Upload tokenizer
Browse files- special_tokens_map.json +41 -0
- tokenizer.json +0 -0
- tokenizer_config.json +6 -0
special_tokens_map.json
ADDED
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"additional_special_tokens": [
|
3 |
+
"#@μ£Όμ#",
|
4 |
+
"#@μ΄λͺ¨ν°μ½#",
|
5 |
+
"#@μ΄λ¦#",
|
6 |
+
"#@URL#",
|
7 |
+
"#@μμ#",
|
8 |
+
"#@κΈ°ν#",
|
9 |
+
"#@μ λ²#",
|
10 |
+
"#@κ³μ #",
|
11 |
+
"#@url#",
|
12 |
+
"#@λ²νΈ#",
|
13 |
+
"#@κΈμ΅#",
|
14 |
+
"#@μ μ#",
|
15 |
+
"#@μ₯μ#",
|
16 |
+
"#@μμ€ν
#μ¬μ§#",
|
17 |
+
"#@μμ€ν
#λμμ#",
|
18 |
+
"#@μμ€ν
#κΈ°ν#",
|
19 |
+
"#@μμ€ν
#κ²μ#",
|
20 |
+
"#@μμ€ν
#μ§λ#",
|
21 |
+
"#@μμ€ν
#μμ #",
|
22 |
+
"#@μμ€ν
#νμΌ#",
|
23 |
+
"#@μμ€ν
#μ‘κΈ#",
|
24 |
+
"#@μμ€ν
#",
|
25 |
+
"#κ°μΈ λ° κ΄κ³#",
|
26 |
+
"#λ―Έμ©κ³Ό 건κ°#",
|
27 |
+
"#μκ±°λ(μΌν)#",
|
28 |
+
"#μμ¬/κ΅μ‘#",
|
29 |
+
"#μμλ£#",
|
30 |
+
"#μ¬κ° μν#",
|
31 |
+
"#μΌκ³Ό μ§μ
#",
|
32 |
+
"#μ£Όκ±°μ μν#",
|
33 |
+
"#νμ¬#",
|
34 |
+
"[sep]"
|
35 |
+
],
|
36 |
+
"bos_token": "</s>",
|
37 |
+
"eos_token": "</s>",
|
38 |
+
"mask_token": "<mask>",
|
39 |
+
"pad_token": "<pad>",
|
40 |
+
"unk_token": "<unk>"
|
41 |
+
}
|
tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_config.json
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"model_max_length": 1000000000000000019884624838656,
|
3 |
+
"name_or_path": "/content/drive/MyDrive/αα
΅α«αα
©αΌαα
΅αα
³αΌ/αα
’αΌαα
₯αΌαα
αα
£α¨αα
³α
α
©αα
¦α¨αα
³/Model/KoBART/checkpoint/domain_adaptation/checkpoint-12500",
|
4 |
+
"special_tokens_map_file": "/root/.cache/huggingface/hub/models--gogamza--kobart-base-v2/snapshots/f9f2ec35d3c32a1ecc7a3281f9626b7ec1913fed/special_tokens_map.json",
|
5 |
+
"tokenizer_class": "PreTrainedTokenizerFast"
|
6 |
+
}
|