Create tokenizer_config.json
Browse files- tokenizer_config.json +16 -0
tokenizer_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
// tokenizer_config.json
|
| 2 |
+
{
|
| 3 |
+
"additional_special_tokens": [
|
| 4 |
+
"ar_AR",
|
| 5 |
+
"cs_CZ",
|
| 6 |
+
"de_DE",
|
| 7 |
+
"en_XX",
|
| 8 |
+
"es_XX",
|
| 9 |
+
"fr_XX",
|
| 10 |
+
"zh_CN"
|
| 11 |
+
],
|
| 12 |
+
"legacy": false,
|
| 13 |
+
"model_max_length": 1024,
|
| 14 |
+
"pad_token": "<pad>",
|
| 15 |
+
"tokenizer_class": "MBart50TokenizerFast"
|
| 16 |
+
}
|