-
Notifications
You must be signed in to change notification settings - Fork 1
/
test_GT_D2G.py
152 lines (140 loc) · 7.69 KB
/
test_GT_D2G.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
"""
Test scripts
Author:
Create Date: Dec 10, 2020
"""
import os
import time
import json
import random
import statistics
import numpy as np
import torch as th
import torch.nn as nn
from torch.utils.data import DataLoader
import sacred
from utils import convert_adj_vec_to_matrix, get_sequence_lens_by_pointers, mask_generated_graph
from model.data_loader import prepare_ingredients, collate_fn
from model.GPT_GRNN import GCNEncoder, GraphClassifier
from model.GPT_GRNN import GPTGRNNDecoderS, GPTGRNNDecoder, GPTGRNNDecoderVariable
from model.GPT_GRNN import GPTGRNNDecoderSVar
# Sacred Setup
ex = sacred.Experiment('test_GT-D2G')
@ex.config
def my_config():
config_path = ''
checkpoint_path = ''
gumbel_tau = 1e5 # high temperaute for evaluate
@ex.automain
def test_model(config_path, checkpoint_path, gumbel_tau, _run, _log):
if not config_path or not checkpoint_path:
_log.error('missing arg=config_path | checkpoint_path')
exit(-1)
# Load config
_log.info('Load config from %s' % (config_path))
with open(config_path) as fopen:
loaded_cfg = json.load(fopen)
opt = loaded_cfg['opt']
random.seed(opt['seed'])
np.random.seed(opt['seed'])
th.manual_seed(opt['seed'])
n_labels = opt['n_labels'][opt['corpus_type']]
# recover newly added params
pretrain_emb_name = opt.get('pretrain_emb_name', 'glove.840B.300d.txt')
pretrain_emb_cache = opt.get('pretrain_emb_cache', None)
pretrain_emb_max_vectors = opt.get('pretrain_emb_max_vectors', 160000)
gcn_encoder_pooling = opt.get('gcn_encoder_pooling', 'mean')
yelp_senti_feat = opt.get('yelp_senti_feat', False)
pretrain_emb_dropout = opt.get('pretrain_emb_dropout', 0.0)
# Load corpus
batch_size = opt['batch_size']
pickle_path = opt['processed_pickle_path']
_log.info('[%s] Start loading %s corpus from %s' % (time.ctime(), opt['corpus_type'], pickle_path))
train_set, val_set, test_set, vocab = prepare_ingredients(pickle_path, corpus_type=opt['corpus_type'],
pretrain_name=pretrain_emb_name,
emb_cache=pretrain_emb_cache,
max_vectors=pretrain_emb_max_vectors,
yelp_senti_feature=yelp_senti_feat)
# train_iter = DataLoader(train_set, batch_size=batch_size, shuffle=False, collate_fn=collate_fn)
test_iter = DataLoader(test_set, batch_size=batch_size, shuffle=False, collate_fn=collate_fn)
_log.info('[%s] Load train, val, test sets Done, len=%d,%d,%d' % (time.ctime(),
len(train_set), len(val_set), len(test_set)))
# Build models
pretrained_emb = vocab.vectors
gcn_encoder = GCNEncoder(pretrained_emb, pretrained_emb.shape[1]+3, opt['gcn_encoder_hidden_size'],
gcn_encoder_pooling, yelp_senti_feat, pretrain_emb_dropout)
if opt['gpt_grnn_variant'] == 'neigh':
gptrnn_decoder = GPTGRNNDecoderS(opt['gcn_encoder_hidden_size'], opt['GPT_attention_unit'],
opt['max_out_node_size'], gumbel_tau)
elif opt['gpt_grnn_variant'] == 'path':
gptrnn_decoder = GPTGRNNDecoder(opt['gcn_encoder_hidden_size'], opt['GPT_attention_unit'],
opt['max_out_node_size'], opt['graph_rnn_num_layers'],
opt['graph_rnn_hidden_size'], opt['edge_rnn_num_layers'],
opt['edge_rnn_hidden_size'],
gumbel_tau, opt['gptrnn_decoder_dropout']
)
elif opt['gpt_grnn_variant'] == 'path-var':
gptrnn_decoder = GPTGRNNDecoderVariable(opt['gcn_encoder_hidden_size'], opt['GPT_attention_unit'],
opt['max_out_node_size'], opt['graph_rnn_num_layers'],
opt['graph_rnn_hidden_size'], opt['edge_rnn_num_layers'],
opt['edge_rnn_hidden_size'],
opt['gumbel_tau_init'], opt['gptrnn_decoder_dropout']
)
elif opt['gpt_grnn_variant'] == 'neigh-var':
gptrnn_decoder = GPTGRNNDecoderSVar(opt['gcn_encoder_hidden_size'], opt['GPT_attention_unit'],
opt['max_out_node_size'], opt['graph_rnn_num_layers'],
opt['graph_rnn_hidden_size'], opt['gumbel_tau_init'],
opt['gptrnn_decoder_dropout']
)
else:
_log.error('invalid gpt_grnn_variant=%s, expected "neigh"|"path"')
exit(-1)
gcn_classifier = GraphClassifier(opt['gcn_encoder_hidden_size'], opt['gcn_classifier_hidden_size'],
n_labels)
class_criterion = nn.CrossEntropyLoss()
# Load checkpoints
checkpoint = th.load(checkpoint_path)
gcn_encoder.load_state_dict(checkpoint['gcn_encoder'])
gptrnn_decoder.load_state_dict(checkpoint['gptrnn_decoder'])
gcn_classifier.load_state_dict(checkpoint['gcn_classifier'])
_log.info('Load state_dict from %s' % (checkpoint_path))
if opt['gpu']:
gcn_encoder = gcn_encoder.cuda()
gptrnn_decoder = gptrnn_decoder.cuda()
gcn_classifier = gcn_classifier.cuda()
class_criterion = class_criterion.cuda()
gcn_encoder.eval()
gptrnn_decoder.eval()
gcn_classifier.eval()
val_node_cnts = []
all_pred = []
all_gold = []
with th.no_grad():
for i_batch, batch in enumerate(test_iter):
batched_graph, nid_mappings, labels, docids = batch
batch_size = labels.shape[0]
if opt['gpu']:
batched_graph = batched_graph.to('cuda:0')
labels = labels.cuda()
h, hg = gcn_encoder(batched_graph)
if opt['gpt_grnn_variant'] in ['variable', 'neigh-var']:
pointer_argmaxs, cov_loss, encoder_out, adj_vecs, att_scores = gptrnn_decoder(batched_graph, h, hg)
generated_node_lens = get_sequence_lens_by_pointers(pointer_argmaxs) # (batch,)
adj_matrix = convert_adj_vec_to_matrix(adj_vecs, add_self_loop=True)
generated_nodes_emb = th.matmul(pointer_argmaxs.transpose(1, 2), encoder_out) # batch*seq_l*hid
generated_nodes_emb, adj_matrix = mask_generated_graph(generated_nodes_emb, adj_matrix,
generated_node_lens)
pred = gcn_classifier(generated_nodes_emb, adj_matrix)
val_node_cnts.extend(generated_node_lens.tolist())
else:
pointer_argmaxs, cov_loss, encoder_out, adj_vecs = gptrnn_decoder(batched_graph, h, hg)
adj_matrix = convert_adj_vec_to_matrix(adj_vecs, add_self_loop=True)
generated_nodes_emb = th.matmul(pointer_argmaxs.transpose(1, 2), encoder_out) # batch*seq_l*hid
pred = gcn_classifier(generated_nodes_emb, adj_matrix)
all_gold.extend(labels.detach().tolist())
all_pred.extend(th.argmax(pred, dim=1).detach().tolist())
acc = (th.LongTensor(all_gold) == th.LongTensor(all_pred)).sum() / len(all_pred)
_log.info('[%s] test acc=%.2f, ' % (time.ctime(), acc*100))
# if opt['gpt_grnn_variant'] in ['variable', 'neigh-var']:
# _log.info('generated node cnt avg=%.3f, stdev=%.3f' % (statistics.mean(val_node_cnts),
# statistics.pstdev(val_node_cnts)))