Struct ModelConfig 

pub struct ModelConfig {

    pub hidden_size: usize,
    pub num_layers: usize,
    pub num_attention_heads: usize,
    pub num_kv_heads: usize,
    pub intermediate_size: usize,
    pub vocab_size: usize,
    pub context_length: usize,
    pub rope_theta: f32,
    pub head_dim: Option<usize>,
    pub layer_types: Option<Vec<String>>,
    pub num_kv_shared_layers: Option<usize>,
    pub use_double_wide_mlp: Option<bool>,
    pub hidden_act: Option<String>,
    pub num_experts: Option<usize>,
    pub num_experts_per_tok: Option<usize>,
    pub tie_word_embeddings: Option<bool>,
    pub conv_l_cache: Option<usize>,
    pub sliding_window: Option<usize>,
    pub partial_rotary_factor: Option<f32>,
    pub global_head_dim: Option<usize>,
}

Fields

hidden_size: usize

num_layers: usize

num_attention_heads: usize

num_kv_heads: usize

intermediate_size: usize

vocab_size: usize

context_length: usize

rope_theta: f32

head_dim: Option<usize>

layer_types: Option<Vec<String>>

num_kv_shared_layers: Option<usize>

use_double_wide_mlp: Option<bool>

hidden_act: Option<String>

num_experts: Option<usize>

num_experts_per_tok: Option<usize>

tie_word_embeddings: Option<bool>

conv_l_cache: Option<usize>

LFM2 short-conv kernel / cache length (HF conv_L_cache, default 3).

sliding_window: Option<usize>

Sliding-window attention length (HF sliding_window). Required for gemma-4.

partial_rotary_factor: Option<f32>

Gemma-4 global p-RoPE fraction (HF partial_rotary_factor). Required for gemma-4.

global_head_dim: Option<usize>

Gemma-4 full-attention head dim (HF global_head_dim). Required for gemma-4.

Trait Implementations

impl Clone for ModelConfig

fn clone(&self) -> ModelConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ModelConfig

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for ModelConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.