Struct ArchitectureOverrides 

pub struct ArchitectureOverrides {
    pub hidden_size: Option<usize>,
    pub num_hidden_layers: Option<usize>,
    pub num_attention_heads: Option<usize>,
    pub num_kv_heads: Option<usize>,
    pub intermediate_size: Option<usize>,
    pub vocab_size: Option<usize>,
    pub max_position_embeddings: Option<usize>,
    pub rms_norm_eps: Option<f32>,
    pub rope_theta: Option<f32>,
    pub use_bias: Option<bool>,
    pub head_dim: Option<usize>,
}

Architecture override parameters from YAML manifest.

These override individual fields of the TransformerConfig resolved from config.json or preset defaults. Only Some fields apply; None fields are left as-is from the base config.

Fields

hidden_size: Option<usize>

Hidden size (embedding dimension)

num_hidden_layers: Option<usize>

Number of transformer layers

num_attention_heads: Option<usize>

Number of attention heads

num_kv_heads: Option<usize>

Number of key-value heads (for grouped-query attention)

intermediate_size: Option<usize>

FFN intermediate dimension

vocab_size: Option<usize>

Vocabulary size

max_position_embeddings: Option<usize>

Maximum sequence/position length

rms_norm_eps: Option<f32>

RMS normalization epsilon

rope_theta: Option<f32>

RoPE theta (rotary positional encoding base)

use_bias: Option<bool>

Whether to use bias in linear layers

head_dim: Option<usize>

Per-head dimension override (for models where head_dim != hidden_size / num_heads)

Implementations

impl ArchitectureOverrides

pub fn is_empty(&self) -> bool

Returns true if no overrides are set.

Trait Implementations

impl Clone for ArchitectureOverrides

fn clone(&self) -> ArchitectureOverrides

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ArchitectureOverrides

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

Formats the value using the given formatter.

impl Default for ArchitectureOverrides

fn default() -> ArchitectureOverrides

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for ArchitectureOverrides

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for ArchitectureOverrides

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.