Closed tvraman closed 6 months ago
It is in llm package. Are you sure that you need it in ellama? Can you explain your use case? It can be implemented in ellama but I don't understand why it should be here.
Sergey Kostyaev @.***> writes:
Agreed that it likely belongs in LLM, but I asked because:
My use-case is to produce auditory feedback (emacspeak), e.g. a short auditory tick as the results stream in.
That feels like a front-end thing, and llm.el is designed to be separate from various front-ends.
see https:github.com/tvraman/emacspeak> It is in llm package. Are you sure that you need it in ellama? Can you Also, I'll create a separate issue for the next question below so you can track it:
Should ellama-mode derive from comint-mode; we'll then get the ability to navigate through various steps in a conversation for free.
explain your use case? It can be implemented in ellama but I don't understand why it should be here.
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.
--
@tvraman I see. So, based on your use case this callback should be function with generated text as an argument?
Sergey Kostyaev @.***> writes:
that would be one possibility. Other possibility would be to provide a functon that I can advice via defadvice.
There are two possible types of behavior I'd like to be able to implement.
@tvraman I see. So, based on your use case this callback should be function with generated text as an argument?
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.
--
@tvraman in version 0.4.5 function ellama-stream
supports :on-done
callback function parameter. Does it helps you?
Sergey Kostyaev @.***> writes:
Not directly -- will need to think through it. Would be better if the callback could be specified by setting a variable.
@tvraman in version 0.4.5 function ellama-stream supports :on-done callback function parameter. Does it helps you?
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.
--
Not sure how we would build that; perhaps it needs to be part of the llm.el package?