Upload model files.

README.md

@@ -0,0 +1,47 @@
+---
+language:
+- zh
+thumbnail: https://ckip.iis.sinica.edu.tw/files/ckip_logo.png
+tags:
+- pytorch
+- token-classification
+- bert
+- zh
+license: gpl-3.0
+datasets:
+metrics:
+---
+
+# CKIP BERT Tiny Chinese
+
+This project provides traditional Chinese transformers models (including ALBERT, BERT, GPT2) and NLP tools (including word segmentation, part-of-speech tagging, named entity recognition).
+
+這個專案提供了繁體中文的 transformers 模型（包含 ALBERT、BERT、GPT2）及自然語言處理工具（包含斷詞、詞性標記、實體辨識）。
+
+## Homepage
+
+* https://github.com/ckiplab/ckip-transformers
+
+## Contributers
+
+* [Mu Yang](https://muyang.pro) at [CKIP](https://ckip.iis.sinica.edu.tw) (Author & Maintainer)
+
+## Usage
+
+Please use BertTokenizerFast as tokenizer instead of AutoTokenizer.
+
+請使用 BertTokenizerFast 而非 AutoTokenizer。
+
+```
+from transformers import (
+  BertTokenizerFast,
+  AutoModel,
+)
+
+tokenizer = BertTokenizerFast.from_pretrained('bert-base-chinese')
+model = AutoModel.from_pretrained('ckiplab/bert-tiny-chinese-pos')
+```
+
+For full usage and more information, please refer to https://github.com/ckiplab/ckip-transformers.
+
+有關完整使用方法及其他資訊，請參見 https://github.com/ckiplab/ckip-transformers 。

config.json

@@ -0,0 +1,152 @@
+{
+  "_name_or_path": "../../../model/bert-tiny-scratch-lm",
+  "architectures": [
+    "BertForTokenClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "directionality": "bidi",
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 312,
+  "id2label": {
+    "0": "A",
+    "1": "Caa",
+    "2": "Cab",
+    "3": "Cba",
+    "4": "Cbb",
+    "5": "D",
+    "6": "Da",
+    "7": "Dfa",
+    "8": "Dfb",
+    "9": "Di",
+    "10": "Dk",
+    "11": "DM",
+    "12": "I",
+    "13": "Na",
+    "14": "Nb",
+    "15": "Nc",
+    "16": "Ncd",
+    "17": "Nd",
+    "18": "Nep",
+    "19": "Neqa",
+    "20": "Neqb",
+    "21": "Nes",
+    "22": "Neu",
+    "23": "Nf",
+    "24": "Ng",
+    "25": "Nh",
+    "26": "Nv",
+    "27": "P",
+    "28": "T",
+    "29": "VA",
+    "30": "VAC",
+    "31": "VB",
+    "32": "VC",
+    "33": "VCL",
+    "34": "VD",
+    "35": "VF",
+    "36": "VE",
+    "37": "VG",
+    "38": "VH",
+    "39": "VHC",
+    "40": "VI",
+    "41": "VJ",
+    "42": "VK",
+    "43": "VL",
+    "44": "V_2",
+    "45": "DE",
+    "46": "SHI",
+    "47": "FW",
+    "48": "COLONCATEGORY",
+    "49": "COMMACATEGORY",
+    "50": "DASHCATEGORY",
+    "51": "DOTCATEGORY",
+    "52": "ETCCATEGORY",
+    "53": "EXCLAMATIONCATEGORY",
+    "54": "PARENTHESISCATEGORY",
+    "55": "PAUSECATEGORY",
+    "56": "PERIODCATEGORY",
+    "57": "QUESTIONCATEGORY",
+    "58": "SEMICOLONCATEGORY",
+    "59": "SPCHANGECATEGORY"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 1248,
+  "label2id": {
+    "A": 0,
+    "COLONCATEGORY": 48,
+    "COMMACATEGORY": 49,
+    "Caa": 1,
+    "Cab": 2,
+    "Cba": 3,
+    "Cbb": 4,
+    "D": 5,
+    "DASHCATEGORY": 50,
+    "DE": 45,
+    "DM": 11,
+    "DOTCATEGORY": 51,
+    "Da": 6,
+    "Dfa": 7,
+    "Dfb": 8,
+    "Di": 9,
+    "Dk": 10,
+    "ETCCATEGORY": 52,
+    "EXCLAMATIONCATEGORY": 53,
+    "FW": 47,
+    "I": 12,
+    "Na": 13,
+    "Nb": 14,
+    "Nc": 15,
+    "Ncd": 16,
+    "Nd": 17,
+    "Nep": 18,
+    "Neqa": 19,
+    "Neqb": 20,
+    "Nes": 21,
+    "Neu": 22,
+    "Nf": 23,
+    "Ng": 24,
+    "Nh": 25,
+    "Nv": 26,
+    "P": 27,
+    "PARENTHESISCATEGORY": 54,
+    "PAUSECATEGORY": 55,
+    "PERIODCATEGORY": 56,
+    "QUESTIONCATEGORY": 57,
+    "SEMICOLONCATEGORY": 58,
+    "SHI": 46,
+    "SPCHANGECATEGORY": 59,
+    "T": 28,
+    "VA": 29,
+    "VAC": 30,
+    "VB": 31,
+    "VC": 32,
+    "VCL": 33,
+    "VD": 34,
+    "VE": 36,
+    "VF": 35,
+    "VG": 37,
+    "VH": 38,
+    "VHC": 39,
+    "VI": 40,
+    "VJ": 41,
+    "VK": 42,
+    "VL": 43,
+    "V_2": 44
+  },
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 4,
+  "pad_token_id": 0,
+  "pooler_fc_size": 312,
+  "pooler_num_attention_heads": 12,
+  "pooler_num_fc_layers": 3,
+  "pooler_size_per_head": 128,
+  "pooler_type": "first_token_transform",
+  "tokenizer_class": "BertTokenizerFast",
+  "type_vocab_size": 2,
+  "vocab_size": 21128
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a35cc03bfebeabbd283208389b29664a7ddf613ffc03cb24f4d266f9f11243e5
+size 45874807

special_tokens_map.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": false, "do_basic_tokenize": true, "never_split": null, "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, "name_or_path": "bert-base-chinese"}