openinterpreter / aifs Goto Github PK

Are PRs welcome?

While searching a Folder with a single pdf, I keep getting error:
Number of requested results 5 is greater than number of elements in index 1, updating n_results = 1

Here is the code I am using:

from aifs import search
search("What is Popular Sovereignty?", path="/Users/km/work/test/Revolutions")
search("What triggered the American Revolution?") # Path defaults to CWD

Any suggestions?

Can someone create its google colab sheet for demo?

Before I make a PR, I've few questions...

Why is it ingesting before every search?
Is storing it in json format necessary?

In my version I'm maintaining persistent vector store at one place with collection for every drive in windows and base directory (like /mnt, /var, ...) for unix based systems.

• Current Scope: "aifs" excels in local semantic search for text-based content.
• Proposed Enhancement: Extend capabilities to include audio and video content.
• Implementation:
  • Generate .srt Files: Utilize tools like Whisper (runs great locally on most computers) for transcribing audio and video into subtitle files.
  • Search Integration: Incorporate these .srt files into the search functionality.
• Outcome:
  • Search Utility: Enable searches to return specific segments from multimedia content.
  • Timestamps Feature: Provide timestamps alongside search results for precise referencing.
• Impact: This enhancement would transform "aifs" into a comprehensive tool for searching across various content formats, greatly expanding its usability.

Any way to make it link to the original document, pulling up the PDF page possibly as well? or does it already do this and I am doing something wrong?

As in https://pypi.org/project/unstructured/ described, it should be:

pip install "unstructured[all-docs]"

great tool and thanks for sharing! it'd be nice to exclude the .git folder and whatever specified in .gitignore by default.

Anything equivalent for image/photo files for automatic detecting faces/images & grouping/labelling/organising?

When you have a moment if you can also publish the releases on github, it makes it easier to build from source
(this is coming as the maintainer from the nixos package).

Also it makes it easy for other people to follow the project.

Thank you for making this open source!

Hi Killian,
does this repo miss some commits or it's just an error?

Cheers