Struct LlamaModel 

pub struct LlamaModel {
    pub embed_tokens: Embedding,
    pub layers: Vec<LlamaBlock>,
    pub norm: RMSNorm,
    pub lm_head: Linear,
}

A Llama-style decoder: embedding, blocks, final norm and LM head.

Fields

embed_tokens: Embedding

Token embeddings.

layers: Vec<LlamaBlock>

Decoder blocks.

norm: RMSNorm

Final norm.

lm_head: Linear

Projection to vocabulary logits.

Implementations

impl LlamaModel

pub fn forward( &self, ctx: &WgpuContext, engine: &ComputeEngine, input_ids: &[u32], start_pos: usize, cache: Option<&mut KVCache>, mask: Option<&UnifiedTensor>, ) -> Tensor

Logits [1, seq, vocab] for input_ids, the first of which is at start_pos.

pub fn params(&self) -> Vec<Tensor>

Trainable tensors (weights, then any LoRA adapters).

pub fn named_params(&self, prefix: &str) -> HashMap<String, Tensor>

Parameters keyed by Hugging Face-style names under prefix.

pub fn lora_params(&self) -> Vec<Tensor>

Only the LoRA adapter tensors (what LoRA fine-tuning updates).