ruda-nn 0.21.53

Ruda neural network layers, activation modules and losses.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

mod cross_attention;
mod mask;
mod mha;
mod packed;
mod dense;
mod cached;
mod sparse_ops;
mod compression;
mod compressed_projection;
mod indexer;
mod compressed;
mod packed_compressed;
pub use compression::*;
pub use compressed_projection::*;
pub use indexer::*;
pub use compressed::*;
pub use packed_compressed::*;
pub use sparse_ops::{sparse_stable_topk, sparse_gather_entries, indexer_kl_loss, SparseRotaryEmbedding};
pub(crate) use cached::append as append_cached_projected;
pub(crate) use cached::cached_masks as cached_projected_masks;
mod packed_projection;
mod packed_mask;
pub use packed_mask::{PackedDocumentAttentionMask,packed_scaled_dot_product_attention_masked};

pub use cross_attention::*;
pub use mask::*;
pub use mha::*;
pub use packed::*;
pub use dense::*;