gahyunlee commited on
Commit
fba0152
ยท
verified ยท
1 Parent(s): a0b78b0

edit postprocessing logic

Browse files
Files changed (1) hide show
  1. postprocessing.py +11 -6
postprocessing.py CHANGED
@@ -1,6 +1,11 @@
1
-
2
- def extract_analysis_results(text):
3
- import re
4
- pattern = r"- ๋ถ„์„ ๊ฒฐ๊ณผ \d+:.*?(?=\n|$)"
5
- matches = re.findall(pattern, text, re.DOTALL)
6
- return "\n".join(matches).strip()
 
 
 
 
 
 
1
+ def extract_relevant_text(text):
2
+ """
3
+ ์ฃผ์–ด์ง„ ํ…์ŠคํŠธ์—์„œ `[/INST]`๋กœ ์‹œ์ž‘ํ•˜๊ณ  ์ข…๋ฃŒ ํƒœ๊ทธ `</>` ์ด์ „๊นŒ์ง€์˜ ํ…์ŠคํŠธ๋ฅผ ์ถ”์ถœํ•˜๋Š” ํ•จ์ˆ˜.
4
+ """
5
+ pattern = r"\[/INST\](.*?</)" # [/INST]๋กœ ์‹œ์ž‘ํ•˜๊ณ  </> ์ด์ „๊นŒ์ง€์˜ ํ…์ŠคํŠธ๋ฅผ ์ถ”์ถœ
6
+ match = re.search(pattern, text, re.DOTALL)
7
+ if match:
8
+ # `[/INST]` ์ดํ›„ `</s>` ์ „๊นŒ์ง€์˜ ํ…์ŠคํŠธ ๋ฐ˜ํ™˜
9
+ return match.group(1).strip().replace("</", "")
10
+ else:
11
+ return "๋งค์นญ๋˜๋Š” ํ…์ŠคํŠธ๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค."