boostr 0.1.0

ML framework built on numr - attention, quantization, model architectures
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13

pub mod flash;
pub mod fused_qkv;
pub mod mla;
#[path = "rope.rs"]
pub mod rope;

pub use flash::multi_head_attention_impl;
pub use fused_qkv::{
    fused_output_projection_residual_bwd_impl, fused_output_projection_residual_impl,
    fused_qkv_projection_bwd_impl, fused_qkv_projection_impl,
};
pub use mla::scaled_dot_product_attention_impl;
pub use rope::{apply_rope_impl, apply_rope_interleaved_impl, apply_rope_yarn_impl};