Crate ruvllm_sparse_attention 

Re-exports

pub use attention::dense_attention;

pub use attention::AttentionBackend;

pub use attention::AttentionError;

pub use attention::IncrementalLandmarks;

pub use attention::KvCache;

pub use attention::SparseAttentionConfig;

pub use attention::SubquadraticSparseAttention;

pub use fastgrnn_gate::FastGrnnGate;

pub use fastgrnn_gate::DEFAULT_HIDDEN_DIM as FASTGRNN_DEFAULT_HIDDEN_DIM;

pub use model::RuvLlmSparseBlock;

pub use model::RuvLlmSparseBlockConfig;

pub use tensor::Tensor3;

Modules

attention

fastgrnn_gate

FastGRNN salience gate for near-linear attention.

model

tensor