eval setup for local models for Kimi

[routearch]

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

After a few runs on Threadripper box, I ended up with decent performance, but the setup only really made sense once I adjusted around consumer GPUs.

Nothing scientific here, just practical notes from a real setup.

My use case is pretty ordinary: local chat, some document work, and occasional batch jobs.

[batchsize]

What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at MoE serving. Not glamorous advice, but repeatable baselines save a lot of time.

[benchplease]

What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at MoE serving. Not glam

My first guess would be this: Before rebuilding anything, I'd test one clean baseline with DeepSeek and compare from there.

[batchsize]

A couple questions before I guess too hard. What model, what runtime, and roughly what context size?

[nanobrook]

Not sure I agree there. I've seen people blame INT4 when the real issue was much more mundane. Not saying you're wrong, just that I wouldn't lock that in as the lesson yet.

[routerunner]

Thanks, that helps. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[benchplease]

Thanks, that helps. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimizat

I'd start with the simpler stack.