Add tokenizer_config.json for AutoTokenizer compatibility
Browse files- tokenizer_config.json +16 -0
tokenizer_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"auto_map": {
|
| 3 |
+
"AutoTokenizer": [
|
| 4 |
+
"tokenization_plasmid_lm.PlasmidLMTokenizer",
|
| 5 |
+
null
|
| 6 |
+
]
|
| 7 |
+
},
|
| 8 |
+
"tokenizer_class": "PlasmidLMTokenizer",
|
| 9 |
+
"bos_token": "<BOS>",
|
| 10 |
+
"eos_token": "<EOS>",
|
| 11 |
+
"unk_token": "<UNK>",
|
| 12 |
+
"pad_token": "<PAD>",
|
| 13 |
+
"sep_token": "<SEP>",
|
| 14 |
+
"model_max_length": 16384,
|
| 15 |
+
"use_fast": false
|
| 16 |
+
}
|