Add AutoTokenizer support (trust_remote_code)
Browse files- tokenizer_config.json +4 -2
tokenizer_config.json
CHANGED
|
@@ -1,10 +1,12 @@
|
|
| 1 |
{
|
| 2 |
"tokenizer_class": "TurkTokenizer",
|
| 3 |
"model_type": "turk-tokenizer",
|
|
|
|
|
|
|
|
|
|
| 4 |
"version": "1.0.0",
|
| 5 |
-
"description": "Turkish morphological tokenizer — TR-MMLU world record 92%",
|
| 6 |
"language": "tr",
|
| 7 |
-
"
|
| 8 |
"requires_java": true,
|
| 9 |
"dependencies": ["turkish-tokenizer", "jpype1"]
|
| 10 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"tokenizer_class": "TurkTokenizer",
|
| 3 |
"model_type": "turk-tokenizer",
|
| 4 |
+
"auto_map": {
|
| 5 |
+
"AutoTokenizer": ["tokenization_turk.TurkTokenizer", null]
|
| 6 |
+
},
|
| 7 |
"version": "1.0.0",
|
|
|
|
| 8 |
"language": "tr",
|
| 9 |
+
"description": "Turkish morphological tokenizer — TR-MMLU world record 92%",
|
| 10 |
"requires_java": true,
|
| 11 |
"dependencies": ["turkish-tokenizer", "jpype1"]
|
| 12 |
}
|