Upload 3 files
Browse files- merges.txt +0 -0
- tokenizer_config.json +24 -0
- vocab.json +0 -0
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 3 |
+
"auto_map": {
|
| 4 |
+
"AutoTokenizer": [
|
| 5 |
+
"tokenizers.Tokenizer",
|
| 6 |
+
null
|
| 7 |
+
]
|
| 8 |
+
},
|
| 9 |
+
"model_type": "bpe",
|
| 10 |
+
"vocab_size": 32000,
|
| 11 |
+
"unk_token": "[UNK]",
|
| 12 |
+
"special_tokens": {
|
| 13 |
+
"unk_token": "[UNK]",
|
| 14 |
+
"additional_special_tokens": [
|
| 15 |
+
"<NUM>",
|
| 16 |
+
"<URL>",
|
| 17 |
+
"<EMAIL>"
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"model_max_length": 512,
|
| 21 |
+
"padding_side": "right",
|
| 22 |
+
"truncation_side": "right",
|
| 23 |
+
"clean_up_tokenization_spaces": true
|
| 24 |
+
}
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|