def __init__(self,
news_config,
ckpt_fn,
thread_num=1,
input_queue=None,
output_queue=None,
batch_size=1,
job_name="ez_transfer_job"):
super(PredictProcess, self).__init__(job_name,
thread_num,
input_queue=input_queue,
output_queue=output_queue,
batch_size=batch_size)
self.graph = tf.Graph()
self.gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.9,
allow_growth=False)
self.session_conf = tf.ConfigProto(intra_op_parallelism_threads=8,
inter_op_parallelism_threads=8,
allow_soft_placement=True,
gpu_options=gpu_options)
with self.graph.as_default():
self.sess = tf.Session(config=self.session_conf)
self.initial_context = tf.placeholder(tf.int32, [batch_size, None])
self.p_for_topp = tf.placeholder(tf.float32, [batch_size])
self.eos_token = tf.placeholder(tf.int32, [])
self.min_len = tf.placeholder(tf.int32, [])
self.max_len = tf.placeholder(tf.int32, [])
self.k_for_topk = tf.placeholder(tf.int32, [])
self.tokens, self.probs = sample(
news_config=news_config,
initial_context=self.initial_context,
eos_token=self.eos_token,
min_len=self.min_len,
max_len=self.max_len,
ignore_ids=None,
p_for_topp=self.p_for_topp,
k_for_topk=self.k_for_topk,
do_topk=False)
self.saver = tf.train.Saver()
self.saver.restore(self.sess, ckpt_fn)
self.input_dict = {
"initial_context": self.initial_context,
"p_for_topp": self.p_for_topp,
"eos_token": self.eos_token,
"min_len": self.min_len,
"max_len": self.max_len,
"k_for_topk": self.k_for_topk
}
self.predictions = {"tokens": self.tokens, "probs": self.probs}
# This controls the top p for each generation.
top_p = np.ones(
(num_chunks, batch_size_per_chunk), dtype=np.float32) * args.top_p
tf_config = tf.ConfigProto(allow_soft_placement=True)
with tf.Session(config=tf_config, graph=tf.Graph()) as sess:
initial_context = tf.placeholder(tf.int32, [batch_size_per_chunk, None])
p_for_topp = tf.placeholder(tf.float32, [batch_size_per_chunk])
eos_token = tf.placeholder(tf.int32, [])
min_len = tf.placeholder(tf.int32, [])
tokens, probs = sample(news_config=news_config,
initial_context=initial_context,
eos_token=eos_token,
min_len=min_len,
ignore_ids=None,
p_for_topp=p_for_topp,
do_topk=False)
saver = tf.train.Saver()
saver.restore(sess, args.ckpt_fn)
print('🍺Model loaded. \nInput something please:⬇️')
text = input()
while text != "":
for i in range(args.samples):
print("Sample,", i + 1, " of ", args.samples)
line = tokenization.convert_to_unicode(text)
bert_tokens = tokenizer.tokenize(line)
encoded = tokenizer.convert_tokens_to_ids(bert_tokens)
context_formatted = []
flush=True)
# This controls the top p for each generation.
top_p = np.ones(
(num_chunks, batch_size_per_chunk), dtype=np.float32) * args.top_p
tf_config = tf.ConfigProto(allow_soft_placement=True)
with tf.Session(config=tf_config, graph=tf.Graph()) as sess:
initial_context = tf.placeholder(tf.int32, [batch_size_per_chunk, None])
p_for_topp = tf.placeholder(tf.float32, [batch_size_per_chunk])
eos_token = tf.placeholder(tf.int32, [])
tokens, probs = sample(news_config=news_config,
initial_context=initial_context,
eos_token=eos_token,
ignore_ids=None,
p_for_topp=p_for_topp,
do_topk=False,
sentence_length=args.length)
saver = tf.train.Saver()
saver.restore(sess, args.model_ckpt)
while True:
text = input('🍺Model[{}] loaded. Input something please:\n'.format(
args.model_ckpt))
if text == "":
break
# init step
_start_time = time.time()
def predict():
##### ignore tf deprecated warning temporarily
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
# mac-specific settings, comment this when exec in other systems
os.environ['KMP_DUPLICATE_LIB_OK']='True'
tf.compat.v1.logging.set_verbosity(tf.compat.v1.logging.DEBUG)
from tensorflow.python.util import deprecation
deprecation._PRINT_DEPRECATION_WARNINGS = False
try:
from tensorflow.python.util import module_wrapper as deprecation
except ImportError:
from tensorflow.python.util import deprecation_wrapper as deprecation
deprecation._PER_MODULE_WARNING_LIMIT = 0
#####
parser = argparse.ArgumentParser(description='Contextual generation (aka given some metadata we will generate articles')
parser.add_argument(
'-metadata_fn',
dest='metadata_fn',
type=str,
help='Path to a JSONL containing metadata',
)
parser.add_argument(
'-out_fn',
dest='out_fn',
type=str,
help='Out jsonl, which will contain the completed jsons',
)
parser.add_argument(
'-input',
dest='input',
type=str,
help='Text to complete',
)
parser.add_argument(
'-model_config_fn',
dest='model_config_fn',
default='configs/mega.json',
type=str,
help='Configuration JSON for the model',
)
parser.add_argument(
'-model_ckpt',
dest='model_ckpt',
default='model.ckpt-220000',
type=str,
help='checkpoint file for the model',
)
parser.add_argument(
'-target',
dest='target',
default='article',
type=str,
help='What to generate for each item in metadata_fn. can be article (body), title, etc.',
)
parser.add_argument(
'-batch_size',
dest='batch_size',
default=1,
type=int,
help='How many things to generate per context. will split into chunks if need be',
)
parser.add_argument(
'-num_folds',
dest='num_folds',
default=1,
type=int,
help='Number of folds. useful if we want to split up a big file into multiple jobs.',
)
parser.add_argument(
'-fold',
dest='fold',
default=0,
type=int,
help='which fold we are on. useful if we want to split up a big file into multiple jobs.'
)
parser.add_argument(
'-max_batch_size',
dest='max_batch_size',
default=None,
type=int,
help='max batch size. You can leave this out and we will infer one based on the number of hidden layers',
)
parser.add_argument(
'-top_p',
dest='top_p',
default=0.95,
type=float,
help='p to use for top p sampling. if this isn\'t none, use this for everthing'
)
parser.add_argument(
'-min_len',
dest='min_len',
default=1024,
type=int,
help='min length of sample',
)
parser.add_argument(
'-eos_token',
dest='eos_token',
default=60000,
type=int,
help='eos token id',
)
parser.add_argument(
'-samples',
dest='samples',
default=5,
type=int,
help='num_samples',
)
def extract_generated_target(output_tokens, tokenizer):
"""
Given some tokens that were generated, extract the target
:param output_tokens: [num_tokens] thing that was generated
:param encoder: how they were encoded
:param target: the piece of metadata we wanted to generate!
:return:
"""
# Filter out first instance of start token
assert output_tokens.ndim == 1
start_ind = 0
end_ind = output_tokens.shape[0]
return {
'extraction': tokenization.printable_text(''.join(tokenizer.convert_ids_to_tokens(output_tokens))),
'start_ind': start_ind,
'end_ind': end_ind,
}
# args = parser.parse_args()
args, unknown = parser.parse_known_args()
proj_root_path = os.path.dirname(os.path.realpath(__file__))
vocab_file_path = os.path.join(proj_root_path, "tokenization/clue-vocab.txt")
tokenizer = tokenization.FullTokenizer(vocab_file=vocab_file_path , do_lower_case=True)
news_config = GroverConfig.from_json_file(args.model_config_fn)
# We might have to split the batch into multiple chunks if the batch size is too large
default_mbs = {12: 32, 24: 16, 48: 3}
max_batch_size = args.max_batch_size if args.max_batch_size is not None else default_mbs[news_config.num_hidden_layers]
# factorize args.batch_size = (num_chunks * batch_size_per_chunk) s.t. batch_size_per_chunk < max_batch_size
num_chunks = int(np.ceil(args.batch_size / max_batch_size))
batch_size_per_chunk = int(np.ceil(args.batch_size / num_chunks))
# This controls the top p for each generation.
top_p = np.ones((num_chunks, batch_size_per_chunk), dtype=np.float32) * args.top_p
tf_config = tf.ConfigProto(allow_soft_placement=True)
with tf.Session(config=tf_config, graph=tf.Graph()) as sess:
initial_context = tf.placeholder(tf.int32, [batch_size_per_chunk, None])
p_for_topp = tf.placeholder(tf.float32, [batch_size_per_chunk])
eos_token = tf.placeholder(tf.int32, [])
min_len = tf.placeholder(tf.int32, [])
tokens, probs = sample(news_config=news_config, initial_context=initial_context,
eos_token=eos_token, min_len=min_len, ignore_ids=None, p_for_topp=p_for_topp,
do_topk=False)
saver = tf.train.Saver()
saver.restore(sess, args.model_ckpt)
'''
如果部署到web上，则所有的print都不需要
input改为web返回的message
不需要while循环
将最后的"\n".join(l) 返回到一个参数，并展示到web中
主要参数（篇数、长度）要用户在web中输入，或者在本代码里写死 -- 有默认值
待解决：
sample有5个，下面代码会for循环分别predict 5次，这5次结果要怎么在网页展示？
min_lens没有用，比如1024的时候还是会生产一两百字的文章
'''
# print('🍺Model loaded. \nInput something please:⬇️')
if request.method == 'POST':
text = request.form['message']
# data = [text] 原spam detection里的代码，不确定此处是否需要
for i in range(args.samples):
# print("Sample,", i + 1, " of ", args.samples)
line = tokenization.convert_to_unicode(text)
bert_tokens = tokenizer.tokenize(line)
encoded = tokenizer.convert_tokens_to_ids(bert_tokens)
context_formatted = []
context_formatted.extend(encoded)
# Format context end
gens = []
gens_raw = []
gen_probs = []
final_result = []
for chunk_i in range(num_chunks):
tokens_out, probs_out = sess.run([tokens, probs],
feed_dict={initial_context: [context_formatted] * batch_size_per_chunk,
eos_token: args.eos_token, min_len: args.min_len,
p_for_topp: top_p[chunk_i]})
for t_i, p_i in zip(tokens_out, probs_out):
extraction = extract_generated_target(output_tokens=t_i, tokenizer=tokenizer)
gens.append(extraction['extraction'])
l = re.findall('.{1,70}', gens[0].replace('[UNK]', '').replace('##', ''))
# 下一句的参应该传给 return
# print("\n".join(l))
# return a for loop
# https://stackoverflow.com/questions/44564414/how-to-use-a-return-statement-in-a-for-loop
final_result.append("\n".join(l))
return render_template('result.html',prediction = final_result)
