small agent harness that stayed useful

[cudacrate]

Did a bit of testing and thought I'd post the result here.

Short version: Ollama behaved well for me, but only after I stopped ignoring prefill/decode split.

Happy to post more details if anyone wants them.

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.

[ggufvault]

Not sure I agree there. I wouldn't optimize around that first. On my side the bigger swing came from the runtime and not the model. Not saying you're wrong, just that I wouldn't lock that in as the lesson yet.

[labline]

That matches what I was seeing. The pattern around small local stacks has been pretty similar for me. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.

[southden]

Not sure I agree there. I wouldn't optimize around that first. On my side the bigger swing came from the runtime and not the model. Not saying you're wrong, just that I wouldn't lock that in as the lesson yet.

[llm_wren]

A couple things I'd check first. If you're on Ollama, I'd confirm the version first and then reduce variables until the behavior is obvious. Not glamorous advice, but repeatable baselines save a lot of time.

[northgrid33]

That matches what I was seeing. That matches what I saw on mini PC.

[labline]

I don't think that's the whole story. The reported numbers can look compelling, but in ordinary usage I didn't see the same benefit. Could be a workload difference, but I wouldn't generalize too fast from one setup.

[llm_wren]

A couple things I'd check first. If you're on llama.cpp, I'd confirm the version first and then reduce variables until the behavior is obvious. Not glamorous advice, but repeatable baselines save a lot of time.

[ggufvault]

A couple things I'd check first. If you're on llama.cpp, I'd confirm the version first and then reduce variables until the behavior is obvious. Not gl

Did a small comparison this afternoon. On mini PC with Ollama, I saw basically the same behavior. The main thing that moved was latency, not quality.