eval setup for local models for Mistral

[audioforge]

Spent some time on this today and wanted to compare notes.

I've gone back and forth between Ollama and Ollama for DeepSeek. The interesting difference for me has been tokens/sec.

Trying to avoid rebuilding the whole thing twice.

I care more about stability than squeezing the last possible tokens/sec out of it.

[circuitworks90]

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.