seanwbren

For the WikipediaReader does anyone else

For the WikipediaReader, does anyone else have it calling both the query phrase and also looking up favicon.ico for some reason? Using OnDemandLoader tool from the Data Agents blog post. Code:

Plain Text

wikiReader = WikipediaReader()
wiki_tool = OnDemandLoaderTool.from_defaults(
 wikiReader,
 name="WikipediaTool",
 description="A tool for loading data about places and querying articles from Wikipedia. Do not call this for math. "
)

1 comment

sseanwbren

Sorry one more question I m creating a

Sorry one more question, I'm creating a FnRetrieverOpenAIAgent with a working wolfram tool and wiki tool, but there's a mapping step that requires metadata.name for the Tools. How do I add/configure that?

1 comment

sseanwbren

For gmail gcal tool having problems with

For gmail / gcal tool, having problems with the example tutorial. Setting the permissions right to get the credentails.json file seems quite tricky?
Navigated a few issues but now the llamahub gmail base package has a refresh issue?

Plain Text

File ".../python3.11/site-packages/llama_hub/tools/gmail/base.py", line 78, in _get_credentials
    creds.refresh(Request())
  File ".../python3.11/site-packages/google/oauth2/credentials.py", line 345, in refresh
    ) = reauth.refresh_grant(
        ^^^^^^^^^^^^^^^^^^^^^
google.auth.exceptions.RefreshError: ('invalid_scope: Bad Request', {'error': 'invalid_scope', 'error_description': 'Bad Request'})

2 comments

sseanwbren

Tools

Following the "Building Better Tools" recent blog post, and running into a module issue here.
Getting ModuleNotFoundError: No module named 'llama_index.tools.base'

Find answers from the community

For the WikipediaReader does anyone else

Sorry one more question I m creating a

For gmail gcal tool having problems with

Tools

I just asked the same question Running 1