Working on a quantum simulator in Rust specifically based on photonics (for now), looking for collaborators to implement the photonics backend and also contribute in general.
So I've been experimenting with a lot of local LLMs lately, tried a bunch of Qwen and Gemma models with different quantisations however I feel I'm still not able to max out the tps I can possibly get out of my machine because of the wrong choice of llm server. I'm using a Macbook M4 Pro with 24 GB unified mem with ollama hooked to claude code and I would like if someone suggests a good combination of a llm server and a cli tool like opencode if they have tried multiple combinations.