Commit
·
1051438
1
Parent(s):
4ebf1d8
add tokenizer
Browse files- tokenizer.json +3 -3
- vocab.txt +2 -2
tokenizer.json
CHANGED
|
@@ -84,10 +84,10 @@
|
|
| 84 |
"C": 6,
|
| 85 |
"G": 7,
|
| 86 |
"T": 8,
|
| 87 |
-
"##
|
| 88 |
-
"##
|
| 89 |
"##C": 11,
|
| 90 |
-
"##
|
| 91 |
"AA": 13,
|
| 92 |
"GA": 14,
|
| 93 |
"TT": 15,
|
|
|
|
| 84 |
"C": 6,
|
| 85 |
"G": 7,
|
| 86 |
"T": 8,
|
| 87 |
+
"##A": 9,
|
| 88 |
+
"##T": 10,
|
| 89 |
"##C": 11,
|
| 90 |
+
"##G": 12,
|
| 91 |
"AA": 13,
|
| 92 |
"GA": 14,
|
| 93 |
"TT": 15,
|
vocab.txt
CHANGED
|
@@ -7,10 +7,10 @@ A
|
|
| 7 |
C
|
| 8 |
G
|
| 9 |
T
|
| 10 |
-
##G
|
| 11 |
##A
|
| 12 |
-
##C
|
| 13 |
##T
|
|
|
|
|
|
|
| 14 |
AA
|
| 15 |
GA
|
| 16 |
TT
|
|
|
|
| 7 |
C
|
| 8 |
G
|
| 9 |
T
|
|
|
|
| 10 |
##A
|
|
|
|
| 11 |
##T
|
| 12 |
+
##C
|
| 13 |
+
##G
|
| 14 |
AA
|
| 15 |
GA
|
| 16 |
TT
|