Acdong commited on
Commit
ed43a08
·
1 Parent(s): 058faf5
Files changed (6) hide show
  1. LICENSE.txt +238 -0
  2. README.txt +18 -0
  3. config.json +26 -0
  4. pytorch_model.bin +3 -0
  5. spiece.model +3 -0
  6. tokenizer_config.json +1 -0
LICENSE.txt ADDED
@@ -0,0 +1,238 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+
3
+ 엑소브레인 한국어 언어모델(ET5) 라이선스
4
+
5
+
6
+
7
+ 1. 엑소브레인 한국어 언어모델(ET5)는
8
+ 연구용으로 활용 시, Apache2.0 라이선스를 따르고
9
+ 상업용으로 활용 시, 상업적 활용 정보 제공 의무를 가진다.
10
+
11
+
12
+ 2. ET5 모델을 상업적으로 활용하고자 할 경우,
13
+ 아래 활용 정보를 ETRI 언어지능연구실에 제공한 협약자만이 상업적 사용권을 가진다.
14
+ [제출처 및 제출 주기]
15
+ - 제출처: ETRI 언어지능연구실 배경만 PL / [email protected]
16
+ - 제출주기: 6개월 단위 제출
17
+
18
+ [상업적 활용 시 제공 정보]
19
+ - 활용 주체 (i.e. 회사명)
20
+ - 활용 서비스 (i.e. 서비스명 및 소개)
21
+ - 예상 서비스 기간 (i.e. 기간)
22
+ - 담당자 및 연락처
23
+
24
+
25
+ 3. ET5 모델의 연구용 활용은 Apache 2.0 라이선스를 따른다.
26
+ - 아래 Apache2.0 LICENSE 참조
27
+
28
+
29
+ 4. 상업적 용도로 특정 도메인에 특화된 ET5 모델을 개발하여 활용하고자 하는 경우의 연락처는 아래와 같다.
30
+ - 문의처: ETRI 언어지능연구실 배경만 PL / [email protected]
31
+ - 유상 기술이전 유형
32
+
33
+
34
+ ------------------------------------------------------------------------------------------
35
+
36
+
37
+ Apache License
38
+ Version 2.0, January 2004
39
+ http://www.apache.org/licenses/
40
+
41
+
42
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
43
+
44
+ 1. Definitions.
45
+
46
+ "License" shall mean the terms and conditions for use, reproduction,
47
+ and distribution as defined by Sections 1 through 9 of this document.
48
+
49
+ "Licensor" shall mean the copyright owner or entity authorized by
50
+ the copyright owner that is granting the License.
51
+
52
+ "Legal Entity" shall mean the union of the acting entity and all
53
+ other entities that control, are controlled by, or are under common
54
+ control with that entity. For the purposes of this definition,
55
+ "control" means (i) the power, direct or indirect, to cause the
56
+ direction or management of such entity, whether by contract or
57
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
58
+ outstanding shares, or (iii) beneficial ownership of such entity.
59
+
60
+ "You" (or "Your") shall mean an individual or Legal Entity
61
+ exercising permissions granted by this License.
62
+
63
+ "Source" form shall mean the preferred form for making modifications,
64
+ including but not limited to software source code, documentation
65
+ source, and configuration files.
66
+
67
+ "Object" form shall mean any form resulting from mechanical
68
+ transformation or translation of a Source form, including but
69
+ not limited to compiled object code, generated documentation,
70
+ and conversions to other media types.
71
+
72
+ "Work" shall mean the work of authorship, whether in Source or
73
+ Object form, made available under the License, as indicated by a
74
+ copyright notice that is included in or attached to the work
75
+ (an example is provided in the Appendix below).
76
+
77
+ "Derivative Works" shall mean any work, whether in Source or Object
78
+ form, that is based on (or derived from) the Work and for which the
79
+ editorial revisions, annotations, elaborations, or other modifications
80
+ represent, as a whole, an original work of authorship. For the purposes
81
+ of this License, Derivative Works shall not include works that remain
82
+ separable from, or merely link (or bind by name) to the interfaces of,
83
+ the Work and Derivative Works thereof.
84
+
85
+ "Contribution" shall mean any work of authorship, including
86
+ the original version of the Work and any modifications or additions
87
+ to that Work or Derivative Works thereof, that is intentionally
88
+ submitted to Licensor for inclusion in the Work by the copyright owner
89
+ or by an individual or Legal Entity authorized to submit on behalf of
90
+ the copyright owner. For the purposes of this definition, "submitted"
91
+ means any form of electronic, verbal, or written communication sent
92
+ to the Licensor or its representatives, including but not limited to
93
+ communication on electronic mailing lists, source code control systems,
94
+ and issue tracking systems that are managed by, or on behalf of, the
95
+ Licensor for the purpose of discussing and improving the Work, but
96
+ excluding communication that is conspicuously marked or otherwise
97
+ designated in writing by the copyright owner as "Not a Contribution."
98
+
99
+ "Contributor" shall mean Licensor and any individual or Legal Entity
100
+ on behalf of whom a Contribution has been received by Licensor and
101
+ subsequently incorporated within the Work.
102
+
103
+ 2. Grant of Copyright License. Subject to the terms and conditions of
104
+ this License, each Contributor hereby grants to You a perpetual,
105
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
106
+ copyright license to reproduce, prepare Derivative Works of,
107
+ publicly display, publicly perform, sublicense, and distribute the
108
+ Work and such Derivative Works in Source or Object form.
109
+
110
+ 3. Grant of Patent License. Subject to the terms and conditions of
111
+ this License, each Contributor hereby grants to You a perpetual,
112
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
113
+ (except as stated in this section) patent license to make, have made,
114
+ use, offer to sell, sell, import, and otherwise transfer the Work,
115
+ where such license applies only to those patent claims licensable
116
+ by such Contributor that are necessarily infringed by their
117
+ Contribution(s) alone or by combination of their Contribution(s)
118
+ with the Work to which such Contribution(s) was submitted. If You
119
+ institute patent litigation against any entity (including a
120
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
121
+ or a Contribution incorporated within the Work constitutes direct
122
+ or contributory patent infringement, then any patent licenses
123
+ granted to You under this License for that Work shall terminate
124
+ as of the date such litigation is filed.
125
+
126
+ 4. Redistribution. You may reproduce and distribute copies of the
127
+ Work or Derivative Works thereof in any medium, with or without
128
+ modifications, and in Source or Object form, provided that You
129
+ meet the following conditions:
130
+
131
+ (a) You must give any other recipients of the Work or
132
+ Derivative Works a copy of this License; and
133
+
134
+ (b) You must cause any modified files to carry prominent notices
135
+ stating that You changed the files; and
136
+
137
+ (c) You must retain, in the Source form of any Derivative Works
138
+ that You distribute, all copyright, patent, trademark, and
139
+ attribution notices from the Source form of the Work,
140
+ excluding those notices that do not pertain to any part of
141
+ the Derivative Works; and
142
+
143
+ (d) If the Work includes a "NOTICE" text file as part of its
144
+ distribution, then any Derivative Works that You distribute must
145
+ include a readable copy of the attribution notices contained
146
+ within such NOTICE file, excluding those notices that do not
147
+ pertain to any part of the Derivative Works, in at least one
148
+ of the following places: within a NOTICE text file distributed
149
+ as part of the Derivative Works; within the Source form or
150
+ documentation, if provided along with the Derivative Works; or,
151
+ within a display generated by the Derivative Works, if and
152
+ wherever such third-party notices normally appear. The contents
153
+ of the NOTICE file are for informational purposes only and
154
+ do not modify the License. You may add Your own attribution
155
+ notices within Derivative Works that You distribute, alongside
156
+ or as an addendum to the NOTICE text from the Work, provided
157
+ that such additional attribution notices cannot be construed
158
+ as modifying the License.
159
+
160
+ You may add Your own copyright statement to Your modifications and
161
+ may provide additional or different license terms and conditions
162
+ for use, reproduction, or distribution of Your modifications, or
163
+ for any such Derivative Works as a whole, provided Your use,
164
+ reproduction, and distribution of the Work otherwise complies with
165
+ the conditions stated in this License.
166
+
167
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
168
+ any Contribution intentionally submitted for inclusion in the Work
169
+ by You to the Licensor shall be under the terms and conditions of
170
+ this License, without any additional terms or conditions.
171
+ Notwithstanding the above, nothing herein shall supersede or modify
172
+ the terms of any separate license agreement you may have executed
173
+ with Licensor regarding such Contributions.
174
+
175
+ 6. Trademarks. This License does not grant permission to use the trade
176
+ names, trademarks, service marks, or product names of the Licensor,
177
+ except as required for reasonable and customary use in describing the
178
+ origin of the Work and reproducing the content of the NOTICE file.
179
+
180
+ 7. Disclaimer of Warranty. Unless required by applicable law or
181
+ agreed to in writing, Licensor provides the Work (and each
182
+ Contributor provides its Contributions) on an "AS IS" BASIS,
183
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
184
+ implied, including, without limitation, any warranties or conditions
185
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
186
+ PARTICULAR PURPOSE. You are solely responsible for determining the
187
+ appropriateness of using or redistributing the Work and assume any
188
+ risks associated with Your exercise of permissions under this License.
189
+
190
+ 8. Limitation of Liability. In no event and under no legal theory,
191
+ whether in tort (including negligence), contract, or otherwise,
192
+ unless required by applicable law (such as deliberate and grossly
193
+ negligent acts) or agreed to in writing, shall any Contributor be
194
+ liable to You for damages, including any direct, indirect, special,
195
+ incidental, or consequential damages of any character arising as a
196
+ result of this License or out of the use or inability to use the
197
+ Work (including but not limited to damages for loss of goodwill,
198
+ work stoppage, computer failure or malfunction, or any and all
199
+ other commercial damages or losses), even if such Contributor
200
+ has been advised of the possibility of such damages.
201
+
202
+ 9. Accepting Warranty or Additional Liability. While redistributing
203
+ the Work or Derivative Works thereof, You may choose to offer,
204
+ and charge a fee for, acceptance of support, warranty, indemnity,
205
+ or other liability obligations and/or rights consistent with this
206
+ License. However, in accepting such obligations, You may act only
207
+ on Your own behalf and on Your sole responsibility, not on behalf
208
+ of any other Contributor, and only if You agree to indemnify,
209
+ defend, and hold each Contributor harmless for any liability
210
+ incurred by, or claims asserted against, such Contributor by reason
211
+ of your accepting any such warranty or additional liability.
212
+
213
+ END OF TERMS AND CONDITIONS
214
+
215
+ APPENDIX: How to apply the Apache License to your work.
216
+
217
+ To apply the Apache License to your work, attach the following
218
+ boilerplate notice, with the fields enclosed by brackets "[]"
219
+ replaced with your own identifying information. (Don't include
220
+ the brackets!) The text should be enclosed in the appropriate
221
+ comment syntax for the file format. We also recommend that a
222
+ file or class name and description of purpose be included on the
223
+ same "printed page" as the copyright notice for easier
224
+ identification within third-party archives.
225
+
226
+ Copyright [yyyy] [name of copyright owner]
227
+
228
+ Licensed under the Apache License, Version 2.0 (the "License");
229
+ you may not use this file except in compliance with the License.
230
+ You may obtain a copy of the License at
231
+
232
+ http://www.apache.org/licenses/LICENSE-2.0
233
+
234
+ Unless required by applicable law or agreed to in writing, software
235
+ distributed under the License is distributed on an "AS IS" BASIS,
236
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
237
+ See the License for the specific language governing permissions and
238
+ limitations under the License.
README.txt ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 한국어 언어모델(ET5) README
2
+
3
+ >>> 파일에 대한 기본 설명
4
+ - "config.json": 언어모델의 구조 정보를 담고 있는 파일
5
+ - "pytorch_model.bin": 언어모델 weight를 저장한 파일
6
+ - "spiece.model": vocab 모델 파일
7
+ - "tokenizer_config.json": 토크나이저 구성 정보 파일
8
+
9
+
10
+ >>> 기초 성능 정보
11
+ - KorQuAD-1.0: F1_94.26-EM_86.37
12
+ - AI Hub 요약 뉴스-abstractive: ROUGE-1/2/L 50.05 / 22.98 / 47.37
13
+
14
+
15
+ >>> 활용 정보(Prerequisite)
16
+ - Huggingface Transformers (v4.3.2) https://github.com/huggingface/transformers
17
+ - SentencePiece (v0.1.95)
18
+
config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "T5ForConditionalGeneration"
4
+ ],
5
+ "d_ff": 3072,
6
+ "d_kv": 64,
7
+ "d_model": 768,
8
+ "decoder_start_token_id": 0,
9
+ "dropout_rate": 0.1,
10
+ "eos_token_id": 1,
11
+ "feed_forward_proj": "gated-gelu",
12
+ "initializer_factor": 1.0,
13
+ "is_encoder_decoder": true,
14
+ "layer_norm_epsilon": 1e-06,
15
+ "model_type": "t5",
16
+ "num_decoder_layers": 12,
17
+ "num_heads": 12,
18
+ "num_layers": 12,
19
+ "pad_token_id": 0,
20
+ "relative_attention_num_buckets": 32,
21
+ "tie_word_embeddings": false,
22
+ "tokenizer_class": "T5Tokenizer",
23
+ "transformers_version": "4.6.0.dev0",
24
+ "use_cache": true,
25
+ "vocab_size": 45100
26
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:23b0a3056db3f7d7033534c5ed6e55bcace0b71be8ef980b1f4dd75c94be7857
3
+ size 648330360
spiece.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ebba0a7b0fadda80677fe980264027fd8c51f822c486201b39dfdfe8804a570
3
+ size 1196021
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "extra_ids": 0}