+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: Optimization & Performance (https://www.datasculpting.ai/inference/forumdisplay.php?fid=10)
+--- Forum: Context / KV Cache (https://www.datasculpting.ai/inference/forumdisplay.php?fid=41)
+--- Thread: draft model choices for Phi (/showthread.php?tid=182)
Pages:
1
2
draft model choices for Phi - cacheline - 09-09-2026
Thought I'd post a small recipe because a lot of threads ask similar things.
This isn't revolutionary, but the combination of LM Studio and 5090 felt surprisingly usable once I tuned around quant drift.
If someone wants the exact settings I can clean them up and post them.
Re: draft model choices for Phi - fastfield4 - 09-11-2026
Quote:Thought I'd post a small recipe because a lot of threads ask similar things.
Tried to reproduce it just now. On 5090 with SGLang, I saw basically the same behavior. I expected the model choice to matter more than it did.
I'm happy to test one more variant if you want a second data point.
Re: draft model choices for Phi - cacheline - 09-11-2026
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.
Re: draft model choices for Phi - vramarch - 09-11-2026
That matches what I was seeing. I don't think your conclusion is far off. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.
Re: draft model choices for Phi - cacheline - 09-12-2026
A couple questions before I guess too hard. Is this local-only or are you routing anything externally?
Re: draft model choices for Phi - cobaltworks25 - 09-12-2026
subscribed, curious where this lands.
Re: draft model choices for Phi - cacheline - 09-12-2026
Can you fill in a bit more context? Do you see the same thing with a smaller batch or shorter prompt?
Re: draft model choices for Phi - promptpoint - 09-15-2026
My first guess would be this: Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at KV cache pressure. Once you know the plain path works, then add the faster path back in piece by piece.
Re: draft model choices for Phi - amberrunner - 09-15-2026
Ran a quick check on my box. My result was mixed: first run looked bad, second run looked normal. The main thing that moved was latency, not quality.
Re: draft model choices for Phi - cacheline - 09-18-2026
A couple things I'd check first. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at HBM constraints. Not glamorous advice, but repeatable baselines save a lot of time.