enricoros
commented
8 months ago This feature could be small (auto-setup OpenAI or another engine and picking the first) or larger (defining the models available to the user on the backend so they're backend controlled all the time and could be a mix of different vendors).
In a simple case, we could auto-setup sources that have a _API_KEY set, but we'll also need one Env var to choose which is the default model (sources and models are unsorted, so we don't have a default model to pick).
In a more controlled and complex case, the IT will need to define a file with the models setup, which every user will use.
Which of the 2 options is more important?
haopengsong
slfhstr
clawfire
Why This project is great in many ways as it incorporates lots of features in accordance with the current development of AGI. However, some users who do not process a background in tech or not fluent in computers (senior people) may have troubles when using. People may find it uneasy to deal with so much settings (api_keys, models, temperatures, etc ... ).
Description A common usecase will be that ITs will setup the api_keys in .env, so that model selection will default to the first modal option in the model list, and the rest models will stay in a selection list for later use. Users can start typing prompts once logged in without having to setup models.
Requirements ITs setup environment variables in .env -> Users logged in -> Typing prompts directly (no need to setup modals)