def infer(args):
id2word_dict = reader.load_dict(args.word_dict_path)
word2id_dict = reader.load_reverse_dict(args.word_dict_path)
id2label_dict = reader.load_dict(args.label_dict_path)
label2id_dict = reader.load_reverse_dict(args.label_dict_path)
q2b_dict = reader.load_dict(args.word_rep_dict_path)
test_data = paddle.batch(reader.test_reader(args.test_data_dir,
word2id_dict, label2id_dict,
q2b_dict),
batch_size=args.batch_size)
place = fluid.CPUPlace()
exe = fluid.Executor(place)
inference_scope = fluid.core.Scope()
with fluid.scope_guard(inference_scope):
[inference_program, feed_target_names,
fetch_targets] = fluid.io.load_inference_model(args.model_path, exe)
for data in test_data():
full_out_str = ""
word_idx = to_lodtensor([x[0] for x in data], place)
word_list = [x[1] for x in data]
(crf_decode, ) = exe.run(inference_program,
feed={"word": word_idx},
fetch_list=fetch_targets,
return_numpy=False)
lod_info = (crf_decode.lod())[0]
np_data = np.array(crf_decode)
assert len(data) == len(lod_info) - 1
for sen_index in xrange(len(data)):
assert len(
data[sen_index][0]) == lod_info[sen_index +
1] - lod_info[sen_index]
word_index = 0
outstr = ""
cur_full_word = ""
cur_full_tag = ""
words = word_list[sen_index]
for tag_index in xrange(lod_info[sen_index],
lod_info[sen_index + 1]):
cur_word = words[word_index]
cur_tag = id2label_dict[str(np_data[tag_index][0])]
if cur_tag.endswith("-B") or cur_tag.endswith("O"):
if len(cur_full_word) != 0:
outstr += cur_full_word.encode(
'utf8') + "/" + cur_full_tag.encode(
'utf8') + " "
cur_full_word = cur_word
cur_full_tag = get_real_tag(cur_tag)
else:
cur_full_word += cur_word
word_index += 1
outstr += cur_full_word.encode(
'utf8') + "/" + cur_full_tag.encode('utf8') + " "
outstr = outstr.strip()
full_out_str += outstr + "\n"
print full_out_str.strip()
def train(args):
"""
Train the network.
"""
if not os.path.exists(args.model_save_dir):
os.mkdir(args.model_save_dir)
word2id_dict = reader.load_reverse_dict(args.word_dict_path)
label2id_dict = reader.load_reverse_dict(args.label_dict_path)
word_rep_dict = reader.load_dict(args.word_rep_dict_path)
word_dict_len = max(map(int, word2id_dict.values())) + 1
label_dict_len = max(map(int, label2id_dict.values())) + 1
avg_cost, crf_decode, word, target = lex_net(args, word_dict_len,
label_dict_len)
sgd_optimizer = fluid.optimizer.SGD(learning_rate=args.base_learning_rate)
sgd_optimizer.minimize(avg_cost)
(precision, recall, f1_score, num_infer_chunks, num_label_chunks,
num_correct_chunks) = fluid.layers.chunk_eval(
input=crf_decode,
label=target,
chunk_scheme="IOB",
num_chunk_types=int(math.ceil((label_dict_len - 1) / 2.0)))
chunk_evaluator = fluid.metrics.ChunkEvaluator()
chunk_evaluator.reset()
train_reader_list = []
corpus_num = len(args.corpus_type_list)
for i in xrange(corpus_num):
train_reader = paddle.batch(
paddle.reader.shuffle(reader.file_reader(args.traindata_dir,
word2id_dict,
label2id_dict,
word_rep_dict,
args.corpus_type_list[i]),
buf_size=args.traindata_shuffle_buffer),
batch_size=int(args.batch_size * args.corpus_proportion_list[i]))
train_reader_list.append(train_reader)
test_reader = paddle.batch(reader.file_reader(args.testdata_dir,
word2id_dict, label2id_dict,
word_rep_dict),
batch_size=args.batch_size)
train_reader_itr_list = []
for train_reader in train_reader_list:
cur_reader_itr = train_reader()
train_reader_itr_list.append(cur_reader_itr)
place = fluid.CUDAPlace(0) if args.use_gpu else fluid.CPUPlace()
feeder = fluid.DataFeeder(feed_list=[word, target], place=place)
exe = fluid.Executor(place)
exe.run(fluid.default_startup_program())
batch_id = 0
start_time = time.time()
eval_list = []
iter = 0
while True:
full_batch = []
cur_batch = []
for i in xrange(corpus_num):
reader_itr = train_reader_itr_list[i]
try:
cur_batch = next(reader_itr)
except StopIteration:
print(args.corpus_type_list[i] +
" corpus finish a pass of training")
new_reader = train_reader_list[i]
train_reader_itr_list[i] = new_reader()
cur_batch = next(train_reader_itr_list[i])
full_batch += cur_batch
random.shuffle(full_batch)
cost_var, nums_infer, nums_label, nums_correct = exe.run(
fluid.default_main_program(),
fetch_list=[
avg_cost, num_infer_chunks, num_label_chunks,
num_correct_chunks
],
feed=feeder.feed(full_batch))
print("batch_id:" + str(batch_id) + ", avg_cost:" + str(cost_var[0]))
chunk_evaluator.update(nums_infer, nums_label, nums_correct)
batch_id += 1
if (batch_id % args.save_model_per_batchs == 1):
save_exe = fluid.Executor(place)
save_dirname = os.path.join(args.model_save_dir,
"params_batch_%d" % batch_id)
fluid.io.save_inference_model(save_dirname, ['word'], [crf_decode],
save_exe)
temp_save_model = os.path.join(args.model_save_dir,
"temp_model_for_test")
fluid.io.save_inference_model(
temp_save_model, ['word', 'target'],
[num_infer_chunks, num_label_chunks, num_correct_chunks],
save_exe)
precision, recall, f1_score = chunk_evaluator.eval()
print("[train] batch_id:" + str(batch_id) + ", precision:" +
str(precision) + ", recall:" + str(recall) + ", f1:" +
str(f1_score))
chunk_evaluator.reset()
p, r, f1 = test(exe, chunk_evaluator, temp_save_model, test_reader,
place)
chunk_evaluator.reset()
print("[test] batch_id:" + str(batch_id) + ", precision:" +
str(p) + ", recall:" + str(r) + ", f1:" + str(f1))
end_time = time.time()
print("cur_batch_id:" + str(batch_id) + ", last " +
str(args.save_model_per_batchs) + " batchs, time_cost:" +
str(end_time - start_time))
start_time = time.time()
if len(eval_list) < 2 * args.eval_window:
eval_list.append(f1)
else:
eval_list.pop(0)
eval_list.append(f1)
last_avg_f1 = sum(
eval_list[0:args.eval_window]) / args.eval_window
cur_avg_f1 = sum(
eval_list[args.eval_window:2 *
args.eval_window]) / args.eval_window
if cur_avg_f1 <= last_avg_f1:
return
else:
print "keep training!"
iter += 1
if (iter == args.num_iterations):
return
results = []
# get out data from output tensor
output_names = predictor.get_output_names()
for i, name in enumerate(output_names):
output_tensor = predictor.get_output_tensor(name)
output_data = output_tensor.copy_to_cpu()
results.append(output_data)
return results
if __name__ == '__main__':
args = parse_args()
word2id_dict = reader.load_reverse_dict(args.word_dict_path)
label2id_dict = reader.load_reverse_dict(args.label_dict_path)
word_rep_dict = reader.load_dict(args.word_rep_dict_path)
word_dict_len = max(map(int, word2id_dict.values())) + 1
label_dict_len = max(map(int, label2id_dict.values())) + 1
pred = create_predictor(args)
test_data = paddle.batch(reader.file_reader(args.testdata_dir,
word2id_dict, label2id_dict,
word_rep_dict),
batch_size=1)
batch_id = 0
id2word = {v: k for k, v in word2id_dict.items()}
id2label = {v: k for k, v in label2id_dict.items()}
for data in test_data():
batch_id += 1
word_data, word_lod = to_lodtensor(list(map(lambda x: x[0], data)))
def train(train_data_path,
test_data_path,
src_dict_path,
trg_dict_path,
enc_conv_blocks,
dec_conv_blocks,
emb_dim=256,
pos_size=200,
drop_rate=0.,
use_bn=False,
batch_size=32,
num_passes=15):
"""
Train the convolution sequence-to-sequence model.
:param train_data_path: The path of the training set.
:type train_data_path: str
:param test_data_path: The path of the test set.
:type test_data_path: str
:param src_dict_path: The path of the source dictionary.
:type src_dict_path: str
:param trg_dict_path: The path of the target dictionary.
:type trg_dict_path: str
:param enc_conv_blocks: The scale list of the encoder's convolution blocks. And each element of
the list contains output dimension and context length of the corresponding
convolution block.
:type enc_conv_blocks: list of tuple
:param dec_conv_blocks: The scale list of the decoder's convolution blocks. And each element of
the list contains output dimension and context length of the corresponding
convolution block.
:type dec_conv_blocks: list of tuple
:param emb_dim: The dimension of the embedding vector.
:type emb_dim: int
:param pos_size: The total number of the position indexes, which means
the maximum value of the index is pos_size - 1.
:type pos_size: int
:param drop_rate: Dropout rate.
:type drop_rate: float
:param use_bn: Whether to use batch normalization or not. False is the default value.
:type use_bn: bool
:param batch_size: The size of a mini-batch.
:type batch_size: int
:param num_passes: The total number of the passes to train.
:type num_passes: int
"""
# load dict
src_dict = reader.load_dict(src_dict_path)
trg_dict = reader.load_dict(trg_dict_path)
src_dict_size = src_dict.__len__()
trg_dict_size = trg_dict.__len__()
optimizer = paddle.optimizer.Adam(learning_rate=1e-3, )
cost = conv_seq2seq(src_dict_size=src_dict_size,
trg_dict_size=trg_dict_size,
pos_size=pos_size,
emb_dim=emb_dim,
enc_conv_blocks=enc_conv_blocks,
dec_conv_blocks=dec_conv_blocks,
drop_rate=drop_rate,
with_bn=use_bn,
is_infer=False)
# create parameters and trainer
parameters = paddle.parameters.create(cost)
trainer = paddle.trainer.SGD(cost=cost,
parameters=parameters,
update_equation=optimizer)
padding_list = [context_len - 1 for (size, context_len) in dec_conv_blocks]
padding_num = reduce(lambda x, y: x + y, padding_list)
train_reader, test_reader = create_reader(padding_num=padding_num,
train_data_path=train_data_path,
test_data_path=test_data_path,
src_dict=src_dict,
trg_dict=trg_dict,
pos_size=pos_size,
batch_size=batch_size)
feeding = {
'src_word': 0,
'src_word_pos': 1,
'trg_word': 2,
'trg_word_pos': 3,
'trg_next_word': 4
}
# create event handler
def event_handler(event):
if isinstance(event, paddle.event.EndIteration):
if event.batch_id % 20 == 0:
cur_time = time.strftime('%Y.%m.%d %H:%M:%S', time.localtime())
print "[%s]: Pass: %d, Batch: %d, TrainCost: %f, %s" % (
cur_time, event.pass_id, event.batch_id, event.cost,
event.metrics)
sys.stdout.flush()
if isinstance(event, paddle.event.EndPass):
if test_reader is not None:
cur_time = time.strftime('%Y.%m.%d %H:%M:%S', time.localtime())
result = trainer.test(reader=test_reader, feeding=feeding)
print "[%s]: Pass: %d, TestCost: %f, %s" % (
cur_time, event.pass_id, result.cost, result.metrics)
sys.stdout.flush()
with gzip.open("output/params.pass-%d.tar.gz" % event.pass_id,
'w') as f:
trainer.save_parameter_to_tar(f)
if not os.path.exists('output'):
os.mkdir('output')
trainer.train(reader=train_reader,
event_handler=event_handler,
num_passes=num_passes,
feeding=feeding)
def infer(infer_data_path,
src_dict_path,
trg_dict_path,
model_path,
enc_conv_blocks,
dec_conv_blocks,
emb_dim=256,
pos_size=200,
drop_rate=0.,
use_bn=False,
max_len=100,
batch_size=1,
beam_size=1,
is_show_attention=False):
"""
Inference.
:param infer_data_path: The path of the data for inference.
:type infer_data_path: str
:param src_dict_path: The path of the source dictionary.
:type src_dict_path: str
:param trg_dict_path: The path of the target dictionary.
:type trg_dict_path: str
:param model_path: The path of a trained model.
:type model_path: str
:param enc_conv_blocks: The scale list of the encoder's convolution blocks. And each element of
the list contains output dimension and context length of the corresponding
convolution block.
:type enc_conv_blocks: list of tuple
:param dec_conv_blocks: The scale list of the decoder's convolution blocks. And each element of
the list contains output dimension and context length of the corresponding
convolution block.
:type dec_conv_blocks: list of tuple
:param emb_dim: The dimension of the embedding vector.
:type emb_dim: int
:param pos_size: The total number of the position indexes, which means
the maximum value of the index is pos_size - 1.
:type pos_size: int
:param drop_rate: Dropout rate.
:type drop_rate: float
:param use_bn: Whether to use batch normalization or not. False is the default value.
:type use_bn: bool
:param max_len: The maximum length of the sentence to be generated.
:type max_len: int
:param beam_size: The width of beam expansion.
:type beam_size: int
:param is_show_attention: Whether to show attention weight or not. False is the default value.
:type is_show_attention: bool
"""
# load dict
src_dict = reader.load_dict(src_dict_path)
trg_dict = reader.load_dict(trg_dict_path)
src_dict_size = src_dict.__len__()
trg_dict_size = trg_dict.__len__()
prob, weight = conv_seq2seq(src_dict_size=src_dict_size,
trg_dict_size=trg_dict_size,
pos_size=pos_size,
emb_dim=emb_dim,
enc_conv_blocks=enc_conv_blocks,
dec_conv_blocks=dec_conv_blocks,
drop_rate=drop_rate,
with_bn=use_bn,
is_infer=True)
# load parameters
parameters = paddle.parameters.Parameters.from_tar(gzip.open(model_path))
padding_list = [context_len - 1 for (size, context_len) in dec_conv_blocks]
padding_num = reduce(lambda x, y: x + y, padding_list)
infer_reader = reader.data_reader(data_file=infer_data_path,
src_dict=src_dict,
trg_dict=trg_dict,
pos_size=pos_size,
padding_num=padding_num)
if is_show_attention:
attention_inferer = paddle.inference.Inference(output_layer=weight,
parameters=parameters)
for i, data in enumerate(infer_reader()):
src_len = len(data[0])
trg_len = len(data[2])
attention_weight = attention_inferer.infer([data],
field='value',
flatten_result=False)
attention_weight = [
weight.reshape((trg_len, src_len))
for weight in attention_weight
]
print attention_weight
break
return
infer_data = []
for i, raw_data in enumerate(infer_reader()):
infer_data.append([raw_data[0], raw_data[1]])
inferer = paddle.inference.Inference(output_layer=prob,
parameters=parameters)
searcher = BeamSearch(inferer=inferer,
trg_dict=trg_dict,
pos_size=pos_size,
padding_num=padding_num,
max_len=max_len,
batch_size=batch_size,
beam_size=beam_size)
searcher.search(infer_data)
return