adagrad') or (int(SCI_optimizer) == 7) :
# optimizer = optim.Adagrad(cnn.parameters(), lr=LR, weight_decay=REGULARIZATION)
if (SCI_optimizer == 'Adamax'):
optimizer = optim.Adamax(cnn.parameters(),
lr=LR,
weight_decay=REGULARIZATION)
if (SCI_optimizer == 'ASGD'):
optimizer = optim.ASGD(cnn.parameters(),
lr=LR,
weight_decay=REGULARIZATION)
#if (SCI_optimizer == 'LBFGS') or (int(SCI_optimizer) == 10) :
#optimizer = optim.LBFGS(cnn.parameters(), lr=LR)
else:
if (int(SCI_optimizer) == 1):
optimizer = optim.Adam(cnn.parameters(),
lr=LR,
betas=(0.01, 0.999),
weight_decay=REGULARIZATION)
if (int(SCI_optimizer) == 2):
optimizer = optim.Adam(cnn.parameters(),
lr=LR,
betas=(0.01, 0.999),
weight_decay=REGULARIZATION,
amsgrad=True)
if (int(SCI_optimizer) == 3):
optimizer = AdamW(cnn.parameters(),
lr=LR,
betas=(0.01, 0.999),
weight_decay=REGULARIZATION)
if (int(SCI_optimizer) == 4):
optimizer = optim.RMSprop(cnn.parameters(), lr=LR)
#if (SCI_optimizer == 'SparseAdam') or (int(SCI_optimizer) == 4) :
#optimizer = optim.SparseAdam(cnn.parameters(), lr=LR)
if (int(SCI_optimizer) == 5):
optimizer = optim.SGD(cnn.parameters(),
lr=LR,
momentum=SCI_SGD_MOMENTUM,
weight_decay=REGULARIZATION)
if (int(SCI_optimizer) == 6):
optimizer = optim.Adadelta(cnn.parameters(),
lr=LR,
weight_decay=REGULARIZATION)
if (int(SCI_optimizer) == 7):
optimizer = optim.Rprop(cnn.parameters(), lr=LR)
#if (SCI_optimizer == 'Adagrad') or (int(SCI_optimizer) == 7) :
# optimizer = optim.Adagrad(cnn.parameters(), lr=LR, weight_decay=REGULARIZATION)
if (int(SCI_optimizer) == 8):
optimizer = optim.Adamax(cnn.parameters(),
lr=LR,
weight_decay=REGULARIZATION)
if (int(SCI_optimizer) == 9):
optimizer = optim.ASGD(cnn.parameters(),
lr=LR,
weight_decay=REGULARIZATION)
#if (SCI_optimizer == 'LBFGS') or (int(SCI_optimizer) == 10) :
#optimizer = optim.LBFGS(cnn.parameters(), lr=LR)
return optimizer
def main(lr, batch_size, epoch, gpu, train_set, valid_set):
# ------------- Part for tensorboard --------------
writer = SummaryWriter(comment="_naive_DENET")
# ------------- Part for tensorboard --------------
# -------------- Some prepare ---------------------
torch.backends.cudnn.enabled = True
torch.cuda.set_device(gpu)
# torch.set_default_tensor_type('torch.cuda.FloatTensor')
# -------------- Some prepare ---------------------
BATCH_SIZE = batch_size
EPOCH = epoch
LEARNING_RATE = lr
belta1 = 0.9
belta2 = 0.999
trainset = mydataset(train_set, transform_train)
valset = mydataset(valid_set)
trainLoader = torch.utils.data.DataLoader(trainset,
batch_size=BATCH_SIZE,
shuffle=True)
valLoader = torch.utils.data.DataLoader(valset,
batch_size=1,
shuffle=False)
opter = Opter(128, 128, batch_size)
SepConvNet = Network(opter).cuda()
SepConvNet.apply(weights_init)
# SepConvNet.load_state_dict(torch.load('/mnt/hdd/xiasifeng/sepconv/sepconv_mutiscale_LD/SepConv_iter33-ltype_fSATD_fs-lr_0.001-trainloss_0.1497-evalloss_0.1357-evalpsnr_29.6497.pkl'))
# SepConvNet_cost = nn.MSELoss().cuda()
# SepConvNet_cost = nn.L1Loss().cuda()
SepConvNet_cost = sepconv.SATDLoss().cuda()
SepConvNet_optimizer = optim.Adamax(SepConvNet.parameters(),
lr=LEARNING_RATE,
betas=(belta1, belta2))
SepConvNet_schedule = optim.lr_scheduler.ReduceLROnPlateau(
SepConvNet_optimizer,
factor=0.1,
patience=3,
verbose=True,
min_lr=1e-5)
# ---------------- Time part -------------------
start_time = time.time()
global_step = 0
# ---------------- Time part -------------------
for epoch in range(0, EPOCH):
SepConvNet.train().cuda()
cnt = 0
sumloss = 0.0 # The sumloss is for the whole training_set
tsumloss = 0.0 # The tsumloss is for the printinterval
printinterval = 300
print("---------------[Epoch%3d]---------------" % (epoch + 1))
for imgL, imgR, label in trainLoader:
global_step = global_step + 1
cnt = cnt + 1
SepConvNet_optimizer.zero_grad()
imgL = var(imgL).cuda()
imgR = var(imgR).cuda()
label = var(label).cuda()
output = SepConvNet(imgL, imgR)
loss = SepConvNet_cost(output, label)
loss.backward()
SepConvNet_optimizer.step()
sumloss = sumloss + loss.data.item()
tsumloss = tsumloss + loss.data.item()
if cnt % printinterval == 0:
writer.add_image("Prev image", imgR[0], cnt)
writer.add_image("Pred image", output[0], cnt)
writer.add_scalar('Train Batch SATD loss', loss.data.item(),
int(global_step / printinterval))
writer.add_scalar('Train Interval SATD loss',
tsumloss / printinterval,
int(global_step / printinterval))
print(
'Epoch [%d/%d], Iter [%d/%d], Time [%4.4f], Batch loss [%.6f], Interval loss [%.6f]'
% (epoch + 1, EPOCH, cnt, len(trainset) // BATCH_SIZE,
time.time() - start_time, loss.data.item(),
tsumloss / printinterval))
tsumloss = 0.0
print('Epoch [%d/%d], iter: %d, Time [%4.4f], Avg Loss [%.6f]' %
(epoch + 1, EPOCH, cnt, time.time() - start_time, sumloss / cnt))
# ---------------- Part for validation ----------------
trainloss = sumloss / cnt
SepConvNet.eval().cuda()
evalcnt = 0
pos = 0.0
sumloss = 0.0
psnr = 0.0
for imgL, imgR, label in valLoader:
imgL = var(imgL).cuda()
imgR = var(imgR).cuda()
label = var(label).cuda()
with torch.no_grad():
output = SepConvNet(imgL, imgR)
loss = SepConvNet_cost(output, label)
sumloss = sumloss + loss.data.item()
psnr = psnr + calcPSNR.calcPSNR(output.cpu().data.numpy(),
label.cpu().data.numpy())
evalcnt = evalcnt + 1
# ------------- Tensorboard part -------------
writer.add_scalar("Valid SATD loss", sumloss / evalcnt, epoch)
writer.add_scalar("Valid PSNR", psnr / valset.__len__(), epoch)
# ------------- Tensorboard part -------------
print('Validation loss [%.6f], Average PSNR [%.4f]' %
(sumloss / evalcnt, psnr / valset.__len__()))
SepConvNet_schedule.step(psnr / valset.__len__())
torch.save(
SepConvNet.state_dict(),
os.path.join(
'.', 'naive_DENET_iter' + str(epoch + 1) + '-ltype_fSATD_fs' +
'-lr_' + str(LEARNING_RATE) + '-trainloss_' +
str(round(trainloss, 4)) + '-evalloss_' +
str(round(sumloss / evalcnt, 4)) + '-evalpsnr_' +
str(round(psnr / valset.__len__(), 4)) + '.pkl'))
writer.close()
for key, value in dict(model.named_parameters()).items():
if value.requires_grad:
if 'cnn' in key:
params += [{'params':[value], 'lr':opt.cnn_learning_rate,
'weight_decay':opt.cnn_weight_decay, 'betas':(opt.cnn_optim_alpha, opt.cnn_optim_beta)}]
else:
params += [{'params':[value], 'lr':opt.learning_rate,
'weight_decay':opt.weight_decay, 'betas':(opt.optim_alpha, opt.optim_beta)}]
print("Use %s as optmization method" %(opt.optim))
if opt.optim == 'sgd':
optimizer = optim.SGD(params, momentum=0.9)
elif opt.optim == 'adam':
optimizer = optim.Adam(params)
elif opt.optim == 'adamax':
optimizer = optim.Adamax(params)
# if opt.cnn_optim == 'sgd':
# cnn_optimizer = optim.SGD(cnn_params, momentum=0.9)
# else:
# cnn_optimizer = optim.Adam(cnn_params)
# load optimizer
# learning_rate_list = np.linspace(opt.learning_rate, 0.0005, opt.max_epochs)
for epoch in range(start_epoch, opt.max_epochs):
if epoch > opt.learning_rate_decay_start and opt.learning_rate_decay_start >= 0:
if (epoch - opt.learning_rate_decay_start) % opt.learning_rate_decay_every == 0:
# decay the learning rate.
utils.set_lr(optimizer, opt.learning_rate_decay_rate)
opt.learning_rate = opt.learning_rate * opt.learning_rate_decay_rate
def main(lr, batch_size, epoch, gpu, train_set, valid_set):
# ------------- Part for tensorboard --------------
# writer = SummaryWriter(log_dir='tb/LSTM_ft1')
# ------------- Part for tensorboard --------------
torch.backends.cudnn.enabled = True
torch.cuda.set_device(gpu)
BATCH_SIZE = batch_size
EPOCH = epoch
LEARNING_RATE = lr
belta1 = 0.9
belta2 = 0.999
trainset = vimeodataset(train_set, 'filelist.txt', transform_train)
valset = vimeodataset(valid_set, 'test.txt')
trainLoader = torch.utils.data.DataLoader(trainset,
batch_size=BATCH_SIZE,
shuffle=True)
valLoader = torch.utils.data.DataLoader(valset,
batch_size=BATCH_SIZE,
shuffle=False)
assert (len(valset) % BATCH_SIZE == 0)
SepConvNet = Network().cuda()
# SepConvNet.apply(weights_init)
SepConvNet.load_my_state_dict(
torch.load(
'ft2_baseline_iter86-ltype_fSATD_fs-lr_0.001-trainloss_0.1249-evalloss_0.1155-evalpsnr_29.9327.pkl',
map_location='cuda:%d' % (gpu)))
# SepConvNet.load_state_dict(torch.load('beta_LSTM_iter8-ltype_fSATD_fs-lr_0.001-trainloss_0.557-evalloss_0.1165-evalpsnr_29.8361.pkl'))
# MSE_cost = nn.MSELoss().cuda()
# SepConvNet_cost = nn.L1Loss().cuda()
SepConvNet_cost = sepconv.SATDLoss().cuda()
SepConvNet_optimizer = optim.Adamax(SepConvNet.parameters(),
lr=LEARNING_RATE,
betas=(belta1, belta2))
SepConvNet_schedule = optim.lr_scheduler.ReduceLROnPlateau(
SepConvNet_optimizer, factor=0.1, patience=3, verbose=True)
# ---------------- Time part -------------------
start_time = time.time()
global_step = 0
# ---------------- Time part -------------------
# ---------------- Opt part -----------------------
# opter = Opter(gpu)
# -------------------------------------------------
for epoch in range(0, EPOCH):
SepConvNet.train().cuda()
cnt = 0
sumloss = 0.0 # The sumloss is for the whole training_set
tsumloss = 0.0 # The tsumloss is for the printinterval
printinterval = 100
print("---------------[Epoch%3d]---------------" % (epoch + 1))
for label_list in trainLoader:
bad_list = label_list[7:]
label_list = label_list[:7]
# IPython.embed()
# exit()
global_step = global_step + 1
cnt = cnt + 1
for i in range(5):
imgL = var(bad_list[i]).cuda()
imgR = var(bad_list[i + 1]).cuda()
label = var(label_list[i + 2]).cuda()
poor_label = var(bad_list[i + 2]).cuda()
label_L = var(label_list[i]).cuda()
SepConvNet_optimizer.zero_grad()
if i == 0:
output_f, output_b, stat = SepConvNet(imgL, imgR)
else:
output_f, output_b, stat = SepConvNet(
imgL, imgR, res_f, res_b, stat)
res_f = poor_label - output_f
res_b = imgL - output_b
loss = 0.5 * SepConvNet_cost(output_f,
label) + 0.5 * SepConvNet_cost(
output_b, label_L)
if i < 4:
loss.backward(retain_graph=True)
else:
loss.backward()
SepConvNet_optimizer.step()
sumloss = sumloss + loss.data.item()
tsumloss = tsumloss + loss.data.item()
if cnt % printinterval == 0:
print(
'Epoch [%d/%d], Iter [%d/%d], Time [%4.4f], Batch loss [%.6f], Interval loss [%.6f]'
% (epoch + 1, EPOCH, cnt, len(trainset) // BATCH_SIZE,
time.time() - start_time, loss.data.item(),
tsumloss / printinterval / 5))
tsumloss = 0.0
print('Epoch [%d/%d], iter: %d, Time [%4.4f], Avg Loss [%.6f]' %
(epoch + 1, EPOCH, cnt, time.time() - start_time,
sumloss / cnt / 5))
# ---------------- Part for validation ----------------
trainloss = sumloss / cnt
SepConvNet.eval().cuda()
evalcnt = 0
pos = 0.0
sumloss = 0.0
psnr = 0.0
for label_list in valLoader:
bad_list = label_list[7:]
label_list = label_list[:7]
with torch.no_grad():
for i in range(5):
imgL = var(bad_list[i]).cuda()
imgR = var(bad_list[i + 1]).cuda()
label = var(label_list[i + 2]).cuda()
poor_label = var(bad_list[i + 2]).cuda()
label_L = var(label_list[i]).cuda()
if i == 0:
output_f, output_b, stat = SepConvNet(imgL, imgR)
else:
output_f, output_b, stat = SepConvNet(
imgL, imgR, res_f, res_b, stat)
psnr = psnr + calcPSNR.calcPSNR(
output_f.cpu().data.numpy(),
label.cpu().data.numpy())
res_f = poor_label - output_f
res_b = label_L - output_b
loss = SepConvNet_cost(output_f, label)
sumloss = sumloss + loss.data.item()
evalcnt = evalcnt + 5
# ------------- Tensorboard part -------------
# writer.add_scalar("Valid SATD loss", sumloss / evalcnt, epoch)
# writer.add_scalar("Valid PSNR", psnr / valset.__len__(), epoch)
# ------------- Tensorboard part -------------
print('Validation loss [%.6f], Average PSNR [%.4f]' %
(sumloss / evalcnt, psnr / evalcnt))
SepConvNet_schedule.step(psnr / evalcnt)
torch.save(
SepConvNet.state_dict(),
os.path.join(
'.', 'minidual_LSTM_iter' + str(epoch + 1) +
'-ltype_fSATD_fs' + '-lr_' + str(LEARNING_RATE) +
'-trainloss_' + str(round(trainloss, 4)) + '-evalloss_' +
str(round(sumloss / evalcnt, 4)) + '-evalpsnr_' +
str(round(psnr / evalcnt, 4)) + '.pkl'))
def lr_lambda_fun(current_iteration: int) -> float:
"""Returns a learning rate multiplier.
Till `warmup_epochs`, learning rate linearly increases to `initial_lr`,
and then gets multiplied by `lr_gamma` every time a milestone is crossed.
"""
current_epoch = float(current_iteration) / iterations
if current_epoch < config["solver"]["warmup_epochs"]:
alpha = current_epoch / float(config["solver"]["warmup_epochs"])
return config["solver"]["warmup_factor"] * (1.0 - alpha) + alpha
else:
idx = bisect(config["solver"]["lr_milestones"], current_epoch)
return pow(config["solver"]["lr_gamma"], idx)
optimizer = optim.Adamax(model.parameters(), lr=config["solver"]["initial_lr"])
scheduler = lr_scheduler.LambdaLR(optimizer, lr_lambda=lr_lambda_fun) # 可以在一个组里面调节lr参数
# =============================================================================
# SETUP BEFORE TRAINING LOOP
# =============================================================================
start_time = datetime.datetime.strftime(datetime.datetime.utcnow(), '%d-%b-%Y-%H:%M:%S')
checkpoint_dirpath = args.save_dirpath
if checkpoint_dirpath == 'checkpoints/':
checkpoint_dirpath += '%s+%s/%s' % (config["model"]["encoder"], config["model"]["decoder"], start_time)
if args.save_model:
summary_writer = SummaryWriter(log_dir=checkpoint_dirpath)
checkpoint_manager = CheckpointManager(model, optimizer, checkpoint_dirpath, config=config)
sparse_metrics = SparseGTMetrics()
ndcg = NDCG()
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--data', default='./data/')
parser.add_argument('--epoch', type=int, default=10000)
parser.add_argument('--batch_size', type=int, default=20)
parser.add_argument('--lr', type=float, default=0.001)
parser.add_argument('--embed_dim', type=int, default=300)
parser.add_argument('--hidden_dim', type=int, default=150)
parser.add_argument('--num_layers', type=int, default=1)
parser.add_argument('--bidirectional', default=True)
parser.add_argument('--glove', default='../Data/glove/glove.840B.300d.txt')
parser.add_argument('--cuda_set', default=True)
args = parser.parse_args()
###############################################################################
# Load data
###############################################################################
print("Load data file...")
train_data, dev_data = load_data(args.data)
print("Preparing batch loader...")
print("============= Train ===============")
train_loader = BatchLoader(train_data, 'train', args.cuda_set,
args.batch_size)
print("============= Valid ===============")
dev_loader = BatchLoader(dev_data, 'dev', args.cuda_set, args.batch_size)
# vocabulary set
vocab_size = len(dev_data['word2idx'])
print("============= Vocab Size ===============")
print(vocab_size)
print("")
idx2word = dev_data['idx2word']
###############################################################################
# Build the model
###############################################################################
cuda.set_device(0)
if args.cuda_set == True:
model = MatchNet(vocab_size, args.embed_dim, args.hidden_dim,
args.cuda_set, args.num_layers,
args.bidirectional).cuda()
criterion = nn.NLLLoss().cuda()
else:
model = MatchNet(vocab_size, args.embed_dim, args.hidden_dim,
args.cuda_set, args.num_layers, args.bidirectional)
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adamax(model.parameters(),
lr=args.lr,
betas=(0.9, 0.999))
print("#" * 15, "Model Info", "#" * 15)
print("Model: ", model)
print("Criterion: ", criterion)
print("Optimizer: ", optimizer)
print("")
###############################################################################
# Load the pretrained word embedding
###############################################################################
print("loading pretrinaed word embedding ...")
emb_file = os.path.join(args.data, 'glove_emb.pth')
if os.path.isfile(emb_file):
W_emb = torch.load(emb_file)
else:
W_emb, embed_dim = load_pretrained_embedding(dev_data['word2idx'],
args.glove)
W_emb = torch.from_numpy(W_emb).cuda()
torch.save(W_emb, emb_file)
if args.cuda_set:
W_emb = W_emb.cuda()
model.embed.embed.state_dict()['weight'].copy_(W_emb)
model.embed.embed.state_dict()['weight'].requires_grad = False
###############################################################################
# Training
###############################################################################
for epoch in range(args.epoch):
start_time = time.time()
train_loss = AverageMeter()
train_acc = AverageMeter()
message = "Epoch: %d training.." % (epoch)
print(message)
print("")
for i, data in enumerate(train_loader):
model.train()
doc = data[0]
qry = data[1]
anss = data[2]
sis = data[3]
eis = data[4]
ids = data[5]
dm = data[9]
qm = data[10]
output1, output2 = model(doc, qry, dm, qm)
_, pred1 = output1.data.cpu().topk(1)
_, pred2 = output2.data.cpu().topk(1)
loss1 = criterion(output1, sis)
loss2 = criterion(output2, eis)
loss = loss1 + loss2
train_loss.update(loss.data[0], doc.size(0))
acc_tmp = accuracy(pred1.numpy(), tensor2np(sis), pred2.numpy(),
tensor2np(eis), ids)
train_acc.update(acc_tmp, doc.size(0))
optimizer.zero_grad()
loss.backward()
optimizer.step()
print(
"=================== Train ======================")
print("doc_len: ", doc.size(1))
random_idx = randomChoice(doc.size(0))
show_question(random_idx, doc, qry, dm, qm, anss, idx2word)
show_answer(random_idx, doc, qry, pred1, pred2, sis, eis, idx2word)
print("")
message = "Train epoch: %d iter: %d train_loss: %1.3f train_acc: %1.3f elapsed: %1.3f " % (
epoch, i, train_loss.avg, train_acc.avg, time.time() - start_time)
print(message)
print("")
###############################################################################
# Validation
###############################################################################
print(
"==================== Evaluation ======================")
val_acc = AverageMeter()
start_time = time.time()
model.eval()
cor_cnt = 0
incor_cnt = 0
pad_cnt = 0
val_out = 0
val_in = 0
val_false = 0
for j, data in enumerate(dev_loader):
doc = data[0]
qry = data[1]
anss = data[2]
sis = data[3]
eis = data[4]
ids = data[5]
dm = data[9]
qm = data[10]
output1, output2 = model(doc, qry, dm, qm)
_, val_pred1 = output1.data.cpu().topk(1)
_, val_pred2 = output2.data.cpu().topk(1)
acc_tmp = accuracy_dev(val_pred1.numpy(), sis, val_pred2.numpy(),
eis, ids)
val_acc.update(acc_tmp, doc.size(0))
message = "Epoch: %d train_iter: %d iter: %d val_acc: %1.3f elapsed: %1.3f " % (
epoch, i, j, val_acc.avg, time.time() - start_time)
print(message)
print("")
###############################################################################
# Show the sample Q&A
###############################################################################
random_idx = randomChoice(doc.size(0))
show_question(random_idx, doc, qry, dm, qm, anss, idx2word)
show_answer_dev(random_idx, doc, qry, val_pred1, val_pred2, sis, eis,
idx2word)
train_loss = AverageMeter()
train_acc = AverageMeter()
start_time = time.time()
def train_model(opt_):
env = TrainEnvironment(opt_)
dictionary = env.dict
if opt_.load_checkpoint:
net, dictionary = load_model(opt_.load_checkpoint, opt_)
env = TrainEnvironment(opt_, dictionary)
env.dict = dictionary
else:
net = create_model(opt_, dictionary["words"])
if opt_.embeddings and opt_.embeddings != "None":
load_embeddings(opt_, dictionary["words"], net)
paramnum = 0
trainable = 0
for name, parameter in net.named_parameters():
if parameter.requires_grad:
trainable += parameter.numel()
paramnum += parameter.numel()
print("TRAINABLE", paramnum, trainable)
if opt_.cuda:
net = torch.nn.DataParallel(net)
net = net.cuda()
if opt_.optimizer == "adamax":
lr = opt_.learning_rate or 0.002
named_params_to_optimize = filter(lambda p: p[1].requires_grad,
net.named_parameters())
params_to_optimize = (p[1] for p in named_params_to_optimize)
optimizer = optim.Adamax(params_to_optimize, lr=lr)
if opt_.epoch_start != 0:
saved_params = torch.load(
opt_.load_checkpoint,
map_location=lambda storage, loc: storage)
optimizer.load_state_dict(saved_params["optim_dict"])
else:
lr = opt_.learning_rate or 0.01
optimizer = optim.SGD(filter(lambda p: p.requires_grad,
net.parameters()),
lr=lr)
start_time = time.time()
# best_loss = float("+inf")
best_loss = 0
test_data_shuffled = env.build_valid_dataloader(True)
test_data_not_shuffled = env.build_valid_dataloader(False)
with torch.no_grad():
validate(
0,
net,
test_data_shuffled,
nb_candidates=opt_.hits_at_nb_cands,
shuffled_str="shuffled",
)
train_data = None
for epoch in range(opt_.epoch_start, opt_.num_epochs):
if train_data is None or opt_.dataset_name == "reddit":
train_data = env.build_train_dataloader(epoch)
train(epoch, start_time, net, optimizer, opt_, train_data)
with torch.no_grad():
# We compute the loss both for shuffled and not shuffled case.
# however, the loss that determines if the model is better is the
# same as the one used for training.
loss_shuffled = validate(
epoch,
net,
test_data_shuffled,
nb_candidates=opt_.hits_at_nb_cands,
shuffled_str="shuffled",
)
loss_not_shuffled = validate(
epoch,
net,
test_data_not_shuffled,
nb_candidates=opt_.hits_at_nb_cands,
shuffled_str="not-shuffled",
)
if opt_.no_shuffle:
loss = loss_not_shuffled
else:
loss = loss_shuffled
# if loss < best_loss:
# ========= change loss to P@1,100 =======
if loss > best_loss:
best_loss = loss
best_loss_epoch = epoch
# logging.info(f"New best loss, saving model to {opt_.model_file}")
logging.info(
f"New best P@1,100, saving model to {opt_.model_file}")
save_model(opt_.model_file, net, dictionary, optimizer)
# Stop if it's been too many epochs since the loss has decreased
if opt_.stop_crit_num_epochs != -1:
if epoch - best_loss_epoch >= opt_.stop_crit_num_epochs:
break
return net, dictionary
def train(model, train_loader, test_loader, gen_loader, configs):
model.train()
# optimizer, it's better to set up lr for some modules separately so that the whole training become more stable
params = [{
'params': model.reader.parameters(),
'lr': 0.2 * configs.lr
}, {
'params': model.h_mean.parameters(),
'lr': 0.1 * configs.lr
}, {
'params': model.h_var.parameters(),
'lr': 0.1 * configs.lr
}, {
'params': model.writer.parameters()
}, {
'params': model.pos_dist.parameters()
}, {
'params': model.combine.parameters()
}, {
'params': model.describe.parameters()
}, {
'params': model.box_vae.parameters(),
'lr': 10 * configs.lr
}, {
'params': model.offset_vae.parameters(),
'lr': 10 * configs.lr
}, {
'params': model.renderer.parameters()
}, {
'params': model.bias_mean.parameters()
}, {
'params': model.bias_var.parameters()
}]
if configs.net == 'PNP':
params.append({'params': model.vis_dist.parameters()})
elif configs.net == 'SIMPLE':
pass
else:
raise ValueError('configs.net ?= ', configs.net, 'not a valid value')
optimizer = optim.Adamax(params, lr=configs.lr)
model.cuda()
trainer = PNPNetTrainer(model=model,
train_loader=train_loader,
val_loader=test_loader,
gen_loader=gen_loader,
optimizer=optimizer,
configs=configs)
minloss = 1000
for epoch_num in range(0, configs.epochs + 1):
timestamp_start = datetime.datetime.now(
pytz.timezone('America/New_York'))
trainer.train_epoch(epoch_num, timestamp_start)
if epoch_num % configs.validate_interval == 0 and epoch_num > 0:
minloss = trainer.validate(epoch_num, timestamp_start, minloss)
if epoch_num % configs.sample_interval == 0 and epoch_num > 0:
trainer.sample(epoch_num,
sample_num=8,
timestamp_start=timestamp_start)
if epoch_num % configs.save_interval == 0 and epoch_num > 0:
torch.save(
model.state_dict(),
osp.join(configs.exp_dir, 'checkpoints',
'model_epoch_{0}.pth'.format(epoch_num)))
if args.cuda != -1:
rnn = rnn.cuda(args.cuda)
print(rnn)
last_save_losses = []
if args.optim == 'adam':
optimizer = optim.Adam(rnn.parameters(),
lr=args.lr,
eps=1e-9,
betas=[0.9, 0.98]) # 0.0001
elif args.optim == 'adamax':
optimizer = optim.Adamax(rnn.parameters(),
lr=args.lr,
eps=1e-9,
betas=[0.9, 0.98]) # 0.0001
elif args.optim == 'rmsprop':
optimizer = optim.RMSprop(rnn.parameters(),
lr=args.lr,
momentum=0.9,
eps=1e-10) # 0.0001
elif args.optim == 'sgd':
optimizer = optim.SGD(rnn.parameters(), lr=args.lr) # 0.01
elif args.optim == 'adagrad':
optimizer = optim.Adagrad(rnn.parameters(), lr=args.lr)
elif args.optim == 'adadelta':
optimizer = optim.Adadelta(rnn.parameters(), lr=args.lr)
last_100_losses = []
def main(
dataset,
dataroot,
download,
augment,
batch_size,
eval_batch_size,
epochs,
saved_model,
seed,
hidden_channels,
K,
L,
actnorm_scale,
flow_permutation,
flow_coupling,
LU_decomposed,
learn_top,
y_condition,
y_weight,
max_grad_clip,
max_grad_norm,
lr,
n_workers,
cuda,
n_init_batches,
output_dir,
saved_optimizer,
warmup,
):
device = "cpu" if (not torch.cuda.is_available() or not cuda) else "cuda:0"
check_manual_seed(seed)
ds = check_dataset(dataset, dataroot, augment, download)
image_shape, num_classes, train_dataset, test_dataset = ds
# Note: unsupported for now
multi_class = False
train_loader = data.DataLoader(
train_dataset,
batch_size=batch_size,
shuffle=True,
num_workers=n_workers,
drop_last=True,
)
test_loader = data.DataLoader(
test_dataset,
batch_size=eval_batch_size,
shuffle=False,
num_workers=n_workers,
drop_last=False,
)
model = Glow(
image_shape,
hidden_channels,
K,
L,
actnorm_scale,
flow_permutation,
flow_coupling,
LU_decomposed,
num_classes,
learn_top,
y_condition,
)
model = model.to(device)
optimizer = optim.Adamax(model.parameters(), lr=lr, weight_decay=5e-5)
lr_lambda = lambda epoch: min(1.0, (epoch + 1) / warmup) # noqa
scheduler = torch.optim.lr_scheduler.LambdaLR(optimizer,
lr_lambda=lr_lambda)
def step(engine, batch):
model.train()
optimizer.zero_grad()
x, y = batch
x = x.to(device)
if y_condition:
y = y.to(device)
z, nll, y_logits = model(x, y)
losses = compute_loss_y(nll, y_logits, y_weight, y, multi_class)
else:
z, nll, y_logits = model(x, None)
losses = compute_loss(nll)
losses["total_loss"].backward()
if max_grad_clip > 0:
torch.nn.utils.clip_grad_value_(model.parameters(), max_grad_clip)
if max_grad_norm > 0:
torch.nn.utils.clip_grad_norm_(model.parameters(), max_grad_norm)
optimizer.step()
return losses
def eval_step(engine, batch):
model.eval()
x, y = batch
x = x.to(device)
with torch.no_grad():
if y_condition:
y = y.to(device)
z, nll, y_logits = model(x, y)
losses = compute_loss_y(nll,
y_logits,
y_weight,
y,
multi_class,
reduction="none")
else:
z, nll, y_logits = model(x, None)
losses = compute_loss(nll, reduction="none")
return losses
trainer = Engine(step)
checkpoint_handler = ModelCheckpoint(output_dir,
"glow",
n_saved=2,
require_empty=False)
trainer.add_event_handler(
Events.EPOCH_COMPLETED,
checkpoint_handler,
{
"model": model,
"optimizer": optimizer
},
)
monitoring_metrics = ["total_loss"]
RunningAverage(output_transform=lambda x: x["total_loss"]).attach(
trainer, "total_loss")
evaluator = Engine(eval_step)
# Note: replace by https://github.com/pytorch/ignite/pull/524 when released
Loss(
lambda x, y: torch.mean(x),
output_transform=lambda x: (
x["total_loss"],
torch.empty(x["total_loss"].shape[0]),
),
).attach(evaluator, "total_loss")
if y_condition:
monitoring_metrics.extend(["nll"])
RunningAverage(output_transform=lambda x: x["nll"]).attach(
trainer, "nll")
# Note: replace by https://github.com/pytorch/ignite/pull/524 when released
Loss(
lambda x, y: torch.mean(x),
output_transform=lambda x:
(x["nll"], torch.empty(x["nll"].shape[0])),
).attach(evaluator, "nll")
pbar = ProgressBar()
pbar.attach(trainer, metric_names=monitoring_metrics)
# load pre-trained model if given
if saved_model:
model.load_state_dict(torch.load(saved_model))
model.set_actnorm_init()
if saved_optimizer:
optimizer.load_state_dict(torch.load(saved_optimizer))
file_name, ext = os.path.splitext(saved_model)
resume_epoch = int(file_name.split("_")[-1])
@trainer.on(Events.STARTED)
def resume_training(engine):
engine.state.epoch = resume_epoch
engine.state.iteration = resume_epoch * len(
engine.state.dataloader)
@trainer.on(Events.STARTED)
def init(engine):
model.train()
init_batches = []
init_targets = []
with torch.no_grad():
for batch, target in islice(train_loader, None, n_init_batches):
init_batches.append(batch)
init_targets.append(target)
init_batches = torch.cat(init_batches).to(device)
assert init_batches.shape[0] == n_init_batches * batch_size
if y_condition:
init_targets = torch.cat(init_targets).to(device)
else:
init_targets = None
model(init_batches, init_targets)
@trainer.on(Events.EPOCH_COMPLETED)
def evaluate(engine):
evaluator.run(test_loader)
scheduler.step()
metrics = evaluator.state.metrics
losses = ", ".join(
[f"{key}: {value:.2f}" for key, value in metrics.items()])
print(f"Validation Results - Epoch: {engine.state.epoch} {losses}")
timer = Timer(average=True)
timer.attach(
trainer,
start=Events.EPOCH_STARTED,
resume=Events.ITERATION_STARTED,
pause=Events.ITERATION_COMPLETED,
step=Events.ITERATION_COMPLETED,
)
@trainer.on(Events.EPOCH_COMPLETED)
def print_times(engine):
pbar.log_message(
f"Epoch {engine.state.epoch} done. Time per batch: {timer.value():.3f}[s]"
)
timer.reset()
trainer.run(train_loader, epochs)
weight_decay=5e-4) # optimizrにMomentumSGDを指定
optKind2 = 'MomentumSGD'
network3 = CNNNet().to(device) # networkにさっき定義したnetworkを代入
optimizer3 = optim.Adam(network3.parameters(), lr=0.001) # optimizrにAdamを指定
optKind3 = 'Adam'
network4 = CNNNet().to(device) # networkにさっき定義したnetworkを代入
optimizer4 = optim.RMSprop(network4.parameters(), lr=0.0005,
eps=1e-06) # optimizrにRMSpropを指定
optKind4 = 'RMSprop'
network5 = CNNNet().to(device) # networkにさっき定義したnetworkを代入
optimizer5 = optim.Adamax(network5.parameters(),
lr=0.002,
betas=(0.9, 0.999),
eps=1e-08,
weight_decay=0) # optimizrにAdamaxを指定
optKind5 = 'Adamax'
times, test_acc_list, train_loss_lists, train_acc_lists, val_loss_lists, val_acc_lists = [], [], [], [], [], []
Optims = [optKind1, optKind2, optKind3, optKind4, optKind5]
Optimizers = [optimizer1, optimizer2, optimizer3, optimizer4, optimizer5]
networks = [network1, network2, network3, network4, network5]
for i in range(len(Optims)):
network = networks[i]
optKind = Optims[i]
optimizer = Optimizers[i]
def train(data, save_model_dir, seg=True):
print("Training with {} model.".format(data.model_type))
#data.show_data_summary()
model = SeqModel(data)
print( "finish building model.")
parameters = filter(lambda p: p.requires_grad, model.parameters())
optimizer = optim.Adamax(parameters, lr=data.HP_lr)
best_dev = -1
best_dev_p = -1
best_dev_r = -1
best_test = -1
best_test_p = -1
best_test_r = -1
## start training
for idx in range(data.HP_iteration):
epoch_start = time.time()
temp_start = epoch_start
print(("Epoch: %s/%s" %(idx,data.HP_iteration)))
optimizer = lr_decay(optimizer, idx, data.HP_lr_decay, data.HP_lr)
instance_count = 0
sample_loss = 0
batch_loss = 0
total_loss = 0
right_token = 0
whole_token = 0
random.shuffle(data.train_Ids)
## set model in train model
model.train()
model.zero_grad()
batch_size = data.HP_batch_size
batch_id = 0
train_num = len(data.train_Ids)
total_batch = train_num//batch_size+1
for batch_id in range(total_batch):
start = batch_id*batch_size
end = (batch_id+1)*batch_size
if end >train_num:
end = train_num
instance = data.train_Ids[start:end]
words = data.train_texts[start:end]
if not instance:
continue
gaz_list, batch_word, batch_biword, batch_wordlen, batch_label, layer_gaz, gaz_count, gaz_chars, gaz_mask, gazchar_mask, mask, batch_bert, bert_mask = batchify_with_label(instance, data.HP_gpu,data.HP_num_layer)
instance_count += 1
loss, tag_seq = model.neg_log_likelihood_loss(gaz_list, batch_word, batch_biword, batch_wordlen, layer_gaz, gaz_count,gaz_chars, gaz_mask, gazchar_mask, mask, batch_label, batch_bert, bert_mask)
right, whole = predict_check(tag_seq, batch_label, mask)
right_token += right
whole_token += whole
sample_loss += loss.data
total_loss += loss.data
batch_loss += loss
if end%500 == 0:
temp_time = time.time()
temp_cost = temp_time - temp_start
temp_start = temp_time
print((" Instance: %s; Time: %.2fs; loss: %.4f; acc: %s/%s=%.4f"%(end, temp_cost, sample_loss, right_token, whole_token,(right_token+0.)/whole_token)))
sys.stdout.flush()
sample_loss = 0
if end%data.HP_batch_size == 0:
batch_loss.backward()
optimizer.step()
model.zero_grad()
batch_loss = 0
temp_time = time.time()
temp_cost = temp_time - temp_start
print((" Instance: %s; Time: %.2fs; loss: %.4f; acc: %s/%s=%.4f"%(end, temp_cost, sample_loss, right_token, whole_token,(right_token+0.)/whole_token)) )
epoch_finish = time.time()
epoch_cost = epoch_finish - epoch_start
print(("Epoch: %s training finished. Time: %.2fs, speed: %.2fst/s, total loss: %s"%(idx, epoch_cost, train_num/epoch_cost, total_loss)))
speed, acc, p, r, f, pred_labels, gazs = evaluate(data, model, "dev")
dev_finish = time.time()
dev_cost = dev_finish - epoch_finish
if seg:
current_score = f
print(("Dev: time: %.2fs, speed: %.2fst/s; acc: %.4f, p: %.4f, r: %.4f, f: %.4f"%(dev_cost, speed, acc, p, r, f)))
else:
current_score = acc
print(("Dev: time: %.2fs speed: %.2fst/s; acc: %.4f"%(dev_cost, speed, acc)))
if current_score > best_dev:
if seg:
print( "Exceed previous best f score:", best_dev)
else:
print( "Exceed previous best acc score:", best_dev)
model_name = save_model_dir
torch.save(model.state_dict(), model_name)
#best_dev = current_score
best_dev_p = p
best_dev_r = r
# ## decode test
speed, acc, p, r, f, pred_labels, gazs = evaluate(data, model, "test")
test_finish = time.time()
test_cost = test_finish - dev_finish
if seg:
current_test_score = f
print(("Test: time: %.2fs, speed: %.2fst/s; acc: %.4f, p: %.4f, r: %.4f, f: %.4f"%(test_cost, speed, acc, p, r, f)))
else:
current_test_score = acc
print(("Test: time: %.2fs, speed: %.2fst/s; acc: %.4f"%(test_cost, speed, acc)))
if current_score > best_dev:
best_dev = current_score
best_test = current_test_score
best_test_p = p
best_test_r = r
print("Best dev score: p:{}, r:{}, f:{}".format(best_dev_p,best_dev_r,best_dev))
print("Test score: p:{}, r:{}, f:{}".format(best_test_p,best_test_r,best_test))
gc.collect()
with open(data.result_file,"a") as f:
f.write(save_model_dir+'\n')
f.write("Best dev score: p:{}, r:{}, f:{}\n".format(best_dev_p,best_dev_r,best_dev))
f.write("Test score: p:{}, r:{}, f:{}\n\n".format(best_test_p,best_test_r,best_test))
f.close()
