jingyeom's picture
Update README.md
5fe35e3 verified
|
raw
history blame
3.22 kB
---
library_name: transformers
license: gemma
---
---
base_model: jingyeom/solar_merge_dpo
tags:
- trl
- sft
- generated_from_trainer
datasets:
- generator
---
## Training
* Base Model
* [google/gemma-2-2b-it](https://huggingface.co./google/gemma-2-2b-it)
* Training dataset
* [ํ‘œ ์ด๋ฏธ์ง€-ํ…์ŠคํŠธ ์Œ ๋ฐ์ดํ„ฐ](https://www.aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=data&dataSetSn=71709)
``` python
from transformers import pipeline
sys_text = """Given the following HTML table, convert it to a descriptive text format.
Summarize the content of the table. and then describe the key features or characteristics of the table."""
table = """WHO์˜ ICD-11 'Gaming Disorder' ์งˆ๋ณ‘์ฝ”๋“œํ™” ์‹œ๊ธฐ๋ณ„ ์ถ”์ง„๊ฒฝ๊ณผ ๋‚ด์šฉ
<table><tr><td colspan="1" rowspan="1">2014๋…„</td><td colspan="1" rowspan="1">ยทWHO ์ •์‹ ๊ฑด๊ฐ•๋ถ€ ์ค‘๋… ์„น์…˜ ์ž๋ฌธ ๊ทธ๋ฃน ํšŒ์˜์ฒด๋ฅผ ํ†ตํ•ด ๊ฒŒ์ž„ ๋“ฑ ๋””์ง€ํ„ธ๋ฏธ๋”” ์–ด์˜๊ณผ๋„ํ•œ ์‚ฌ์šฉ์ด ๊ณต์ค‘๋ณด๊ฑดํ•™์  ๋ฌธ์ œ๋กœ ๋Œ€์‘์ด ํ•„์š”ํ•˜๋‹ค๋Š” ์˜๊ฒฌ ๋„์ถœ </td></tr><tr><td colspan="1" rowspan="1">2015๋…„</td><td colspan="1" rowspan="1">ยท2์ฐจ TF ํšŒ์˜ ํ†ตํ•ด Gaming Disorder๋กœ ๋ช…๋ช…ํ•˜์—ฌ ICD-11 ๋“ฑ์žฌ๋ฅผ ์ถ”์ง„ํ•˜๊ธฐ๋กœ ์ „๋ฌธ๊ฐ€ํ•ฉ์˜ ๋„์ถœ </td></tr><tr><td colspan="1" rowspan="1">2016๋…„</td><td colspan="1" rowspan="1">ยทICD-11 ๊ฐœ์ • ์‚ฌ์ดํŠธ์— ์ง„๋‹จ๊ธฐ์ค€ ๊ฒŒ์‹œ ๋ณด๊ฑด์ „๋ฌธ๊ฐ€์˜ ์˜๊ฒฌ ์ˆ˜๋ ด ์‹œ์ž‘ </td></tr><tr><td colspan="1" rowspan="1">2017๋…„12์›” </td><td colspan="1" rowspan="1">ยทICD-11 ๊ฐœ์ •์•ˆ ๊ณต๊ฐœ </td></tr><tr><td colspan="1" rowspan="1">2018๋…„5์›” </td><td colspan="1" rowspan="1">ยท71์ฐจ ์„ธ๊ณ„๋ณด๊ฑด์ดํšŒ์—์„œ ์•ˆ๊ฑด ์ƒ์ • ์˜ˆ๊ณ  ยท๊ณต๊ฐœ๋œ ๊ฐœ์ •์•ˆ์— ๋Œ€ํ•œ ๊ฐ ํšŒ์›๊ตญ์˜ ๊ฒ€ํ†  ์‹œ๊ฐ„ ํ™•๋ณด๋ฅผ ์œ„ํ•ด ICD-11์ฐจ ๊ฐœ์ •์•ˆ ์˜์ดํšŒ ์•ˆ๊ฑด ์ƒ์ •์„ 1๋…„๊ฐ„ ์—ฐ๊ธฐ(2019๋…„ 5์›” 72์ฐจ ์„ธ๊ณ„๋ณด๊ฑด์ดํšŒ) </td></tr><tr><td colspan="1" rowspan="1">2018๋…„ 6์›”</td><td colspan="1" rowspan="1">ยทICD-11 ์ตœ์ข…์•ˆ์„ WHO ํ™ˆํŽ˜์ด์ง€ ๊ฒŒ์žฌ </td></tr><tr><td colspan="1" rowspan="1">2019๋…„5์›” </td><td colspan="1" rowspan="1">ยทICD-11 ๊ฐœ์ •์•ˆ ๋งŒ์žฅ์ผ์น˜ ํ†ต๊ณผ </td></tr></table>"""
messages = [{'role': 'system', 'content': sys_text}, {'role': 'user', 'content': table}]
output = pipeline((messages))
print(output[0]['generated_text'][-1]['content'])
```
* ๊ฒฐ๊ณผ
```text
ํ‘œ๋Š” WHO์˜ ICD-11 'Gaming Disorder' ์งˆ๋ณ‘์ฝ”๋“œํ™” ์‹œ๊ธฐ๋ณ„ ์ถ”์ง„๊ฒฝ๊ณผ ๋‚ด์šฉ์— ๊ด€ํ•œ ํ‘œ๋กœ 7ํ–‰ 2์—ด์ด๋ฉฐ 2014๋…„๋ถ€ํ„ฐ 2019๋…„๊นŒ์ง€ WHO์˜ ์ •์‹ ๊ฑด๊ฐ•๋ถ€ ์ค‘๋… ์„น์…˜ ์ž๋ฌธ ๊ทธ๋ฃน ํšŒ์˜์ฒด๋ฅผ ํ†ตํ•ด ๊ฒŒ์ž„ ๋“ฑ ๋””์ง€ํ„ธ ๋ฏธ๋””์–ด์˜ ๊ณผ๋„ํ•œ ์‚ฌ์šฉ์ด ๊ณต์ค‘๋ณด๊ฑดํ•™์  ๋ฌธ์ œ๋กœ ๋Œ€์‘์ด ํ•„์š”ํ•˜๋‹ค๋Š” ์˜๊ฒฌ์„ ๋„์ถœํ•˜๊ณ , 2015๋…„์—๋Š” Gaming Disorder๋กœ ๋ช…๋ช…ํ•˜์—ฌ ICD-11 ๋“ฑ์žฌ๋ฅผ ์ถ”์ง„ํ•˜๊ธฐ๋กœ ์ „๋ฌธ๊ฐ€ ํ•ฉ์˜๋ฅผ ๋„์ถœํ•˜์˜€๋‹ค.
1. 2014๋…„์—๋Š” WHO ์ •์‹ ๊ฑด๊ฐ•๋ถ€ ์ค‘๋… ์„น์…˜ ์ž๋ฌธ ๊ทธ๋ฃน ํšŒ์˜์ฒด๋ฅผ ํ†ตํ•ด ๊ฒŒ์ž„ ๋“ฑ ๋””์ง€ํ„ธ ๋ฏธ๋””์–ด์˜ ๊ณผ๋„ํ•œ ์‚ฌ์šฉ์ด ๊ณต์ค‘๋ณด๊ฑดํ•™์  ๋ฌธ์ œ๋กœ ๋Œ€์‘์ด ํ•„์š”ํ•˜๋‹ค๋Š” ์˜๊ฒฌ์„ ๋„์ถœํ•˜์˜€๋‹ค. 2. 2015๋…„์—๋Š” 2์ฐจ TF ํšŒ์˜๋ฅผ ํ†ตํ•ด Gaming Disorder๋กœ ๋ช…๋ช…ํ•˜์—ฌ ICD-11 ๋“ฑ์žฌ๋ฅผ ์ถ”์ง„ํ•˜๊ธฐ๋กœ ์ „๋ฌธ๊ฐ€ ํ•ฉ์˜๋ฅผ ๋„์ถœํ•˜์˜€๋‹ค. 3. 2018๋…„ 6์›”์—๋Š” ICD-11 ์ตœ์ข…์•ˆ์„ WHO ํ™ˆํŽ˜์ด์ง€์— ๊ฒŒ์žฌํ•˜์˜€๋‹ค.
```