Join Nostr
2025-02-26 05:46:04 UTC
in reply to

picofarad on Nostr: >prodded to think a long time right before it's about to return from "thinking" you ...

>prodded to think a long time

right before it's about to return from "thinking" you inject a "wait..." into the assistant context. My interface screws up "editing" the assistant using deepseek (or it did last week) but i was able to test it with the llama thinking model (iirc)