back to the huggingface API embeddings, remove non-usable embeddings

update libraries and code to support Modal as OpenAI-based server

update dependencies, remove biblio from search space

add query analyzer with min and avg similarity

fix breaking change in API

Merge branch 'main' into question-coefficient

return embeddings from storage retrieval

refactoring grobid client generic

refactor grobid processors (#29)

fix coordinates colors when working with chunks

add merger that preserve the coordinates and aggregate them meaningfully

include only bibliographics data that have potentially coordinates

Merge branch 'main' into add-pdf-viewer

use paragraphs instead of sentences

remove sqlite3 fix in the development branch

fix sqlite3 and chroma problem in streamlit.io

enable extraction of coordinates from pdf, using sentences

cosmetics

Merge branch 'main' into pdf-render

fix dimensionality error when switching from one model to the other and using the same document

fix another regression

fix regression

fix condition to establish when we need to cleanup the storage

more constraints on the memory

trying ot harness the model to avoid hallucinating due to the chat memory

cleanup and update documentation

Merge branch 'main' into fix-conversational-memory

fix signature

re-implement the conversational memory access

fix signature

change from year to publication year

include title, authors and year in the data store

Merge branch 'main' into pdf-render

cosmetics

fix chroma embeddings when an already seen file is uplaoded

fix chroma embeddings when an already seen file is uplaoded

fix env variables (#9)

fix engine to deal with the latest chroma

fix import, and reformat

Setuptools build, pypy release, github action improvements (#5)