Extracting the model’s thoughts
The model response content from the previous exercise's math_problem has been stored under response_content.
Note that this may differ from what you saw when you ran the exercise, as LLM outputs are inherently random (more on this in Chapter 2).
Your job is to separate thought from answer using regular expressions (RegEx).
Deze oefening maakt deel uit van de cursus
Working with DeepSeek in Python
Oefeninstructies
- Import the
rebuilt-in Python module. - Search for the model's thoughts in the response by looking for strings between two
<think>tags using the RegEx provided. - Extract the first group from
matchusing the.group()method ,and strip these thoughts of any leading and trailing whitespace.
Praktische interactieve oefening
Probeer deze oefening eens door deze voorbeeldcode in te vullen.
# Import the re module
import re
# Search for strings between think tags
match = re.____(r'(.*?) ', ____, re.DOTALL)
# Extract the group from the match and strip whitespace
think_content = match.____(1).____()
print(think_content)