Jesse Zhang
Jesse Zhang
When creating a video source via monitor capture, it seems that monitors are 0-indexed, but for each given index, I'm unable to figure out how to get any info on...
I built the package locally and was able to initialize osn successfully in Electron 6. However, once `OBS_service_startRecording` is called, the app immediately crashes and fills the logs with this...
Previously, the `recursive` attribute was not being used. The reader always recursively read subdirectories even though the attribute is `False` by default. This is fixed. I also added a new...
Enables Document metadata to be included in queries by prepending it to each node. Includes the work from #216. Metadata can now also be printed when listing the source documents...
Some improvements to what info is provided in the user-facing `SourceNodes`: - Similarity score is now included, addressing #243 - Metadata is now included (so you can immediately display info...
Added a new file reader for .pptx files. Notably, it demonstrates the first use case of tokenizing non-text items (i.e. images) from within a file. Images inside the Powerpoint are...
Download and use loaders from the new loader hub. Super simple usage: ``` from gpt_index import download_loader GoogleDocsReader = download_loader('GoogleDocsReader') reader = GoogleDocsReader() reader.load_data(document_ids=['1wf-y2pd9C878Oh-FmLH7Q_BQkljdm6TQal-c1pUfrec']) ```
This is a first stab at refactoring text splitters. The primary changes are: - All indices accept a `text_splitter` which can be anything that inherits LangChain's `TextSplitter`. All the Text...
# Thank you for contributing an eval! ♥️ 🚨 Please make sure your PR follows these guidelines, __failure to follow the guidelines below will result in the PR being closed...