-
Notifications
You must be signed in to change notification settings - Fork 1
/
train.py
315 lines (281 loc) · 12 KB
/
train.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
import pickle
import sys
import dgl
import numpy as np
import torch
import torch.nn.functional as F
from sklearn.metrics import mean_squared_error
from torch.utils.data import DataLoader
from torch_geometric.data import Data
import torch.nn.functional as F
import signal
from graph_generator import CalibrationDataset, HumanGraph
from nets.gat import GAT
from nets.rgcnDGL import RGCN
if torch.cuda.is_available() is True:
device = torch.device('cuda')
else:
device = torch.device('cpu')
def activation_functions(activation_tuple_src):
ret = []
for x in activation_tuple_src:
if x == 'relu':
ret.append(F.relu)
elif x == 'elu':
ret.append(F.elu)
elif x == 'tanh':
ret.append(torch.tanh)
elif x == 'leaky_relu':
ret.append(F.leaky_relu)
else:
print('Unknown activation function {}.'.format(x))
sys.exit(-1)
return tuple(ret)
def describe_model(model):
# Print model's state_dict
print("Model's state_dict:")
for param_tensor in model.state_dict():
print(param_tensor, "\t", model.state_dict()[param_tensor].size())
def collate(sample):
graphs, feats, labels = map(list, zip(*sample))
graph = dgl.batch(graphs)
feats = torch.from_numpy(np.concatenate(feats))
labels = torch.from_numpy(np.concatenate(labels))
return graph, feats, labels
def evaluate(feats, model, subgraph, labels, loss_fcn, fw, net_class):
with torch.no_grad():
model.eval()
if fw == 'dgl' :
model.g = subgraph
for layer in model.layers:
layer.g = subgraph
if net_class in [RGCN]:
output = model(feats.float(), subgraph.edata['rel_type'].to(device).squeeze())
else:
output = model(feats.float())
oput = output[getMaskForBatch(subgraph)]
loss_data = loss_fcn(oput, labels.float())
predict = output[getMaskForBatch(subgraph)].data.cpu().numpy()
score = mean_squared_error(labels.data.cpu().numpy(), predict)
return score, loss_data.item()
def getMaskForBatch(subgraph):
first_node_index_in_the_next_graph = 0
indexes = []
for g in dgl.unbatch(subgraph):
indexes.append(first_node_index_in_the_next_graph)
first_node_index_in_the_next_graph += g.number_of_nodes()
return indexes
def num_of_params(model):
model_parameters = filter(lambda p: p.requires_grad, model.parameters())
return sum([np.prod(p.size()) for p in model_parameters])
stop_training = False
ctrl_c_counter = 0
def signal_handler(sig, frame):
global ctrl_c_counter
ctrl_c_counter += 1
if ctrl_c_counter >= 6:
sys.exit(-1)
elif ctrl_c_counter >= 3:
global stop_training
stop_training = True
print('\nIf you press Ctr+c 3 times we will stop _SAVING_ the training information ({} times)'.format(ctrl_c_counter))
print( 'If you press Ctr+c 6+ times we will stop _NOT SAVING_ the training information ({} times)'.format(ctrl_c_counter))
signal.signal(signal.SIGTERM, signal_handler)
signal.signal(signal.SIGINT, signal_handler)
# MAIN
def main(training_file, dev_file, test_file, epochs=None, patience=None, heads=None, num_layers=None,
num_hidden=None, residual=None, in_drop=None, attn_drop=None, lr=None, weight_decay=None,
alpha=None, batch_size=None, graph_type=None, net=None, activations=('elu', 'tanh'), fw='dgl'):
if net.lower() == 'GAT'.lower():
net_class = GAT
elif net.lower() == 'RGCN'.lower():
net_class = RGCN
print('DEVICE', device)
# define loss function
loss_fcn = torch.nn.MSELoss()
print('=========================')
print('HEADS', heads)
print('LAYERS', num_layers)
print('HIDDEN', num_hidden)
print('RESIDUAL', residual)
print('inDROP', in_drop)
print('atDROP', attn_drop)
print('LR', lr)
print('DECAY', weight_decay)
print('ALPHA', alpha)
print('BATCH', batch_size)
print('GRAPH_ALT', graph_type)
print('ARCHITECTURE', net)
print('=========================')
# create the dataset
print('Loading training set...')
train_dataset = CalibrationDataset(training_file, mode='train', alt=graph_type)
print('Loading dev set...')
valid_dataset = CalibrationDataset(dev_file, mode='valid', alt=graph_type)
print('Loading test set...')
test_dataset = CalibrationDataset(test_file, mode='test', alt=graph_type)
print('Done loading files')
train_dataloader = DataLoader(train_dataset, batch_size=batch_size, collate_fn=collate)
valid_dataloader = DataLoader(valid_dataset, batch_size=batch_size, collate_fn=collate)
test_dataloader = DataLoader(test_dataset, batch_size=batch_size, collate_fn=collate)
num_rels = len(HumanGraph.get_rels())
cur_step = 0
best_loss = -1
n_classes = train_dataset.labels.shape[1]
print('Number of classes: {}'.format(n_classes))
num_feats = train_dataset.features.shape[1]
print('Number of features: {}'.format(num_feats))
g = train_dataset.graph
# define the model
print('LAST', fw, net)
if fw == 'dgl':
if net_class in [GAT]:
model = net_class(g, num_layers, num_feats, num_hidden, n_classes, heads, activation_functions(activations),
in_drop, attn_drop, alpha, residual)
elif net_class in [RGCN]:
model = RGCN(g, gnn_layers=num_layers, in_dim=num_feats, hidden_dimensions=num_hidden, num_rels=num_rels,
activations=activation_functions(activations), feat_drop=in_drop)
print(f'CREATING RGCN(GRAPH, gnn_layers:{num_layers}, num_feats:{num_feats}, num_hidden:{num_hidden}, num_rels:{num_rels}, non-linearity:{activation_functions(activations)}, drop:{in_drop})')
else:
print('Unhandled', net)
sys.exit(1)
#Describe the model
#describe_model(model)
# define the optimizer
optimizer = torch.optim.Adam(model.parameters(), lr=lr, weight_decay=weight_decay)
# for name, param in model.named_parameters():
# if param.requires_grad:
# print(name, param.data.shape)
model = model.to(device)
for epoch in range(epochs):
if stop_training:
print("Stopping training. Please wait.")
break
model.train()
loss_list = []
for batch, data in enumerate(train_dataloader):
subgraph, feats, labels = data
subgraph.set_n_initializer(dgl.init.zero_initializer)
subgraph.set_e_initializer(dgl.init.zero_initializer)
feats = feats.to(device)
labels = labels.to(device)
if fw == 'dgl':
model.g = subgraph
for layer in model.layers:
layer.g = subgraph
if net_class in [RGCN]:
logits = model(feats.float(), subgraph.edata['rel_type'].squeeze().to(device))
else:
logits = model(feats.float())
loss = loss_fcn(logits[getMaskForBatch(subgraph)], labels.float())
optimizer.zero_grad()
a = list(model.parameters())[0].clone()
loss.backward()
optimizer.step()
b = list(model.parameters())[0].clone()
not_learning = torch.equal(a.data, b.data)
if not_learning:
print('Not learning')
sys.exit(1)
loss_list.append(loss.item())
loss_data = np.array(loss_list).mean()
print('Loss: {}'.format(loss_data))
if epoch % 5 == 0:
if epoch % 5 == 0:
print("Epoch {:05d} | Loss: {:.4f} | Patience: {} | ".format(epoch, loss_data, cur_step), end='')
score_list = []
val_loss_list = []
for batch, valid_data in enumerate(valid_dataloader):
subgraph, feats, labels = valid_data
subgraph.set_n_initializer(dgl.init.zero_initializer)
subgraph.set_e_initializer(dgl.init.zero_initializer)
feats = feats.to(device)
labels = labels.to(device)
score, val_loss = evaluate(feats.float(), model, subgraph, labels.float(), loss_fcn, fw, net_class)
score_list.append(score)
val_loss_list.append(val_loss)
mean_score = np.array(score_list).mean()
mean_val_loss = np.array(val_loss_list).mean()
if epoch % 5 == 0:
print("Score: {:.4f} MEAN: {:.4f} BEST: {:.4f}".format(mean_score, mean_val_loss, best_loss))
# early stop
if best_loss > mean_val_loss or best_loss < 0:
print('Saving...')
best_loss = mean_val_loss
# Save the model
torch.save(model.state_dict(), 'calibration_' + fw + '_' + net + '.tch')
params = {'loss': best_loss,
'net': net,
'fw': fw,
'num_layers': num_layers,
'num_feats': num_feats,
'num_hidden': num_hidden,
'graph_type' : graph_type,
'n_classes': n_classes,
'heads': heads,
'F': F.relu,
'in_drop': in_drop,
'attn_drop': attn_drop,
'alpha': alpha,
'residual': residual,
'non-linearity': activations,
'num_rels': num_rels
}
pickle.dump(params, open('calibration_' + fw + '_' + net + '_tmp.prms', 'wb'))
cur_step = 0
else:
cur_step += 1
if cur_step >= patience:
break
model.eval()
test_score_list = []
for batch, test_data in enumerate(test_dataloader):
subgraph, feats, labels = test_data
subgraph.set_n_initializer(dgl.init.zero_initializer)
subgraph.set_e_initializer(dgl.init.zero_initializer)
feats = feats.to(device)
labels = labels.to(device)
test_score_list.append(evaluate(feats, model, subgraph, labels.float(), loss_fcn, fw, net_class)[1])
print("MSE for the test set {}".format(np.array(test_score_list).mean()))
params = {'loss': best_loss,
'test_loss': np.array(test_score_list).mean(),
'net': net,
'fw': fw,
'num_gnn_layers': num_layers,
'num_feats': num_feats,
'num_hidden': num_hidden,
'graph_type' : graph_type,
'n_classes': n_classes,
'heads': heads,
'F': F.relu,
'in_drop': in_drop,
'attn_drop': attn_drop,
'alpha': alpha,
'residual': residual,
'non-linearity': activations,
'num_rels': num_rels,
'parameters': num_of_params(model)
}
pickle.dump(params, open('calibration_' + fw + '_' + net + '.prms', 'wb'))
print('Number of parameters in the model {}.'.format(num_of_params(model)))
return best_loss, np.array(test_score_list).mean()
if __name__ == '__main__':
best_loss, test_loss = main('datasets/training_DS1.json',
'datasets/dev.json',
'datasets/test.json',
epochs=1000,
patience=15,
heads=[18, 17, 16],
num_layers=3,
num_hidden=[11, 8, 5],
residual=False,
in_drop=0.,
attn_drop=0.,
lr=0.0001,
weight_decay=0.00000001,
alpha=0.12,
batch_size=10,
graph_type='1',
net='gat',
activations=['relu', 'relu', 'tanh'],
fw='dgl')