Create tokenizer.json
Browse files- tokenizer.json +14 -0
tokenizer.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": "1.0",
|
| 3 |
+
"model_type": "GPT-2",
|
| 4 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 5 |
+
"vocab_file": "vocab.json",
|
| 6 |
+
"merges_file": "merges.txt",
|
| 7 |
+
"special_tokens_map": {
|
| 8 |
+
"pad_token": "<PAD>",
|
| 9 |
+
"unk_token": "<UNK>",
|
| 10 |
+
"cls_token": "<CLS>",
|
| 11 |
+
"sep_token": "<SEP>",
|
| 12 |
+
"mask_token": "<MASK>"
|
| 13 |
+
}
|
| 14 |
+
}
|