Hey Tudor, your doing it! The orphaned chat context is from the setting UI implementation. I used the notebook to generate the notebook! You can scrub it out your history by adding "viewof context_viz" to the "excludes" list. I want to centralize the tip sharing here and updates here https://www.reddit.com/r/robocoop/ Please send me your wants/needs and we can make it better together.

Thanks, I'll join the reddit, I'm new to observable with a tiny bit of notebook experience, so quite a steep curve for me, I guess I'll need to make it private to hide my openAI key and limit usage? edit: I'll partly answer my own question, yes, in settings/secrets you can set a key value pair and store the openAI key out of sight. However, I'm guessing my public notebook will still be 'live' and interactions by others with the notebook will make an OpenAI request, Is that correct?

Off hand, would observable allow us to set the API endpoint to a local instance of GPT4ALL, it has a drop in replacement for OpenAI calls afaik? Open source AI is interesting to me, I like control :) Some of the latest models offer fairly decent performance especially in creative tasks. https://github.com/keldenl/gpt-llama.cpp '"gpt-llama.cpp is an API wrapper around llama.cpp. It runs a local API server that simulates OpenAI's API GPT endpoints but uses local llama-based models to process requests."'

no you don't need to make it private. If you look at src for the UI element containing the key (https://observablehq.com/d/4a253a1af4aa88eb#OPENAI_API_KEY) you will see its binding the UI to local-storage-view (https://observablehq.com/@tomlarkworthy/local-storage-view). I.e. your browser is the persistence. Notebooks are run in iframes, so the only JS that can access that variable are those served from your isolated domain associated with your user name. So its safe as long as you don't fork dodgy notebooks (no history of that occurring on the Observable platform but it is a vector). Stuff written in UI elements is not part of the notebook that is forked. > Off hand, would observable allow us to set the API endpoint to a local instance of GPT4ALL, it has a drop in replacement for OpenAI calls afaik? You can do it! This cell contains the API call https://observablehq.com/d/4a253a1af4aa88eb#openAiResponse, you can edit anything. The whole system is implemented within the notebook ecosystem! If you add the API endpoint as a choice in the settings UI (https://observablehq.com/d/4a253a1af4aa88eb#settings) , you can send me "suggestions" (a.k.a. change requests) to the notebook you forked off so we could make that a general 1st class feature. Id be very interesting seeing how llama performs. Share on the reddit, it's also a good place to ask generic questions coz other people might has the same concerns.

I added the ability to customize the API endpoint now https://www.reddit.com/r/robocoop/comments/17sphfx/q_can_i_change_the_api_endpoint_to_a_local/