Embeddings discussion for FLUX

[vectorshed43]

That matches what I was seeing. That matches what I saw on mini PC. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[quantnest]

That matches what I was seeing. That matches what I saw on mini PC. Worth rechecking your defaults though, because one bad assumption can make the res

I tested this tonight. My result was mixed: first run looked bad, second run looked normal. The surprise was that memory pressure showed up before throughput collapsed.

[routerunner]

Good point. I ended up in about the same place using Ollama. The biggest change for me was simplifying the stack instead of adding one more optimization layer.