Allow re-generating model messages #81
Description
Requirements
- I have searched the issues of this repository and believe that this is not a duplicate
- I can confirm this feature request is not already implemented in the app
Description
Currently, if the model responds in a way that is undesirable, the only thing the user can do to fix that is to edit the response. This can be tedious when compared to generating additional responses to find one that is acceptable. Further, there is no way to roll back the chat by multiple messages (e.g. to before an idea came up that the AI fixated on) other than, again, editing the entire subsequent conversation.
Possible solution
The ideal, IMO, would be storing the conversation as a tree, allowing the user to regenerate any past response to create a "new timeline" that can be switched between at the branching point. However, even without being able to preserve multiple timelines, it would be useful to be able to regenerate any AI message, effectively rolling back to the previous user message and re-sampling the AI.
Additional context
No response