Struct LlamaContextParams 

pub struct LlamaContextParams {
    pub context_params: llama_context_params,
}

Fields

context_params: llama_context_params

Implementations

impl LlamaContextParams

pub fn with_n_ctx(self, n_ctx: Option<NonZeroU32>) -> Self

pub const fn n_ctx(&self) -> Option<NonZeroU32>

pub const fn with_n_batch(self, n_batch: u32) -> Self

pub const fn n_batch(&self) -> u32

pub const fn with_n_ubatch(self, n_ubatch: u32) -> Self

pub const fn n_ubatch(&self) -> u32

pub const fn with_flash_attention_policy( self, policy: llama_flash_attn_type, ) -> Self

pub const fn flash_attention_policy(&self) -> llama_flash_attn_type

pub const fn with_offload_kqv(self, enabled: bool) -> Self

pub const fn offload_kqv(&self) -> bool

pub fn with_rope_scaling_type(self, rope_scaling_type: RopeScalingType) -> Self

pub fn rope_scaling_type(&self) -> RopeScalingType

pub const fn with_rope_freq_base(self, rope_freq_base: f32) -> Self

pub const fn rope_freq_base(&self) -> f32

pub const fn with_rope_freq_scale(self, rope_freq_scale: f32) -> Self

pub const fn rope_freq_scale(&self) -> f32

pub const fn n_threads(&self) -> i32

pub const fn n_threads_batch(&self) -> i32

pub const fn with_n_threads(self, n_threads: i32) -> Self

pub const fn with_n_threads_batch(self, n_threads: i32) -> Self

pub const fn embeddings(&self) -> bool

pub const fn with_embeddings(self, embedding: bool) -> Self

pub fn with_cb_eval(self, cb_eval: ggml_backend_sched_eval_callback) -> Self

pub const fn with_cb_eval_user_data( self, cb_eval_user_data: *mut c_void, ) -> Self

pub fn with_pooling_type(self, pooling_type: LlamaPoolingType) -> Self

pub fn pooling_type(&self) -> LlamaPoolingType

pub const fn with_swa_full(self, enabled: bool) -> Self

pub const fn swa_full(&self) -> bool

pub const fn with_n_seq_max(self, n_seq_max: u32) -> Self

pub const fn n_seq_max(&self) -> u32

pub fn with_type_k(self, type_k: KvCacheType) -> Self

pub fn type_k(&self) -> KvCacheType

pub fn with_type_v(self, type_v: KvCacheType) -> Self

pub fn type_v(&self) -> KvCacheType

pub fn with_attention_type(self, attention_type: LlamaAttentionType) -> Self

pub fn attention_type(&self) -> LlamaAttentionType

pub const fn with_yarn_ext_factor(self, yarn_ext_factor: f32) -> Self

pub const fn yarn_ext_factor(&self) -> f32

pub const fn with_yarn_attn_factor(self, yarn_attn_factor: f32) -> Self

pub const fn yarn_attn_factor(&self) -> f32

pub const fn with_yarn_beta_fast(self, yarn_beta_fast: f32) -> Self

pub const fn yarn_beta_fast(&self) -> f32

pub const fn with_yarn_beta_slow(self, yarn_beta_slow: f32) -> Self

pub const fn yarn_beta_slow(&self) -> f32

pub const fn with_yarn_orig_ctx(self, yarn_orig_ctx: u32) -> Self

pub const fn yarn_orig_ctx(&self) -> u32

pub const fn with_defrag_thold(self, defrag_thold: f32) -> Self

pub const fn defrag_thold(&self) -> f32

pub const fn with_no_perf(self, no_perf: bool) -> Self

pub const fn no_perf(&self) -> bool

pub const fn with_op_offload(self, op_offload: bool) -> Self

pub const fn op_offload(&self) -> bool

pub const fn with_kv_unified(self, kv_unified: bool) -> Self

pub const fn kv_unified(&self) -> bool

Trait Implementations

impl Clone for LlamaContextParams

fn clone(&self) -> LlamaContextParams

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for LlamaContextParams

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for LlamaContextParams

fn default() -> Self

Returns the “default value” for a type.

impl Send for LlamaContextParams

impl Sync for LlamaContextParams