libmir-cuda 0.2.0

CUDA inference backend for libmir
1
2
3
4
5
6
7
8
9
10
11
mod gated;
mod nvfp4;
mod pair;
mod reduce;

pub use gated::{
    GatedActivation, SelectedAffineGated, SelectedAffineGatedLaunch, SelectedAffineGatedSpec,
};
pub use nvfp4::{NvFp4BankView, SelectedNvFp4Gated, SelectedNvFp4Reduce, SelectedNvFp4Spec};
pub use pair::{SelectedAffinePair, SelectedAffinePairLaunch, SelectedAffinePairSpec};
pub use reduce::{SelectedAffineReduce, SelectedAffineReduceLaunch, SelectedAffineReduceSpec};