Evan Almloff

This doesn't appear to be much faster in kalosm-llama at least on the CPU. We can revisit this once GPU support is added

Accelerator support is now enabled. This may provide more performance benefits on CUDA/metal

You can now train a classifier in Kalosm. We should support fine tuning models in Kalosm first before we provide an interface in Floneum

That would be great to support. Kalosm uses `async-openai` to interact with the open ai API. I think we can add support by just allowing a custom model [here](https://github.com/floneum/floneum/blob/393ca97ae83287ec63d8751e79c55b67cf6ed73b/interfaces/language-model/src/remote/open_ai.rs#L98)

We can draw on some of the algorithms https://github.com/pemistahl/grex uses here

This looks like exactly what I was envisioning: https://github.com/cantino/selectorgadget

👋 We built a similar system for Dioxus and published it as a separate library: [generational-box](https://lib.rs/crates/generational-box) is similar to Leptos' copy sync system except: - You can include your own...

> @ealmloff Btw are there any plans for making EventHandler etc. use that? Would be ⭐amazing⭐ if it was possible to just move that around more freely That is something...

The liveview, and tui platforms don't exist in the CLI, so this will currently break those platforms in the template. It looks like cargo generate [provides some utilities to convert...

We format the toml files [here](https://github.com/DioxusLabs/dioxus/blob/61e2478fbf1b9c7fbdb6f7e9644d00acc50a10bf/packages/cli/src/cli/create.rs#L40-L58) in the dioxus cli to clean up the output of the template