boostr 0.1.0

ML framework built on numr - attention, quantization, model architectures
Documentation
1
2
3
4
5
6
7
8
9
pub mod dequant;
pub mod fused_int4_qkv;
pub mod fused_int4_swiglu;
pub mod helpers;
pub mod int4_gemm;
pub mod nf4;
pub mod quant_matmul;

pub use helpers::{common_helpers, read_f16_inline, read_i8_inline, read_u8_inline};