Upload tokenizer
Browse files- special_tokens_map.json +1 -0
- tokenizer.json +9 -0
special_tokens_map.json
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"bos_token": "<|endoftext|>",
|
|
|
|
| 3 |
"eos_token": "<|endoftext|>",
|
| 4 |
"pad_token": "[PAD]",
|
| 5 |
"sep_token": "[SEP]",
|
|
|
|
| 1 |
{
|
| 2 |
"bos_token": "<|endoftext|>",
|
| 3 |
+
"cls_token": "[CLS]",
|
| 4 |
"eos_token": "<|endoftext|>",
|
| 5 |
"pad_token": "[PAD]",
|
| 6 |
"sep_token": "[SEP]",
|
tokenizer.json
CHANGED
|
@@ -9245,6 +9245,15 @@
|
|
| 9245 |
"rstrip": false,
|
| 9246 |
"normalized": false,
|
| 9247 |
"special": true
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9248 |
}
|
| 9249 |
],
|
| 9250 |
"normalizer": {
|
|
|
|
| 9245 |
"rstrip": false,
|
| 9246 |
"normalized": false,
|
| 9247 |
"special": true
|
| 9248 |
+
},
|
| 9249 |
+
{
|
| 9250 |
+
"id": 1027,
|
| 9251 |
+
"content": "[CLS]",
|
| 9252 |
+
"single_word": false,
|
| 9253 |
+
"lstrip": false,
|
| 9254 |
+
"rstrip": false,
|
| 9255 |
+
"normalized": false,
|
| 9256 |
+
"special": true
|
| 9257 |
}
|
| 9258 |
],
|
| 9259 |
"normalizer": {
|