Is your feature request related to a problem? Please describe.
A future change to the solution of #34, and by extension #70, both of which evaluates experience per message, which can be wasteful.
Describe the solution you'd like
Potentially using the LLM to shorten the whole chat log to just the important part (potentially including the AI messages for context). Afterwards, some changes need to be made in the experience categorization pipeline (most is simply changing Messages to Chats). Also, a good change to consider is moving experience categorization to the Chat-Categories table introduced in #70, which stores the categorization of every chats (designed for use by researchers).
Describe alternatives you've considered
Not doing it at all, but this can be wasteful (money wise and processing time wise). Especially if the user approach this as a chat and answer questions without providing any context of the question, we lose context + some messages will be off-track.
BasicallyOk
commented
5 months ago
Is your feature request related to a problem? Please describe. A future change to the solution of #34, and by extension #70, both of which evaluates experience per message, which can be wasteful.
Describe the solution you'd like Potentially using the LLM to shorten the whole chat log to just the important part (potentially including the AI messages for context). Afterwards, some changes need to be made in the experience categorization pipeline (most is simply changing Messages to Chats). Also, a good change to consider is moving experience categorization to the Chat-Categories table introduced in #70, which stores the categorization of every chats (designed for use by researchers).
Describe alternatives you've considered Not doing it at all, but this can be wasteful (money wise and processing time wise). Especially if the user approach this as a chat and answer questions without providing any context of the question, we lose context + some messages will be off-track.