quality tests for local assistants for Mistral

[audioforge]

Posting this in case it saves someone else a little time.

I'm not trying to build a perfect architecture diagram. I just want a sane starting point around vLLM and 4090.

Simple and boring is a valid answer.

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.

[ggufvault]

Interesting, I hadn't tried that. The pattern around VRAM headroom has been pretty similar for me. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.

[plaingarden18]

A couple things I'd check first. On M3 Max, the easiest win is often to back off the aggressive settings and see whether the problem survives.

[bytesupply76]

That matches what I was seeing. I ended up in about the same place using PyTorch. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[edgequill]

That matches what I was seeing. I ended up in about the same place using PyTorch. The biggest change for me was simplifying the stack instead of addin

A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Gemma and compare from there. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[cacheworks]

Appreciate the detail. 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.

[sparrowshed]

Interesting, I hadn't tried that. That matches what I saw on M3 Max.

[cacheline]

What worked for me was pretty simple. If you're on vLLM, I'd confirm the version first and then reduce variables until the behavior is obvious. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[tinycontext]

Did a small comparison this afternoon. The issue went away once I changed one setting and restarted the stack cleanly. The main thing that moved was latency, not quality.

[marbleyard32]

Good point. 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.