do you even need a vector DB here

[torchsmith25]

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

Short version: vLLM behaved well for me, but only after I stopped ignoring embeddings.

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

[sparrowhall]

Tried to reproduce it just now. On dual 3090 with llama.cpp, I saw basically the same behavior.

[cacheline]

I'd start with the simpler stack.

[rocmrunner40]

Can you fill in a bit more context? What model, what runtime, and roughly what context size?

There are a few different conversations hiding inside this thread, so being explicit about the goal helps.

[phasescope]

What exact setup are you running? Is this local-only or are you routing anything externally?

[mangopeak72]

I'd approach it in a boring order. Before rebuilding anything, I'd test one clean baseline with Gemma and compare from there.

[visionforge]

I'd approach it in a boring order. On B200, the easiest win is often to back off the aggressive settings and see whether the problem survives. Not glamorous advice, but repeatable baselines save a lot of time.

[circuitworks90]

I'd approach it in a boring order. On B200, the easiest win is often to back off the aggressive settings and see whether the problem survives. Not gla

A couple questions before I guess too hard. Is this local-only or are you routing anything externally?

[byteyard]

What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at batch size.

[emberpeak]

What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at batch size.

Not sure I agree there. I've seen people blame prefix caching when the real issue was much more mundane.