Hello,
Thank you for this project, it's very interesting. I've looked at the code and spotted some possible improvements and refactoring points.
Before diving deeper into its use, I'd like to know what's currently available and fully functional, given that the training time (and resource consumption) seems to be a significant investment:
-
Is it heads-up only? I saw pub const N but it seems some values are hardcoded to 2 in other project places.
-
What hypothetical clustering/training times have you obtained? (with the initial parameters)
-
Are there still known bottlenecks ?
I'm available to help improve everything, but I'd appreciate this information before I get started!
Hello,
Thank you for this project, it's very interesting. I've looked at the code and spotted some possible improvements and refactoring points.
Before diving deeper into its use, I'd like to know what's currently available and fully functional, given that the training time (and resource consumption) seems to be a significant investment:
Is it heads-up only? I saw
pub const Nbut it seems some values are hardcoded to 2 in other project places.What hypothetical clustering/training times have you obtained? (with the initial parameters)
Are there still known bottlenecks ?
I'm available to help improve everything, but I'd appreciate this information before I get started!