do you even need a vector DB here for Mistral

[vectorhouse71]

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

Short version: PyTorch behaved well for me, but only after I stopped ignoring NVFP4.

Curious if anyone else landed in the same place.

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

[embersmith]

that fixed it for me too.

[rocmplane]

that fixed it for me too.

Good point. The pattern around MoE serving has been pretty similar for me. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[kernelghost]

that fixed it for me too.

That matches what I was seeing. The pattern around long context has been pretty similar for me. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[kernelgrid]

I'd approach it in a boring order. On M3 Max, 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.

[audiopath66]

I'd approach it in a boring order. On 5090, 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.

[labpath]

My first guess would be this: If you're on SGLang, I'd confirm the version first and then reduce variables until the behavior is obvious.

[batchember]

A couple things I'd check first. On B200, the easiest win is often to back off the aggressive settings and see whether the problem survives.

I hit almost the same issue last night.

[tinycontext]

I tested this tonight. On 5090 with MLX, I saw basically the same behavior. The main thing that moved was latency, not quality.

[modbench]

I'd start with the simpler stack.