ragna
ragna copied to clipboard
Quansight
RAG orchestration framework ⛵️
This is a work in progress. This is a continuation of PR #358. It was created by rebasing PR #358 to include the merged commits from PR #362 that removes...
### Feature description Oftentimes, I'll start up Ragna like so to create a log: `ragna api 2>&1 | tee ragna.log` I can then `grep` the log to find errors or...
We currently suggest people install `ragna[all]` by default and only drop the `[all]` extra if they want fine-grained control https://github.com/Quansight/ragna/blob/e0fe0145a5d7428923be778542d5211e8359c174/docs/install.md?plain=1#L9-L23 `pip install ragna`, which is what most people will try...
~Work in progress - do not merge~ Ready for review This PR uses a better approach for CSS styling in Panel. The method we used so far showed its limitations...
Was released today: https://blog.holoviz.org/posts/panel_release_1.4/ There are a few chat improvements that we should take a look at: https://blog.holoviz.org/posts/panel_release_1.4/#chat-improvements.
### Bug description ```sh ragna init ``` Select all storages and both ChatGPT assistants, then ```sh OPENAI_API_KEY="..." ragna ui ``` Try to add a PDF document and start a chat....
### Feature description Currently chats are listed oldest first (highest) in `ragna ui`. It seems more helpful more most use cases to reverse that order and have newer chats appear...
Our web UI is currently sits on top of the REST API and cannot be deployed as a standalone application. This architecture was chosen to avoid giving us a benefit...
This PR adds support for Ollama assistants.
The Gemma2B model is provided by Ollama, vLLM, and possibly others. Originally, we included it as `ragna.assistants.Gemma2B` from Ollama, but there could be a naming conflict if vLLM is included...