Embedding recalculation occurs even when only the bot's “name”, “description”, and “instruction” is wanted to be updated, so waiting time occurs. I want to skip the embedding phase for updates that don't require recalculation.
Why the solution needed
As the number of knowledge files increases, processing time increases. I want to quickly tune instructions.
Additional context
None
Implementation feasibility
Are you willing to discuss the solution with us, decide on the approach, and assist with the implementation?
Describe the solution you'd like
Embedding recalculation occurs even when only the bot's “name”, “description”, and “instruction” is wanted to be updated, so waiting time occurs. I want to skip the embedding phase for updates that don't require recalculation.
Why the solution needed
As the number of knowledge files increases, processing time increases. I want to quickly tune instructions.
Additional context
None
Implementation feasibility
Are you willing to discuss the solution with us, decide on the approach, and assist with the implementation?