* Added Q4_0 llamav2 models (7B, 13B) to the models.json
* Added support for Llama 70B
* Cleaned up some debugging statements.
* Update README.md to include Llamas
* Fix names in README
* Added some centering for the bars.
* Added some padding to the bottom of the button.
* Fixed some formatting bugs.
* Changed margins on larger screens to make large screen the same as original.
* Added a hamburger icon to the sidebar to allow users to have it on or off.
* Added ability to load previous session theme & automatically adjust bar height by using an onmount
* Formatted the code better
* Made checkmark switcher default value dependent on theme.
---------
Co-authored-by: Juan Calderon-Perez <835733+gaby@users.noreply.github.com>
* add CI for web project
* run prettier over whole app
* ci fix
* sigh
* fix linting
* Do not run Docker Build until all checks have passed
* clean up delete feature
* fix index
---------
Co-authored-by: Juan Calderon-Perez <835733+gaby@users.noreply.github.com>
* make it so you can just click send and it will send "Reformulate your last answer"
* new route for deleting a single prompt
* add a little trashcan button to last prompt so it can be deleted from history
* black formatting and removed unused import json
* + causes problems when deleting prompts
* log debug the final_prompt
* remove unused "let container;"
* add a uuid to every prompt and share that between backend and frontend
* delete prompts using an id, fallback to using content for old chats
* fix f string without placeholders
---------
Co-authored-by: Juan Calderon-Perez <835733+gaby@users.noreply.github.com>
* add api/static/* to .githignore
* add init_prompt to params so that we can set n_ctx to len(init_prompt) + params.n_ctx
* change get_prompt so it properly builds a instruction prompt from the recent history
- keeps the prompt below n_ctx
- goes backwards and adds only the most recent sentences
- always prepends the init_prompt
* ignore eventSource.onerror
I don't know why it does this, but this gets called locally for me and messes up the history, just doing nothing here works fine for me
* fix split on ! not *
* run black formatter on stream.py
* revert previous black format, just do the one change it wants manually
* third times the charm
---------
Co-authored-by: Juan Calderon-Perez <835733+gaby@users.noreply.github.com>
* Style: Web UI/UX improvements
* Docs: add credits for the svg
* Style: fix in header nav tittle always center
* Style: fix chat padding md:responsive
* Style: fix system chat padding md:responsive
* Style: text-lg to text-base... [for me look better]. (suggestion only).. if you dont like it I remove it
* Style: [suggestion] header title format one line 'Serge: <modelName>' second line '<date>'
* Style: fix color using daisyui semantic color
* Style: fix color user avatar
* Style: data-theme=dark
* Style: colors by compatible with all themes
* Style: colors by compatible with all themes [second and last review]
* Style: svg and colors on download models [replace icons to svg icons]
* Style: input h-auto
* Feature: add render markdown on user and serge
* Style: code tag color and background for any theme
* Style: copy to clipboard button background and overflow-y hidden in chat render parent element
* Style: code block background adaptive to daisy themes
* Style: remove images and add rubik-pixels font
* Style: add rubik-pixels font from fontsource
---------
Co-authored-by: Juan Calderon-Perez <835733+gaby@users.noreply.github.com>
* integrate langchain
get rid of mongodb
use llama-cpp-python bindings
* fixed most chat endpoints except posting questions
* Working post endpoint !
* everything works except streaming
* current state
* streaming as is
* got rid of langchain wrapper for calling llm, went back to using bindings directly
* working streaming
* sort chats by time
* cleaned up styling and added back loading indicator
* Add persistence support to redis
* fixed tooltips
* fixed default prompts
* added link to api docs (closes How to use the api #155 )
* Adding a new session button which uses current session settings / ALT+N for a new session
* Make it look better
* Adding autofocus to input box
* Adding requested changes moving button
---------
Co-authored-by: Raven Scott <commits@ssh.surf>
* When a user hits CTRL+ENTER send input
* Fixing Input and Send Button Overlay on Mobile
* add a small margin at the bottom
---------
Co-authored-by: raven <you@example.com>
Co-authored-by: Nathan Sarrazin <sarrazin.nathan@gmail.com>
* begin work on dev environment
* more work on dev image
* working dev + prod images with SPA front-end
* reworked dockerfile
* make CI point to the right action
* Improvements to github actions (#79)
* Improvements to github actions
* Change username to repo owner username
* Add fix for login into ghcr (#81)
* Update bug_report.yml
* added dev instructions to readme
* reduced number of steps in dockerfile
---------
Co-authored-by: Juan Calderon-Perez <835733+gaby@users.noreply.github.com>
* initial work on linting & templates
* moved everyone into a nice dockerfile
* move everyone into a single dockerfile
* update sample .env file
* got rid of .env file
* rename db volume to avoid confusion and conflicts with previous version
* added bug report template
- context window size, lets you select how far back the model should consider in its prompt.
- n_threads for multithreading
- init_prompt for customizing the initial prompt before the conversation.
- Added nginx, api & web app on the same port now.
- Allowed CSR, through sveltekit, with a hook for redirecting server side api requests.
- Implemented menu to pass model parameters on start page.
- Added a loading indicator while the model is computing
- Added convert script, thanks to @eiz, will catch unconverted .bin files and convert them on startup.
- Switched back to main branch of llama.cpp
- Got rid of code to handle magic.dat
