distillation that keeps useful behavior with vLLM

[quiettoken]

Curious what people here are seeing.

I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling tokens/sec.

Would appreciate practical advice rather than theory.

This is mostly for local work rather than public serving.

[plainember52]

What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at tokens/sec. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[vramridge]

that fixed it for me too.

[thetagrid15]

Need a little more detail here. Is this local-only or are you routing anything externally?

[turboyard94]

That matches what I was seeing. I ended up in about the same place using SGLang. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[quietstack]

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. Worth testing the simpler explanation before adopting the more interesting one.

[amberpatch25]

I tested this tonight. My result was mixed: first run looked bad, second run looked normal. I expected the model choice to matter more than it did.

[turboyard94]

Good point. That matches what I saw on 7900 XTX. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[quietstack]

What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at tail latency. Once you know the plain path works, then add the faster path back in piece by piece.

[batchscope72]

I think the bottleneck is somewhere else. I've seen people blame tokens/sec when the real issue was much more mundane.