initial release

README.md

@@ -0,0 +1,26 @@
+---
+language:
+- "zh"
+tags:
+- "chinese"
+- "masked-lm"
+- "wikipedia"
+license: "cc-by-sa-4.0"
+pipeline_tag: "fill-mask"
+mask_token: "[MASK]"
+---
+
+# deberta-base-chinese
+
+## Model Description
+
+This is a DeBERTa(V2) model pre-trained on Chinese Wikipedia texts (both simplified and traditional). NVIDIA A100-SXM4-40GB took 52 hours 10 minutes for training. You can fine-tune `deberta-base-chinese` for downstream tasks, such as [POS-tagging](https://huggingface.co/KoichiYasuoka/deberta-base-chinese-upos), [dependency-parsing](https://huggingface.co/KoichiYasuoka/deberta-base-chinese-ud-goeswith), and so on.
+
+## How to Use
+
+```py
+from transformers import AutoTokenizer,AutoModelForMaskedLM
+tokenizer=AutoTokenizer.from_pretrained("KoichiYasuoka/deberta-base-chinese")
+model=AutoModelForMaskedLM.from_pretrained("KoichiYasuoka/deberta-base-chinese")
+```
+

config.json

@@ -0,0 +1,29 @@
+{
+  "architectures": [
+    "DebertaV2ForMaskedLM"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-07,
+  "max_position_embeddings": 512,
+  "max_relative_positions": -1,
+  "model_type": "deberta-v2",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "pooler_dropout": 0,
+  "pooler_hidden_act": "gelu",
+  "pooler_hidden_size": 768,
+  "pos_att_type": null,
+  "position_biased_input": true,
+  "relative_attention": false,
+  "tokenizer_class": "BertTokenizerFast",
+  "torch_dtype": "float32",
+  "transformers_version": "4.22.1",
+  "type_vocab_size": 0,
+  "vocab_size": 26582
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e62658c30ba60e956f8958c6c44bddc57db9dbf85b38235f040e38afbb711f47
+size 426001005

special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

tokenizer_config.json

@@ -0,0 +1,20 @@
+{
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": false,
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "never_split": [
+    "[CLS]",
+    "[PAD]",
+    "[SEP]",
+    "[UNK]",
+    "[MASK]"
+  ],
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": false,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizerFast",
+  "unk_token": "[UNK]"
+}