KoichiYasuoka commited on
Commit
8d33077
β€’
1 Parent(s): b020f81

model improved

Browse files
Files changed (5) hide show
  1. README.md +1 -1
  2. config.json +1 -7
  3. pytorch_model.bin +2 -2
  4. tokenizer_config.json +9 -3
  5. vocab.txt +0 -0
README.md CHANGED
@@ -16,7 +16,7 @@ pipeline_tag: "token-classification"
16
 
17
  ## Model Description
18
 
19
- This is a RoBERTa model pre-trained on Chinese texts (both simplified and traditional) for POS-tagging and dependency-parsing (using `goeswith` for subwords), derived from [roberta-base-chinese-upos](https://huggingface.co/KoichiYasuoka/roberta-base-chinese-upos).
20
 
21
  ## How to Use
22
 
 
16
 
17
  ## Model Description
18
 
19
+ This is a RoBERTa model pre-trained on Chinese Wikipedia texts (both simplified and traditional) for POS-tagging and dependency-parsing (using `goeswith` for subwords), derived from [roberta-base-chinese-upos](https://huggingface.co/KoichiYasuoka/roberta-base-chinese-upos).
20
 
21
  ## How to Use
22
 
config.json CHANGED
@@ -10,7 +10,6 @@
10
  "impl": "ud.UniversalDependenciesPipeline"
11
  }
12
  },
13
- "directionality": "bidi",
14
  "eos_token_id": 2,
15
  "hidden_act": "gelu",
16
  "hidden_dropout_prob": 0.1,
@@ -633,16 +632,11 @@
633
  "num_attention_heads": 12,
634
  "num_hidden_layers": 12,
635
  "pad_token_id": 1,
636
- "pooler_fc_size": 768,
637
- "pooler_num_attention_heads": 12,
638
- "pooler_num_fc_layers": 3,
639
- "pooler_size_per_head": 128,
640
- "pooler_type": "first_token_transform",
641
  "position_embedding_type": "absolute",
642
  "tokenizer_class": "BertTokenizerFast",
643
  "torch_dtype": "float32",
644
  "transformers_version": "4.22.1",
645
  "type_vocab_size": 2,
646
  "use_cache": true,
647
- "vocab_size": 21597
648
  }
 
10
  "impl": "ud.UniversalDependenciesPipeline"
11
  }
12
  },
 
13
  "eos_token_id": 2,
14
  "hidden_act": "gelu",
15
  "hidden_dropout_prob": 0.1,
 
632
  "num_attention_heads": 12,
633
  "num_hidden_layers": 12,
634
  "pad_token_id": 1,
 
 
 
 
 
635
  "position_embedding_type": "absolute",
636
  "tokenizer_class": "BertTokenizerFast",
637
  "torch_dtype": "float32",
638
  "transformers_version": "4.22.1",
639
  "type_vocab_size": 2,
640
  "use_cache": true,
641
+ "vocab_size": 26582
642
  }
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d46ab28f34d65e8cd28c870bc9ca6e8d10d012e9a900df272620a292b67f5825
3
- size 409151985
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b4195111b8295c7cfa9abd1def9e84fef30abb99a4e61d33285b8642ee3bd1f
3
+ size 424465905
tokenizer_config.json CHANGED
@@ -1,13 +1,19 @@
1
  {
2
  "cls_token": "[CLS]",
3
  "do_basic_tokenize": true,
4
- "do_lower_case": true,
5
  "mask_token": "[MASK]",
6
  "model_max_length": 512,
7
- "never_split": null,
 
 
 
 
 
 
8
  "pad_token": "[PAD]",
9
  "sep_token": "[SEP]",
10
- "strip_accents": null,
11
  "tokenize_chinese_chars": true,
12
  "tokenizer_class": "BertTokenizerFast",
13
  "unk_token": "[UNK]"
 
1
  {
2
  "cls_token": "[CLS]",
3
  "do_basic_tokenize": true,
4
+ "do_lower_case": false,
5
  "mask_token": "[MASK]",
6
  "model_max_length": 512,
7
+ "never_split": [
8
+ "[CLS]",
9
+ "[PAD]",
10
+ "[SEP]",
11
+ "[UNK]",
12
+ "[MASK]"
13
+ ],
14
  "pad_token": "[PAD]",
15
  "sep_token": "[SEP]",
16
+ "strip_accents": false,
17
  "tokenize_chinese_chars": true,
18
  "tokenizer_class": "BertTokenizerFast",
19
  "unk_token": "[UNK]"
vocab.txt CHANGED
The diff for this file is too large to render. See raw diff