- Decouple configuring backend from starting server.
Backend search and processors can be configured after the backend
server has started
- Set global state in main instead of in configure_server method.
This allows the app to start even if configure_server exits early in
the first run scenario, where no config available to configure server
- Now start server, even if no config, before GUI started in main
- This refactor of app startup flow will allow users to configure
backend using the configure screen after server start
- Search is being reconfigured multiple times in /regenerate and
n/reload. More appropriate name is configure_ rather than initialize_
for it
- Standardize name of methods under configure.py
- Main.py was becoming too big to manage. It had both
controllers/routers and component configurations (search, processors)
in it
- Now that the native app GUI code is also getting added to the main
path, good time to split/modularize/clean main.py
- Put global state into a separate file to share across modules
- Run FastAPI server in a separate thread.
- This allows starting both the server and gui in parallel
- Create System Tray for Khoj
- Contains menu items that open search or config pages in browser
- Rearrange code to have only the code required to start Backend and
GUI in the run() method
- Move the backend setup code into a separate method
- Most concretely right now,
it eliminates the re-rank latency hit
on re-rank triggered on user hitting enter
after re-rank is already done on user idle
in the emacs interface
- Improves search latency of (incremental) search
- Improve code layout by ensuring all web interface specific code
under the src/interface/web directory
- Rename config API to more specifi /config instead of /ui
- Rename config data GET, POST api to /config/data instead of /config
- Improve search speed by ~10x
Tested on corpus of 125K lines, 12.5K entries
- Allow cross-encoder to re-rank results by settings &?r=true when querying /search API
- It's an optional param that default to False
- Earlier all results were re-ranked by cross-encoder
- Making this configurable allows for much faster results, if desired
but for lower accuracy
- Formalize filters into class with can_filter() and filter() methods
- Use can_filter() method to decide whether to apply filter and
create deep copies of entries and embeddings for it
- Improve search speed for queries with no filters
as deep copying entries, embeddings takes the most time
after cross-encodes scoring when calling the /search API
Earlier we would create deep copies of entries, embeddings
even if the query did not contain any filter keywords
- Reason:
Allow natural search on markdown based notes, documentation,
websites etc
- Details:
- Create markdown processor to extract Markdown entries (identified by
Heading) into standard jsonl format required by text_search
- Update API, Configs to support interfacing with new markdown type
- Update Emacs, Web clients to support interfacing with new markdown
type via API
- Update Readme to mentiond markdown is also supported
Closes#35
- The code for both the text search types were mostly the same
It was earlier done this way for expedience while experimenting
- The minor differences were reconciled and merged into a single
text_search type
- This simplifies the app and making it easier to process other
text types
- Had already made some progress on this earlier by updating the image
search responses. But needed to update the text search responses to
use lowercase entry and score
- Update khoj.el to consume the updated json response keys for text
search
- Allow viewing image results returned by Semantic Search.
Until now there wasn't any interface within the app to view image
search results. For text results, we at least had the emacs interface
- This should help with debugging issues with image search too
For text the Swagger interface was good enough
- Copy images to accessible directory
- Return URL paths to them to ease access
- This is to be used in the web interface to render image results
directly in browser
- Return image, metadata scores for each image in response as well
This should help get a better sense of image scores along both
XMP metadata and whole image axis
Details
--
- The filters to apply are configured for each type in the search controller
- Muliple filters can be applied on the query, entries etc before search
- The asymmetric query method now just applies the passed filters to the
query, entries and embeddings before semantic search is performed
Reason
--
This abstraction will simplify adding other pre-search filters. E.g datetime filter
- Issue
- Explicit filtering was earlier being done after search by bi-encoder
but before re-ranking by cross-encoder
- This was limiting the quality of results being returned. As the
bi-encoder returned results which were going to be excluded. So the
burden of improving those limited results post filtering was on the
cross-encoder by re-ranking the remaining results based on query
- Fix
- Given the embeddings corresponding to an entry are at the same index
in their respective lists. We can run the filter for blocked,
required words before the search by the bi-encoder model. And limit
entries, embeddings being considered for the current query
- Result
- Semantic search by the bi-encoder gets to return most relevant
results for the query, knowing that the results aren't going to be
filtered out after. So the cross-encoder shoulders less of the
burden of improving results
- Corollary
- This pre-filtering technique allows us to apply other explicit
filters on entries relevant for the current query
- E.g limit search for entries within date/time specified in query
- Use local variable to pass device to asymmetric.setup method via /reload, /regenerate API
- Set default argument to torch.device('cpu') instead of 'cpu' to be more formal
- The reload API adds the ability to separate out the loading of
embeddings from file without having to restart app or (re-)generate embeddings
- Before this the only way to load model from file was by restarting app
- The other way to reload the model embeddings by regenerating them
was to expensive for larger datasets
- This unlocks at least 1 use-case, where
- we regenerate model via an app instance running on a separate server and
- just reload the generated embeddings on the client device
- This allows us to offload the expensive embedding generation
compute to a background server while letting
- This avoids having to (re-)restart application on client device or
be forced to generate embeddings on the client device itself
- But it requires the model relevant files to be synced to the client device
This can be done with any file syncing application like Syncthing
- We can then call /regenerate on server and /reload client on a
regular schedule to keep our data up to date on semantic search
- Introduce prompt for GPT to automatically extract user's search intent
- Expose new search api endpoint to use that to set SearchType being
passed to search API
- Currently meant as an experimental API to gauge usefulness,
extendability. Evaluating for phone or voice use-case
