8B vs 14B sweet spot

[lablab]

Thought I'd post a small recipe because a lot of threads ask similar things.

Maybe useful as a starting point: vLLM on GB200, keeping KV cache pressure in mind from the start.

If someone wants the exact settings I can clean them up and post them.

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

[latentscope35]

What worked for me was pretty simple. If you're on TensorRT-LLM, 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.

[modgarden]

Need a little more detail here. What model, what runtime, and roughly what context size?

[thetagrid15]

My first guess would be this: On GB200, the easiest win is often to back off the aggressive settings and see whether the problem survives. Once you know the plain path works, then add the faster path back in piece by piece.

[homelabhenry]

I'd start with the simpler stack.

[quantyard]

Good point. I ended up in about the same place using Ollama. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[cachelab]

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

[fastlab]

Not sure I agree there. I get why that seems right, but on GB200 the bottleneck felt elsewhere. Worth testing the simpler explanation before adopting the more interesting one.

[edgekite]

What worked for me was pretty simple. If you're on TensorRT-LLM, I'd confirm the version first and then reduce variables until the behavior is obvious.

[amberrunner]

What worked for me was pretty simple. If you're on TensorRT-LLM, I'd confirm the version first and then reduce variables until the behavior is obvious

Good point. I ended up in about the same place using vLLM. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.