Version upped

generate clusters only if more than 3 clusters would be created

optionally search one language only

keep last question/answer as additional context

This should help with questions relating to the last anwser when they
haven't been rephrased well enough

Version upped

option to prefer a configured language instead of guessing

Added dumping of TSV files to SQLite store

This allows visualizing the embed vectors

finish similar pages feature

Version upped

use a k-means based cluster approach to speed up similarity searches

simplify cosine distance calculation

Since all OpenAI vectors are normalized, only the dotproduct needs to be
calculated for the distance. This saves a couple of floating point ops
per chunk, but do

simplify cosine distance calculation

Since all OpenAI vectors are normalized, only the dotproduct needs to be
calculated for the distance. This saves a couple of floating point ops
per chunk, but doesn't make a huge difference overall.

show more ...

Version upped

output info on similar chunk fetching

helps with figuring out how fast the store is

First go at syntax to display similar pages

reuse link colors when styling chat and buttons

brainier button icon

implement Pinecone based storage

First go at syntax to display similar pages

Version upped

support GPT4

Version upped

fix visibility changes in chat dialog

lazy load embeddings, storage and model

This should fix an immediate problem with the sqlite db initializing
slowly, but the underlying problem has to be identified.

added button to reset the conversation

This is needed to start conversing on a new topic

use a cut-off point when considering similar documents