TODO.txt

PRIORITY:
- fix update function for chroma

IN PROGRESS:
- finish qa with sources chain support

OTHER:
- handling multiple indexes in one vectorstore (pinecone)
- handling multiple vectorstoretypes
- selecting filetypes to download by checkboxes; additionaly, groups of filetypes
- add streaming of responses (low priority)
- automatic text encoding recognition (https://charset-normalizer.readthedocs.io/en/latest/) (done for python)
- finish qa with sources chain support:
    - display probabilty of each source, so that hybrid search can be fine tuned
- add option to enable/disable providing the whole conversation history together with the question to the doc retriever vs just the question
- ui: if update: 
    - set chunk size/ overlap
    - file extensions to download
- search options:
    - chroma
        - search_type "mmr" or "similarity"
    - pinecone
        - (hybrid search alternatives)
- option to save current parameters to preset
- loading screen when chatbot is being initialized
- combine url and subdirectory input into one field