Commit
·
089d35f
1
Parent(s):
622cffe
Upload tokenizer
Browse files- tokenizer_config.json +7 -0
tokenizer_config.json
CHANGED
|
@@ -37,7 +37,9 @@
|
|
| 37 |
"rstrip": false,
|
| 38 |
"single_word": false
|
| 39 |
},
|
|
|
|
| 40 |
"model_max_length": 512,
|
|
|
|
| 41 |
"pad_token": {
|
| 42 |
"__type": "AddedToken",
|
| 43 |
"content": "<pad>",
|
|
@@ -46,6 +48,8 @@
|
|
| 46 |
"rstrip": false,
|
| 47 |
"single_word": false
|
| 48 |
},
|
|
|
|
|
|
|
| 49 |
"sep_token": {
|
| 50 |
"__type": "AddedToken",
|
| 51 |
"content": "</s>",
|
|
@@ -54,8 +58,11 @@
|
|
| 54 |
"rstrip": false,
|
| 55 |
"single_word": false
|
| 56 |
},
|
|
|
|
| 57 |
"tokenizer_class": "RobertaTokenizer",
|
| 58 |
"trim_offsets": true,
|
|
|
|
|
|
|
| 59 |
"unk_token": {
|
| 60 |
"__type": "AddedToken",
|
| 61 |
"content": "<unk>",
|
|
|
|
| 37 |
"rstrip": false,
|
| 38 |
"single_word": false
|
| 39 |
},
|
| 40 |
+
"max_length": 384,
|
| 41 |
"model_max_length": 512,
|
| 42 |
+
"pad_to_multiple_of": null,
|
| 43 |
"pad_token": {
|
| 44 |
"__type": "AddedToken",
|
| 45 |
"content": "<pad>",
|
|
|
|
| 48 |
"rstrip": false,
|
| 49 |
"single_word": false
|
| 50 |
},
|
| 51 |
+
"pad_token_type_id": 0,
|
| 52 |
+
"padding_side": "right",
|
| 53 |
"sep_token": {
|
| 54 |
"__type": "AddedToken",
|
| 55 |
"content": "</s>",
|
|
|
|
| 58 |
"rstrip": false,
|
| 59 |
"single_word": false
|
| 60 |
},
|
| 61 |
+
"stride": 0,
|
| 62 |
"tokenizer_class": "RobertaTokenizer",
|
| 63 |
"trim_offsets": true,
|
| 64 |
+
"truncation_side": "right",
|
| 65 |
+
"truncation_strategy": "only_second",
|
| 66 |
"unk_token": {
|
| 67 |
"__type": "AddedToken",
|
| 68 |
"content": "<unk>",
|