Stop and Continue response like llmstudio

[electro199]

In LLM studio there is option to stop generation edit the response and then continue so LLM use the provided text/token for completion. How to replicate that in the llama-server ?

[bobleer]

the same question.
how to stop a request in llama-server?

[ExtReMLapin]

In llama-server, to stop a request, just close the connection.

To continue it, just resent the whole prompt (so it doesn't work with chat endpoint, eh) with the partially generated message you saved