Commit
•
a0121d1
1
Parent(s):
e1197ad
model improved
Browse files- README.md +14 -1
- config.json +1303 -85
- pytorch_model.bin +2 -2
- tokenizer_config.json +1 -1
README.md
CHANGED
|
@@ -6,6 +6,7 @@ tags:
|
|
| 6 |
- "token-classification"
|
| 7 |
- "pos"
|
| 8 |
- "wikipedia"
|
|
|
|
| 9 |
datasets:
|
| 10 |
- "universal_dependencies"
|
| 11 |
license: "apache-2.0"
|
|
@@ -18,7 +19,7 @@ widget:
|
|
| 18 |
|
| 19 |
## Model Description
|
| 20 |
|
| 21 |
-
This is a BERT model pre-trained on Thai Wikipedia texts for POS-tagging, derived from [bert-base-th-cased](https://huggingface.co/Geotrend/bert-base-th-cased). Every word is tagged by [UPOS](https://universaldependencies.org/u/pos/) (Universal Part-Of-Speech).
|
| 22 |
|
| 23 |
## How to Use
|
| 24 |
|
|
@@ -33,3 +34,15 @@ p=[model.config.id2label[q] for q in torch.argmax(model(tokenizer.encode(s,retur
|
|
| 33 |
print(list(zip(t,p)))
|
| 34 |
```
|
| 35 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
- "token-classification"
|
| 7 |
- "pos"
|
| 8 |
- "wikipedia"
|
| 9 |
+
- "dependency-parsing"
|
| 10 |
datasets:
|
| 11 |
- "universal_dependencies"
|
| 12 |
license: "apache-2.0"
|
|
|
|
| 19 |
|
| 20 |
## Model Description
|
| 21 |
|
| 22 |
+
This is a BERT model pre-trained on Thai Wikipedia texts for POS-tagging and dependency-parsing, derived from [bert-base-th-cased](https://huggingface.co/Geotrend/bert-base-th-cased). Every word is tagged by [UPOS](https://universaldependencies.org/u/pos/) (Universal Part-Of-Speech).
|
| 23 |
|
| 24 |
## How to Use
|
| 25 |
|
|
|
|
| 34 |
print(list(zip(t,p)))
|
| 35 |
```
|
| 36 |
|
| 37 |
+
or
|
| 38 |
+
|
| 39 |
+
```py
|
| 40 |
+
import esupar
|
| 41 |
+
nlp=esupar.load("KoichiYasuoka/bert-base-thai-upos")
|
| 42 |
+
print(nlp("หลายหัวดีกว่าหัวเดียว"))
|
| 43 |
+
```
|
| 44 |
+
|
| 45 |
+
## See Also
|
| 46 |
+
|
| 47 |
+
[esupar](https://github.com/KoichiYasuoka/esupar): Tokenizer POS-tagger and Dependency-parser with BERT/RoBERTa models
|
| 48 |
+
|
config.json
CHANGED
|
@@ -3,100 +3,355 @@
|
|
| 3 |
"BertForTokenClassification"
|
| 4 |
],
|
| 5 |
"attention_probs_dropout_prob": 0.1,
|
|
|
|
| 6 |
"directionality": "bidi",
|
| 7 |
"gradient_checkpointing": false,
|
| 8 |
"hidden_act": "gelu",
|
| 9 |
"hidden_dropout_prob": 0.1,
|
| 10 |
"hidden_size": 768,
|
| 11 |
"id2label": {
|
| 12 |
-
"0": "
|
| 13 |
-
"1": "
|
| 14 |
-
"2": "
|
| 15 |
-
"3": "
|
| 16 |
-
"4": "
|
| 17 |
-
"5": "
|
| 18 |
-
"6": "
|
| 19 |
-
"7": "
|
| 20 |
-
"8": "
|
| 21 |
-
"9": "B-ADJ",
|
| 22 |
-
"10": "
|
| 23 |
-
"11": "
|
| 24 |
-
"12": "B-
|
| 25 |
-
"13": "B-
|
| 26 |
-
"14": "
|
| 27 |
-
"15": "
|
| 28 |
-
"16": "
|
| 29 |
-
"17": "ADP",
|
| 30 |
-
"18": "B-
|
| 31 |
-
"19": "B-
|
| 32 |
-
"20": "B-
|
| 33 |
-
"21": "
|
| 34 |
-
"22": "
|
| 35 |
-
"23": "
|
| 36 |
-
"24": "
|
| 37 |
-
"25": "
|
| 38 |
-
"26": "
|
| 39 |
-
"27": "B-
|
| 40 |
-
"28": "
|
| 41 |
-
"29": "B-
|
| 42 |
-
"30": "
|
| 43 |
-
"31": "B-
|
| 44 |
-
"32": "
|
| 45 |
-
"33": "
|
| 46 |
-
"34": "
|
| 47 |
-
"35": "
|
| 48 |
-
"36": "B-
|
| 49 |
-
"37": "
|
| 50 |
-
"38": "
|
| 51 |
-
"39": "B-
|
| 52 |
-
"40": "
|
| 53 |
-
"41": "
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
},
|
| 55 |
"initializer_range": 0.02,
|
| 56 |
"intermediate_size": 3072,
|
| 57 |
"label2id": {
|
| 58 |
-
"ADJ":
|
| 59 |
-
"ADP":
|
| 60 |
-
"
|
| 61 |
-
"
|
| 62 |
-
"
|
| 63 |
-
"
|
| 64 |
-
"B-
|
| 65 |
-
"B-
|
| 66 |
-
"B-
|
| 67 |
-
"B-
|
| 68 |
-
"B-
|
| 69 |
-
"B-
|
| 70 |
-
"B-
|
| 71 |
-
"B-
|
| 72 |
-
"B-
|
| 73 |
-
"B-
|
| 74 |
-
"B-
|
| 75 |
-
"B-
|
| 76 |
-
"
|
| 77 |
-
"
|
| 78 |
-
"
|
| 79 |
-
"
|
| 80 |
-
"
|
| 81 |
-
"
|
| 82 |
-
"
|
| 83 |
-
"
|
| 84 |
-
"
|
| 85 |
-
"
|
| 86 |
-
"
|
| 87 |
-
"
|
| 88 |
-
"
|
| 89 |
-
"
|
| 90 |
-
"
|
| 91 |
-
"
|
| 92 |
-
"NOUN":
|
| 93 |
-
"
|
| 94 |
-
"
|
| 95 |
-
"
|
| 96 |
-
"
|
| 97 |
-
"
|
| 98 |
-
"
|
| 99 |
-
"VERB":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 100 |
},
|
| 101 |
"layer_norm_eps": 1e-12,
|
| 102 |
"max_position_embeddings": 512,
|
|
@@ -110,8 +365,971 @@
|
|
| 110 |
"pooler_size_per_head": 128,
|
| 111 |
"pooler_type": "first_token_transform",
|
| 112 |
"position_embedding_type": "absolute",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 113 |
"torch_dtype": "float32",
|
| 114 |
-
"transformers_version": "4.
|
| 115 |
"type_vocab_size": 2,
|
| 116 |
"use_cache": true,
|
| 117 |
"vocab_size": 8506
|
|
|
|
| 3 |
"BertForTokenClassification"
|
| 4 |
],
|
| 5 |
"attention_probs_dropout_prob": 0.1,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
"directionality": "bidi",
|
| 8 |
"gradient_checkpointing": false,
|
| 9 |
"hidden_act": "gelu",
|
| 10 |
"hidden_dropout_prob": 0.1,
|
| 11 |
"hidden_size": 768,
|
| 12 |
"id2label": {
|
| 13 |
+
"0": "ADJ",
|
| 14 |
+
"1": "ADP",
|
| 15 |
+
"2": "ADP+ADP+PRON",
|
| 16 |
+
"3": "ADP+NOUN",
|
| 17 |
+
"4": "ADV",
|
| 18 |
+
"5": "AUX",
|
| 19 |
+
"6": "B-ADJ",
|
| 20 |
+
"7": "B-ADJ+ADV",
|
| 21 |
+
"8": "B-ADJ+DET",
|
| 22 |
+
"9": "B-ADJ+NOUN",
|
| 23 |
+
"10": "B-ADP",
|
| 24 |
+
"11": "B-ADP+ADJ",
|
| 25 |
+
"12": "B-ADP+ADV+ADV+ADV",
|
| 26 |
+
"13": "B-ADP+NOUN",
|
| 27 |
+
"14": "B-ADP+PRON",
|
| 28 |
+
"15": "B-ADP+PROPN",
|
| 29 |
+
"16": "B-ADV",
|
| 30 |
+
"17": "B-ADV+ADP",
|
| 31 |
+
"18": "B-ADV+ADV",
|
| 32 |
+
"19": "B-ADV+VERB",
|
| 33 |
+
"20": "B-AUX",
|
| 34 |
+
"21": "B-AUX+VERB",
|
| 35 |
+
"22": "B-CCONJ",
|
| 36 |
+
"23": "B-CCONJ+ADJ",
|
| 37 |
+
"24": "B-CCONJ+ADV+ADV+ADV",
|
| 38 |
+
"25": "B-CCONJ+AUX",
|
| 39 |
+
"26": "B-CCONJ+NOUN",
|
| 40 |
+
"27": "B-CCONJ+PRON",
|
| 41 |
+
"28": "B-CCONJ+PROPN",
|
| 42 |
+
"29": "B-CCONJ+VERB",
|
| 43 |
+
"30": "B-DET",
|
| 44 |
+
"31": "B-DET+NOUN",
|
| 45 |
+
"32": "B-NOUN",
|
| 46 |
+
"33": "B-NOUN+ADJ",
|
| 47 |
+
"34": "B-NOUN+ADP",
|
| 48 |
+
"35": "B-NOUN+ADV",
|
| 49 |
+
"36": "B-NOUN+AUX",
|
| 50 |
+
"37": "B-NOUN+DET",
|
| 51 |
+
"38": "B-NOUN+DET+VERB",
|
| 52 |
+
"39": "B-NOUN+NOUN",
|
| 53 |
+
"40": "B-NOUN+PROPN",
|
| 54 |
+
"41": "B-NOUN+VERB",
|
| 55 |
+
"42": "B-NUM",
|
| 56 |
+
"43": "B-PART",
|
| 57 |
+
"44": "B-PART+ADJ",
|
| 58 |
+
"45": "B-PART+ADP",
|
| 59 |
+
"46": "B-PART+ADV",
|
| 60 |
+
"47": "B-PART+AUX",
|
| 61 |
+
"48": "B-PART+VERB",
|
| 62 |
+
"49": "B-PRON",
|
| 63 |
+
"50": "B-PRON+ADP",
|
| 64 |
+
"51": "B-PRON+ADP+PRON",
|
| 65 |
+
"52": "B-PRON+ADV",
|
| 66 |
+
"53": "B-PRON+AUX",
|
| 67 |
+
"54": "B-PRON+CCONJ",
|
| 68 |
+
"55": "B-PRON+DET",
|
| 69 |
+
"56": "B-PRON+PRON",
|
| 70 |
+
"57": "B-PRON+VERB",
|
| 71 |
+
"58": "B-PROPN",
|
| 72 |
+
"59": "B-PROPN+ADV",
|
| 73 |
+
"60": "B-PROPN+AUX",
|
| 74 |
+
"61": "B-PROPN+AUX+VERB",
|
| 75 |
+
"62": "B-PROPN+CCONJ",
|
| 76 |
+
"63": "B-PROPN+DET",
|
| 77 |
+
"64": "B-PROPN+DET+VERB",
|
| 78 |
+
"65": "B-PROPN+PROPN",
|
| 79 |
+
"66": "B-PROPN+VERB",
|
| 80 |
+
"67": "B-PUNCT",
|
| 81 |
+
"68": "B-SCONJ",
|
| 82 |
+
"69": "B-VERB",
|
| 83 |
+
"70": "B-VERB+ADJ",
|
| 84 |
+
"71": "B-VERB+ADP",
|
| 85 |
+
"72": "B-VERB+ADV",
|
| 86 |
+
"73": "B-VERB+DET",
|
| 87 |
+
"74": "B-VERB+NOUN",
|
| 88 |
+
"75": "B-VERB+NOUN+DET",
|
| 89 |
+
"76": "B-VERB+PART",
|
| 90 |
+
"77": "B-VERB+PRON",
|
| 91 |
+
"78": "B-VERB+PROPN",
|
| 92 |
+
"79": "B-VERB+VERB",
|
| 93 |
+
"80": "CCONJ",
|
| 94 |
+
"81": "CCONJ+ADP+PRON",
|
| 95 |
+
"82": "DET",
|
| 96 |
+
"83": "I-ADJ",
|
| 97 |
+
"84": "I-ADJ+ADV",
|
| 98 |
+
"85": "I-ADJ+DET",
|
| 99 |
+
"86": "I-ADJ+NOUN",
|
| 100 |
+
"87": "I-ADP",
|
| 101 |
+
"88": "I-ADP+ADJ",
|
| 102 |
+
"89": "I-ADP+ADV+ADV+ADV",
|
| 103 |
+
"90": "I-ADP+NOUN",
|
| 104 |
+
"91": "I-ADP+PRON",
|
| 105 |
+
"92": "I-ADP+PROPN",
|
| 106 |
+
"93": "I-ADV",
|
| 107 |
+
"94": "I-ADV+ADP",
|
| 108 |
+
"95": "I-ADV+ADV",
|
| 109 |
+
"96": "I-ADV+VERB",
|
| 110 |
+
"97": "I-AUX",
|
| 111 |
+
"98": "I-AUX+VERB",
|
| 112 |
+
"99": "I-CCONJ",
|
| 113 |
+
"100": "I-CCONJ+ADJ",
|
| 114 |
+
"101": "I-CCONJ+ADV+ADV+ADV",
|
| 115 |
+
"102": "I-CCONJ+AUX",
|
| 116 |
+
"103": "I-CCONJ+NOUN",
|
| 117 |
+
"104": "I-CCONJ+PRON",
|
| 118 |
+
"105": "I-CCONJ+PROPN",
|
| 119 |
+
"106": "I-CCONJ+VERB",
|
| 120 |
+
"107": "I-DET",
|
| 121 |
+
"108": "I-DET+NOUN",
|
| 122 |
+
"109": "I-NOUN",
|
| 123 |
+
"110": "I-NOUN+ADJ",
|
| 124 |
+
"111": "I-NOUN+ADP",
|
| 125 |
+
"112": "I-NOUN+ADV",
|
| 126 |
+
"113": "I-NOUN+AUX",
|
| 127 |
+
"114": "I-NOUN+DET",
|
| 128 |
+
"115": "I-NOUN+DET+VERB",
|
| 129 |
+
"116": "I-NOUN+NOUN",
|
| 130 |
+
"117": "I-NOUN+PROPN",
|
| 131 |
+
"118": "I-NOUN+VERB",
|
| 132 |
+
"119": "I-NUM",
|
| 133 |
+
"120": "I-PART",
|
| 134 |
+
"121": "I-PART+ADJ",
|
| 135 |
+
"122": "I-PART+ADP",
|
| 136 |
+
"123": "I-PART+ADV",
|
| 137 |
+
"124": "I-PART+AUX",
|
| 138 |
+
"125": "I-PART+VERB",
|
| 139 |
+
"126": "I-PRON",
|
| 140 |
+
"127": "I-PRON+ADP",
|
| 141 |
+
"128": "I-PRON+ADP+PRON",
|
| 142 |
+
"129": "I-PRON+ADV",
|
| 143 |
+
"130": "I-PRON+AUX",
|
| 144 |
+
"131": "I-PRON+CCONJ",
|
| 145 |
+
"132": "I-PRON+DET",
|
| 146 |
+
"133": "I-PRON+PRON",
|
| 147 |
+
"134": "I-PRON+VERB",
|
| 148 |
+
"135": "I-PROPN",
|
| 149 |
+
"136": "I-PROPN+ADV",
|
| 150 |
+
"137": "I-PROPN+AUX",
|
| 151 |
+
"138": "I-PROPN+AUX+VERB",
|
| 152 |
+
"139": "I-PROPN+CCONJ",
|
| 153 |
+
"140": "I-PROPN+DET",
|
| 154 |
+
"141": "I-PROPN+DET+VERB",
|
| 155 |
+
"142": "I-PROPN+PROPN",
|
| 156 |
+
"143": "I-PROPN+VERB",
|
| 157 |
+
"144": "I-PUNCT",
|
| 158 |
+
"145": "I-SCONJ",
|
| 159 |
+
"146": "I-VERB",
|
| 160 |
+
"147": "I-VERB+ADJ",
|
| 161 |
+
"148": "I-VERB+ADP",
|
| 162 |
+
"149": "I-VERB+ADV",
|
| 163 |
+
"150": "I-VERB+DET",
|
| 164 |
+
"151": "I-VERB+NOUN",
|
| 165 |
+
"152": "I-VERB+NOUN+DET",
|
| 166 |
+
"153": "I-VERB+PART",
|
| 167 |
+
"154": "I-VERB+PRON",
|
| 168 |
+
"155": "I-VERB+PROPN",
|
| 169 |
+
"156": "I-VERB+VERB",
|
| 170 |
+
"157": "NOUN",
|
| 171 |
+
"158": "NOUN+DET",
|
| 172 |
+
"159": "NOUN+NOUN",
|
| 173 |
+
"160": "NUM",
|
| 174 |
+
"161": "PART",
|
| 175 |
+
"162": "PRON",
|
| 176 |
+
"163": "PROPN",
|
| 177 |
+
"164": "PUNCT",
|
| 178 |
+
"165": "SYM",
|
| 179 |
+
"166": "VERB",
|
| 180 |
+
"167": "VERB+ADP",
|
| 181 |
+
"168": "VERB+ADP+NOUN"
|
| 182 |
},
|
| 183 |
"initializer_range": 0.02,
|
| 184 |
"intermediate_size": 3072,
|
| 185 |
"label2id": {
|
| 186 |
+
"ADJ": 0,
|
| 187 |
+
"ADP": 1,
|
| 188 |
+
"ADP+ADP+PRON": 2,
|
| 189 |
+
"ADP+NOUN": 3,
|
| 190 |
+
"ADV": 4,
|
| 191 |
+
"AUX": 5,
|
| 192 |
+
"B-ADJ": 6,
|
| 193 |
+
"B-ADJ+ADV": 7,
|
| 194 |
+
"B-ADJ+DET": 8,
|
| 195 |
+
"B-ADJ+NOUN": 9,
|
| 196 |
+
"B-ADP": 10,
|
| 197 |
+
"B-ADP+ADJ": 11,
|
| 198 |
+
"B-ADP+ADV+ADV+ADV": 12,
|
| 199 |
+
"B-ADP+NOUN": 13,
|
| 200 |
+
"B-ADP+PRON": 14,
|
| 201 |
+
"B-ADP+PROPN": 15,
|
| 202 |
+
"B-ADV": 16,
|
| 203 |
+
"B-ADV+ADP": 17,
|
| 204 |
+
"B-ADV+ADV": 18,
|
| 205 |
+
"B-ADV+VERB": 19,
|
| 206 |
+
"B-AUX": 20,
|
| 207 |
+
"B-AUX+VERB": 21,
|
| 208 |
+
"B-CCONJ": 22,
|
| 209 |
+
"B-CCONJ+ADJ": 23,
|
| 210 |
+
"B-CCONJ+ADV+ADV+ADV": 24,
|
| 211 |
+
"B-CCONJ+AUX": 25,
|
| 212 |
+
"B-CCONJ+NOUN": 26,
|
| 213 |
+
"B-CCONJ+PRON": 27,
|
| 214 |
+
"B-CCONJ+PROPN": 28,
|
| 215 |
+
"B-CCONJ+VERB": 29,
|
| 216 |
+
"B-DET": 30,
|
| 217 |
+
"B-DET+NOUN": 31,
|
| 218 |
+
"B-NOUN": 32,
|
| 219 |
+
"B-NOUN+ADJ": 33,
|
| 220 |
+
"B-NOUN+ADP": 34,
|
| 221 |
+
"B-NOUN+ADV": 35,
|
| 222 |
+
"B-NOUN+AUX": 36,
|
| 223 |
+
"B-NOUN+DET": 37,
|
| 224 |
+
"B-NOUN+DET+VERB": 38,
|
| 225 |
+
"B-NOUN+NOUN": 39,
|
| 226 |
+
"B-NOUN+PROPN": 40,
|
| 227 |
+
"B-NOUN+VERB": 41,
|
| 228 |
+
"B-NUM": 42,
|
| 229 |
+
"B-PART": 43,
|
| 230 |
+
"B-PART+ADJ": 44,
|
| 231 |
+
"B-PART+ADP": 45,
|
| 232 |
+
"B-PART+ADV": 46,
|
| 233 |
+
"B-PART+AUX": 47,
|
| 234 |
+
"B-PART+VERB": 48,
|
| 235 |
+
"B-PRON": 49,
|
| 236 |
+
"B-PRON+ADP": 50,
|
| 237 |
+
"B-PRON+ADP+PRON": 51,
|
| 238 |
+
"B-PRON+ADV": 52,
|
| 239 |
+
"B-PRON+AUX": 53,
|
| 240 |
+
"B-PRON+CCONJ": 54,
|
| 241 |
+
"B-PRON+DET": 55,
|
| 242 |
+
"B-PRON+PRON": 56,
|
| 243 |
+
"B-PRON+VERB": 57,
|
| 244 |
+
"B-PROPN": 58,
|
| 245 |
+
"B-PROPN+ADV": 59,
|
| 246 |
+
"B-PROPN+AUX": 60,
|
| 247 |
+
"B-PROPN+AUX+VERB": 61,
|
| 248 |
+
"B-PROPN+CCONJ": 62,
|
| 249 |
+
"B-PROPN+DET": 63,
|
| 250 |
+
"B-PROPN+DET+VERB": 64,
|
| 251 |
+
"B-PROPN+PROPN": 65,
|
| 252 |
+
"B-PROPN+VERB": 66,
|
| 253 |
+
"B-PUNCT": 67,
|
| 254 |
+
"B-SCONJ": 68,
|
| 255 |
+
"B-VERB": 69,
|
| 256 |
+
"B-VERB+ADJ": 70,
|
| 257 |
+
"B-VERB+ADP": 71,
|
| 258 |
+
"B-VERB+ADV": 72,
|
| 259 |
+
"B-VERB+DET": 73,
|
| 260 |
+
"B-VERB+NOUN": 74,
|
| 261 |
+
"B-VERB+NOUN+DET": 75,
|
| 262 |
+
"B-VERB+PART": 76,
|
| 263 |
+
"B-VERB+PRON": 77,
|
| 264 |
+
"B-VERB+PROPN": 78,
|
| 265 |
+
"B-VERB+VERB": 79,
|
| 266 |
+
"CCONJ": 80,
|
| 267 |
+
"CCONJ+ADP+PRON": 81,
|
| 268 |
+
"DET": 82,
|
| 269 |
+
"I-ADJ": 83,
|
| 270 |
+
"I-ADJ+ADV": 84,
|
| 271 |
+
"I-ADJ+DET": 85,
|
| 272 |
+
"I-ADJ+NOUN": 86,
|
| 273 |
+
"I-ADP": 87,
|
| 274 |
+
"I-ADP+ADJ": 88,
|
| 275 |
+
"I-ADP+ADV+ADV+ADV": 89,
|
| 276 |
+
"I-ADP+NOUN": 90,
|
| 277 |
+
"I-ADP+PRON": 91,
|
| 278 |
+
"I-ADP+PROPN": 92,
|
| 279 |
+
"I-ADV": 93,
|
| 280 |
+
"I-ADV+ADP": 94,
|
| 281 |
+
"I-ADV+ADV": 95,
|
| 282 |
+
"I-ADV+VERB": 96,
|
| 283 |
+
"I-AUX": 97,
|
| 284 |
+
"I-AUX+VERB": 98,
|
| 285 |
+
"I-CCONJ": 99,
|
| 286 |
+
"I-CCONJ+ADJ": 100,
|
| 287 |
+
"I-CCONJ+ADV+ADV+ADV": 101,
|
| 288 |
+
"I-CCONJ+AUX": 102,
|
| 289 |
+
"I-CCONJ+NOUN": 103,
|
| 290 |
+
"I-CCONJ+PRON": 104,
|
| 291 |
+
"I-CCONJ+PROPN": 105,
|
| 292 |
+
"I-CCONJ+VERB": 106,
|
| 293 |
+
"I-DET": 107,
|
| 294 |
+
"I-DET+NOUN": 108,
|
| 295 |
+
"I-NOUN": 109,
|
| 296 |
+
"I-NOUN+ADJ": 110,
|
| 297 |
+
"I-NOUN+ADP": 111,
|
| 298 |
+
"I-NOUN+ADV": 112,
|
| 299 |
+
"I-NOUN+AUX": 113,
|
| 300 |
+
"I-NOUN+DET": 114,
|
| 301 |
+
"I-NOUN+DET+VERB": 115,
|
| 302 |
+
"I-NOUN+NOUN": 116,
|
| 303 |
+
"I-NOUN+PROPN": 117,
|
| 304 |
+
"I-NOUN+VERB": 118,
|
| 305 |
+
"I-NUM": 119,
|
| 306 |
+
"I-PART": 120,
|
| 307 |
+
"I-PART+ADJ": 121,
|
| 308 |
+
"I-PART+ADP": 122,
|
| 309 |
+
"I-PART+ADV": 123,
|
| 310 |
+
"I-PART+AUX": 124,
|
| 311 |
+
"I-PART+VERB": 125,
|
| 312 |
+
"I-PRON": 126,
|
| 313 |
+
"I-PRON+ADP": 127,
|
| 314 |
+
"I-PRON+ADP+PRON": 128,
|
| 315 |
+
"I-PRON+ADV": 129,
|
| 316 |
+
"I-PRON+AUX": 130,
|
| 317 |
+
"I-PRON+CCONJ": 131,
|
| 318 |
+
"I-PRON+DET": 132,
|
| 319 |
+
"I-PRON+PRON": 133,
|
| 320 |
+
"I-PRON+VERB": 134,
|
| 321 |
+
"I-PROPN": 135,
|
| 322 |
+
"I-PROPN+ADV": 136,
|
| 323 |
+
"I-PROPN+AUX": 137,
|
| 324 |
+
"I-PROPN+AUX+VERB": 138,
|
| 325 |
+
"I-PROPN+CCONJ": 139,
|
| 326 |
+
"I-PROPN+DET": 140,
|
| 327 |
+
"I-PROPN+DET+VERB": 141,
|
| 328 |
+
"I-PROPN+PROPN": 142,
|
| 329 |
+
"I-PROPN+VERB": 143,
|
| 330 |
+
"I-PUNCT": 144,
|
| 331 |
+
"I-SCONJ": 145,
|
| 332 |
+
"I-VERB": 146,
|
| 333 |
+
"I-VERB+ADJ": 147,
|
| 334 |
+
"I-VERB+ADP": 148,
|
| 335 |
+
"I-VERB+ADV": 149,
|
| 336 |
+
"I-VERB+DET": 150,
|
| 337 |
+
"I-VERB+NOUN": 151,
|
| 338 |
+
"I-VERB+NOUN+DET": 152,
|
| 339 |
+
"I-VERB+PART": 153,
|
| 340 |
+
"I-VERB+PRON": 154,
|
| 341 |
+
"I-VERB+PROPN": 155,
|
| 342 |
+
"I-VERB+VERB": 156,
|
| 343 |
+
"NOUN": 157,
|
| 344 |
+
"NOUN+DET": 158,
|
| 345 |
+
"NOUN+NOUN": 159,
|
| 346 |
+
"NUM": 160,
|
| 347 |
+
"PART": 161,
|
| 348 |
+
"PRON": 162,
|
| 349 |
+
"PROPN": 163,
|
| 350 |
+
"PUNCT": 164,
|
| 351 |
+
"SYM": 165,
|
| 352 |
+
"VERB": 166,
|
| 353 |
+
"VERB+ADP": 167,
|
| 354 |
+
"VERB+ADP+NOUN": 168
|
| 355 |
},
|
| 356 |
"layer_norm_eps": 1e-12,
|
| 357 |
"max_position_embeddings": 512,
|
|
|
|
| 365 |
"pooler_size_per_head": 128,
|
| 366 |
"pooler_type": "first_token_transform",
|
| 367 |
"position_embedding_type": "absolute",
|
| 368 |
+
"task_specific_params": {
|
| 369 |
+
"upos_multiword": {
|
| 370 |
+
"ADJ+ADV": {
|
| 371 |
+
"\u0e40\u0e2b\u0e19\u0e37\u0e2d\u0e01\u0e27\u0e48\u0e32": [
|
| 372 |
+
"\u0e40\u0e2b\u0e19\u0e37\u0e2d",
|
| 373 |
+
"\u0e01\u0e27\u0e48\u0e32"
|
| 374 |
+
]
|
| 375 |
+
},
|
| 376 |
+
"ADJ+DET": {
|
| 377 |
+
"\u0e40\u0e17\u0e32\u0e17\u0e35\u0e48": [
|
| 378 |
+
"\u0e40\u0e17\u0e32",
|
| 379 |
+
"\u0e17\u0e35\u0e48"
|
| 380 |
+
]
|
| 381 |
+
},
|
| 382 |
+
"ADJ+NOUN": {
|
| 383 |
+
"\u0e2d\u0e14\u0e35\u0e15\u0e23\u0e31\u0e10\u0e21\u0e19\u0e15\u0e23\u0e35": [
|
| 384 |
+
"\u0e2d\u0e14\u0e35\u0e15",
|
| 385 |
+
"\u0e23\u0e31\u0e10\u0e21\u0e19\u0e15\u0e23\u0e35"
|
| 386 |
+
],
|
| 387 |
+
"\u0e43\u0e2b\u0e0d\u0e48\u0e21\u0e13\u0e11\u0e25": [
|
| 388 |
+
"\u0e43\u0e2b\u0e0d\u0e48",
|
| 389 |
+
"\u0e21\u0e13\u0e11\u0e25"
|
| 390 |
+
]
|
| 391 |
+
},
|
| 392 |
+
"ADP+ADJ": {
|
| 393 |
+
"\u0e41\u0e15\u0e48\u0e19\u0e49\u0e2d\u0e22": [
|
| 394 |
+
"\u0e41\u0e15\u0e48",
|
| 395 |
+
"\u0e19\u0e49\u0e2d\u0e22"
|
| 396 |
+
]
|
| 397 |
+
},
|
| 398 |
+
"ADP+ADP+PRON": {
|
| 399 |
+
"\u0e19\u0e2d\u0e01\u0e08\u0e32\u0e01\u0e19\u0e35\u0e49": [
|
| 400 |
+
"\u0e19\u0e2d\u0e01",
|
| 401 |
+
"\u0e08\u0e32\u0e01",
|
| 402 |
+
"\u0e19\u0e35\u0e49"
|
| 403 |
+
]
|
| 404 |
+
},
|
| 405 |
+
"ADP+ADV+ADV+ADV": {
|
| 406 |
+
"\u0e2d\u0e22\u0e48\u0e32\u0e07\u0e44\u0e23\u0e01\u0e47\u0e15\u0e32\u0e21": [
|
| 407 |
+
"\u0e2d\u0e22\u0e48\u0e32\u0e07",
|
| 408 |
+
"\u0e44\u0e23",
|
| 409 |
+
"\u0e01\u0e47",
|
| 410 |
+
"\u0e15\u0e32\u0e21"
|
| 411 |
+
]
|
| 412 |
+
},
|
| 413 |
+
"ADP+NOUN": {
|
| 414 |
+
"\u0e40\u0e21\u0e37\u0e48\u0e2d\u0e27\u0e31\u0e19\u0e17\u0e35\u0e48": [
|
| 415 |
+
"\u0e40\u0e21\u0e37\u0e48\u0e2d",
|
| 416 |
+
"\u0e27\u0e31\u0e19\u0e17\u0e35\u0e48"
|
| 417 |
+
],
|
| 418 |
+
"\u0e41\u0e01\u0e48\u0e2d\u0e2a\u0e31\u0e0d\u0e01\u0e23\u0e23\u0e21": [
|
| 419 |
+
"\u0e41\u0e01\u0e48",
|
| 420 |
+
"\u0e2d\u0e2a\u0e31\u0e0d\u0e01\u0e23\u0e23\u0e21"
|
| 421 |
+
],
|
| 422 |
+
"\u0e43\u0e19\u0e1b\u0e35": [
|
| 423 |
+
"\u0e43\u0e19",
|
| 424 |
+
"\u0e1b\u0e35"
|
| 425 |
+
],
|
| 426 |
+
"\u0e43\u0e19\u0e27\u0e31\u0e19\u0e17\u0e35\u0e48": [
|
| 427 |
+
"\u0e43\u0e19",
|
| 428 |
+
"\u0e27\u0e31\u0e19\u0e17\u0e35\u0e48"
|
| 429 |
+
]
|
| 430 |
+
},
|
| 431 |
+
"ADP+PRON": {
|
| 432 |
+
"\u0e41\u0e15\u0e48\u0e19\u0e31\u0e49\u0e19": [
|
| 433 |
+
"\u0e41\u0e15\u0e48",
|
| 434 |
+
"\u0e19\u0e31\u0e49\u0e19"
|
| 435 |
+
]
|
| 436 |
+
},
|
| 437 |
+
"ADP+PROPN": {
|
| 438 |
+
"\u0e41\u0e15\u0e48\u0e21\u0e32\u0e23\u0e4c\u0e04": [
|
| 439 |
+
"\u0e41\u0e15\u0e48",
|
| 440 |
+
"\u0e21\u0e32\u0e23\u0e4c\u0e04"
|
| 441 |
+
]
|
| 442 |
+
},
|
| 443 |
+
"ADV+ADP": {
|
| 444 |
+
"\u0e02\u0e36\u0e49\u0e19\u0e40\u0e21\u0e37\u0e48\u0e2d": [
|
| 445 |
+
"\u0e02\u0e36\u0e49\u0e19",
|
| 446 |
+
"\u0e40\u0e21\u0e37\u0e48\u0e2d"
|
| 447 |
+
]
|
| 448 |
+
},
|
| 449 |
+
"ADV+ADV": {
|
| 450 |
+
"\u0e1a\u0e48\u0e2d\u0e22\u0e21\u0e32\u0e01": [
|
| 451 |
+
"\u0e1a\u0e48\u0e2d\u0e22",
|
| 452 |
+
"\u0e21\u0e32\u0e01"
|
| 453 |
+
],
|
| 454 |
+
"\u0e41\u0e15\u0e48\u0e2d\u0e22\u0e48\u0e32\u0e07\u0e43\u0e14": [
|
| 455 |
+
"\u0e41\u0e15\u0e48",
|
| 456 |
+
"\u0e2d\u0e22\u0e48\u0e32\u0e07\u0e43\u0e14"
|
| 457 |
+
]
|
| 458 |
+
},
|
| 459 |
+
"ADV+VERB": {
|
| 460 |
+
"\u0e40\u0e25\u0e22\u0e19\u0e31\u0e1a": [
|
| 461 |
+
"\u0e40\u0e25\u0e22",
|
| 462 |
+
"\u0e19\u0e31\u0e1a"
|
| 463 |
+
]
|
| 464 |
+
},
|
| 465 |
+
"AUX+VERB": {
|
| 466 |
+
"\u0e04\u0e27\u0e23\u0e21\u0e35": [
|
| 467 |
+
"\u0e04\u0e27\u0e23",
|
| 468 |
+
"\u0e21\u0e35"
|
| 469 |
+
],
|
| 470 |
+
"\u0e16\u0e39\u0e01\u0e23\u0e32\u0e22\u0e07\u0e32\u0e19": [
|
| 471 |
+
"\u0e16\u0e39\u0e01",
|
| 472 |
+
"\u0e23\u0e32\u0e22\u0e07\u0e32\u0e19"
|
| 473 |
+
],
|
| 474 |
+
"\u0e16\u0e39\u0e01\u0e23\u0e37\u0e49\u0e2d\u0e16\u0e2d\u0e19": [
|
| 475 |
+
"\u0e16\u0e39\u0e01",
|
| 476 |
+
"\u0e23\u0e37\u0e49\u0e2d\u0e16\u0e2d\u0e19"
|
| 477 |
+
],
|
| 478 |
+
"\u0e40\u0e04\u0e22\u0e21\u0e35": [
|
| 479 |
+
"\u0e40\u0e04\u0e22",
|
| 480 |
+
"\u0e21\u0e35"
|
| 481 |
+
]
|
| 482 |
+
},
|
| 483 |
+
"CCONJ+ADJ": {
|
| 484 |
+
"\u0e41\u0e15\u0e48\u0e2d\u0e22\u0e48\u0e32\u0e07\u0e19\u0e49\u0e2d\u0e22": [
|
| 485 |
+
"\u0e41\u0e15\u0e48",
|
| 486 |
+
"\u0e2d\u0e22\u0e48\u0e32\u0e07\u0e19\u0e49\u0e2d\u0e22"
|
| 487 |
+
]
|
| 488 |
+
},
|
| 489 |
+
"CCONJ+ADP+PRON": {
|
| 490 |
+
"\u0e19\u0e2d\u0e01\u0e08\u0e32\u0e01\u0e19\u0e35\u0e49": [
|
| 491 |
+
"\u0e19\u0e2d\u0e01",
|
| 492 |
+
"\u0e08\u0e32\u0e01",
|
| 493 |
+
"\u0e19\u0e35\u0e49"
|
| 494 |
+
]
|
| 495 |
+
},
|
| 496 |
+
"CCONJ+ADV+ADV+ADV": {
|
| 497 |
+
"\u0e2d\u0e22\u0e48\u0e32\u0e07\u0e44\u0e23\u0e01\u0e47\u0e15\u0e32\u0e21": [
|
| 498 |
+
"\u0e2d\u0e22\u0e48\u0e32\u0e07",
|
| 499 |
+
"\u0e44\u0e23",
|
| 500 |
+
"\u0e01\u0e47",
|
| 501 |
+
"\u0e15\u0e32\u0e21"
|
| 502 |
+
]
|
| 503 |
+
},
|
| 504 |
+
"CCONJ+AUX": {
|
| 505 |
+
"\u0e41\u0e15\u0e48\u0e2d\u0e32\u0e08": [
|
| 506 |
+
"\u0e41\u0e15\u0e48",
|
| 507 |
+
"\u0e2d\u0e32\u0e08"
|
| 508 |
+
]
|
| 509 |
+
},
|
| 510 |
+
"CCONJ+NOUN": {
|
| 511 |
+
"\u0e41\u0e15\u0e48\u0e2d\u0e31\u0e25\u0e1a\u0e31\u0e49\u0e21": [
|
| 512 |
+
"\u0e41\u0e15\u0e48",
|
| 513 |
+
"\u0e2d\u0e31\u0e25\u0e1a\u0e31\u0e49\u0e21"
|
| 514 |
+
]
|
| 515 |
+
},
|
| 516 |
+
"CCONJ+PRON": {
|
| 517 |
+
"\u0e41\u0e15\u0e48\u0e19\u0e31\u0e48\u0e19": [
|
| 518 |
+
"\u0e41\u0e15\u0e48",
|
| 519 |
+
"\u0e19\u0e31\u0e48\u0e19"
|
| 520 |
+
],
|
| 521 |
+
"\u0e41\u0e15\u0e48\u0e21\u0e31\u0e19": [
|
| 522 |
+
"\u0e41\u0e15\u0e48",
|
| 523 |
+
"\u0e21\u0e31\u0e19"
|
| 524 |
+
]
|
| 525 |
+
},
|
| 526 |
+
"CCONJ+PROPN": {
|
| 527 |
+
"\u0e41\u0e15\u0e48\u0e19\u0e34\u0e27\u0e15\u0e31\u0e19": [
|
| 528 |
+
"\u0e41\u0e15\u0e48",
|
| 529 |
+
"\u0e19\u0e34\u0e27\u0e15\u0e31\u0e19"
|
| 530 |
+
]
|
| 531 |
+
},
|
| 532 |
+
"CCONJ+VERB": {
|
| 533 |
+
"\u0e41\u0e15\u0e48\u0e21\u0e35": [
|
| 534 |
+
"\u0e41\u0e15\u0e48",
|
| 535 |
+
"\u0e21\u0e35"
|
| 536 |
+
]
|
| 537 |
+
},
|
| 538 |
+
"DET+NOUN": {
|
| 539 |
+
"\u0e17\u0e38\u0e01\u0e23\u0e31\u0e10": [
|
| 540 |
+
"\u0e17\u0e38\u0e01",
|
| 541 |
+
"\u0e23\u0e31\u0e10"
|
| 542 |
+
],
|
| 543 |
+
"\u0e17\u0e38\u0e01\u0e23\u0e32\u0e22\u0e25\u0e30\u0e40\u0e2d\u0e35\u0e22\u0e14": [
|
| 544 |
+
"\u0e17\u0e38\u0e01",
|
| 545 |
+
"\u0e23\u0e32\u0e22\u0e25\u0e30\u0e40\u0e2d\u0e35\u0e22\u0e14"
|
| 546 |
+
],
|
| 547 |
+
"\u0e2b\u0e25\u0e32\u0e22\u0e19\u0e32\u0e22": [
|
| 548 |
+
"\u0e2b\u0e25\u0e32\u0e22",
|
| 549 |
+
"\u0e19\u0e32\u0e22"
|
| 550 |
+
],
|
| 551 |
+
"\u0e2d\u0e35\u0e01\u0e23\u0e32\u0e07\u0e27\u0e31\u0e25": [
|
| 552 |
+
"\u0e2d\u0e35\u0e01",
|
| 553 |
+
"\u0e23\u0e32\u0e07\u0e27\u0e31\u0e25"
|
| 554 |
+
]
|
| 555 |
+
},
|
| 556 |
+
"NOUN+ADJ": {
|
| 557 |
+
"\u0e22\u0e38\u0e04\u0e21\u0e37\u0e14": [
|
| 558 |
+
"\u0e22\u0e38\u0e04",
|
| 559 |
+
"\u0e21\u0e37\u0e14"
|
| 560 |
+
]
|
| 561 |
+
},
|
| 562 |
+
"NOUN+ADP": {
|
| 563 |
+
"\u0e1b\u0e31\u0e0d\u0e2b\u0e32\u0e01\u0e31\u0e1a": [
|
| 564 |
+
"\u0e1b\u0e31\u0e0d\u0e2b\u0e32",
|
| 565 |
+
"\u0e01\u0e31\u0e1a"
|
| 566 |
+
],
|
| 567 |
+
"\u0e1c\u0e25\u0e01\u0e31\u0e1a": [
|
| 568 |
+
"\u0e1c\u0e25",
|
| 569 |
+
"\u0e01\u0e31\u0e1a"
|
| 570 |
+
],
|
| 571 |
+
"\u0e2a\u0e31\u0e0d\u0e0d\u0e32\u0e01\u0e31\u0e1a": [
|
| 572 |
+
"\u0e2a\u0e31\u0e0d\u0e0d\u0e32",
|
| 573 |
+
"\u0e01\u0e31\u0e1a"
|
| 574 |
+
],
|
| 575 |
+
"\u0e2d\u0e34\u0e17\u0e18\u0e34\u0e1e\u0e25\u0e01\u0e31\u0e1a": [
|
| 576 |
+
"\u0e2d\u0e34\u0e17\u0e18\u0e34\u0e1e\u0e25",
|
| 577 |
+
"\u0e01\u0e31\u0e1a"
|
| 578 |
+
]
|
| 579 |
+
},
|
| 580 |
+
"NOUN+ADV": {
|
| 581 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e1e\u0e23\u0e23\u0e14\u0e34\u0e01\u0e47": [
|
| 582 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e1e\u0e23\u0e23\u0e14\u0e34",
|
| 583 |
+
"\u0e01\u0e47"
|
| 584 |
+
],
|
| 585 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e1e\u0e23\u0e23\u0e14\u0e34\u0e21\u0e31\u0e01": [
|
| 586 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e1e\u0e23\u0e23\u0e14\u0e34",
|
| 587 |
+
"\u0e21\u0e31\u0e01"
|
| 588 |
+
],
|
| 589 |
+
"\u0e1c\u0e25\u0e01\u0e47": [
|
| 590 |
+
"\u0e1c\u0e25",
|
| 591 |
+
"\u0e01\u0e47"
|
| 592 |
+
],
|
| 593 |
+
"\u0e23\u0e31\u0e10\u0e1a\u0e32\u0e25\u0e01\u0e47": [
|
| 594 |
+
"\u0e23\u0e31\u0e10\u0e1a\u0e32\u0e25",
|
| 595 |
+
"\u0e01\u0e47"
|
| 596 |
+
],
|
| 597 |
+
"\u0e23\u0e32\u0e04\u0e32\u0e01\u0e47": [
|
| 598 |
+
"\u0e23\u0e32\u0e04\u0e32",
|
| 599 |
+
"\u0e01\u0e47"
|
| 600 |
+
],
|
| 601 |
+
"\u0e25\u0e39\u0e01\u0e40\u0e23\u0e37\u0e2d\u0e01\u0e27\u0e48\u0e32": [
|
| 602 |
+
"\u0e25\u0e39\u0e01\u0e40\u0e23\u0e37\u0e2d",
|
| 603 |
+
"\u0e01\u0e27\u0e48\u0e32"
|
| 604 |
+
],
|
| 605 |
+
"\u0e25\u0e49\u0e2d\u0e01\u0e47": [
|
| 606 |
+
"\u0e25\u0e49\u0e2d",
|
| 607 |
+
"\u0e01\u0e47"
|
| 608 |
+
],
|
| 609 |
+
"\u0e40\u0e17\u0e04\u0e42\u0e19\u0e42\u0e25\u0e22\u0e35\u0e19\u0e49\u0e2d\u0e22": [
|
| 610 |
+
"\u0e40\u0e17\u0e04\u0e42\u0e19\u0e42\u0e25\u0e22\u0e35",
|
| 611 |
+
"\u0e19\u0e49\u0e2d\u0e22"
|
| 612 |
+
],
|
| 613 |
+
"\u0e42\u0e06\u0e29\u0e13\u0e32\u0e01\u0e47": [
|
| 614 |
+
"\u0e42\u0e06\u0e29\u0e13\u0e32",
|
| 615 |
+
"\u0e01\u0e47"
|
| 616 |
+
]
|
| 617 |
+
},
|
| 618 |
+
"NOUN+AUX": {
|
| 619 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e27\u0e32\u0e25\u0e01\u0e33\u0e25\u0e31\u0e07": [
|
| 620 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e27\u0e32\u0e25",
|
| 621 |
+
"\u0e01\u0e33\u0e25\u0e31\u0e07"
|
| 622 |
+
],
|
| 623 |
+
"\u0e1c\u0e39\u0e49\u0e0a\u0e32\u0e22\u0e19\u0e48\u0e32": [
|
| 624 |
+
"\u0e1c\u0e39\u0e49\u0e0a\u0e32\u0e22",
|
| 625 |
+
"\u0e19\u0e48\u0e32"
|
| 626 |
+
],
|
| 627 |
+
"\u0e23\u0e31\u0e10\u0e1a\u0e32\u0e25\u0e01\u0e33\u0e25\u0e31\u0e07": [
|
| 628 |
+
"\u0e23\u0e31\u0e10\u0e1a\u0e32\u0e25",
|
| 629 |
+
"\u0e01\u0e33\u0e25\u0e31\u0e07"
|
| 630 |
+
]
|
| 631 |
+
},
|
| 632 |
+
"NOUN+DET": {
|
| 633 |
+
"\u0e01\u0e25\u0e27\u0e34\u0e18\u0e35\u0e19\u0e35\u0e49": [
|
| 634 |
+
"\u0e01\u0e25\u0e27\u0e34\u0e18\u0e35",
|
| 635 |
+
"\u0e19\u0e35\u0e49"
|
| 636 |
+
],
|
| 637 |
+
"\u0e14\u0e32\u0e23\u0e32\u0e19\u0e31\u0e49\u0e19": [
|
| 638 |
+
"\u0e14\u0e32\u0e23\u0e32",
|
| 639 |
+
"\u0e19\u0e31\u0e49\u0e19"
|
| 640 |
+
],
|
| 641 |
+
"\u0e19\u0e34\u0e22\u0e32\u0e22\u0e19\u0e31\u0e49\u0e19": [
|
| 642 |
+
"\u0e19\u0e34\u0e22\u0e32\u0e22",
|
| 643 |
+
"\u0e19\u0e31\u0e49\u0e19"
|
| 644 |
+
],
|
| 645 |
+
"\u0e27\u0e31\u0e19\u0e17\u0e35\u0e48": [
|
| 646 |
+
"\u0e27\u0e31\u0e19",
|
| 647 |
+
"\u0e17\u0e35\u0e48"
|
| 648 |
+
],
|
| 649 |
+
"\u0e2a\u0e31\u0e0d\u0e0d\u0e32\u0e17\u0e35\u0e48": [
|
| 650 |
+
"\u0e2a\u0e31\u0e0d\u0e0d\u0e32",
|
| 651 |
+
"\u0e17\u0e35\u0e48"
|
| 652 |
+
],
|
| 653 |
+
"\u0e2a\u0e32\u0e02\u0e32\u0e19\u0e35\u0e49": [
|
| 654 |
+
"\u0e2a\u0e32\u0e02\u0e32",
|
| 655 |
+
"\u0e19\u0e35\u0e49"
|
| 656 |
+
],
|
| 657 |
+
"\u0e2d\u0e32\u0e0d\u0e32\u0e17\u0e38\u0e01": [
|
| 658 |
+
"\u0e2d\u0e32\u0e0d\u0e32",
|
| 659 |
+
"\u0e17\u0e38\u0e01"
|
| 660 |
+
],
|
| 661 |
+
"\u0e40\u0e02\u0e32\u0e17\u0e35\u0e48": [
|
| 662 |
+
"\u0e40\u0e02\u0e32",
|
| 663 |
+
"\u0e17\u0e35\u0e48"
|
| 664 |
+
],
|
| 665 |
+
"\u0e41\u0e16\u0e27\u0e19\u0e31\u0e49\u0e19": [
|
| 666 |
+
"\u0e41\u0e16\u0e27",
|
| 667 |
+
"\u0e19\u0e31\u0e49\u0e19"
|
| 668 |
+
],
|
| 669 |
+
"\u0e43\u0e1a\u0e1b\u0e25\u0e34\u0e27\u0e19\u0e31\u0e49\u0e19": [
|
| 670 |
+
"\u0e43\u0e1a\u0e1b\u0e25\u0e34\u0e27",
|
| 671 |
+
"\u0e19\u0e31\u0e49\u0e19"
|
| 672 |
+
]
|
| 673 |
+
},
|
| 674 |
+
"NOUN+DET+VERB": {
|
| 675 |
+
"\u0e0b\u0e34\u0e25\u0e34\u0e01\u0e32\u0e17\u0e35\u0e48\u0e21\u0e35": [
|
| 676 |
+
"\u0e0b\u0e34\u0e25\u0e34\u0e01\u0e32",
|
| 677 |
+
"\u0e17\u0e35\u0e48",
|
| 678 |
+
"\u0e21\u0e35"
|
| 679 |
+
]
|
| 680 |
+
},
|
| 681 |
+
"NOUN+NOUN": {
|
| 682 |
+
"\u0e01\u0e35\u0e2c\u0e32\u0e21\u0e2b\u0e32\u0e27\u0e34\u0e17\u0e22\u0e32\u0e25\u0e31\u0e22": [
|
| 683 |
+
"\u0e01\u0e35\u0e2c\u0e32",
|
| 684 |
+
"\u0e21\u0e2b\u0e32\u0e27\u0e34\u0e17\u0e22\u0e32\u0e25\u0e31\u0e22"
|
| 685 |
+
],
|
| 686 |
+
"\u0e01\u0e35\u0e2c\u0e32\u0e27\u0e31\u0e22": [
|
| 687 |
+
"\u0e01\u0e35\u0e2c\u0e32",
|
| 688 |
+
"\u0e27\u0e31\u0e22"
|
| 689 |
+
],
|
| 690 |
+
"\u0e19\u0e32\u0e22\u0e01\u0e23\u0e31\u0e10\u0e21\u0e19\u0e15\u0e23\u0e35": [
|
| 691 |
+
"\u0e19\u0e32\u0e22\u0e01",
|
| 692 |
+
"\u0e23\u0e31\u0e10\u0e21\u0e19\u0e15\u0e23\u0e35"
|
| 693 |
+
],
|
| 694 |
+
"\u0e1b\u0e23\u0e30\u0e21\u0e27\u0e25\u0e01\u0e0e\u0e2b\u0e21\u0e32\u0e22": [
|
| 695 |
+
"\u0e1b\u0e23\u0e30\u0e21\u0e27\u0e25",
|
| 696 |
+
"\u0e01\u0e0e\u0e2b\u0e21\u0e32\u0e22"
|
| 697 |
+
],
|
| 698 |
+
"\u0e1d\u0e48\u0e32\u0e22\u0e19\u0e42\u0e22\u0e1a\u0e32\u0e22": [
|
| 699 |
+
"\u0e1d\u0e48\u0e32\u0e22",
|
| 700 |
+
"\u0e19\u0e42\u0e22\u0e1a\u0e32\u0e22"
|
| 701 |
+
],
|
| 702 |
+
"\u0e22\u0e38\u0e04\u0e23\u0e32\u0e0a\u0e27\u0e07\u0e28\u0e4c": [
|
| 703 |
+
"\u0e22\u0e38\u0e04",
|
| 704 |
+
"\u0e23\u0e32\u0e0a\u0e27\u0e07\u0e28\u0e4c"
|
| 705 |
+
],
|
| 706 |
+
"\u0e23\u0e31\u0e10\u0e1a\u0e32\u0e25\u0e01\u0e25\u0e32\u0e07": [
|
| 707 |
+
"\u0e23\u0e31\u0e10\u0e1a\u0e32\u0e25",
|
| 708 |
+
"\u0e01\u0e25\u0e32\u0e07"
|
| 709 |
+
],
|
| 710 |
+
"\u0e23\u0e32\u0e04\u0e32\u0e19\u0e49\u0e33\u0e21\u0e31\u0e19": [
|
| 711 |
+
"\u0e23\u0e32\u0e04\u0e32",
|
| 712 |
+
"\u0e19\u0e49\u0e33\u0e21\u0e31\u0e19"
|
| 713 |
+
],
|
| 714 |
+
"\u0e27\u0e31\u0e19\u0e17\u0e35\u0e48": [
|
| 715 |
+
"\u0e27\u0e31\u0e19",
|
| 716 |
+
"\u0e17\u0e35\u0e48"
|
| 717 |
+
],
|
| 718 |
+
"\u0e2a\u0e20\u0e32\u0e19\u0e34\u0e15\u0e34\u0e1a\u0e31\u0e0d\u0e0d\u0e31\u0e15\u0e34": [
|
| 719 |
+
"\u0e2a\u0e20\u0e32",
|
| 720 |
+
"\u0e19\u0e34\u0e15\u0e34\u0e1a\u0e31\u0e0d\u0e0d\u0e31\u0e15\u0e34"
|
| 721 |
+
],
|
| 722 |
+
"\u0e2a\u0e42\u0e21\u0e2a\u0e23\u0e21\u0e2b\u0e32\u0e19\u0e04\u0e23": [
|
| 723 |
+
"\u0e2a\u0e42\u0e21\u0e2a\u0e23",
|
| 724 |
+
"\u0e21\u0e2b\u0e32\u0e19\u0e04\u0e23"
|
| 725 |
+
],
|
| 726 |
+
"\u0e40\u0e19\u0e37\u0e49\u0e2d\u0e2b\u0e32\u0e27\u0e34\u0e17\u0e22\u0e32\u0e28\u0e32\u0e2a\u0e15\u0e23\u0e4c": [
|
| 727 |
+
"\u0e40\u0e19\u0e37\u0e49\u0e2d\u0e2b\u0e32",
|
| 728 |
+
"\u0e27\u0e34\u0e17\u0e22\u0e32\u0e28\u0e32\u0e2a\u0e15\u0e23\u0e4c"
|
| 729 |
+
]
|
| 730 |
+
},
|
| 731 |
+
"NOUN+PROPN": {
|
| 732 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e1e\u0e23\u0e23\u0e14\u0e34\u0e01\u0e32\u0e23\u0e32\u0e01\u0e31\u0e25\u0e25\u0e32": [
|
| 733 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e1e\u0e23\u0e23\u0e14\u0e34",
|
| 734 |
+
"\u0e01\u0e32\u0e23\u0e32\u0e01\u0e31\u0e25\u0e25\u0e32"
|
| 735 |
+
],
|
| 736 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e1e\u0e23\u0e23\u0e14\u0e34\u0e19\u0e42\u0e1b\u0e40\u0e25\u0e35\u0e22\u0e19": [
|
| 737 |
+
"\u0e08\u0e31\u0e01\u0e23\u0e1e\u0e23\u0e23\u0e14\u0e34",
|
| 738 |
+
"\u0e19\u0e42\u0e1b\u0e40\u0e25\u0e35\u0e22\u0e19"
|
| 739 |
+
],
|
| 740 |
+
"\u0e1b\u0e23\u0e30\u0e40\u0e17\u0e28\u0e44\u0e17\u0e22": [
|
| 741 |
+
"\u0e1b\u0e23\u0e30\u0e40\u0e17\u0e28",
|
| 742 |
+
"\u0e44\u0e17\u0e22"
|
| 743 |
+
],
|
| 744 |
+
"\u0e1e\u0e23\u0e23\u0e04\u0e23\u0e34\u0e1e\u0e31\u0e1a\u0e25\u0e34\u0e01\u0e31\u0e19": [
|
| 745 |
+
"\u0e1e\u0e23\u0e23\u0e04",
|
| 746 |
+
"\u0e23\u0e34\u0e1e\u0e31\u0e1a\u0e25\u0e34\u0e01\u0e31\u0e19"
|
| 747 |
+
],
|
| 748 |
+
"\u0e1e\u0e23\u0e23\u0e04\u0e23\u0e35\u0e1e\u0e31\u0e1a\u0e25\u0e34\u0e01\u0e31\u0e19": [
|
| 749 |
+
"\u0e1e\u0e23\u0e23\u0e04",
|
| 750 |
+
"\u0e23\u0e35\u0e1e\u0e31\u0e1a\u0e25\u0e34\u0e01\u0e31\u0e19"
|
| 751 |
+
],
|
| 752 |
+
"\u0e22\u0e38\u0e04\u0e21\u0e32\u0e22\u0e32": [
|
| 753 |
+
"\u0e22\u0e38\u0e04",
|
| 754 |
+
"\u0e21\u0e32\u0e22\u0e32"
|
| 755 |
+
],
|
| 756 |
+
"\u0e25\u0e31\u0e17\u0e18\u0e34\u0e21\u0e32\u0e23\u0e4c\u0e01\u0e0b\u0e34\u0e2a\u0e21\u0e4c": [
|
| 757 |
+
"\u0e25\u0e31\u0e17\u0e18\u0e34",
|
| 758 |
+
"\u0e21\u0e32\u0e23\u0e4c\u0e01\u0e0b\u0e34\u0e2a\u0e21\u0e4c"
|
| 759 |
+
],
|
| 760 |
+
"\u0e2a\u0e2b\u0e23\u0e31\u0e10\u0e2d\u0e40\u0e21\u0e23\u0e34\u0e01\u0e32": [
|
| 761 |
+
"\u0e2a\u0e2b\u0e23\u0e31\u0e10",
|
| 762 |
+
"\u0e2d\u0e40\u0e21\u0e23\u0e34\u0e01\u0e32"
|
| 763 |
+
]
|
| 764 |
+
},
|
| 765 |
+
"NOUN+VERB": {
|
| 766 |
+
"\u0e02\u0e49\u0e2d\u0e01\u0e33\u0e2b\u0e19\u0e14": [
|
| 767 |
+
"\u0e02\u0e49\u0e2d",
|
| 768 |
+
"\u0e01\u0e33\u0e2b\u0e19\u0e14"
|
| 769 |
+
],
|
| 770 |
+
"\u0e04\u0e33\u0e27\u0e48\u0e32": [
|
| 771 |
+
"\u0e04\u0e33",
|
| 772 |
+
"\u0e27\u0e48\u0e32"
|
| 773 |
+
],
|
| 774 |
+
"\u0e15\u0e36\u0e01\u0e23\u0e30": [
|
| 775 |
+
"\u0e15\u0e36\u0e01",
|
| 776 |
+
"\u0e23\u0e30"
|
| 777 |
+
],
|
| 778 |
+
"\u0e1b\u0e31\u0e0d\u0e2b\u0e32\u0e01\u0e32\u0e23\u0e02\u0e19\u0e2a\u0e48\u0e07": [
|
| 779 |
+
"\u0e1b\u0e31\u0e0d\u0e2b\u0e32",
|
| 780 |
+
"\u0e01\u0e32\u0e23\u0e02\u0e19\u0e2a\u0e48\u0e07"
|
| 781 |
+
],
|
| 782 |
+
"\u0e1c\u0e25\u0e01\u0e23\u0e30\u0e17\u0e1a": [
|
| 783 |
+
"\u0e1c\u0e25",
|
| 784 |
+
"\u0e01\u0e23\u0e30\u0e17\u0e1a"
|
| 785 |
+
],
|
| 786 |
+
"\u0e1c\u0e25\u0e01\u0e32\u0e23\u0e14\u0e33\u0e40\u0e19\u0e34\u0e19\u0e07\u0e32\u0e19": [
|
| 787 |
+
"\u0e1c\u0e25",
|
| 788 |
+
"\u0e01\u0e32\u0e23\u0e14\u0e33\u0e40\u0e19\u0e34\u0e19\u0e07\u0e32\u0e19"
|
| 789 |
+
],
|
| 790 |
+
"\u0e1c\u0e25\u0e01\u0e32\u0e23\u0e28\u0e36\u0e01\u0e29\u0e32": [
|
| 791 |
+
"\u0e1c\u0e25",
|
| 792 |
+
"\u0e01\u0e32\u0e23\u0e28\u0e36\u0e01\u0e29\u0e32"
|
| 793 |
+
],
|
| 794 |
+
"\u0e1c\u0e25\u0e01\u0e32\u0e23\u0e2a\u0e41\u0e01\u0e19": [
|
| 795 |
+
"\u0e1c\u0e25",
|
| 796 |
+
"\u0e01\u0e32\u0e23\u0e2a\u0e41\u0e01\u0e19"
|
| 797 |
+
],
|
| 798 |
+
"\u0e1c\u0e39\u0e49\u0e19\u0e33": [
|
| 799 |
+
"\u0e1c\u0e39\u0e49",
|
| 800 |
+
"\u0e19\u0e33"
|
| 801 |
+
],
|
| 802 |
+
"\u0e1e\u0e31\u0e19\u0e18\u0e21\u0e34\u0e15\u0e23\u0e21\u0e35": [
|
| 803 |
+
"\u0e1e\u0e31\u0e19\u0e18\u0e21\u0e34\u0e15\u0e23",
|
| 804 |
+
"\u0e21\u0e35"
|
| 805 |
+
],
|
| 806 |
+
"\u0e27\u0e35\u0e25\u0e01\u0e25\u0e31\u0e1a": [
|
| 807 |
+
"\u0e27\u0e35\u0e25",
|
| 808 |
+
"\u0e01\u0e25\u0e31\u0e1a"
|
| 809 |
+
],
|
| 810 |
+
"\u0e2a\u0e16\u0e34\u0e15\u0e34\u0e01\u0e32\u0e23\u0e22\u0e34\u0e07": [
|
| 811 |
+
"\u0e2a\u0e16\u0e34\u0e15\u0e34",
|
| 812 |
+
"\u0e01\u0e32\u0e23\u0e22\u0e34\u0e07"
|
| 813 |
+
],
|
| 814 |
+
"\u0e2d\u0e31\u0e15\u0e23\u0e32\u0e01\u0e32\u0e23\u0e40\u0e15\u0e34\u0e1a\u0e42\u0e15": [
|
| 815 |
+
"\u0e2d\u0e31\u0e15\u0e23\u0e32",
|
| 816 |
+
"\u0e01\u0e32\u0e23\u0e40\u0e15\u0e34\u0e1a\u0e42\u0e15"
|
| 817 |
+
],
|
| 818 |
+
"\u0e2d\u0e31\u0e15\u0e23\u0e32\u0e01\u0e32\u0e23\u0e40\u0e25\u0e34\u0e01": [
|
| 819 |
+
"\u0e2d\u0e31\u0e15\u0e23\u0e32",
|
| 820 |
+
"\u0e01\u0e32\u0e23\u0e40\u0e25\u0e34\u0e01"
|
| 821 |
+
],
|
| 822 |
+
"\u0e40\u0e02\u0e32\u0e17\u0e23\u0e38\u0e14": [
|
| 823 |
+
"\u0e40\u0e02\u0e32",
|
| 824 |
+
"\u0e17\u0e23\u0e38\u0e14"
|
| 825 |
+
],
|
| 826 |
+
"\u0e40\u0e14\u0e47\u0e01\u0e23\u0e30\u0e1a\u0e38": [
|
| 827 |
+
"\u0e40\u0e14\u0e47\u0e01",
|
| 828 |
+
"\u0e23\u0e30\u0e1a\u0e38"
|
| 829 |
+
],
|
| 830 |
+
"\u0e40\u0e19\u0e37\u0e49\u0e2d\u0e07\u0e2d\u0e01": [
|
| 831 |
+
"\u0e40\u0e19\u0e37\u0e49\u0e2d",
|
| 832 |
+
"\u0e07\u0e2d\u0e01"
|
| 833 |
+
],
|
| 834 |
+
"\u0e42\u0e23\u0e04\u0e23\u0e30\u0e1a\u0e32\u0e14": [
|
| 835 |
+
"\u0e42\u0e23\u0e04",
|
| 836 |
+
"\u0e23\u0e30\u0e1a\u0e32\u0e14"
|
| 837 |
+
]
|
| 838 |
+
},
|
| 839 |
+
"PART+ADJ": {
|
| 840 |
+
"\u0e44\u0e21\u0e48\u0e19\u0e32\u0e19": [
|
| 841 |
+
"\u0e44\u0e21\u0e48",
|
| 842 |
+
"\u0e19\u0e32\u0e19"
|
| 843 |
+
]
|
| 844 |
+
},
|
| 845 |
+
"PART+ADP": {
|
| 846 |
+
"\u0e02\u0e36\u0e49\u0e19\u0e40\u0e21\u0e37\u0e48\u0e2d": [
|
| 847 |
+
"\u0e02\u0e36\u0e49\u0e19",
|
| 848 |
+
"\u0e40\u0e21\u0e37\u0e48\u0e2d"
|
| 849 |
+
],
|
| 850 |
+
"\u0e44\u0e21\u0e48\u0e2d\u0e22\u0e48\u0e32\u0e07": [
|
| 851 |
+
"\u0e44\u0e21\u0e48",
|
| 852 |
+
"\u0e2d\u0e22\u0e48\u0e32\u0e07"
|
| 853 |
+
]
|
| 854 |
+
},
|
| 855 |
+
"PART+ADV": {
|
| 856 |
+
"\u0e44\u0e21\u0e48\u0e19\u0e32\u0e19": [
|
| 857 |
+
"\u0e44\u0e21\u0e48",
|
| 858 |
+
"\u0e19\u0e32\u0e19"
|
| 859 |
+
],
|
| 860 |
+
"\u0e44\u0e21\u0e48\u0e19\u0e49\u0e2d\u0e22": [
|
| 861 |
+
"\u0e44\u0e21\u0e48",
|
| 862 |
+
"\u0e19\u0e49\u0e2d\u0e22"
|
| 863 |
+
]
|
| 864 |
+
},
|
| 865 |
+
"PART+AUX": {
|
| 866 |
+
"\u0e44\u0e21\u0e48\u0e2d\u0e32\u0e08": [
|
| 867 |
+
"\u0e44\u0e21\u0e48",
|
| 868 |
+
"\u0e2d\u0e32\u0e08"
|
| 869 |
+
]
|
| 870 |
+
},
|
| 871 |
+
"PART+VERB": {
|
| 872 |
+
"\u0e01\u0e32\u0e23\u0e44\u0e21\u0e48\u0e2d\u0e19\u0e38\u0e0d\u0e32\u0e15": [
|
| 873 |
+
"\u0e01\u0e32\u0e23\u0e44\u0e21\u0e48",
|
| 874 |
+
"\u0e2d\u0e19\u0e38\u0e0d\u0e32\u0e15"
|
| 875 |
+
],
|
| 876 |
+
"\u0e44\u0e21\u0e48\u0e21\u0e32": [
|
| 877 |
+
"\u0e44\u0e21\u0e48",
|
| 878 |
+
"\u0e21\u0e32"
|
| 879 |
+
],
|
| 880 |
+
"\u0e44\u0e21\u0e48\u0e21\u0e35": [
|
| 881 |
+
"\u0e44\u0e21\u0e48",
|
| 882 |
+
"\u0e21\u0e35"
|
| 883 |
+
],
|
| 884 |
+
"\u0e44\u0e21\u0e48\u0e2d\u0e22\u0e32\u0e01": [
|
| 885 |
+
"\u0e44\u0e21\u0e48",
|
| 886 |
+
"\u0e2d\u0e22\u0e32\u0e01"
|
| 887 |
+
],
|
| 888 |
+
"\u0e44\u0e21\u0e48\u0e2d\u0e22\u0e39\u0e48": [
|
| 889 |
+
"\u0e44\u0e21\u0e48",
|
| 890 |
+
"\u0e2d\u0e22\u0e39\u0e48"
|
| 891 |
+
]
|
| 892 |
+
},
|
| 893 |
+
"PRON+ADP": {
|
| 894 |
+
"\u0e40\u0e02\u0e32\u0e17\u0e35\u0e48": [
|
| 895 |
+
"\u0e40\u0e02\u0e32",
|
| 896 |
+
"\u0e17\u0e35\u0e48"
|
| 897 |
+
],
|
| 898 |
+
"\u0e40\u0e02\u0e32\u0e27\u0e48\u0e32": [
|
| 899 |
+
"\u0e40\u0e02\u0e32",
|
| 900 |
+
"\u0e27\u0e48\u0e32"
|
| 901 |
+
]
|
| 902 |
+
},
|
| 903 |
+
"PRON+ADP+PRON": {
|
| 904 |
+
"\u0e40\u0e02\u0e32\u0e17\u0e35\u0e48\u0e19\u0e31\u0e48\u0e19": [
|
| 905 |
+
"\u0e40\u0e02\u0e32",
|
| 906 |
+
"\u0e17\u0e35\u0e48",
|
| 907 |
+
"\u0e19\u0e31\u0e48\u0e19"
|
| 908 |
+
]
|
| 909 |
+
},
|
| 910 |
+
"PRON+ADV": {
|
| 911 |
+
"\u0e2d\u0e30\u0e44\u0e23\u0e21\u0e32\u0e01": [
|
| 912 |
+
"\u0e2d\u0e30\u0e44\u0e23",
|
| 913 |
+
"\u0e21\u0e32\u0e01"
|
| 914 |
+
],
|
| 915 |
+
"\u0e40\u0e02\u0e32\u0e01\u0e47": [
|
| 916 |
+
"\u0e40\u0e02\u0e32",
|
| 917 |
+
"\u0e01\u0e47"
|
| 918 |
+
],
|
| 919 |
+
"\u0e40\u0e02\u0e32\u0e21\u0e31\u0e01": [
|
| 920 |
+
"\u0e40\u0e02\u0e32",
|
| 921 |
+
"\u0e21\u0e31\u0e01"
|
| 922 |
+
],
|
| 923 |
+
"\u0e40\u0e02\u0e32\u0e22\u0e31\u0e07": [
|
| 924 |
+
"\u0e40\u0e02\u0e32",
|
| 925 |
+
"\u0e22\u0e31\u0e07"
|
| 926 |
+
],
|
| 927 |
+
"\u0e40\u0e18\u0e2d\u0e01\u0e47": [
|
| 928 |
+
"\u0e40\u0e18\u0e2d",
|
| 929 |
+
"\u0e01\u0e47"
|
| 930 |
+
],
|
| 931 |
+
"\u0e40\u0e23\u0e32\u0e01\u0e47": [
|
| 932 |
+
"\u0e40\u0e23\u0e32",
|
| 933 |
+
"\u0e01\u0e47"
|
| 934 |
+
]
|
| 935 |
+
},
|
| 936 |
+
"PRON+AUX": {
|
| 937 |
+
"\u0e40\u0e23\u0e32\u0e01\u0e33\u0e25\u0e31\u0e07": [
|
| 938 |
+
"\u0e40\u0e23\u0e32",
|
| 939 |
+
"\u0e01\u0e33\u0e25\u0e31\u0e07"
|
| 940 |
+
],
|
| 941 |
+
"\u0e40\u0e23\u0e32\u0e22\u0e31\u0e07": [
|
| 942 |
+
"\u0e40\u0e23\u0e32",
|
| 943 |
+
"\u0e22\u0e31\u0e07"
|
| 944 |
+
]
|
| 945 |
+
},
|
| 946 |
+
"PRON+CCONJ": {
|
| 947 |
+
"\u0e40\u0e02\u0e32\u0e01\u0e31\u0e1a": [
|
| 948 |
+
"\u0e40\u0e02\u0e32",
|
| 949 |
+
"\u0e01\u0e31\u0e1a"
|
| 950 |
+
]
|
| 951 |
+
},
|
| 952 |
+
"PRON+DET": {
|
| 953 |
+
"\u0e40\u0e02\u0e32\u0e17\u0e35\u0e48": [
|
| 954 |
+
"\u0e40\u0e02\u0e32",
|
| 955 |
+
"\u0e17\u0e35\u0e48"
|
| 956 |
+
],
|
| 957 |
+
"\u0e40\u0e02\u0e32\u0e19\u0e31\u0e49\u0e19": [
|
| 958 |
+
"\u0e40\u0e02\u0e32",
|
| 959 |
+
"\u0e19\u0e31\u0e49\u0e19"
|
| 960 |
+
]
|
| 961 |
+
},
|
| 962 |
+
"PRON+PRON": {
|
| 963 |
+
"\u0e19\u0e35\u0e48\u0e21\u0e31\u0e19": [
|
| 964 |
+
"\u0e19\u0e35\u0e48",
|
| 965 |
+
"\u0e21\u0e31\u0e19"
|
| 966 |
+
],
|
| 967 |
+
"\u0e2d\u0e30\u0e44\u0e23\u0e21\u0e31\u0e19": [
|
| 968 |
+
"\u0e2d\u0e30\u0e44\u0e23",
|
| 969 |
+
"\u0e21\u0e31\u0e19"
|
| 970 |
+
]
|
| 971 |
+
},
|
| 972 |
+
"PRON+VERB": {
|
| 973 |
+
"\u0e40\u0e02\u0e32\u0e01\u0e23\u0e30\u0e15\u0e38\u0e49\u0e19": [
|
| 974 |
+
"\u0e40\u0e02\u0e32",
|
| 975 |
+
"\u0e01\u0e23\u0e30\u0e15\u0e38\u0e49\u0e19"
|
| 976 |
+
],
|
| 977 |
+
"\u0e40\u0e02\u0e32\u0e01\u0e25\u0e31\u0e1a": [
|
| 978 |
+
"\u0e40\u0e02\u0e32",
|
| 979 |
+
"\u0e01\u0e25\u0e31\u0e1a"
|
| 980 |
+
],
|
| 981 |
+
"\u0e40\u0e02\u0e32\u0e01\u0e25\u0e48\u0e32\u0e27": [
|
| 982 |
+
"\u0e40\u0e02\u0e32",
|
| 983 |
+
"\u0e01\u0e25\u0e48\u0e32\u0e27"
|
| 984 |
+
],
|
| 985 |
+
"\u0e40\u0e02\u0e32\u0e17\u0e33": [
|
| 986 |
+
"\u0e40\u0e02\u0e32",
|
| 987 |
+
"\u0e17\u0e33"
|
| 988 |
+
],
|
| 989 |
+
"\u0e40\u0e02\u0e32\u0e21\u0e35": [
|
| 990 |
+
"\u0e40\u0e02\u0e32",
|
| 991 |
+
"\u0e21\u0e35"
|
| 992 |
+
],
|
| 993 |
+
"\u0e40\u0e02\u0e32\u0e22\u0e37\u0e19": [
|
| 994 |
+
"\u0e40\u0e02\u0e32",
|
| 995 |
+
"\u0e22\u0e37\u0e19"
|
| 996 |
+
],
|
| 997 |
+
"\u0e40\u0e02\u0e32\u0e22\u0e37\u0e19\u0e22\u0e31\u0e19": [
|
| 998 |
+
"\u0e40\u0e02\u0e32",
|
| 999 |
+
"\u0e22\u0e37\u0e19\u0e22\u0e31\u0e19"
|
| 1000 |
+
],
|
| 1001 |
+
"\u0e40\u0e02\u0e32\u0e23\u0e31\u0e1a\u0e23\u0e39\u0e49": [
|
| 1002 |
+
"\u0e40\u0e02\u0e32",
|
| 1003 |
+
"\u0e23\u0e31\u0e1a\u0e23\u0e39\u0e49"
|
| 1004 |
+
],
|
| 1005 |
+
"\u0e40\u0e02\u0e32\u0e23\u0e32\u0e22\u0e07\u0e32\u0e19": [
|
| 1006 |
+
"\u0e40\u0e02\u0e32",
|
| 1007 |
+
"\u0e23\u0e32\u0e22\u0e07\u0e32\u0e19"
|
| 1008 |
+
],
|
| 1009 |
+
"\u0e40\u0e02\u0e32\u0e23\u0e39\u0e49\u0e2a\u0e36\u0e01": [
|
| 1010 |
+
"\u0e40\u0e02\u0e32",
|
| 1011 |
+
"\u0e23\u0e39\u0e49\u0e2a\u0e36\u0e01"
|
| 1012 |
+
],
|
| 1013 |
+
"\u0e40\u0e02\u0e32\u0e27\u0e48\u0e32": [
|
| 1014 |
+
"\u0e40\u0e02\u0e32",
|
| 1015 |
+
"\u0e27\u0e48\u0e32"
|
| 1016 |
+
],
|
| 1017 |
+
"\u0e40\u0e18\u0e2d\u0e19\u0e33": [
|
| 1018 |
+
"\u0e40\u0e18\u0e2d",
|
| 1019 |
+
"\u0e19\u0e33"
|
| 1020 |
+
],
|
| 1021 |
+
"\u0e40\u0e23\u0e32\u0e01\u0e25\u0e31\u0e1a": [
|
| 1022 |
+
"\u0e40\u0e23\u0e32",
|
| 1023 |
+
"\u0e01\u0e25\u0e31\u0e1a"
|
| 1024 |
+
],
|
| 1025 |
+
"\u0e40\u0e23\u0e32\u0e01\u0e49\u0e32\u0e27": [
|
| 1026 |
+
"\u0e40\u0e23\u0e32",
|
| 1027 |
+
"\u0e01\u0e49\u0e32\u0e27"
|
| 1028 |
+
],
|
| 1029 |
+
"\u0e40\u0e23\u0e32\u0e17\u0e33": [
|
| 1030 |
+
"\u0e40\u0e23\u0e32",
|
| 1031 |
+
"\u0e17\u0e33"
|
| 1032 |
+
],
|
| 1033 |
+
"\u0e40\u0e23\u0e32\u0e21\u0e35": [
|
| 1034 |
+
"\u0e40\u0e23\u0e32",
|
| 1035 |
+
"\u0e21\u0e35"
|
| 1036 |
+
],
|
| 1037 |
+
"\u0e40\u0e23\u0e32\u0e23\u0e39\u0e49": [
|
| 1038 |
+
"\u0e40\u0e23\u0e32",
|
| 1039 |
+
"\u0e23\u0e39\u0e49"
|
| 1040 |
+
]
|
| 1041 |
+
},
|
| 1042 |
+
"PROPN+ADV": {
|
| 1043 |
+
"\u0e04\u0e32\u0e40\u0e23\u0e25\u0e01\u0e47": [
|
| 1044 |
+
"\u0e04\u0e32\u0e40\u0e23\u0e25",
|
| 1045 |
+
"\u0e01\u0e47"
|
| 1046 |
+
],
|
| 1047 |
+
"\u0e1f\u0e25\u0e2d\u0e23\u0e34\u0e14\u0e32\u0e01\u0e27\u0e48\u0e32": [
|
| 1048 |
+
"\u0e1f\u0e25\u0e2d\u0e23\u0e34\u0e14\u0e32",
|
| 1049 |
+
"\u0e01\u0e27\u0e48\u0e32"
|
| 1050 |
+
],
|
| 1051 |
+
"\u0e40\u0e04\u0e35\u0e22\u0e23\u0e48\u0e32\u0e21\u0e32\u0e01": [
|
| 1052 |
+
"\u0e40\u0e04\u0e35\u0e22\u0e23\u0e48\u0e32",
|
| 1053 |
+
"\u0e21\u0e32\u0e01"
|
| 1054 |
+
],
|
| 1055 |
+
"\u0e40\u0e14\u0e2d\u0e40\u0e21\u0e0b\u0e32\u0e01\u0e47": [
|
| 1056 |
+
"\u0e40\u0e14\u0e2d\u0e40\u0e21\u0e0b\u0e32",
|
| 1057 |
+
"\u0e01\u0e47"
|
| 1058 |
+
]
|
| 1059 |
+
},
|
| 1060 |
+
"PROPN+AUX": {
|
| 1061 |
+
"\u0e2d\u0e40\u0e21\u0e23\u0e34\u0e01\u0e32\u0e21\u0e32": [
|
| 1062 |
+
"\u0e2d\u0e40\u0e21\u0e23\u0e34\u0e01\u0e32",
|
| 1063 |
+
"\u0e21\u0e32"
|
| 1064 |
+
]
|
| 1065 |
+
},
|
| 1066 |
+
"PROPN+AUX+VERB": {
|
| 1067 |
+
"\u0e2d\u0e30\u0e41\u0e25\u0e2a\u0e01\u0e32\u0e21\u0e32\u0e19\u0e31\u0e1a": [
|
| 1068 |
+
"\u0e2d\u0e30\u0e41\u0e25\u0e2a\u0e01\u0e32",
|
| 1069 |
+
"\u0e21\u0e32",
|
| 1070 |
+
"\u0e19\u0e31\u0e1a"
|
| 1071 |
+
]
|
| 1072 |
+
},
|
| 1073 |
+
"PROPN+CCONJ": {
|
| 1074 |
+
"\u0e42\u0e23\u0e21\u0e34\u0e42\u0e2d\u0e01\u0e31\u0e1a": [
|
| 1075 |
+
"\u0e42\u0e23\u0e21\u0e34\u0e42\u0e2d",
|
| 1076 |
+
"\u0e01\u0e31\u0e1a"
|
| 1077 |
+
],
|
| 1078 |
+
"\u0e44\u0e2d\u0e23\u0e4c\u0e41\u0e25\u0e19\u0e14\u0e4c\u0e40\u0e2b\u0e19\u0e37\u0e2d\u0e01\u0e31\u0e1a": [
|
| 1079 |
+
"\u0e44\u0e2d\u0e23\u0e4c\u0e41\u0e25\u0e19\u0e14\u0e4c\u0e40\u0e2b\u0e19\u0e37\u0e2d",
|
| 1080 |
+
"\u0e01\u0e31\u0e1a"
|
| 1081 |
+
]
|
| 1082 |
+
},
|
| 1083 |
+
"PROPN+DET": {
|
| 1084 |
+
"\u0e2d\u0e2d\u0e2a\u0e40\u0e15\u0e23\u0e40\u0e25\u0e35\u0e22\u0e19\u0e35\u0e49": [
|
| 1085 |
+
"\u0e2d\u0e2d\u0e2a\u0e40\u0e15\u0e23\u0e40\u0e25\u0e35\u0e22",
|
| 1086 |
+
"\u0e19\u0e35\u0e49"
|
| 1087 |
+
]
|
| 1088 |
+
},
|
| 1089 |
+
"PROPN+DET+VERB": {
|
| 1090 |
+
"\u0e2d\u0e30\u0e41\u0e25\u0e2a\u0e01\u0e32\u0e17\u0e35\u0e48\u0e21\u0e35": [
|
| 1091 |
+
"\u0e2d\u0e30\u0e41\u0e25\u0e2a\u0e01\u0e32",
|
| 1092 |
+
"\u0e17\u0e35\u0e48",
|
| 1093 |
+
"\u0e21\u0e35"
|
| 1094 |
+
]
|
| 1095 |
+
},
|
| 1096 |
+
"PROPN+PROPN": {
|
| 1097 |
+
"\u0e42\u0e0b\u0e19\u0e35\u0e48\u0e2d\u0e2d\u0e2a\u0e40\u0e15\u0e23\u0e40\u0e25\u0e35\u0e22": [
|
| 1098 |
+
"\u0e42\u0e0b\u0e19\u0e35\u0e48",
|
| 1099 |
+
"\u0e2d\u0e2d\u0e2a\u0e40\u0e15\u0e23\u0e40\u0e25\u0e35\u0e22"
|
| 1100 |
+
]
|
| 1101 |
+
},
|
| 1102 |
+
"PROPN+VERB": {
|
| 1103 |
+
"\u0e01\u0e23\u0e35\u0e01\u0e23\u0e31\u0e1a": [
|
| 1104 |
+
"\u0e01\u0e23\u0e35\u0e01",
|
| 1105 |
+
"\u0e23\u0e31\u0e1a"
|
| 1106 |
+
],
|
| 1107 |
+
"\u0e1e\u0e35\u0e1a\u0e35\u0e22\u0e49\u0e32\u0e22": [
|
| 1108 |
+
"\u0e1e\u0e35\u0e1a\u0e35",
|
| 1109 |
+
"\u0e22\u0e49\u0e32\u0e22"
|
| 1110 |
+
],
|
| 1111 |
+
"\u0e2d\u0e30\u0e1e\u0e39\u0e40\u0e25\u0e35\u0e22\u0e19\u0e33": [
|
| 1112 |
+
"\u0e2d\u0e30\u0e1e\u0e39\u0e40\u0e25\u0e35\u0e22",
|
| 1113 |
+
"\u0e19\u0e33"
|
| 1114 |
+
],
|
| 1115 |
+
"\u0e2e\u0e34\u0e25\u0e25\u0e32\u0e23\u0e35\u0e48\u0e21\u0e35": [
|
| 1116 |
+
"\u0e2e\u0e34\u0e25\u0e25\u0e32\u0e23\u0e35\u0e48",
|
| 1117 |
+
"\u0e21\u0e35"
|
| 1118 |
+
],
|
| 1119 |
+
"\u0e40\u0e14\u0e42\u0e21\u0e41\u0e04\u0e23\u0e15\u0e23\u0e31\u0e01\u0e29\u0e32": [
|
| 1120 |
+
"\u0e40\u0e14\u0e42\u0e21\u0e41\u0e04\u0e23\u0e15",
|
| 1121 |
+
"\u0e23\u0e31\u0e01\u0e29\u0e32"
|
| 1122 |
+
],
|
| 1123 |
+
"\u0e44\u0e04\u0e42\u0e23\u0e21\u0e35": [
|
| 1124 |
+
"\u0e44\u0e04\u0e42\u0e23",
|
| 1125 |
+
"\u0e21\u0e35"
|
| 1126 |
+
]
|
| 1127 |
+
},
|
| 1128 |
+
"VERB+ADJ": {
|
| 1129 |
+
"\u0e0a\u0e14\u0e40\u0e0a\u0e22\u0e21\u0e2b\u0e32\u0e28\u0e32\u0e25": [
|
| 1130 |
+
"\u0e0a\u0e14\u0e40\u0e0a\u0e22",
|
| 1131 |
+
"\u0e21\u0e2b\u0e32\u0e28\u0e32\u0e25"
|
| 1132 |
+
]
|
| 1133 |
+
},
|
| 1134 |
+
"VERB+ADP": {
|
| 1135 |
+
"\u0e01\u0e25\u0e48\u0e32\u0e27\u0e2b\u0e32\u0e27\u0e48\u0e32": [
|
| 1136 |
+
"\u0e01\u0e25\u0e48\u0e32\u0e27\u0e2b\u0e32",
|
| 1137 |
+
"\u0e27\u0e48\u0e32"
|
| 1138 |
+
],
|
| 1139 |
+
"\u0e01\u0e32\u0e23\u0e40\u0e19\u0e49\u0e19\u0e22\u0e49\u0e33\u0e27\u0e48\u0e32": [
|
| 1140 |
+
"\u0e01\u0e32\u0e23\u0e40\u0e19\u0e49\u0e19\u0e22\u0e49\u0e33",
|
| 1141 |
+
"\u0e27\u0e48\u0e32"
|
| 1142 |
+
],
|
| 1143 |
+
"\u0e23\u0e31\u0e1a\u0e21\u0e37\u0e2d\u0e01\u0e31\u0e1a": [
|
| 1144 |
+
"\u0e23\u0e31\u0e1a\u0e21\u0e37\u0e2d",
|
| 1145 |
+
"\u0e01\u0e31\u0e1a"
|
| 1146 |
+
],
|
| 1147 |
+
"\u0e2d\u0e22\u0e39\u0e48\u0e17\u0e35\u0e48": [
|
| 1148 |
+
"\u0e2d\u0e22\u0e39\u0e48",
|
| 1149 |
+
"\u0e17\u0e35\u0e48"
|
| 1150 |
+
],
|
| 1151 |
+
"\u0e41\u0e22\u0e01\u0e23\u0e30\u0e2b\u0e27\u0e48\u0e32\u0e07": [
|
| 1152 |
+
"\u0e41\u0e22\u0e01",
|
| 1153 |
+
"\u0e23\u0e30\u0e2b\u0e27\u0e48\u0e32\u0e07"
|
| 1154 |
+
],
|
| 1155 |
+
"\u0e41\u0e2a\u0e14\u0e07\u0e42\u0e14\u0e22": [
|
| 1156 |
+
"\u0e41\u0e2a\u0e14\u0e07",
|
| 1157 |
+
"\u0e42\u0e14\u0e22"
|
| 1158 |
+
],
|
| 1159 |
+
"\u0e44\u0e14\u0e49\u0e41\u0e01\u0e48": [
|
| 1160 |
+
"\u0e44\u0e14\u0e49",
|
| 1161 |
+
"\u0e41\u0e01\u0e48"
|
| 1162 |
+
]
|
| 1163 |
+
},
|
| 1164 |
+
"VERB+ADP+NOUN": {
|
| 1165 |
+
"\u0e40\u0e01\u0e34\u0e14\u0e40\u0e21\u0e37\u0e48\u0e2d\u0e27\u0e31\u0e19\u0e17\u0e35\u0e48": [
|
| 1166 |
+
"\u0e40\u0e01\u0e34\u0e14",
|
| 1167 |
+
"\u0e40\u0e21\u0e37\u0e48\u0e2d",
|
| 1168 |
+
"\u0e27\u0e31\u0e19\u0e17\u0e35\u0e48"
|
| 1169 |
+
]
|
| 1170 |
+
},
|
| 1171 |
+
"VERB+ADV": {
|
| 1172 |
+
"\u0e2b\u0e32\u0e22\u0e32\u0e01": [
|
| 1173 |
+
"\u0e2b\u0e32",
|
| 1174 |
+
"\u0e22\u0e32\u0e01"
|
| 1175 |
+
]
|
| 1176 |
+
},
|
| 1177 |
+
"VERB+DET": {
|
| 1178 |
+
"\u0e02\u0e31\u0e1a\u0e44\u0e25\u0e48\u0e2d\u0e35\u0e01": [
|
| 1179 |
+
"\u0e02\u0e31\u0e1a\u0e44\u0e25\u0e48",
|
| 1180 |
+
"\u0e2d\u0e35\u0e01"
|
| 1181 |
+
],
|
| 1182 |
+
"\u0e40\u0e2a\u0e19\u0e2d\u0e1a\u0e32\u0e07": [
|
| 1183 |
+
"\u0e40\u0e2a\u0e19\u0e2d",
|
| 1184 |
+
"\u0e1a\u0e32\u0e07"
|
| 1185 |
+
]
|
| 1186 |
+
},
|
| 1187 |
+
"VERB+NOUN": {
|
| 1188 |
+
"\u0e17\u0e33\u0e43\u0e2b\u0e49\u0e19\u0e31\u0e01": [
|
| 1189 |
+
"\u0e17\u0e33\u0e43\u0e2b\u0e49",
|
| 1190 |
+
"\u0e19\u0e31\u0e01"
|
| 1191 |
+
],
|
| 1192 |
+
"\u0e1b\u0e25\u0e48\u0e2d\u0e22\u0e21\u0e25\u0e20\u0e32\u0e27\u0e30": [
|
| 1193 |
+
"\u0e1b\u0e25\u0e48\u0e2d\u0e22",
|
| 1194 |
+
"\u0e21\u0e25\u0e20\u0e32\u0e27\u0e30"
|
| 1195 |
+
],
|
| 1196 |
+
"\u0e1e\u0e36\u0e48\u0e07\u0e1e\u0e32\u0e19\u0e49\u0e33": [
|
| 1197 |
+
"\u0e1e\u0e36\u0e48\u0e07\u0e1e\u0e32",
|
| 1198 |
+
"\u0e19\u0e49\u0e33"
|
| 1199 |
+
],
|
| 1200 |
+
"\u0e2b\u0e32\u0e17\u0e30\u0e40\u0e25": [
|
| 1201 |
+
"\u0e2b\u0e32",
|
| 1202 |
+
"\u0e17\u0e30\u0e40\u0e25"
|
| 1203 |
+
],
|
| 1204 |
+
"\u0e2b\u0e32\u0e17\u0e32\u0e07": [
|
| 1205 |
+
"\u0e2b\u0e32",
|
| 1206 |
+
"\u0e17\u0e32\u0e07"
|
| 1207 |
+
],
|
| 1208 |
+
"\u0e2b\u0e32\u0e17\u0e38\u0e19": [
|
| 1209 |
+
"\u0e2b\u0e32",
|
| 1210 |
+
"\u0e17\u0e38\u0e19"
|
| 1211 |
+
],
|
| 1212 |
+
"\u0e2b\u0e32\u0e19\u0e32\u0e07\u0e1f\u0e49\u0e32": [
|
| 1213 |
+
"\u0e2b\u0e32",
|
| 1214 |
+
"\u0e19\u0e32\u0e07\u0e1f\u0e49\u0e32"
|
| 1215 |
+
],
|
| 1216 |
+
"\u0e2b\u0e32\u0e23\u0e32\u0e01": [
|
| 1217 |
+
"\u0e2b\u0e32",
|
| 1218 |
+
"\u0e23\u0e32\u0e01"
|
| 1219 |
+
],
|
| 1220 |
+
"\u0e43\u0e0a\u0e49\u0e19\u0e42\u0e22\u0e1a\u0e32\u0e22": [
|
| 1221 |
+
"\u0e43\u0e0a\u0e49",
|
| 1222 |
+
"\u0e19\u0e42\u0e22\u0e1a\u0e32\u0e22"
|
| 1223 |
+
],
|
| 1224 |
+
"\u0e43\u0e2b\u0e49\u0e19\u0e31\u0e01": [
|
| 1225 |
+
"\u0e43\u0e2b\u0e49",
|
| 1226 |
+
"\u0e19\u0e31\u0e01"
|
| 1227 |
+
],
|
| 1228 |
+
"\u0e43\u0e2b\u0e49\u0e19\u0e49\u0e33": [
|
| 1229 |
+
"\u0e43\u0e2b\u0e49",
|
| 1230 |
+
"\u0e19\u0e49\u0e33"
|
| 1231 |
+
]
|
| 1232 |
+
},
|
| 1233 |
+
"VERB+NOUN+DET": {
|
| 1234 |
+
"\u0e01\u0e32\u0e23\u0e42\u0e06\u0e29\u0e13\u0e32\u0e27\u0e34\u0e14\u0e35\u0e42\u0e2d\u0e19\u0e35\u0e49": [
|
| 1235 |
+
"\u0e01\u0e32\u0e23\u0e42\u0e06\u0e29\u0e13\u0e32",
|
| 1236 |
+
"\u0e27\u0e34\u0e14\u0e35\u0e42\u0e2d",
|
| 1237 |
+
"\u0e19\u0e35\u0e49"
|
| 1238 |
+
]
|
| 1239 |
+
},
|
| 1240 |
+
"VERB+PART": {
|
| 1241 |
+
"\u0e01\u0e25\u0e32\u0e22\u0e21\u0e32": [
|
| 1242 |
+
"\u0e01\u0e25\u0e32\u0e22",
|
| 1243 |
+
"\u0e21\u0e32"
|
| 1244 |
+
],
|
| 1245 |
+
"\u0e40\u0e2d\u0e32\u0e21\u0e32": [
|
| 1246 |
+
"\u0e40\u0e2d\u0e32",
|
| 1247 |
+
"\u0e21\u0e32"
|
| 1248 |
+
],
|
| 1249 |
+
"\u0e42\u0e1c\u0e25\u0e48\u0e21\u0e32": [
|
| 1250 |
+
"\u0e42\u0e1c\u0e25\u0e48",
|
| 1251 |
+
"\u0e21\u0e32"
|
| 1252 |
+
],
|
| 1253 |
+
"\u0e44\u0e25\u0e48\u0e2d\u0e2d\u0e01": [
|
| 1254 |
+
"\u0e44\u0e25\u0e48",
|
| 1255 |
+
"\u0e2d\u0e2d\u0e01"
|
| 1256 |
+
]
|
| 1257 |
+
},
|
| 1258 |
+
"VERB+PRON": {
|
| 1259 |
+
"\u0e1e\u0e36\u0e48\u0e07\u0e1e\u0e32\u0e21\u0e31\u0e19": [
|
| 1260 |
+
"\u0e1e\u0e36\u0e48\u0e07\u0e1e\u0e32",
|
| 1261 |
+
"\u0e21\u0e31\u0e19"
|
| 1262 |
+
],
|
| 1263 |
+
"\u0e40\u0e08\u0e2d\u0e01\u0e31\u0e19": [
|
| 1264 |
+
"\u0e40\u0e08\u0e2d",
|
| 1265 |
+
"\u0e01\u0e31\u0e19"
|
| 1266 |
+
]
|
| 1267 |
+
},
|
| 1268 |
+
"VERB+PROPN": {
|
| 1269 |
+
"\u0e01\u0e32\u0e23\u0e1b\u0e25\u0e48\u0e2d\u0e22\u0e21\u0e25\u0e20\u0e32\u0e27\u0e30": [
|
| 1270 |
+
"\u0e01\u0e32\u0e23\u0e1b\u0e25\u0e48\u0e2d\u0e22",
|
| 1271 |
+
"\u0e21\u0e25\u0e20\u0e32\u0e27\u0e30"
|
| 1272 |
+
]
|
| 1273 |
+
},
|
| 1274 |
+
"VERB+VERB": {
|
| 1275 |
+
"\u0e01\u0e32\u0e23\u0e04\u0e49\u0e19\u0e2b\u0e32\u0e22\u0e49\u0e2d\u0e19": [
|
| 1276 |
+
"\u0e01\u0e32\u0e23\u0e04\u0e49\u0e19\u0e2b\u0e32",
|
| 1277 |
+
"\u0e22\u0e49\u0e2d\u0e19"
|
| 1278 |
+
],
|
| 1279 |
+
"\u0e01\u0e32\u0e23\u0e41\u0e1a\u0e01\u0e23\u0e31\u0e1a": [
|
| 1280 |
+
"\u0e01\u0e32\u0e23\u0e41\u0e1a\u0e01",
|
| 1281 |
+
"\u0e23\u0e31\u0e1a"
|
| 1282 |
+
],
|
| 1283 |
+
"\u0e02\u0e2d\u0e01\u0e32\u0e23\u0e2a\u0e19\u0e31\u0e1a\u0e2a\u0e19\u0e38\u0e19": [
|
| 1284 |
+
"\u0e02\u0e2d",
|
| 1285 |
+
"\u0e01\u0e32\u0e23\u0e2a\u0e19\u0e31\u0e1a\u0e2a\u0e19\u0e38\u0e19"
|
| 1286 |
+
],
|
| 1287 |
+
"\u0e08\u0e31\u0e1a\u0e15\u0e32\u0e21\u0e2d\u0e07": [
|
| 1288 |
+
"\u0e08\u0e31\u0e1a\u0e15\u0e32",
|
| 1289 |
+
"\u0e21\u0e2d\u0e07"
|
| 1290 |
+
],
|
| 1291 |
+
"\u0e0a\u0e48\u0e27\u0e22\u0e40\u0e2b\u0e25\u0e37\u0e2d\u0e01\u0e32\u0e23\u0e41\u0e01\u0e49": [
|
| 1292 |
+
"\u0e0a\u0e48\u0e27\u0e22\u0e40\u0e2b\u0e25\u0e37\u0e2d",
|
| 1293 |
+
"\u0e01\u0e32\u0e23\u0e41\u0e01\u0e49"
|
| 1294 |
+
],
|
| 1295 |
+
"\u0e14\u0e39\u0e41\u0e25\u0e01\u0e32\u0e23\u0e40\u0e25\u0e37\u0e2d\u0e01\u0e15\u0e31\u0e49\u0e07": [
|
| 1296 |
+
"\u0e14\u0e39\u0e41\u0e25",
|
| 1297 |
+
"\u0e01\u0e32\u0e23\u0e40\u0e25\u0e37\u0e2d\u0e01\u0e15\u0e31\u0e49\u0e07"
|
| 1298 |
+
],
|
| 1299 |
+
"\u0e16\u0e37\u0e2d\u0e01\u0e32\u0e23\u0e1b\u0e0f\u0e34\u0e40\u0e2a\u0e18": [
|
| 1300 |
+
"\u0e16\u0e37\u0e2d",
|
| 1301 |
+
"\u0e01\u0e32\u0e23\u0e1b\u0e0f\u0e34\u0e40\u0e2a\u0e18"
|
| 1302 |
+
],
|
| 1303 |
+
"\u0e17\u0e33\u0e43\u0e2b\u0e49\u0e19\u0e48\u0e32\u0e2a\u0e19\u0e43\u0e08": [
|
| 1304 |
+
"\u0e17\u0e33\u0e43\u0e2b\u0e49",
|
| 1305 |
+
"\u0e19\u0e48\u0e32\u0e2a\u0e19\u0e43\u0e08"
|
| 1306 |
+
],
|
| 1307 |
+
"\u0e23\u0e49\u0e2d\u0e07\u0e02\u0e2d\u0e01\u0e25\u0e31\u0e1a": [
|
| 1308 |
+
"\u0e23\u0e49\u0e2d\u0e07\u0e02\u0e2d",
|
| 1309 |
+
"\u0e01\u0e25\u0e31\u0e1a"
|
| 1310 |
+
],
|
| 1311 |
+
"\u0e40\u0e1c\u0e32\u0e27\u0e2d\u0e14\u0e27\u0e32\u0e22": [
|
| 1312 |
+
"\u0e40\u0e1c\u0e32",
|
| 1313 |
+
"\u0e27\u0e2d\u0e14\u0e27\u0e32\u0e22"
|
| 1314 |
+
],
|
| 1315 |
+
"\u0e40\u0e23\u0e35\u0e22\u0e01\u0e27\u0e48\u0e32": [
|
| 1316 |
+
"\u0e40\u0e23\u0e35\u0e22\u0e01",
|
| 1317 |
+
"\u0e27\u0e48\u0e32"
|
| 1318 |
+
],
|
| 1319 |
+
"\u0e43\u0e2b\u0e49\u0e19\u0e31\u0e1a": [
|
| 1320 |
+
"\u0e43\u0e2b\u0e49",
|
| 1321 |
+
"\u0e19\u0e31\u0e1a"
|
| 1322 |
+
],
|
| 1323 |
+
"\u0e43\u0e2b\u0e49\u0e19\u0e36\u0e01": [
|
| 1324 |
+
"\u0e43\u0e2b\u0e49",
|
| 1325 |
+
"\u0e19\u0e36\u0e01"
|
| 1326 |
+
]
|
| 1327 |
+
}
|
| 1328 |
+
}
|
| 1329 |
+
},
|
| 1330 |
+
"tokenizer_class": "BertTokenizer",
|
| 1331 |
"torch_dtype": "float32",
|
| 1332 |
+
"transformers_version": "4.11.3",
|
| 1333 |
"type_vocab_size": 2,
|
| 1334 |
"use_cache": true,
|
| 1335 |
"vocab_size": 8506
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5285017898549a6e2c0dc7909ebea9f4b396bc6760c42d0fd0a292a90b21d0ce
|
| 3 |
+
size 368539791
|
tokenizer_config.json
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
{"do_lower_case": false, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "
|
|
|
|
| 1 |
+
{"do_lower_case": false, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "do_basic_tokenize": true, "never_split": null, "tokenizer_class": "BertTokenizer"}
|