Support for local sweepai/sweep-next-edit-1.5B model planned?

[jonadoe]

Is there a plan to add support for running sweepai/sweep‑next‑edit‑1.5B with a local LLM?

[versecafe]

@jonadoe Hey Jonathan it's on the list right now internally likely going to use the same approach as the cursortab.nvim local sweep support requiring the user runs it externally with llama cpp (we'll share the command) and then just set to local mode, would that meet your needs?

[jonadoe]

@versecafe That sounds good. I'm using llama.cpp already anyway, so that approach works perfectly for me.

[resynth]

+1 on this, local models all the way!

[ooJaan]

+1 would love to be able to host in using vllm

[sr-tream]

You can see my fork: https://github.com/sr-tream/vscode-nes

I added support of local models over openai-compatible endpoint - you can use sweep 0.5b, 1.5b and 7b models. Additionally, you also can use zeta2 and zeta2.1 models