libmir-cuda 0.3.0

CUDA inference backend for libmir
1
2
3
4
5
6
7
mod attention;
mod math;

pub use attention::{
    ClampedRoutedAttention, ClampedRoutedBatchSplitDecode, ClampedRoutedSplitDecode,
};
pub use math::{ClampedRoutedKernels, ClampedRoutedSpec};