danny-avila/LibreChat v0.3.3 on GitHub

What's Changed

An openAI preset used with this project helped me write this! Check it out on our discord

Created Docker Hub repo, addressing #182 [f5ff91c]
Changed docker-compose.yml to use pre-built image for api service instead of building from local file [f5ff91c]
Use gen_title response for askChatGPTBrowser.js (official title). [017447b]
Export conversation: csv, json, txt, markdown, screenshot. [6f0b559] [9691438]
Set document title when conversation is switched. [e0b0b68]
Add support for deleting individual presets. [83df28f]
Add RQ tokenizer. [88aea81]
Add support for saving messages to database. [a81bd27]
Support user-provided token to bingAI and chatgptBrowser by @wtlyu in #177

Fix abort messages and continue conversation on abort , prevent page refresh on stop generating button by @danny-avila #176

fix nginx container not containing the actual build output by @llk89 in #178
fix: fix infinite query failure when conversationId is not found by @danorlando in #179

Clean code with newConversationId in askbingai. [fc91ed4]
Refactor Message.jsx to remove cancelled message bubble and improve wording of unfinished message. [0bd2409]
Use new conversation Id [3b94a98], remove blank in screenshot [5664a0c], and remove use-screenshot [c271f04].
Fix withAuthentication on post requests [94e0636], fix conversations in nav (put refreshConvoHint back) and pagination [9a0e380].
Fix error and default tone for Sydney [e5e4ee2], and fix bug when clicking on title in search results [c983670]. by @HyunggyuJang and @danorlando
Resolve JavaScript heap out of memory error from Vite [6e42d4f], and fix isNotAppendable condition [78ae220].
Address infinite query failure when conversationId is not found [fb7542c], and correct typo in environment variable name [6aa540c].

User system is in progress
Currently Implementing a reverse engineering of the chatgpt plugins functionality. Official plugins are proving tricky and error prone but a good agent solution that can use community created plugins is much more likely and functional. Possibly also autoGpt/agentGpt in the future
Models supported by llama.cpp can be easily integrated and will get around to this optional feature. The idea is to have users install them as desired.
- LLaMA 🦙, Alpaca, GPT4All, Chinese LLaMA / Alpaca, Vigogne (French), Koala
the official site UI has received a good update and will get around to matching it, possibly as a configurable setting
Optimization of search function and database CRUD for multi-user system

Full Changelog: v0.3.0...v0.3.3