ferrum-kv 0.7.7

KV cache management with PagedAttention for Ferrum inference
Documentation
1
2
3
4
5
6
7
8
9
10
11
pub mod handle;
pub mod pool;
pub mod storage;
pub mod table;

pub use handle::DefaultKvCacheHandle;
pub use pool::{
    Block, BlockAllocation, BlockPool, BlockPoolStats, BlockState, LogicalBlockId, PhysicalBlockId,
};
pub use storage::{BlockStorage, BlockStorageConfig};
pub use table::DefaultBlockTable;