Taishi-N324 commited on
Commit
f053604
1 Parent(s): b6e75aa

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +23 -2
README.md CHANGED
@@ -153,12 +153,21 @@ This format must be adhered to strictly, as deviations may result in less optima
153
  The template used to construct a prompt for the Instruct model is specified as follows:
154
 
155
  ```
156
- <s>[INST] <<SYS>>\n{Instruction}\n<</SYS>>\n\n{USER_MESSAGE_1} [INST] {BOT_MESSAGE_1} </s>[INST] {USER_MESSAGE_2}[/INST]
157
  ```
158
 
 
159
  Please be aware that ``<s>`` and ``</s>`` are special tokens used for the beginning of string (BOS) and end of string (EOS), respectively, while [INST] and [/INST] are considered regular strings.
160
 
161
- For the "{Instruction}" part, We recommend using "あなたは誠実で優秀な日本人のアシスタントです。"
 
 
 
 
 
 
 
 
162
 
163
 
164
  ### Use the instruct model Ver0.1
@@ -228,3 +237,15 @@ Here are the team members:
228
  - [Taishi Nakamura](https://twitter.com/Setuna7777_2)
229
  - [Takumi Okamoto](https://www.linkedin.com/in/takumi-okamoto)
230
  - [Ishida Shigeki](https://www.wantedly.com/id/reborn27)
 
 
 
 
 
 
 
 
 
 
 
 
 
153
  The template used to construct a prompt for the Instruct model is specified as follows:
154
 
155
  ```
156
+ <s>[INST] <<SYS>>\n{SYSTEM_PROMPT}\n<</SYS>>\n\n{USER_MESSAGE_1} [/INST] {BOT_MESSAGE_1} </s>[INST] {USER_MESSAGE_2}[/INST]
157
  ```
158
 
159
+
160
  Please be aware that ``<s>`` and ``</s>`` are special tokens used for the beginning of string (BOS) and end of string (EOS), respectively, while [INST] and [/INST] are considered regular strings.
161
 
162
+ For the "{SYSTEM_PROMPT}" part, We recommend using "あなたは誠実で優秀な日本人のアシスタントです。"
163
+
164
+ For the "{USER_MESSAGE_1}" part, We recommend using {instruction}\n{input}
165
+
166
+ In other words, We recommend the following:
167
+
168
+ ```
169
+ <s>[INST] <<SYS>>\nあなたは誠実で優秀な日本人のアシスタントです。\n<</SYS>>\n\n{instruction1}\n{input1} [/INST] {BOT_MESSAGE_1}</s>[INST] \n\n{instruction2}\n{input2} [/INST]
170
+ ```
171
 
172
 
173
  ### Use the instruct model Ver0.1
 
237
  - [Taishi Nakamura](https://twitter.com/Setuna7777_2)
238
  - [Takumi Okamoto](https://www.linkedin.com/in/takumi-okamoto)
239
  - [Ishida Shigeki](https://www.wantedly.com/id/reborn27)
240
+
241
+ ## How to cite
242
+ ```
243
+ @misc{fujii2024continual,
244
+ title={Continual Pre-Training for Cross-Lingual LLM Adaptation: Enhancing Japanese Language Capabilities},
245
+ author={Kazuki Fujii and Taishi Nakamura and Mengsay Loem and Hiroki Iida and Masanari Ohi and Kakeru Hattori and Hirai Shota and Sakae Mizuki and Rio Yokota and Naoaki Okazaki},
246
+ year={2024},
247
+ eprint={2404.17790},
248
+ archivePrefix={arXiv},
249
+ primaryClass={cs.CL}
250
+ }
251
+ ```