Create tokenizer.json
Browse files- tokenizer.json +22 -0
tokenizer.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens": [],
|
| 3 |
+
"do_lower_case": true,
|
| 4 |
+
"encoder": {
|
| 5 |
+
"cls_token": "[CLS]",
|
| 6 |
+
"eos_token": "[EOS]",
|
| 7 |
+
"unk_token": "[UNK]",
|
| 8 |
+
"pad_token": "[PAD]",
|
| 9 |
+
"sep_token": "[SEP]",
|
| 10 |
+
"mask_token": "[MASK]",
|
| 11 |
+
"vocab_size": 30522,
|
| 12 |
+
"tokens": {
|
| 13 |
+
"[CLS]": 0,
|
| 14 |
+
"[EOS]": 1,
|
| 15 |
+
"[UNK]": 2,
|
| 16 |
+
"[PAD]": 3,
|
| 17 |
+
"[SEP]": 4,
|
| 18 |
+
"[MASK]": 5
|
| 19 |
+
}
|
| 20 |
+
},
|
| 21 |
+
"version": "1.0"
|
| 22 |
+
}
|