Commit
·
d775864
1
Parent(s):
1bbd033
add tokenizer
Browse files- tokenizer.json +24 -4
- vocab.txt +21 -1
tokenizer.json
CHANGED
|
@@ -85,9 +85,9 @@
|
|
| 85 |
"G": 7,
|
| 86 |
"T": 8,
|
| 87 |
"##A": 9,
|
| 88 |
-
"##
|
| 89 |
-
"##
|
| 90 |
-
"##
|
| 91 |
"AA": 13,
|
| 92 |
"GA": 14,
|
| 93 |
"TT": 15,
|
|
@@ -144,7 +144,27 @@
|
|
| 144 |
"CTT": 66,
|
| 145 |
"GTA": 67,
|
| 146 |
"GTC": 68,
|
| 147 |
-
"GGA": 69
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 148 |
}
|
| 149 |
}
|
| 150 |
}
|
|
|
|
| 85 |
"G": 7,
|
| 86 |
"T": 8,
|
| 87 |
"##A": 9,
|
| 88 |
+
"##C": 10,
|
| 89 |
+
"##T": 11,
|
| 90 |
+
"##G": 12,
|
| 91 |
"AA": 13,
|
| 92 |
"GA": 14,
|
| 93 |
"TT": 15,
|
|
|
|
| 144 |
"CTT": 66,
|
| 145 |
"GTA": 67,
|
| 146 |
"GTC": 68,
|
| 147 |
+
"GGA": 69,
|
| 148 |
+
"GTG": 70,
|
| 149 |
+
"CTG": 71,
|
| 150 |
+
"TGG": 72,
|
| 151 |
+
"AGC": 73,
|
| 152 |
+
"GGC": 74,
|
| 153 |
+
"AGG": 75,
|
| 154 |
+
"TCG": 76,
|
| 155 |
+
"ACG": 77,
|
| 156 |
+
"TGT": 78,
|
| 157 |
+
"CAC": 79,
|
| 158 |
+
"CCC": 80,
|
| 159 |
+
"CGT": 81,
|
| 160 |
+
"GCG": 82,
|
| 161 |
+
"GGG": 83,
|
| 162 |
+
"CCG": 84,
|
| 163 |
+
"CTC": 85,
|
| 164 |
+
"TGC": 86,
|
| 165 |
+
"CGA": 87,
|
| 166 |
+
"CGC": 88,
|
| 167 |
+
"CGG": 89
|
| 168 |
}
|
| 169 |
}
|
| 170 |
}
|
vocab.txt
CHANGED
|
@@ -8,9 +8,9 @@ C
|
|
| 8 |
G
|
| 9 |
T
|
| 10 |
##A
|
|
|
|
| 11 |
##T
|
| 12 |
##G
|
| 13 |
-
##C
|
| 14 |
AA
|
| 15 |
GA
|
| 16 |
TT
|
|
@@ -68,3 +68,23 @@ CTT
|
|
| 68 |
GTA
|
| 69 |
GTC
|
| 70 |
GGA
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
G
|
| 9 |
T
|
| 10 |
##A
|
| 11 |
+
##C
|
| 12 |
##T
|
| 13 |
##G
|
|
|
|
| 14 |
AA
|
| 15 |
GA
|
| 16 |
TT
|
|
|
|
| 68 |
GTA
|
| 69 |
GTC
|
| 70 |
GGA
|
| 71 |
+
GTG
|
| 72 |
+
CTG
|
| 73 |
+
TGG
|
| 74 |
+
AGC
|
| 75 |
+
GGC
|
| 76 |
+
AGG
|
| 77 |
+
TCG
|
| 78 |
+
ACG
|
| 79 |
+
TGT
|
| 80 |
+
CAC
|
| 81 |
+
CCC
|
| 82 |
+
CGT
|
| 83 |
+
GCG
|
| 84 |
+
GGG
|
| 85 |
+
CCG
|
| 86 |
+
CTC
|
| 87 |
+
TGC
|
| 88 |
+
CGA
|
| 89 |
+
CGC
|
| 90 |
+
CGG
|