def train(args):
# load data
num_nodes, num_rels = utils.get_total_number('./data/' + args.dataset, 'stat.txt')
if args.dataset == 'icews_know':
train_data, train_times = utils.load_quadruples('./data/' + args.dataset, 'train.txt')
valid_data, valid_times = utils.load_quadruples('./data/' + args.dataset, 'test.txt')
test_data, test_times = utils.load_quadruples('./data/' + args.dataset, 'test.txt')
total_data, total_times = utils.load_quadruples('./data/' + args.dataset, 'train.txt', 'test.txt')
else:
train_data, train_times = utils.load_quadruples('./data/' + args.dataset, 'train.txt')
valid_data, valid_times = utils.load_quadruples('./data/' + args.dataset, 'valid.txt')
test_data, test_times = utils.load_quadruples('./data/' + args.dataset, 'test.txt')
total_data, total_times = utils.load_quadruples('./data/' + args.dataset, 'train.txt', 'valid.txt','test.txt')
# check cuda
use_cuda = args.gpu >= 0 and torch.cuda.is_available()
seed = 999
np.random.seed(seed)
torch.manual_seed(seed)
if use_cuda:
torch.cuda.set_device(args.gpu)
os.makedirs('models', exist_ok=True)
os.makedirs('models/'+ args.dataset, exist_ok=True)
model_state_file = 'models/' + args.dataset + '/rgcn.pth'
model_graph_file = 'models/' + args.dataset + '/rgcn_graph.pth'
model_state_global_file2 = 'models/' + args.dataset + '/max' + str(args.maxpool) + 'rgcn_global2.pth'
model_state_global_file = 'models/' + args.dataset + '/max' + str(args.maxpool) + 'rgcn_global.pth'
model_state_file_backup = 'models/' + args.dataset + '/rgcn_backup.pth'
print("start training...")
model = RENet(num_nodes,
args.n_hidden,
num_rels,
dropout=args.dropout,
model=args.model,
seq_len=args.seq_len,
num_k=args.num_k)
global_model = RENet_global(num_nodes,
args.n_hidden,
num_rels,
dropout=args.dropout,
model=args.model,
seq_len=args.seq_len,
num_k=args.num_k, maxpool=args.maxpool)
optimizer = torch.optim.Adam(model.parameters(), lr=args.lr, weight_decay=0.00001)
checkpoint_global = torch.load(model_state_global_file, map_location=lambda storage, loc: storage)
global_model.load_state_dict(checkpoint_global['state_dict'])
global_emb = checkpoint_global['global_emb']
model.global_emb = global_emb
if use_cuda:
model.cuda()
global_model.cuda()
train_sub = '/train_history_sub.txt'
train_ob = '/train_history_ob.txt'
if args.dataset == 'icews_know':
valid_sub = '/test_history_sub.txt'
valid_ob = '/test_history_ob.txt'
else:
valid_sub = '/dev_history_sub.txt'
valid_ob = '/dev_history_ob.txt'
with open('./data/' + args.dataset+'/train_graphs.txt', 'rb') as f:
graph_dict = pickle.load(f)
model.graph_dict = graph_dict
with open('data/' + args.dataset+'/test_history_sub.txt', 'rb') as f:
s_history_test_data = pickle.load(f)
with open('data/' + args.dataset+'/test_history_ob.txt', 'rb') as f:
o_history_test_data = pickle.load(f)
s_history_test = s_history_test_data[0]
s_history_test_t = s_history_test_data[1]
o_history_test = o_history_test_data[0]
o_history_test_t = o_history_test_data[1]
with open('./data/' + args.dataset+train_sub, 'rb') as f:
s_history_data = pickle.load(f)
with open('./data/' + args.dataset+train_ob, 'rb') as f:
o_history_data = pickle.load(f)
with open('./data/' + args.dataset+valid_sub, 'rb') as f:
s_history_valid_data = pickle.load(f)
with open('./data/' + args.dataset+valid_ob, 'rb') as f:
o_history_valid_data = pickle.load(f)
valid_data = torch.from_numpy(valid_data)
s_history = s_history_data[0]
s_history_t = s_history_data[1]
o_history = o_history_data[0]
o_history_t = o_history_data[1]
s_history_valid = s_history_valid_data[0]
s_history_valid_t = s_history_valid_data[1]
o_history_valid = o_history_valid_data[0]
o_history_valid_t = o_history_valid_data[1]
total_data = torch.from_numpy(total_data)
if use_cuda:
total_data = total_data.cuda()
epoch = 0
best_mrr = 0
while True:
print('training starting')
model.train()
if epoch == args.max_epochs:
break
epoch += 1
loss_epoch = 0
t0 = time.time()
print('training time captured')
train_data_shuffle, s_history_shuffle, s_history_t_shuffle, o_history_shuffle, o_history_t_shuffle = shuffle(train_data, s_history, s_history_t, o_history, o_history_t)
print('training data formatted')
for batch_data, s_hist, s_hist_t, o_hist, o_hist_t in utils.make_batch2(train_data_shuffle, s_history_shuffle, s_history_t_shuffle, o_history_shuffle, o_history_t_shuffle, args.batch_size):
# break
batch_data = torch.from_numpy(batch_data).long()
if use_cuda:
batch_data = batch_data.cuda()
print('batch instance preprocessing')
loss_s = model(batch_data, (s_hist, s_hist_t), (o_hist, o_hist_t), graph_dict, subject=True)
loss_o = model(batch_data, (s_hist, s_hist_t), (o_hist, o_hist_t), graph_dict, subject=False)
loss = loss_s + loss_o
loss.backward()
torch.nn.utils.clip_grad_norm_(model.parameters(), args.grad_norm) # clip gradients
optimizer.step()
optimizer.zero_grad()
loss_epoch += loss.item()
t3 = time.time()
print("Epoch {:04d} | Loss {:.4f} | time {:.4f}".
format(epoch, loss_epoch/(len(train_data)/args.batch_size), t3 - t0))
## VALIDATION
if epoch % args.valid_every == 0 and epoch >= int(args.max_epochs/2):
model.eval()
global_model.eval()
total_loss = 0
total_ranks = np.array([])
model.init_history(train_data, (s_history, s_history_t), (o_history, o_history_t), valid_data,
(s_history_valid, s_history_valid_t), (o_history_valid, o_history_valid_t), test_data,
(s_history_test, s_history_test_t), (o_history_test, o_history_test_t))
model.latest_time = valid_data[0][3]
for i in range(len(valid_data)):
batch_data = valid_data[i]
s_hist = s_history_valid[i]
o_hist = o_history_valid[i]
s_hist_t = s_history_valid_t[i]
o_hist_t = o_history_valid_t[i]
if use_cuda:
batch_data = batch_data.cuda()
with torch.no_grad():
ranks, loss = model.evaluate_filter(batch_data, (s_hist, s_hist_t), (o_hist, o_hist_t), global_model, total_data)
total_ranks = np.concatenate((total_ranks, ranks))
total_loss += loss.item()
mrr = np.mean(1.0 / total_ranks)
mr = np.mean(total_ranks)
hits = []
for hit in [1, 3, 10]:
avg_count = np.mean((total_ranks <= hit))
hits.append(avg_count)
print("valid Hits (filtered) @ {}: {:.6f}".format(hit, avg_count))
print("valid MRR (filtered): {:.6f}".format(mrr))
print("valid MR (filtered): {:.6f}".format(mr))
print("valid Loss: {:.6f}".format(total_loss / (len(valid_data))))
if mrr > best_mrr:
best_mrr = mrr
torch.save({'state_dict': model.state_dict(), 'epoch': epoch,
's_hist': model.s_hist_test, 's_cache': model.s_his_cache,
'o_hist': model.o_hist_test, 'o_cache': model.o_his_cache,
's_hist_t': model.s_hist_test_t, 's_cache_t': model.s_his_cache_t,
'o_hist_t': model.o_hist_test_t, 'o_cache_t': model.o_his_cache_t,
'latest_time': model.latest_time, 'global_emb': model.global_emb},
model_state_file)
torch.save({'state_dict': global_model.state_dict(), 'epoch': epoch,
's_hist': model.s_hist_test, 's_cache': model.s_his_cache,
'o_hist': model.o_hist_test, 'o_cache': model.o_his_cache,
's_hist_t': model.s_hist_test_t, 's_cache_t': model.s_his_cache_t,
'o_hist_t': model.o_hist_test_t, 'o_cache_t': model.o_his_cache_t,
'latest_time': model.latest_time, 'global_emb': global_model.global_emb},
model_state_global_file2)
with open(model_graph_file, 'wb') as fp:
pickle.dump(model.graph_dict, fp)
print("training done")
def test(args):
# load data
num_nodes, num_rels = utils.get_total_number('./data/' + args.dataset,
'stat.txt')
if args.dataset == 'icews_know':
train_data, train_times = utils.load_quadruples(
'./data/' + args.dataset, 'train.txt')
valid_data, valid_times = utils.load_quadruples(
'./data/' + args.dataset, 'test.txt')
test_data, test_times = utils.load_quadruples('./data/' + args.dataset,
'test.txt')
total_data, total_times = utils.load_quadruples(
'./data/' + args.dataset, 'train.txt', 'test.txt')
else:
train_data, train_times = utils.load_quadruples(
'./data/' + args.dataset, 'train.txt')
valid_data, valid_times = utils.load_quadruples(
'./data/' + args.dataset, 'valid.txt')
test_data, test_times = utils.load_quadruples('./data/' + args.dataset,
'test.txt')
total_data, total_times = utils.load_quadruples(
'./data/' + args.dataset, 'train.txt', 'valid.txt', 'test.txt')
# check cuda
use_cuda = args.gpu >= 0 and torch.cuda.is_available()
if use_cuda:
torch.cuda.set_device(args.gpu)
torch.cuda.manual_seed_all(999)
model_state_file = 'models/' + args.dataset + '/rgcn.pth'
model_graph_file = 'models/' + args.dataset + '/rgcn_graph.pth'
model_state_global_file2 = 'models/' + args.dataset + '/max' + str(
args.maxpool) + 'rgcn_global2.pth'
model = RENet(num_nodes,
args.n_hidden,
num_rels,
model=args.model,
seq_len=args.seq_len,
num_k=args.num_k)
global_model = RENet_global(num_nodes,
args.n_hidden,
num_rels,
model=args.model,
seq_len=args.seq_len,
num_k=args.num_k,
maxpool=args.maxpool)
if use_cuda:
model.cuda()
global_model.cuda()
with open('data/' + args.dataset + '/test_history_sub.txt', 'rb') as f:
s_history_test_data = pickle.load(f)
with open('data/' + args.dataset + '/test_history_ob.txt', 'rb') as f:
o_history_test_data = pickle.load(f)
s_history_test = s_history_test_data[0]
s_history_test_t = s_history_test_data[1]
o_history_test = o_history_test_data[0]
o_history_test_t = o_history_test_data[1]
print("\nstart testing:")
checkpoint = torch.load(model_state_file,
map_location=lambda storage, loc: storage)
model.load_state_dict(checkpoint['state_dict'])
model.s_hist_test = checkpoint['s_hist']
model.s_his_cache = checkpoint['s_cache']
model.o_hist_test = checkpoint['o_hist']
model.o_his_cache = checkpoint['o_cache']
model.latest_time = checkpoint['latest_time']
if args.dataset == "icews_know":
model.latest_time = torch.LongTensor([4344])[0]
model.global_emb = checkpoint['global_emb']
model.s_hist_test_t = checkpoint['s_hist_t']
model.s_his_cache_t = checkpoint['s_cache_t']
model.o_hist_test_t = checkpoint['o_hist_t']
model.o_his_cache_t = checkpoint['o_cache_t']
with open(model_graph_file, 'rb') as f:
model.graph_dict = pickle.load(f)
checkpoint_global = torch.load(model_state_global_file2,
map_location=lambda storage, loc: storage)
global_model.load_state_dict(checkpoint_global['state_dict'])
print("Using best epoch: {}".format(checkpoint['epoch']))
total_data = torch.from_numpy(total_data)
test_data = torch.from_numpy(test_data)
model.eval()
global_model.eval()
total_loss = 0
total_ranks = np.array([])
total_ranks_filter = np.array([])
ranks = []
for ee in range(num_nodes):
while len(model.s_hist_test[ee]) > args.seq_len:
model.s_hist_test[ee].pop(0)
model.s_hist_test_t[ee].pop(0)
while len(model.o_hist_test[ee]) > args.seq_len:
model.o_hist_test[ee].pop(0)
model.o_hist_test_t[ee].pop(0)
if use_cuda:
total_data = total_data.cuda()
latest_time = test_times[0]
for i in range(len(test_data)):
batch_data = test_data[i]
s_hist = s_history_test[i]
o_hist = o_history_test[i]
if args.model == 3:
s_hist_t = s_history_test_t[i]
o_hist_t = o_history_test_t[i]
if latest_time != batch_data[3]:
ranks.append(total_ranks_filter)
latest_time = batch_data[3]
total_ranks_filter = np.array([])
if use_cuda:
batch_data = batch_data.cuda()
with torch.no_grad():
# Filtered metric
if args.raw:
ranks_filter, loss = model.evaluate(batch_data,
(s_hist, s_hist_t),
(o_hist, o_hist_t),
global_model)
else:
ranks_filter, loss = model.evaluate_filter(
batch_data, (s_hist, s_hist_t), (o_hist, o_hist_t),
global_model, total_data)
total_ranks_filter = np.concatenate(
(total_ranks_filter, ranks_filter))
total_loss += loss.item()
ranks.append(total_ranks_filter)
for rank in ranks:
total_ranks = np.concatenate((total_ranks, rank))
mrr = np.mean(1.0 / total_ranks)
mr = np.mean(total_ranks)
hits = []
for hit in [1, 3, 10]:
avg_count = np.mean((total_ranks <= hit))
hits.append(avg_count)
print("Hits (filtered) @ {}: {:.6f}".format(hit, avg_count))
print("MRR (filtered): {:.6f}".format(mrr))
print("MR (filtered): {:.6f}".format(mr))
def train(args):
# load data
num_nodes, num_rels = utils.get_total_number('./data/' + args.dataset,
'stat.txt')
train_data, train_times_origin = utils.load_quadruples(
'./data/' + args.dataset, 'train.txt')
# check cuda
use_cuda = args.gpu >= 0 and torch.cuda.is_available()
seed = 999
np.random.seed(seed)
torch.manual_seed(seed)
if use_cuda:
torch.cuda.set_device(args.gpu)
os.makedirs('models', exist_ok=True)
os.makedirs('models/' + args.dataset, exist_ok=True)
if args.model == 0:
model_state_file = 'models/' + args.dataset + 'attn.pth'
elif args.model == 1:
model_state_file = 'models/' + args.dataset + 'mean.pth'
elif args.model == 2:
model_state_file = 'models/' + args.dataset + 'gcn.pth'
elif args.model == 3:
model_state_file = 'models/' + args.dataset + '/max' + str(
args.maxpool) + 'rgcn_global.pth'
# model_graph_file = 'models/' + args.dataset + 'rgcn_graph.pth'
model_state_file_backup = 'models/' + args.dataset + '/max' + str(
args.maxpool) + 'rgcn__global_backup.pth'
# model_graph_file_backup = 'models/' + args.dataset + 'rgcn_graph_backup.pth'
print("start training...")
model = RENet_global(num_nodes,
args.n_hidden,
num_rels,
dropout=args.dropout,
model=args.model,
seq_len=args.seq_len,
num_k=args.num_k,
maxpool=args.maxpool)
optimizer = torch.optim.Adam(model.parameters(),
lr=args.lr,
weight_decay=0.00001)
if use_cuda:
model.cuda()
# train_times = torch.from_numpy(train_times)
with open('./data/' + args.dataset + '/train_graphs.txt', 'rb') as f:
graph_dict = pickle.load(f)
true_prob_s, true_prob_o = utils.get_true_distribution(
train_data, num_nodes)
epoch = 0
loss_small = 10000
while True:
model.train()
if epoch == args.max_epochs:
break
epoch += 1
loss_epoch = 0
t0 = time.time()
# print(graph_dict.keys())
# print(train_times_origin)
train_times, true_prob_s, true_prob_o = shuffle(
train_times_origin, true_prob_s, true_prob_o)
for batch_data, true_s, true_o in utils.make_batch(
train_times, true_prob_s, true_prob_o, args.batch_size):
batch_data = torch.from_numpy(batch_data)
true_s = torch.from_numpy(true_s)
true_o = torch.from_numpy(true_o)
if use_cuda:
batch_data = batch_data.cuda()
true_s = true_s.cuda()
true_o = true_o.cuda()
loss = model(batch_data, true_s, true_o, graph_dict)
loss.backward()
torch.nn.utils.clip_grad_norm_(model.parameters(),
args.grad_norm) # clip gradients
optimizer.step()
optimizer.zero_grad()
loss_epoch += loss.item()
t3 = time.time()
model.global_emb = model.get_global_emb(train_times_origin, graph_dict)
print("Epoch {:04d} | Loss {:.4f} | time {:.4f}".format(
epoch, loss_epoch / (len(train_times) / args.batch_size), t3 - t0))
if loss_epoch < loss_small:
loss_small = loss_epoch
if args.model == 3:
torch.save(
{
'state_dict': model.state_dict(),
'global_emb': model.global_emb
}, model_state_file)
# with open(model_graph_file, 'wb') as fp:
# pickle.dump(model.graph_dict, fp)
else:
torch.save(
{
'state_dict': model.state_dict(),
'epoch': epoch,
's_hist': model.s_hist_test,
's_cache': model.s_his_cache,
'o_hist': model.o_hist_test,
'o_cache': model.o_his_cache,
'latest_time': model.latest_time
}, model_state_file)
print("training done")