Upload preprocessing_config.json with huggingface_hub
Browse files- preprocessing_config.json +90 -0
preprocessing_config.json
ADDED
|
@@ -0,0 +1,90 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"char_map": {
|
| 3 |
+
"0": "o",
|
| 4 |
+
"1": "i",
|
| 5 |
+
"3": "e",
|
| 6 |
+
"4": "a",
|
| 7 |
+
"5": "s",
|
| 8 |
+
"6": "g",
|
| 9 |
+
"7": "t",
|
| 10 |
+
"8": "b",
|
| 11 |
+
"9": "g",
|
| 12 |
+
"@": "a",
|
| 13 |
+
"$": "s",
|
| 14 |
+
"!": "i",
|
| 15 |
+
"#": "h",
|
| 16 |
+
"đ": "d",
|
| 17 |
+
"Đ": "d"
|
| 18 |
+
},
|
| 19 |
+
"obfuscation_patterns": {
|
| 20 |
+
"d[i1!][t7]": "dit",
|
| 21 |
+
"l[o0]n": "lon",
|
| 22 |
+
"c[a@4]c": "cac",
|
| 23 |
+
"c[u\\*]": "cu",
|
| 24 |
+
"d[e3]o": "deo",
|
| 25 |
+
"v[c\\*]l": "vcl",
|
| 26 |
+
"d[m\\*]": "dm",
|
| 27 |
+
"đ[m\\*]": "dm",
|
| 28 |
+
"đ[c\\*]m": "dcm",
|
| 29 |
+
"ch[o0]": "cho"
|
| 30 |
+
},
|
| 31 |
+
"nsfw_keywords": {
|
| 32 |
+
"địt": 100,
|
| 33 |
+
"đụ": 100,
|
| 34 |
+
"đéo": 100,
|
| 35 |
+
"lồn": 100,
|
| 36 |
+
"buồi": 100,
|
| 37 |
+
"cặc": 100,
|
| 38 |
+
"cu": 100,
|
| 39 |
+
"bú": 90,
|
| 40 |
+
"bím": 100,
|
| 41 |
+
"cứt": 80,
|
| 42 |
+
"chịch": 100,
|
| 43 |
+
"doggy": 90,
|
| 44 |
+
"blowjob": 100,
|
| 45 |
+
"địt mẹ": 100,
|
| 46 |
+
"đụ mẹ": 100,
|
| 47 |
+
"lồn mẹ": 100,
|
| 48 |
+
"dmm": 100,
|
| 49 |
+
"đmm": 100,
|
| 50 |
+
"dm": 90,
|
| 51 |
+
"đm": 90,
|
| 52 |
+
"vcl": 90,
|
| 53 |
+
"vkl": 90,
|
| 54 |
+
"dcm": 100,
|
| 55 |
+
"sex": 100,
|
| 56 |
+
"porn": 100,
|
| 57 |
+
"xxx": 100,
|
| 58 |
+
"nude": 100,
|
| 59 |
+
"đĩ": 100,
|
| 60 |
+
"điếm": 100,
|
| 61 |
+
"fuck": 100,
|
| 62 |
+
"shit": 80
|
| 63 |
+
},
|
| 64 |
+
"nsfw_patterns": [
|
| 65 |
+
[
|
| 66 |
+
"đ[i1!][t7]",
|
| 67 |
+
100
|
| 68 |
+
],
|
| 69 |
+
[
|
| 70 |
+
"l[o0]n",
|
| 71 |
+
100
|
| 72 |
+
],
|
| 73 |
+
[
|
| 74 |
+
"c[ặa@][c]",
|
| 75 |
+
100
|
| 76 |
+
],
|
| 77 |
+
[
|
| 78 |
+
"[cđ][u\\*]",
|
| 79 |
+
90
|
| 80 |
+
],
|
| 81 |
+
[
|
| 82 |
+
"v[c\\*]l",
|
| 83 |
+
90
|
| 84 |
+
],
|
| 85 |
+
[
|
| 86 |
+
"đ[mM\\*]+",
|
| 87 |
+
90
|
| 88 |
+
]
|
| 89 |
+
]
|
| 90 |
+
}
|