Upload seamless_communication/cli/toxicity/etox.py with huggingface_hub
Browse files
seamless_communication/cli/toxicity/etox.py
ADDED
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Copyright (c) Meta Platforms, Inc. and affiliates
|
2 |
+
# All rights reserved.
|
3 |
+
#
|
4 |
+
# This source code is licensed under the license found in the
|
5 |
+
# MIT_LICENSE file in the root directory of this source tree.
|
6 |
+
|
7 |
+
import argparse
|
8 |
+
import sys
|
9 |
+
|
10 |
+
from seamless_communication.toxicity import load_etox_bad_word_checker
|
11 |
+
|
12 |
+
|
13 |
+
def main() -> None:
|
14 |
+
parser = argparse.ArgumentParser(
|
15 |
+
description="ETOX will compute the toxicity level of text inputs (STDIN > STDOUT)."
|
16 |
+
)
|
17 |
+
parser.add_argument(
|
18 |
+
"lang",
|
19 |
+
type=str,
|
20 |
+
help="Language, language of the speech to transcribe",
|
21 |
+
)
|
22 |
+
parser.add_argument(
|
23 |
+
"input", nargs="?", type=argparse.FileType("r"), default=sys.stdin
|
24 |
+
)
|
25 |
+
parser.add_argument(
|
26 |
+
"output", nargs="?", type=argparse.FileType("w"), default=sys.stdout
|
27 |
+
)
|
28 |
+
args, _unknown = parser.parse_known_args()
|
29 |
+
|
30 |
+
bad_word_checker = load_etox_bad_word_checker("mintox")
|
31 |
+
|
32 |
+
print("text", "toxicity", "bad_words", sep="\t", file=args.output)
|
33 |
+
for line in args.input:
|
34 |
+
l = line.rstrip()
|
35 |
+
bad_words = bad_word_checker.get_bad_words(
|
36 |
+
text=l,
|
37 |
+
lang=args.lang,
|
38 |
+
)
|
39 |
+
print(l, len(bad_words), ",".join(bad_words), sep="\t", file=args.output)
|
40 |
+
|
41 |
+
|
42 |
+
if __name__ == "__main__":
|
43 |
+
main()
|