RemVdH commited on
Commit
6a000e2
·
verified ·
1 Parent(s): c4de9ec

Update app.py

Browse files

Try to make agent more robust based on the lessons learned with difference between expectation of search wiki and reality (just strings not page names)

Files changed (1) hide show
  1. app.py +6 -3
app.py CHANGED
@@ -9,9 +9,11 @@ import wikipedia
9
 
10
  @tool
11
  def search_wikipedia_pages(search_string: str) -> List[str]:
12
- """This tool helps to search for the names of wikipedia pages that relate to a search string.
13
- This will not retrieve the actual pages, but gives oversight of the different pages on wikipedia to find more information.
14
- This tool will output the first 10 pages related to the search string as a list of strings.
 
 
15
  Args:
16
  search_string: the search string
17
  """
@@ -27,6 +29,7 @@ def search_wikipedia_pages(search_string: str) -> List[str]:
27
  @tool
28
  def get_wikipedia_page_content(page_name: str) -> str:
29
  """This tool helps to get the content of wikipedia pages by their name.
 
30
  The output is a string with the actual content of the wikipedia page.
31
  Args:
32
  page_name: the name of the page in wikipedia
 
9
 
10
  @tool
11
  def search_wikipedia_pages(search_string: str) -> List[str]:
12
+ """This tool helps to search for the names of potential names of wikipedia pages that relate to a search string.
13
+ This doesn't guarantee that every returned string is a perfect, exact page title that wikipedia.page() will accept without any issue.
14
+ This tool sends a query to the Wikipedia API's search endpoint. This endpoint performs a full-text search across Wikipedia's article titles and content.
15
+ This will not retrieve the actual pages, but gives oversight of the different keywords on wikipedia to find more information.
16
+ This tool will output the first 10 strings related to the search string as a list of strings.
17
  Args:
18
  search_string: the search string
19
  """
 
29
  @tool
30
  def get_wikipedia_page_content(page_name: str) -> str:
31
  """This tool helps to get the content of wikipedia pages by their name.
32
+ If you use this tool based on the output of search_wikipedia_pages remember to try a next string that was the output of search_wikipedia_pages if you get a error: Page id
33
  The output is a string with the actual content of the wikipedia page.
34
  Args:
35
  page_name: the name of the page in wikipedia