Upload tokenizer_data.json with huggingface_hub
Browse files- tokenizer_data.json +203 -0
tokenizer_data.json
ADDED
@@ -0,0 +1,203 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"vocab": {
|
3 |
+
"<PAD>": 0,
|
4 |
+
"<UNK>": 1,
|
5 |
+
"a1": 2,
|
6 |
+
"a2": 3,
|
7 |
+
"a3": 4,
|
8 |
+
"a4": 5,
|
9 |
+
"a5": 6,
|
10 |
+
"a6": 7,
|
11 |
+
"a7": 8,
|
12 |
+
"a8": 9,
|
13 |
+
"b1": 10,
|
14 |
+
"b2": 11,
|
15 |
+
"b3": 12,
|
16 |
+
"b4": 13,
|
17 |
+
"b5": 14,
|
18 |
+
"b6": 15,
|
19 |
+
"b7": 16,
|
20 |
+
"b8": 17,
|
21 |
+
"c1": 18,
|
22 |
+
"c2": 19,
|
23 |
+
"c3": 20,
|
24 |
+
"c4": 21,
|
25 |
+
"c5": 22,
|
26 |
+
"c6": 23,
|
27 |
+
"c7": 24,
|
28 |
+
"c8": 25,
|
29 |
+
"d1": 26,
|
30 |
+
"d2": 27,
|
31 |
+
"d3": 28,
|
32 |
+
"d4": 29,
|
33 |
+
"d5": 30,
|
34 |
+
"d6": 31,
|
35 |
+
"d7": 32,
|
36 |
+
"d8": 33,
|
37 |
+
"e1": 34,
|
38 |
+
"e2": 35,
|
39 |
+
"e3": 36,
|
40 |
+
"e4": 37,
|
41 |
+
"e5": 38,
|
42 |
+
"e6": 39,
|
43 |
+
"e7": 40,
|
44 |
+
"e8": 41,
|
45 |
+
"f1": 42,
|
46 |
+
"f2": 43,
|
47 |
+
"f3": 44,
|
48 |
+
"f4": 45,
|
49 |
+
"f5": 46,
|
50 |
+
"f6": 47,
|
51 |
+
"f7": 48,
|
52 |
+
"f8": 49,
|
53 |
+
"g1": 50,
|
54 |
+
"g2": 51,
|
55 |
+
"g3": 52,
|
56 |
+
"g4": 53,
|
57 |
+
"g5": 54,
|
58 |
+
"g6": 55,
|
59 |
+
"g7": 56,
|
60 |
+
"g8": 57,
|
61 |
+
"h1": 58,
|
62 |
+
"h2": 59,
|
63 |
+
"h3": 60,
|
64 |
+
"h4": 61,
|
65 |
+
"h5": 62,
|
66 |
+
"h6": 63,
|
67 |
+
"h7": 64,
|
68 |
+
"h8": 65,
|
69 |
+
"1": 66,
|
70 |
+
".": 67,
|
71 |
+
" ": 68,
|
72 |
+
"2": 69,
|
73 |
+
"N": 70,
|
74 |
+
"3": 71,
|
75 |
+
"c": 72,
|
76 |
+
"x": 73,
|
77 |
+
"4": 74,
|
78 |
+
"5": 75,
|
79 |
+
"B": 76,
|
80 |
+
"6": 77,
|
81 |
+
"7": 78,
|
82 |
+
"8": 79,
|
83 |
+
"b": 80,
|
84 |
+
"9": 81,
|
85 |
+
"0": 82,
|
86 |
+
"e": 83,
|
87 |
+
"O": 84,
|
88 |
+
"-": 85,
|
89 |
+
"Q": 86,
|
90 |
+
"R": 87,
|
91 |
+
"f": 88,
|
92 |
+
"a": 89,
|
93 |
+
"d": 90,
|
94 |
+
"K": 91,
|
95 |
+
"g": 92,
|
96 |
+
"+": 93,
|
97 |
+
"#": 94,
|
98 |
+
"h": 95,
|
99 |
+
"=": 96,
|
100 |
+
"/": 97
|
101 |
+
},
|
102 |
+
"ids_to_tokens": {
|
103 |
+
"0": "<PAD>",
|
104 |
+
"1": "<UNK>",
|
105 |
+
"2": "a1",
|
106 |
+
"3": "a2",
|
107 |
+
"4": "a3",
|
108 |
+
"5": "a4",
|
109 |
+
"6": "a5",
|
110 |
+
"7": "a6",
|
111 |
+
"8": "a7",
|
112 |
+
"9": "a8",
|
113 |
+
"10": "b1",
|
114 |
+
"11": "b2",
|
115 |
+
"12": "b3",
|
116 |
+
"13": "b4",
|
117 |
+
"14": "b5",
|
118 |
+
"15": "b6",
|
119 |
+
"16": "b7",
|
120 |
+
"17": "b8",
|
121 |
+
"18": "c1",
|
122 |
+
"19": "c2",
|
123 |
+
"20": "c3",
|
124 |
+
"21": "c4",
|
125 |
+
"22": "c5",
|
126 |
+
"23": "c6",
|
127 |
+
"24": "c7",
|
128 |
+
"25": "c8",
|
129 |
+
"26": "d1",
|
130 |
+
"27": "d2",
|
131 |
+
"28": "d3",
|
132 |
+
"29": "d4",
|
133 |
+
"30": "d5",
|
134 |
+
"31": "d6",
|
135 |
+
"32": "d7",
|
136 |
+
"33": "d8",
|
137 |
+
"34": "e1",
|
138 |
+
"35": "e2",
|
139 |
+
"36": "e3",
|
140 |
+
"37": "e4",
|
141 |
+
"38": "e5",
|
142 |
+
"39": "e6",
|
143 |
+
"40": "e7",
|
144 |
+
"41": "e8",
|
145 |
+
"42": "f1",
|
146 |
+
"43": "f2",
|
147 |
+
"44": "f3",
|
148 |
+
"45": "f4",
|
149 |
+
"46": "f5",
|
150 |
+
"47": "f6",
|
151 |
+
"48": "f7",
|
152 |
+
"49": "f8",
|
153 |
+
"50": "g1",
|
154 |
+
"51": "g2",
|
155 |
+
"52": "g3",
|
156 |
+
"53": "g4",
|
157 |
+
"54": "g5",
|
158 |
+
"55": "g6",
|
159 |
+
"56": "g7",
|
160 |
+
"57": "g8",
|
161 |
+
"58": "h1",
|
162 |
+
"59": "h2",
|
163 |
+
"60": "h3",
|
164 |
+
"61": "h4",
|
165 |
+
"62": "h5",
|
166 |
+
"63": "h6",
|
167 |
+
"64": "h7",
|
168 |
+
"65": "h8",
|
169 |
+
"66": "1",
|
170 |
+
"67": ".",
|
171 |
+
"68": " ",
|
172 |
+
"69": "2",
|
173 |
+
"70": "N",
|
174 |
+
"71": "3",
|
175 |
+
"72": "c",
|
176 |
+
"73": "x",
|
177 |
+
"74": "4",
|
178 |
+
"75": "5",
|
179 |
+
"76": "B",
|
180 |
+
"77": "6",
|
181 |
+
"78": "7",
|
182 |
+
"79": "8",
|
183 |
+
"80": "b",
|
184 |
+
"81": "9",
|
185 |
+
"82": "0",
|
186 |
+
"83": "e",
|
187 |
+
"84": "O",
|
188 |
+
"85": "-",
|
189 |
+
"86": "Q",
|
190 |
+
"87": "R",
|
191 |
+
"88": "f",
|
192 |
+
"89": "a",
|
193 |
+
"90": "d",
|
194 |
+
"91": "K",
|
195 |
+
"92": "g",
|
196 |
+
"93": "+",
|
197 |
+
"94": "#",
|
198 |
+
"95": "h",
|
199 |
+
"96": "=",
|
200 |
+
"97": "/"
|
201 |
+
},
|
202 |
+
"next_id": 98
|
203 |
+
}
|