Upload 7 files

README.md

@@ -0,0 +1,18 @@
+---
+language: ko
+tags:
+- text-generation
+---
+
+# Bert base model for Korean
+
+* 70GB Korean text dataset and 42000 lower-cased subwords are used
+* Check the model performance and other language models for Korean in [github](https://github.com/kiyoungkim1/LM-kor)
+
+```python
+from transformers import BertTokenizerFast, GPT2LMHeadModel
+tokenizer_gpt3 = BertTokenizerFast.from_pretrained("kykim/gpt3-kor-small_based_on_gpt2")
+input_ids = tokenizer_gpt3.encode("text to tokenize")[1:]  # remove cls token
+        
+model_gpt3 = GPT2LMHeadModel.from_pretrained("kykim/gpt3-kor-small_based_on_gpt2")
+```

config.json

@@ -0,0 +1,28 @@
+{
+  "activation_function": "gelu_new",
+  "architectures": [
+    "GPT2LMHeadModel"
+  ],
+  "attn_pdrop": 0.1,
+  "bos_token_id": 3,
+  "embd_pdrop": 0.1,
+  "eos_token_id": 3,
+  "gradient_checkpointing": false,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "model_type": "gpt2",
+  "n_ctx": 2048,
+  "n_embd": 768,
+  "n_head": 12,
+  "n_inner": null,
+  "n_layer": 12,
+  "n_positions": 2048,
+  "resid_pdrop": 0.1,
+  "summary_activation": null,
+  "summary_first_dropout": 0.1,
+  "summary_proj_to_labels": true,
+  "summary_type": "cls_index",
+  "summary_use_proj": true,
+  "use_cache": true,
+  "vocab_size": 42000
+}

flax_model.msgpack

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e85928f3a07718f939b31cd8d638d403d9b5cdf3c4fb034df4bfcb2c6146710b
+size 475544331

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee4d397c62f57ae051a730dc611a67720bbcd180c600c738232190da5e2dc3f6
+size 525899459

tf_model.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b51889e01d4e7f8686fd79de2bce64834d07bab85c7306ea8c7fee3cf4615d0
+size 475713872

tokenizer_config.json

@@ -0,0 +1,6 @@
+{
+  "do_lower_case": true,
+  "strip_accents": false,
+  "model_max_length": 2048,
+  "tokenizer_class": "BertTokenizer"
+}