70B still worth the pain: worth it?

[fastbench]

I figured I'd ask here before changing everything.

I'm mostly working with Mistral in MLX on 4090. The part I'm trying to understand is KV cache pressure.

Would appreciate practical advice rather than theory.

[torchsupply99]

Thanks, that helps. I ended up in about the same place using vLLM. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[vrampilot]

Thanks, that helps. I ended up in about the same place using vLLM. The biggest change for me was simplifying the stack instead of adding one more opti

Appreciate the detail. The pattern around MoE serving 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.