NLP-Tutorials/utils.py at master · MorvanZhou/NLP-Tutorials · GitHub

History

258 lines (218 loc) · 9.03 KB

Raw

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

import numpy as np

import datetime

import os

import requests

import pandas as pd

import re

import itertools

PAD_ID = 0

class DateData:

def __init__(self, n):

np.random.seed(1)

self.date_cn = []

self.date_en = []

for timestamp in np.random.randint(143835585, 2043835585, n):

date = datetime.datetime.fromtimestamp(timestamp)

self.date_cn.append(date.strftime("%y-%m-%d"))

self.date_en.append(date.strftime("%d/%b/%Y"))

self.vocab = set(

[str(i) for i in range(0, 10)] + ["-", "/", "<GO>", "<EOS>"] + [

i.split("/")[1] for i in self.date_en])

self.v2i = {v: i for i, v in enumerate(sorted(list(self.vocab)), start=1)}

self.v2i["<PAD>"] = PAD_ID

self.vocab.add("<PAD>")

self.i2v = {i: v for v, i in self.v2i.items()}

self.x, self.y = [], []

for cn, en in zip(self.date_cn, self.date_en):

self.x.append([self.v2i[v] for v in cn])

self.y.append(

[self.v2i["<GO>"], ] + [self.v2i[v] for v in en[:3]] + [

self.v2i[en[3:6]], ] + [self.v2i[v] for v in en[6:]] + [

self.v2i["<EOS>"], ])

self.x, self.y = np.array(self.x), np.array(self.y)

self.start_token = self.v2i["<GO>"]

self.end_token = self.v2i["<EOS>"]

def sample(self, n=64):

bi = np.random.randint(0, len(self.x), size=n)

bx, by = self.x[bi], self.y[bi]

decoder_len = np.full((len(bx),), by.shape[1] - 1, dtype=np.int32)

return bx, by, decoder_len

def idx2str(self, idx):

x = []

for i in idx:

x.append(self.i2v[i])

if i == self.end_token:

break

return "".join(x)

@property

def num_word(self):

return len(self.vocab)

def pad_zero(seqs, max_len):

padded = np.full((len(seqs), max_len), fill_value=PAD_ID, dtype=np.long)

for i, seq in enumerate(seqs):

padded[i, :len(seq)] = seq

return padded

def maybe_download_mrpc(save_dir="./MRPC/", proxy=None):

train_url = 'https://mofanpy.com/static/files/MRPC/msr_paraphrase_train.txt'

test_url = 'https://mofanpy.com/static/files/MRPC/msr_paraphrase_test.txt'

os.makedirs(save_dir, exist_ok=True)

proxies = {"http": proxy, "https": proxy}

for url in [train_url, test_url]:

raw_path = os.path.join(save_dir, url.split("/")[-1])

if not os.path.isfile(raw_path):

print("downloading from %s" % url)

r = requests.get(url, proxies=proxies)

with open(raw_path, "w", encoding="utf-8") as f:

f.write(r.text.replace('"', "<QUOTE>"))

print("completed")

def _text_standardize(text):

text = re.sub(r'—', '-', text)

text = re.sub(r'–', '-', text)

text = re.sub(r'―', '-', text)

text = re.sub(r" \d+(,\d+)?(\.\d+)? ", " <NUM> ", text)

text = re.sub(r" \d+-+?\d*", " <NUM>-", text)

return text.strip()

def _process_mrpc(dir="./MRPC", rows=None):

data = {"train": None, "test": None}

files = os.listdir(dir)

for f in files:

df = pd.read_csv(os.path.join(dir, f), sep='\t', nrows=rows)

k = "train" if "train" in f else "test"

data[k] = {"is_same": df.iloc[:, 0].values, "s1": df["#1 String"].values, "s2": df["#2 String"].values}

vocab = set()

for n in ["train", "test"]:

for m in ["s1", "s2"]:

for i in range(len(data[n][m])):

data[n][m][i] = _text_standardize(data[n][m][i].lower())

cs = data[n][m][i].split(" ")

vocab.update(set(cs))

v2i = {v: i for i, v in enumerate(sorted(vocab), start=1)}

v2i["<PAD>"] = PAD_ID

v2i["<MASK>"] = len(v2i)

v2i["<SEP>"] = len(v2i)

v2i["<GO>"] = len(v2i)

i2v = {i: v for v, i in v2i.items()}

for n in ["train", "test"]:

for m in ["s1", "s2"]:

data[n][m+"id"] = [[v2i[v] for v in c.split(" ")] for c in data[n][m]]

return data, v2i, i2v

class MRPCData:

num_seg = 3

pad_id = PAD_ID

def __init__(self, data_dir="./MRPC/", rows=None, proxy=None):

maybe_download_mrpc(save_dir=data_dir, proxy=proxy)

data, self.v2i, self.i2v = _process_mrpc(data_dir, rows)

self.max_len = max(

[len(s1) + len(s2) + 3 for s1, s2 in zip(

data["train"]["s1id"] + data["test"]["s1id"], data["train"]["s2id"] + data["test"]["s2id"])])

self.xlen = np.array([

[

len(data["train"]["s1id"][i]), len(data["train"]["s2id"][i])

] for i in range(len(data["train"]["s1id"]))], dtype=int)

x = [

[self.v2i["<GO>"]] + data["train"]["s1id"][i] + [self.v2i["<SEP>"]] + data["train"]["s2id"][i] + [self.v2i["<SEP>"]]

for i in range(len(self.xlen))

]

self.x = pad_zero(x, max_len=self.max_len)

self.nsp_y = data["train"]["is_same"][:, None]

self.seg = np.full(self.x.shape, self.num_seg-1, np.int32)

for i in range(len(x)):

si = self.xlen[i][0] + 2

self.seg[i, :si] = 0

si_ = si + self.xlen[i][1] + 1

self.seg[i, si:si_] = 1

self.word_ids = np.array(list(set(self.i2v.keys()).difference(

[self.v2i[v] for v in ["<PAD>", "<MASK>", "<SEP>"]])))

def sample(self, n):

bi = np.random.randint(0, self.x.shape[0], size=n)

bx, bs, bl, by = self.x[bi], self.seg[bi], self.xlen[bi], self.nsp_y[bi]

return bx, bs, bl, by

@property

def num_word(self):

return len(self.v2i)

@property

def mask_id(self):

return self.v2i["<MASK>"]

class MRPCSingle:

pad_id = PAD_ID

def __init__(self, data_dir="./MRPC/", rows=None, proxy=None):

maybe_download_mrpc(save_dir=data_dir, proxy=proxy)

data, self.v2i, self.i2v = _process_mrpc(data_dir, rows)

self.max_len = max([len(s) + 2 for s in data["train"]["s1id"] + data["train"]["s2id"]])

x = [

[self.v2i["<GO>"]] + data["train"]["s1id"][i] + [self.v2i["<SEP>"]]

for i in range(len(data["train"]["s1id"]))

]

x += [

[self.v2i["<GO>"]] + data["train"]["s2id"][i] + [self.v2i["<SEP>"]]

for i in range(len(data["train"]["s2id"]))

]

self.x = pad_zero(x, max_len=self.max_len)

self.word_ids = np.array(list(set(self.i2v.keys()).difference([self.v2i["<PAD>"]])))

def sample(self, n):

bi = np.random.randint(0, self.x.shape[0], size=n)

bx = self.x[bi]

return bx

@property

def num_word(self):

return len(self.v2i)

class Dataset:

def __init__(self, x, y, v2i, i2v):

self.x, self.y = x, y

self.v2i, self.i2v = v2i, i2v

self.vocab = v2i.keys()

def sample(self, n):

b_idx = np.random.randint(0, len(self.x), n)

bx, by = self.x[b_idx], self.y[b_idx]

return bx, by

@property

def num_word(self):

return len(self.v2i)

def process_w2v_data(corpus, skip_window=2, method="skip_gram"):

all_words = [sentence.split(" ") for sentence in corpus]

all_words = np.array(list(itertools.chain(*all_words)))

# vocab sort by decreasing frequency for the negative sampling below (nce_loss).

vocab, v_count = np.unique(all_words, return_counts=True)

vocab = vocab[np.argsort(v_count)[::-1]]

print("all vocabularies sorted from more frequent to less frequent:\n", vocab)

v2i = {v: i for i, v in enumerate(vocab)}

i2v = {i: v for v, i in v2i.items()}

# pair data

pairs = []

js = [i for i in range(-skip_window, skip_window + 1) if i != 0]

for c in corpus:

words = c.split(" ")

w_idx = [v2i[w] for w in words]

if method == "skip_gram":

for i in range(len(w_idx)):

for j in js:

if i + j < 0 or i + j >= len(w_idx):

continue

pairs.append((w_idx[i], w_idx[i + j])) # (center, context) or (feature, target)

elif method.lower() == "cbow":

for i in range(skip_window, len(w_idx) - skip_window):

context = []

for j in js:

context.append(w_idx[i + j])

pairs.append(context + [w_idx[i]]) # (contexts, center) or (feature, target)

else:

raise ValueError

pairs = np.array(pairs)

print("5 example pairs:\n", pairs[:5])

if method.lower() == "skip_gram":

x, y = pairs[:, 0], pairs[:, 1]

elif method.lower() == "cbow":

x, y = pairs[:, :-1], pairs[:, -1]

else:

raise ValueError

return Dataset(x, y, v2i, i2v)

def set_soft_gpu(soft_gpu):

import tensorflow as tf

if soft_gpu:

gpus = tf.config.experimental.list_physical_devices('GPU')

if gpus:

# Currently, memory growth needs to be the same across GPUs

for gpu in gpus:

tf.config.experimental.set_memory_growth(gpu, True)

logical_gpus = tf.config.experimental.list_logical_devices('GPU')

print(len(gpus), "Physical GPUs,", len(logical_gpus), "Logical GPUs")

NLP-Tutorials/utils.py at master · MorvanZhou/NLP-Tutorials · GitHub

Expand file tree

Breadcrumbs

utils.py

Latest commit

History

Breadcrumbs

utils.py

File metadata and controls