Module context 

Safe wrapper around llama_context.

Submodules:

• tensor_capture — hook cb_eval during LlamaContext::decode to copy intermediate tensors (per-layer hidden states, norms, …).
• memory_breakdown — per-buffer memory usage after load/decode.
• kv_cache — sequence copy, shift, and clear helpers.

Re-exports

pub use memory_breakdown::MemoryBreakdownEntry;

pub use tensor_capture::CapturedTensor;

pub use tensor_capture::TensorCapture;

pub use tensor_transaction::CapturedTensorData;

pub use tensor_transaction::TensorAccess;

pub use tensor_transaction::TensorBatchRow;

pub use tensor_transaction::TensorCallbackFailure;

pub use tensor_transaction::TensorDataMut;

pub use tensor_transaction::TensorElementType;

pub use tensor_transaction::TensorFiniteValidation;

pub use tensor_transaction::TensorRowMapping;

pub use tensor_transaction::TensorSelector;

pub use tensor_transaction::TensorShape;

pub use tensor_transaction::TensorTransaction;

pub use tensor_transaction::TensorTransactionError;

pub use tensor_transaction::TensorTransactionHandler;

pub use tensor_transaction::TensorTransactions;

pub use tensor_transaction::TensorWriteback;

pub use tensor_transaction::TransactionalTensorCapture;

Modules

kv_cache

utilities for working with the kv cache

memory_breakdown

Per-buffer-type memory usage reported by llama.cpp.

params

A safe wrapper around llama_context_params.

perf

Safe wrapper around llama_perf_context_data.

session

utilities for working with session files

tensor_capture

Capture intermediate tensor outputs during crate::LlamaContext::decode.

tensor_transaction

Bounded, owned tensor capture and transactional mutation during decode.

Structs

LlamaContext

A safe wrapper around the llama_context C++ context.