The idea is:
I notice that most AI flows are for external AI resources like openai etc.. but more companies and enterprises are in need of nodes for local AI resources. The big problem is embedding. When embedding with openai you do not need to apply “Task instruction prefixes” it does it for you. So when you run a local nomic embed model you cannot add task instructions.
My use case:
Can we make it possible to add a code module to the embedding option in vector? or allow the option for inserting embedding prefixes?
I think it would be beneficial to add this because:
This will increase the scope of uses for the platform, there is a very large growing concern about having data leave a property. Making systems that can utilize on proem solutions will be in high demand.
Any resources to support this?
This is just a link to the embedding model that shares how to use prefixes
Are you willing to work on this?
Sure.