-
Notifications
You must be signed in to change notification settings - Fork 10
/
srl_model.py
217 lines (168 loc) · 9.44 KB
/
srl_model.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
# Multi-predicate span-based SRL based on the e2e-coref model.
import math
import numpy as np
import operator
import os
import random
import tensorflow as tf
import util
from lsgn_data import LSGNData
from embedding_helper import get_embeddings
from input_utils import *
from model_utils import *
import attention
from layers import layer_norm, linear
from attention import multihead_attention
def _residual_fn(x, y, residual_dropout):
if residual_dropout > 0.0:
y = tf.nn.dropout(y, 1.0 - residual_dropout)
return layer_norm(x + y)
class SRLModel(object):
def __init__(self, lsgn_data, config):
self.config = config
self.data = lsgn_data
# TODO: Make labels_dict = None at test time.
self.predictions, self.loss = self.get_predictions_and_loss(
self.data.input_dict, self.data.labels_dict)
self.global_step = tf.Variable(0, name="global_step", trainable=False)
self.reset_global_step = tf.assign(self.global_step, 0)
learning_rate = tf.train.exponential_decay(
self.config["learning_rate"], self.global_step, self.config["decay_frequency"],
self.config["decay_rate"], staircase=True)
trainable_params = tf.trainable_variables()
gradients = tf.gradients(self.loss, trainable_params)
gradients, _ = tf.clip_by_global_norm(gradients, self.config["max_gradient_norm"])
optimizers = {
"adam" : tf.train.AdamOptimizer,
"sgd" : tf.train.GradientDescentOptimizer
}
optimizer = optimizers[self.config["optimizer"]](learning_rate)
self.train_op = optimizer.apply_gradients(zip(gradients, trainable_params),
global_step=self.global_step)
# For debugging.
# for var in tf.trainable_variables():
# print var
def get_predictions_and_loss(self, inputs, labels):
# This little thing got batched.
is_training = inputs["is_training"][0]
self.dropout = 1 - (tf.to_float(is_training) * self.config["dropout_rate"])
self.lexical_dropout = 1 - (tf.to_float(is_training) * self.config["lexical_dropout_rate"])
self.lstm_dropout = 1 - (tf.to_float(is_training) * self.config["lstm_dropout_rate"])
sentences = inputs["tokens"]
text_len = inputs["text_len"] # [num_sentences]
context_word_emb = inputs["context_word_emb"] # [num_sentences, max_sentence_length, emb]
head_word_emb = inputs["head_word_emb"] # [num_sentences, max_sentence_length, emb]
num_sentences = tf.shape(context_word_emb)[0]
max_sentence_length = tf.shape(context_word_emb)[1]
context_emb, head_emb, self.lm_weights, self.lm_scaling = get_embeddings(
self.data, sentences, text_len, context_word_emb, head_word_emb, inputs["char_idx"],
inputs["lm_emb"], self.lexical_dropout) # [num_sentences, max_sentence_length, emb]
context_outputs = lstm_contextualize(
context_emb, text_len, self.config, self.lstm_dropout) # [num_sentences, max_sentence_length, emb]
text_len_mask = tf.sequence_mask(text_len, maxlen=max_sentence_length) # [num_sentences, max_sentence_length]
# if self.config["self_attention"]:
# attn_bias = attention.attention_bias(tf.to_float(text_len_mask), "masking")
# with tf.variable_scope("self_attention"):
# y = multihead_attention(
# context_outputs,
# None,
# attn_bias,
# util.shape(context_outputs, 2),#params.attention_key_channels or params.hidden_size,
# util.shape(context_outputs, 2),#params.attention_value_channels or params.hidden_size,
# util.shape(context_outputs, 2),#params.hidden_size,
# 8,#params.num_heads,
# 1.0-0.0,#1.0 - params.attention_dropout,
# attention_function="dot_product"
# )
# context_outputs = _residual_fn(context_outputs, y, 0.1)
# [num_sentences, max_num_candidates], ...
candidate_starts, candidate_ends, candidate_mask = get_span_candidates(
text_len, max_sentence_length, self.config["max_arg_width"])
flat_candidate_mask = tf.reshape(candidate_mask, [-1]) # [num_sentences, max_num_candidates]
batch_word_offset = tf.expand_dims(tf.cumsum(text_len, exclusive=True), 1) # [num_sentences, 1]
flat_candidate_starts = tf.boolean_mask(
tf.reshape(candidate_starts + batch_word_offset, [-1]), flat_candidate_mask) # [num_candidates]
flat_candidate_ends = tf.boolean_mask(
tf.reshape(candidate_ends + batch_word_offset, [-1]), flat_candidate_mask) # [num_candidates]
flat_context_outputs = flatten_emb_by_sentence(context_outputs, text_len_mask) # context_outputs [num_doc_words]
flat_head_emb = flatten_emb_by_sentence(head_emb, text_len_mask) # [num_doc_words]
doc_len = util.shape(flat_context_outputs, 0)
candidate_span_emb, head_scores, span_head_emb, head_indices, head_indices_log_mask = get_span_emb(
flat_head_emb, flat_context_outputs, flat_candidate_starts, flat_candidate_ends,
self.config, self.dropout
) # [num_candidates, emb], [num_candidates, max_span_width, emb], [num_candidates, max_span_width]
num_candidates = util.shape(candidate_span_emb, 0)
max_num_candidates_per_sentence = util.shape(candidate_mask, 1)
candidate_span_ids = tf.sparse_to_dense(
sparse_indices=tf.where(tf.equal(candidate_mask, True)),
output_shape=tf.cast(tf.stack([num_sentences, max_num_candidates_per_sentence]), tf.int64),
sparse_values=tf.range(num_candidates, dtype=tf.int32),
default_value=0,
validate_indices=True) # [num_sentences, max_num_candidates]
spans_log_mask = tf.log(tf.to_float(candidate_mask)) # [num_sentences, max_num_candidates]
predict_dict = {"candidate_starts": candidate_starts, "candidate_ends": candidate_ends}
if head_scores is not None:
predict_dict["head_scores"] = head_scores
# Compute SRL representation.
flat_candidate_arg_scores = get_unary_scores(
candidate_span_emb, self.config, self.dropout, 1, "arg_scores") # [num_candidates,]
candidate_arg_scores = tf.gather(
flat_candidate_arg_scores, candidate_span_ids) + spans_log_mask # [num_sents, max_num_candidates]
# [num_sentences, max_num_args], ... [num_sentences,], [num_sentences, max_num_args]
arg_starts, arg_ends, arg_scores, num_args, top_arg_indices = get_batch_topk(
candidate_starts, candidate_ends, candidate_arg_scores, self.config["argument_ratio"], text_len,
max_sentence_length, sort_spans=False, enforce_non_crossing=False)
candidate_pred_ids = tf.tile(tf.expand_dims(tf.range(max_sentence_length), 0),
[num_sentences, 1]) # [num_sentences, max_sentence_length]
candidate_pred_emb = context_outputs # context_outputs [num_sentences, max_sentence_length, emb]
candidate_pred_scores = get_unary_scores(
candidate_pred_emb, self.config, self.dropout, 1, "pred_scores"
) + tf.log(tf.to_float(text_len_mask)) # [num_sentences, max_sentence_length]
#
if self.config["use_gold_predicates"]:
predicates = inputs["gold_predicates"]
num_preds = inputs["num_gold_predicates"]
pred_scores = tf.zeros_like(predicates, dtype=tf.float32)
top_pred_indices = predicates
else:
# [num_sentences, max_num_preds] ... [num_sentences,]
predicates, _, pred_scores, num_preds, top_pred_indices = get_batch_topk(
candidate_pred_ids, candidate_pred_ids, candidate_pred_scores, self.config["predicate_ratio"],
text_len, max_sentence_length, sort_spans=False, enforce_non_crossing=False)
arg_span_indices = batch_gather(candidate_span_ids, top_arg_indices) # [num_sentences, max_num_args]
arg_emb = tf.gather(candidate_span_emb, arg_span_indices) # [num_sentences, max_num_args, emb]
pred_emb = batch_gather(candidate_pred_emb, top_pred_indices) # [num_sentences, max_num_preds, emb]
max_num_args = util.shape(arg_scores, 1)
max_num_preds = util.shape(pred_scores, 1)
# Compute SRL loss.
srl_labels = get_srl_labels(
arg_starts, arg_ends, predicates, labels, max_sentence_length
) # [num_sentences, max_num_args, max_num_preds]
srl_scores = get_srl_scores(
arg_emb, pred_emb, arg_scores, pred_scores, len(self.data.srl_labels), self.config, self.dropout
) # [num_sentences, max_num_args, max_num_preds, num_labels]
srl_loss = get_srl_softmax_loss(
srl_scores, srl_labels, num_args, num_preds) # [num_sentences, max_num_args, max_num_preds]
#
# if self.config['srl_mode'] == 'dependency':
# pred_disamb_scores = get_unary_scores(
# pred_emb, self.config, self.dropout, len(self.data.srl_labels), "pred_disamb_scores"
# ) + tf.log(tf.to_float(text_len_mask)) # [num_sentences, max_sentence_length]
# pred_disamb_loss = get_softmax_loss(pred_disamb_scores, pred_disamb_labels)
# srl_loss += pred_disamb_loss
predict_dict.update({
"candidate_arg_scores": candidate_arg_scores,
"candidate_pred_scores": candidate_pred_scores,
"arg_starts": arg_starts,
"arg_ends": arg_ends,
"predicates": predicates,
"arg_scores": arg_scores, # New ...
"pred_scores": pred_scores,
"num_args": num_args,
"num_preds": num_preds,
"arg_labels": tf.argmax(srl_scores, -1), # [num_sentences, num_args, num_preds]
"srl_scores": srl_scores,
})
tf.summary.scalar("SRL_loss", srl_loss)
loss = srl_loss
return predict_dict, loss