Spaces:
Runtime error
Runtime error
Replicate default cc_net preprocessing at inference time on KenlmModel.get_perplexity
Browse files
perplexity_lenses/perplexity.py
CHANGED
|
@@ -1,10 +1,53 @@
|
|
| 1 |
import os
|
|
|
|
|
|
|
| 2 |
import urllib.request
|
|
|
|
| 3 |
|
| 4 |
import kenlm
|
| 5 |
|
| 6 |
|
| 7 |
class KenlmModel:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
def __init__(self, language):
|
| 9 |
download_kenlm_model(language)
|
| 10 |
try:
|
|
@@ -19,7 +62,9 @@ class KenlmModel:
|
|
| 19 |
def from_pretrained(cls, language: str):
|
| 20 |
return cls(language)
|
| 21 |
|
| 22 |
-
def get_perplexity(self, doc: str):
|
|
|
|
|
|
|
| 23 |
doc_log_score, doc_length = 0, 0
|
| 24 |
for line in doc.split("\n"):
|
| 25 |
log_score = self.model.score(line)
|
|
@@ -28,6 +73,48 @@ class KenlmModel:
|
|
| 28 |
doc_length += length
|
| 29 |
return 10.0 ** (-doc_log_score / doc_length)
|
| 30 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
|
| 32 |
def download_kenlm_model(language: str):
|
| 33 |
root_url = "http://dl.fbaipublicfiles.com/cc_net/lm"
|
|
|
|
| 1 |
import os
|
| 2 |
+
import re
|
| 3 |
+
import unicodedata
|
| 4 |
import urllib.request
|
| 5 |
+
from typing import Dict
|
| 6 |
|
| 7 |
import kenlm
|
| 8 |
|
| 9 |
|
| 10 |
class KenlmModel:
|
| 11 |
+
digit_re: re.Pattern = re.compile(r"\d")
|
| 12 |
+
unicode_punct: Dict[str, str] = {
|
| 13 |
+
",": ",",
|
| 14 |
+
"。": ".",
|
| 15 |
+
"、": ",",
|
| 16 |
+
"„": '"',
|
| 17 |
+
"”": '"',
|
| 18 |
+
"“": '"',
|
| 19 |
+
"«": '"',
|
| 20 |
+
"»": '"',
|
| 21 |
+
"1": '"',
|
| 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 |
+
unicode_punct_re = re.compile(f"[{''.join(unicode_punct.keys())}]")
|
| 49 |
+
non_printing_chars_re = re.compile(f"[{''.join(map(chr, list(range(0,32)) + list(range(127,160))))}]")
|
| 50 |
+
|
| 51 |
def __init__(self, language):
|
| 52 |
download_kenlm_model(language)
|
| 53 |
try:
|
|
|
|
| 62 |
def from_pretrained(cls, language: str):
|
| 63 |
return cls(language)
|
| 64 |
|
| 65 |
+
def get_perplexity(self, doc: str, normalize_cc_net: bool = True):
|
| 66 |
+
if normalize_cc_net:
|
| 67 |
+
doc = self.normalize(doc)
|
| 68 |
doc_log_score, doc_length = 0, 0
|
| 69 |
for line in doc.split("\n"):
|
| 70 |
log_score = self.model.score(line)
|
|
|
|
| 73 |
doc_length += length
|
| 74 |
return 10.0 ** (-doc_log_score / doc_length)
|
| 75 |
|
| 76 |
+
def normalize(
|
| 77 |
+
self,
|
| 78 |
+
line: str,
|
| 79 |
+
accent: bool = True,
|
| 80 |
+
case: bool = True,
|
| 81 |
+
numbers: bool = True,
|
| 82 |
+
punct: int = 1,
|
| 83 |
+
) -> str:
|
| 84 |
+
line = line.strip()
|
| 85 |
+
if not line:
|
| 86 |
+
return line
|
| 87 |
+
if case:
|
| 88 |
+
line = line.lower()
|
| 89 |
+
if accent:
|
| 90 |
+
line = self.strip_accents(line)
|
| 91 |
+
if numbers:
|
| 92 |
+
line = self.digit_re.sub("0", line)
|
| 93 |
+
if punct == 1:
|
| 94 |
+
line = self.replace_unicode_punct(line)
|
| 95 |
+
elif punct == 2:
|
| 96 |
+
line = self.remove_unicode_punct(line)
|
| 97 |
+
line = self.remove_non_printing_char(line)
|
| 98 |
+
return line
|
| 99 |
+
|
| 100 |
+
def strip_accents(self, line: str) -> str:
|
| 101 |
+
"""Strips accents from a piece of text."""
|
| 102 |
+
nfd = unicodedata.normalize("NFD", line)
|
| 103 |
+
output = [c for c in nfd if unicodedata.category(c) != "Mn"]
|
| 104 |
+
if len(output) == line:
|
| 105 |
+
return line
|
| 106 |
+
return "".join(output)
|
| 107 |
+
|
| 108 |
+
def replace_unicode_punct(self, text: str) -> str:
|
| 109 |
+
return "".join((self.unicode_punct.get(c, c) for c in text))
|
| 110 |
+
|
| 111 |
+
def remove_unicode_punct(self, text: str) -> str:
|
| 112 |
+
"""More aggressive version of replace_unicode_punct but also faster."""
|
| 113 |
+
return self.unicode_punct_re.sub("", text)
|
| 114 |
+
|
| 115 |
+
def remove_non_printing_char(self, text: str) -> str:
|
| 116 |
+
return self.non_printing_chars_re.sub("", text)
|
| 117 |
+
|
| 118 |
|
| 119 |
def download_kenlm_model(language: str):
|
| 120 |
root_url = "http://dl.fbaipublicfiles.com/cc_net/lm"
|