Module params 

A safe wrapper around llama_context_params.

Use LlamaContextParams to configure context size, batching, KV layout, RoPE / YaRN scaling, flash attention, per-sequence samplers, and pairing with another context (ctx_other).

Structs

LlamaContextParams

Builder for llama_context_params.

Enums

LlamaAttentionType

Attention mask mode used when the context runs in embedding mode.

LlamaContextType

A rusty wrapper around llama_context_type.

LlamaFlashAttnType

Flash-attention enablement policy for the context.

LlamaPoolingType

A rusty wrapper around LLAMA_POOLING_TYPE.

ParamsCloneError

Error returned when LlamaContextParams::try_clone cannot duplicate state.

RopeScalingType

A rusty wrapper around rope_scaling_type.