Upload tokenizer.json with huggingface_hub
Browse files- tokenizer.json +4 -39
tokenizer.json
CHANGED
|
@@ -1,42 +1,7 @@
|
|
| 1 |
{
|
| 2 |
"version": "1.0",
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
| 7 |
-
"type": "Sequence",
|
| 8 |
-
"normalizers": [
|
| 9 |
-
{
|
| 10 |
-
"type": "NFD"
|
| 11 |
-
},
|
| 12 |
-
{
|
| 13 |
-
"type": "Lowercase"
|
| 14 |
-
},
|
| 15 |
-
{
|
| 16 |
-
"type": "StripAccents"
|
| 17 |
-
}
|
| 18 |
-
]
|
| 19 |
-
},
|
| 20 |
-
"pre_tokenizer": {
|
| 21 |
-
"type": "Whitespace"
|
| 22 |
-
},
|
| 23 |
-
"post_processor": null,
|
| 24 |
-
"decoder": null,
|
| 25 |
-
"model": {
|
| 26 |
-
"type": "WordPiece",
|
| 27 |
-
"vocab": {
|
| 28 |
-
"[PAD]": 0,
|
| 29 |
-
"[UNK]": 1,
|
| 30 |
-
"a": 2,
|
| 31 |
-
"the": 3,
|
| 32 |
-
"cat": 4,
|
| 33 |
-
"dog": 5,
|
| 34 |
-
"house": 6,
|
| 35 |
-
"mountain": 7,
|
| 36 |
-
"landscape": 8,
|
| 37 |
-
"beautiful": 9,
|
| 38 |
-
"sunset": 10
|
| 39 |
-
},
|
| 40 |
-
"unk_token": "[UNK]"
|
| 41 |
-
}
|
| 42 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"version": "1.0",
|
| 3 |
+
"model_type": "clip",
|
| 4 |
+
"bos_token": "<|startoftext|>",
|
| 5 |
+
"eos_token": "<|endoftext|>",
|
| 6 |
+
"unk_token": "<|endoftext|>"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
}
|