-
Notifications
You must be signed in to change notification settings - Fork 36
/
GRec_TF_Pretrain.py
400 lines (331 loc) · 16.9 KB
/
GRec_TF_Pretrain.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
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
import tensorflow as tf
import data_loader_recsys
import generator_peterrec_non as generator_recsys
import utils
import shutil
import time
import math
import eval
import numpy as np
import argparse
import collections
import random
def shuffleseq(train_set,padtoken):
shuffle_seqtrain = []
for i in range(len(train_set)):
# print x_train[i]
seq = train_set[i][1:]
lenseq = len(seq)
# split=np.split(padtoken)
copyseq=list(seq)
padcount = copyseq.count(padtoken) #the number of padding elements
copyseq = copyseq[padcount:] # the remaining elements
# copyseq=seq
shuffle_indices = np.random.permutation(np.arange(len(copyseq)))
# list to array
copyseq= np.array(copyseq)
copyseq_shuffle=copyseq[shuffle_indices]
padtoken_list=[padtoken]*padcount
# array to list, + means concat in list and real plus in array
seq=list(train_set[i][0:1])+padtoken_list+list(copyseq_shuffle)
shuffle_seqtrain.append(seq)
x_train = np.array(shuffle_seqtrain) # list to ndarray
print "shuffling is done!"
return x_train
def generatesubsequence(train_set):
# create subsession only for training
subseqtrain = []
for i in range(len(train_set)):
# print x_train[i]
seq = train_set[i]
lenseq = len(seq)
# session lens=100 shortest subsession=5 realvalue+95 0
for j in range(lenseq - 2):
subseqend = seq[:len(seq) - j]
subseqbeg = [0] * j
subseq = np.append(subseqbeg, subseqend)
subseqtrain.append(subseq)
x_train = np.array(subseqtrain) # list to ndarray
del subseqtrain
# Randomly shuffle data
np.random.seed(10)
shuffle_train = np.random.permutation(np.arange(len(x_train)))
x_train = x_train[shuffle_train]
print "generating subsessions is done!"
return x_train
MaskedLmInstance = collections.namedtuple("MaskedLmInstance",
["index", "label"])
def create_masked_lm_predictions_frombatch(item_batch, masked_lm_prob,
max_predictions_per_seq, items, rng,item_size):
rng = random.Random()
output_tokens_batch=[]
maskedpositions_batch=[]
maskedlabels_batch=[]
masked_lm_weights_batch=[]
for line_list in range(item_batch.shape[0]):
# output_tokens, masked_lm_positions, masked_lm_labels=create_masked_lm_predictions(item_batch[line_list],masked_lm_prob,max_predictions_per_seq,items,rng,item_size)
output_tokens, masked_lm_positions, masked_lm_labels = create_masked_lm_predictions(item_batch[line_list],
masked_lm_prob,
max_predictions_per_seq,
items, rng, item_size)
# print output_tokens
output_tokens_batch.append(output_tokens)
maskedpositions_batch.append(masked_lm_positions)
maskedlabels_batch.append(masked_lm_labels)
masked_lm_weights = [1.0] * len(masked_lm_labels)
# note you can not change here since it should be consistent with 'num_to_predict' in create_masked_lm_predictions
num_to_predict = min(max_predictions_per_seq,
max(1, int(round(len(item_batch[line_list]) * masked_lm_prob))))
while len(masked_lm_weights) < num_to_predict:
masked_lm_weights.append(0.0)
masked_lm_weights_batch.append(masked_lm_weights)
return output_tokens_batch,maskedpositions_batch,maskedlabels_batch,masked_lm_weights_batch
def create_masked_predictions_frombatch(item_batch):
output_tokens_batch = []
maskedpositions_batch = []
maskedlabels_batch = []
for line_list in range(item_batch.shape[0]):
output_tokens,masked_lm_positions,masked_lm_labels=create_endmask(item_batch[line_list])
output_tokens_batch.append(output_tokens)
maskedpositions_batch.append(masked_lm_positions)
maskedlabels_batch.append(masked_lm_labels)
return output_tokens_batch,maskedpositions_batch,maskedlabels_batch
def create_endmask(tokens):
masked_lm_positions = []
masked_lm_labels = []
lens=len(tokens)
masked_token = 0
dutokens=list(tokens)
dutokens[-1]=masked_token
masked_lm_positions.append(lens-1)
masked_lm_labels.append(tokens[-1])
return dutokens,masked_lm_positions,masked_lm_labels
def create_masked_lm_predictions(tokens, masked_lm_prob,
max_predictions_per_seq, vocab_words, rng,item_size):
cand_indexes = []
for (i, token) in enumerate(tokens):
if token == "[CLS]" or token == "[SEP]":
continue
cand_indexes.append(i)
rng.shuffle(cand_indexes)
output_tokens = list(tokens)
num_to_predict = min(max_predictions_per_seq,
max(1, int(round(len(tokens) * masked_lm_prob))))
masked_lms = []
covered_indexes = set()
for index in cand_indexes:
if len(masked_lms) >= num_to_predict:
break
if index in covered_indexes:
continue
covered_indexes.add(index)
masked_token = None
if rng.random() < 0.8:
masked_token=0 #item_size is "[MASK]" 0 represents '<unk>'
else:
# 10% of the time, keep original
if rng.random() < 0.5:
masked_token = tokens[index]
# 10% of the time, replace with random word
else:
masked_token = vocab_words[rng.randint(0, len(vocab_words) - 1)]
output_tokens[index] = masked_token
masked_lms.append(MaskedLmInstance(index=index, label=tokens[index]))
masked_lms = sorted(masked_lms, key=lambda x: x.index)
masked_lm_positions = []
masked_lm_labels = []
for p in masked_lms:
masked_lm_positions.append(p.index)
masked_lm_labels.append(p.label)
return (output_tokens, masked_lm_positions, masked_lm_labels)
#Using GRec Encoder for pretraining
#Future Data Helps Training: Modeling Future Contexts for Session-based Recommendation. WWW2020. F.Yuan. et al
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--top_k', type=int, default=5,
help='Sample from top k predictions')
parser.add_argument('--beta1', type=float, default=0.9,
help='hyperpara-Adam')
parser.add_argument('--datapath', type=str, default='Data/Session/history_sequences_20181014_fajie_transfer_pretrain_small.csv',
help='data path')
parser.add_argument('--eval_iter', type=int, default=2,
help='Sample generator output evry x steps')
parser.add_argument('--save_para_every', type=int, default=2,
help='save model parameters every')
parser.add_argument('--tt_percentage', type=float, default=0.1,
help='0.2 means 80% training 20% testing')
parser.add_argument('--masked_lm_prob', type=float, default=0.3,
help='0.2 means 20% items are masked')
parser.add_argument('--max_predictions_per_seq', type=int, default=30,
help='maximum number of masked tokens')
parser.add_argument('--max_position', type=int, default=100,
help='maximum number of for positional embedding, it has to be larger than the sequence lens')
parser.add_argument('--is_generatesubsession', type=bool, default=False,
help='whether generating a subsessions, e.g., 12345-->01234,00123,00012 It may be useful for very some very long sequences')
parser.add_argument('--has_positionalembedding', type=bool, default=False,
help='whether contains positional embedding before performing cnnn')
parser.add_argument('--padtoken', type=str, default='-1',
help='is the padding token in the beggining of the sequence')
parser.add_argument('--is_shuffle', type=bool, default=False,
help='whether shuffle the training and testing dataset, e.g., 012345-->051324')
args = parser.parse_args()
dl = data_loader_recsys.Data_Loader({'model_type': 'generator', 'dir_name': args.datapath})
all_samples = dl.item
items = dl.item_dict#key is the original token, value is the mapped value, i.e., 0, 1,2,3...
itemlist=items.values()
item_size=len(items) # the first token is 'unk'
print "len(items)",item_size
if items.has_key(args.padtoken):
padtoken = items[args.padtoken] # is the padding token in the beggining of the sentence
else:
padtoken = item_size+1
max_predictions_per_seq=args.max_predictions_per_seq
masked_lm_prob=args.masked_lm_prob
np.random.seed(10)
shuffle_indices = np.random.permutation(np.arange(len(all_samples)))
all_samples = all_samples[shuffle_indices]
dev_sample_index = -1 * int(args.tt_percentage * float(len(all_samples)))
train_set, valid_set = all_samples[:dev_sample_index], all_samples[dev_sample_index:]
if args.is_generatesubsession:
train_set = generatesubsequence(train_set)
if args.is_shuffle:
train_set = shuffleseq(train_set, padtoken)
model_para = {
#if you changed the parameters here, also do not forget to change paramters in nextitrec_generate.py
'item_size': item_size,
'dilated_channels': 64,
'dilations': [1,4,1,4,1,4,1,4,],
'kernel_size': 3,
'learning_rate':0.001,
'batch_size':2,
'iterations':2,
'max_position':args.max_position,#maximum number of for positional embedding, it has to be larger than the sequence lens
'has_positionalembedding':args.has_positionalembedding,
'is_negsample':True #False denotes no negative sampling
}
itemrec = generator_recsys.NextItNet_Decoder(model_para)
itemrec.train_graph()
optimizer = tf.train.AdamOptimizer(model_para['learning_rate'], beta1=args.beta1).minimize(itemrec.loss)
itemrec.predict_graph(reuse=True)
tf.add_to_collection("dilate_input", itemrec.dilate_input)
tf.add_to_collection("context_embedding", itemrec.context_embedding)
sess= tf.Session()
init=tf.global_variables_initializer()
sess.run(init)
saver = tf.train.Saver()
numIters = 1
for iter in range(model_para['iterations']):
batch_no = 0
batch_size = model_para['batch_size']
while (batch_no + 1) * batch_size < train_set.shape[0]:
start = time.time()
item_batch = train_set[batch_no * batch_size: (batch_no + 1) * batch_size, :]
output_tokens_batch, maskedpositions_batch, maskedlabels_batch,masked_lm_weights_batch= create_masked_lm_predictions_frombatch(
item_batch,masked_lm_prob,max_predictions_per_seq,items=itemlist,rng=None,item_size=item_size
)
_, loss = sess.run(
[optimizer, itemrec.loss],
feed_dict={
itemrec.itemseq_input: output_tokens_batch,
itemrec.masked_position: maskedpositions_batch,
itemrec.masked_items: maskedlabels_batch,
itemrec.label_weights: masked_lm_weights_batch
})
end = time.time()
if numIters % args.eval_iter == 0:
print "-------------------------------------------------------train1"
print "LOSS: {}\tITER: {}\tBATCH_NO: {}\t STEP:{}\t total_batches:{}".format(
loss, iter, batch_no, numIters, train_set.shape[0] / batch_size)
print "TIME FOR BATCH", end - start
if numIters % args.eval_iter == 0:
print "-------------------------------------------------------test1"
batch_no_valid=0
batch_size_valid=batch_size
if (batch_no_valid + 1) * batch_size_valid < valid_set.shape[0]:
start = time.time()
item_batch = valid_set[(batch_no_valid) * batch_size_valid: (batch_no_valid + 1) * batch_size_valid, :]
output_tokens_batch, maskedpositions_batch, maskedlabels_batch, masked_lm_weights_batch = create_masked_lm_predictions_frombatch(
item_batch, masked_lm_prob, max_predictions_per_seq, items=itemlist, rng=None,
item_size=item_size
)
loss = sess.run(
[itemrec.loss],
feed_dict={
itemrec.itemseq_input: output_tokens_batch,
itemrec.masked_position: maskedpositions_batch,
itemrec.masked_items: maskedlabels_batch,
itemrec.label_weights: masked_lm_weights_batch
})
end = time.time()
print "LOSS: {}\tITER: {}\tBATCH_NO: {}\t STEP:{}\t total_batches:{}".format(
loss, iter, batch_no_valid, numIters, valid_set.shape[0] / batch_size_valid)
print "TIME FOR BATCH", end - start
batch_no += 1
if numIters % args.eval_iter == 0:
batch_no_test = 0
batch_size_test = batch_size*1
curr_preds_5=[]
rec_preds_5=[] #1
ndcg_preds_5=[] #1
curr_preds_20 = []
rec_preds_20 = [] # 1
ndcg_preds_20 = [] # 1
while (batch_no_test + 1) * batch_size_test < valid_set.shape[0]:
if (numIters / (args.eval_iter) < 10):
if (batch_no_test > 20):
break
else:
if (batch_no_test > 1000):
break
item_batch = valid_set[batch_no_test * batch_size_test: (batch_no_test + 1) * batch_size_test, :]
output_tokens_batch,maskedpositions_batch,maskedlabels_batch=create_masked_predictions_frombatch(item_batch)
[probs] = sess.run(
[itemrec.log_probs],
feed_dict={
itemrec.itemseq_input: output_tokens_batch,
itemrec.masked_position: maskedpositions_batch
})
for bi in range(probs.shape[0]):
pred_items_5 = utils.sample_top_k(probs[bi], top_k=args.top_k)#top_k=5
pred_items_20 = utils.sample_top_k(probs[bi], top_k=args.top_k+15)
true_item=item_batch[bi][-1]
predictmap_5={ch : i for i, ch in enumerate(pred_items_5)}
pred_items_20 = {ch: i for i, ch in enumerate(pred_items_20)}
rank_5=predictmap_5.get(true_item)
rank_20 = pred_items_20.get(true_item)
if rank_5 ==None:
curr_preds_5.append(0.0)
rec_preds_5.append(0.0)#2
ndcg_preds_5.append(0.0)#2
else:
MRR_5 = 1.0/(rank_5+1)
Rec_5=1.0#3
ndcg_5 = 1.0 / math.log(rank_5 + 2, 2) # 3
curr_preds_5.append(MRR_5)
rec_preds_5.append(Rec_5)#4
ndcg_preds_5.append(ndcg_5) # 4
if rank_20 ==None:
curr_preds_20.append(0.0)
rec_preds_20.append(0.0)#2
ndcg_preds_20.append(0.0)#2
else:
MRR_20 = 1.0/(rank_20+1)
Rec_20=1.0#3
ndcg_20 = 1.0 / math.log(rank_20 + 2, 2) # 3
curr_preds_20.append(MRR_20)
rec_preds_20.append(Rec_20)#4
ndcg_preds_20.append(ndcg_20) # 4
batch_no_test += 1
print "BATCH_NO: {}".format(batch_no_test)
print "Accuracy mrr_5:",sum(curr_preds_5) / float(len(curr_preds_5))#5
print "Accuracy mrr_20:", sum(curr_preds_20) / float(len(curr_preds_20)) # 5
print "Accuracy hit_5:", sum(rec_preds_5) / float(len(rec_preds_5))#5
print "Accuracy hit_20:", sum(rec_preds_20) / float(len(rec_preds_20)) # 5
print "Accuracy ndcg_5:", sum(ndcg_preds_5) / float(len(ndcg_preds_5)) # 5
print "Accuracy ndcg_20:", sum(ndcg_preds_20) / float(len(ndcg_preds_20)) #
numIters += 1
if numIters % args.save_para_every == 0:
save_path = saver.save(sess,
"Data/Models/generation_model/model_nextitnet_cloze".format(iter, numIters))
if __name__ == '__main__':
main()