root commited on
Commit
ca595c6
1 Parent(s): e8f8d66
README.md CHANGED
@@ -1,3 +1,64 @@
1
  ---
2
- license: c-uda
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ language: "pt"
3
+ widget:
4
+ - text: "Tinha uma pedra no meio do caminho."
5
+ - text: "Vamos tomar um café quentinho?"
6
+ - text: "Como você se chama?"
7
+ datasets:
8
+ - MacMorpho
9
  ---
10
+
11
+ # POS-Tagger Portuguese
12
+
13
+ We fine-tuned the [BERTimbau](https://github.com/neuralmind-ai/portuguese-bert/) model with the [MacMorpho](http://nilc.icmc.usp.br/macmorpho/) corpus for the Post-Tagger task, with 10 epochs, achieving a general F1-Score of 0.9826.
14
+
15
+ Metrics:
16
+
17
+ ```
18
+ Precision Recall F1 Suport
19
+ accuracy 0.98 33729
20
+ macro avg 0.96 0.95 0.95 33729
21
+ weighted avg 0.98 0.98 0.98 33729
22
+
23
+ F1: 0.9826 Accuracy: 0.9826
24
+ ```
25
+ Parameters:
26
+ ```
27
+ nclasses = 27
28
+ nepochs = 30
29
+ batch_size = 32
30
+ batch_status = 32
31
+ learning_rate = 1e-5
32
+ early_stop = 3
33
+ max_length = 200
34
+ ```
35
+ Tags:
36
+
37
+ | Tag | Meaning |
38
+ | ------------------- | ------------------- |
39
+ | ADJ | Adjetivo |
40
+ | ADV | Advérbio |
41
+ | ADV-KS | Advérbio conjuntivo subordinado |
42
+ | ADV-KS-REL | Advérbio relativo subordinado |
43
+ | ART | Artigo |
44
+ | CUR | Moeda |
45
+ | IN | Interjeição |
46
+ | KC | Conjunção coordenativa |
47
+ | KS | Conjunção subordinativa |
48
+ | N | Substantivo |
49
+ | NPROP | Substantivo próprio |
50
+ | NUM | Número |
51
+ | PCP | Particípio |
52
+ | PDEN | Palavra denotativa |
53
+ | PREP | Preposição |
54
+ | PROADJ | Pronome Adjetivo |
55
+ | PRO-KS | Pronome conjuntivo subordinado |
56
+ | PRO-KS-REL | Pronome relativo conectivo subordinado |
57
+ | PROPESS | Pronome pessoal |
58
+ | PROSUB | Pronome nominal |
59
+ | V | Verbo |
60
+ | VAUX | Verbo auxiliar |
61
+
62
+ ## Questions?
63
+
64
+ Please, post a Github issue on the [NLP Portuguese POS-Tagger](https://github.com/lisaterumi/nlp-portuguese-postagger).
added_tokens.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {}
config.json ADDED
@@ -0,0 +1,90 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "neuralmind/bert-base-portuguese-cased",
3
+ "architectures": [
4
+ "BertForTokenClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "classifier_dropout": null,
8
+ "directionality": "bidi",
9
+ "hidden_act": "gelu",
10
+ "hidden_dropout_prob": 0.1,
11
+ "hidden_size": 768,
12
+ "id2label": {
13
+ "0": "NUM",
14
+ "1": "KS",
15
+ "2": "PREP+PROADJ",
16
+ "3": "ADV-KS",
17
+ "4": "NPROP",
18
+ "5": "PDEN",
19
+ "6": "PROADJ",
20
+ "7": "PCP",
21
+ "8": "KC",
22
+ "9": "PU",
23
+ "10": "PREP",
24
+ "11": "ADV",
25
+ "12": "PRO-KS",
26
+ "13": "ART",
27
+ "14": "N",
28
+ "15": "PROPESS",
29
+ "16": "PREP+PROPESS",
30
+ "17": "CUR",
31
+ "18": "ADJ",
32
+ "19": "IN",
33
+ "20": "PREP+ART",
34
+ "21": "PROSUB",
35
+ "22": "PREP+PRO-KS",
36
+ "23": "PREP+PROSUB",
37
+ "24": "V",
38
+ "25": "PREP+ADV",
39
+ "26": "<pad>"
40
+ },
41
+ "initializer_range": 0.02,
42
+ "intermediate_size": 3072,
43
+ "label2id": {
44
+ "<pad>": 26,
45
+ "ADJ": 18,
46
+ "ADV": 11,
47
+ "ADV-KS": 3,
48
+ "ART": 13,
49
+ "CUR": 17,
50
+ "IN": 19,
51
+ "KC": 8,
52
+ "KS": 1,
53
+ "N": 14,
54
+ "NPROP": 4,
55
+ "NUM": 0,
56
+ "PCP": 7,
57
+ "PDEN": 5,
58
+ "PREP": 10,
59
+ "PREP+ADV": 25,
60
+ "PREP+ART": 20,
61
+ "PREP+PRO-KS": 22,
62
+ "PREP+PROADJ": 2,
63
+ "PREP+PROPESS": 16,
64
+ "PREP+PROSUB": 23,
65
+ "PRO-KS": 12,
66
+ "PROADJ": 6,
67
+ "PROPESS": 15,
68
+ "PROSUB": 21,
69
+ "PU": 9,
70
+ "V": 24
71
+ },
72
+ "layer_norm_eps": 1e-12,
73
+ "max_position_embeddings": 512,
74
+ "model_type": "bert",
75
+ "num_attention_heads": 12,
76
+ "num_hidden_layers": 12,
77
+ "output_past": true,
78
+ "pad_token_id": 0,
79
+ "pooler_fc_size": 768,
80
+ "pooler_num_attention_heads": 12,
81
+ "pooler_num_fc_layers": 3,
82
+ "pooler_size_per_head": 128,
83
+ "pooler_type": "first_token_transform",
84
+ "position_embedding_type": "absolute",
85
+ "torch_dtype": "float32",
86
+ "transformers_version": "4.20.1",
87
+ "type_vocab_size": 2,
88
+ "use_cache": true,
89
+ "vocab_size": 29794
90
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3f2e1ccd99a2a28417be61caacc92297b97f08e7ff0591d0adc35b2564b984b
3
+ size 433440760
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:136fe2d35d147f464be9f1a08c2e225aab419019a65d5267fd246d6dde79d9b2
3
+ size 433483633
special_tokens_map.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": "[CLS]",
3
+ "mask_token": "[MASK]",
4
+ "pad_token": "[PAD]",
5
+ "sep_token": "[SEP]",
6
+ "unk_token": "[UNK]"
7
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": "[CLS]",
3
+ "do_basic_tokenize": true,
4
+ "do_lower_case": false,
5
+ "mask_token": "[MASK]",
6
+ "name_or_path": "neuralmind/bert-base-portuguese-cased",
7
+ "never_split": null,
8
+ "pad_token": "[PAD]",
9
+ "sep_token": "[SEP]",
10
+ "special_tokens_map_file": "/root/.cache/huggingface/transformers/eecc45187d085a1169eed91017d358cc0e9cbdd5dc236bcd710059dbf0a2f816.dd8bd9bfd3664b530ea4e645105f557769387b3da9f79bdb55ed556bdd80611d",
11
+ "strip_accents": null,
12
+ "tokenize_chinese_chars": true,
13
+ "tokenizer_class": "BertTokenizer",
14
+ "unk_token": "[UNK]"
15
+ }
vocab.txt ADDED
The diff for this file is too large to render. See raw diff