Embeddings discussion

[torchdesk]

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

Maybe useful as a starting point: PyTorch on Threadripper box, keeping quant drift in mind from the start.

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

[routearch]

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

worth checking the driver first.

[ggufwire22]

I'd approach it in a boring order. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at small local stacks. Not glamorous advice, but repeatable baselines save a lot of time.

[edgecoast]

I'd approach it in a boring order. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at small local stacks. Not

Tried to reproduce it just now. My result was mixed: first run looked bad, second run looked normal. The surprise was that memory pressure showed up before throughput collapsed.

[torchdesk]

What exact setup are you running? Is this local-only or are you routing anything externally?

[vectorshed43]

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

[quantnest]

Ran a quick check on my box. The issue went away once I changed one setting and restarted the stack cleanly. The main thing that moved was latency, not quality.

[vectorshed43]

Good point. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[vramscope]

Good point. That matches what I saw on A6000.

[quantnest]

Good point. That matches what I saw on A6000.

A couple things I'd check first. If you're on Ollama, I'd confirm the version first and then reduce variables until the behavior is obvious. Once you know the plain path works, then add the faster path back in piece by piece.