Enum ModelArchitecture 

pub enum ModelArchitecture {
    Mistral,
    Llama,
    Phi,
    Phi3,
    Qwen,
    Gemma,
    Gemma2,
}

Model architecture types supported by RuvLLM.

RuvLLM supports multiple transformer architectures with varying characteristics optimized for different use cases.

Supported Architectures

Architecture	Parameter Sizes	Best For
Llama	1B-70B	General purpose, chat
Mistral	7B	Code, instruction following
Phi	1.5-3B	Efficient edge deployment
Phi3	3B-14B	Extended context, SuRoPE
Qwen	0.5B-72B	Multilingual, reasoning
Gemma	2B-7B	Efficient, instruction-tuned
Gemma2	2B-27B	Soft-capping, alternating attention

Example

use ruvllm::backends::ModelArchitecture;

let arch = ModelArchitecture::Mistral;
assert_eq!(arch.config_name(), "mistral");

let phi3 = ModelArchitecture::Phi3;
assert_eq!(phi3.config_name(), "phi3");

Variants

Mistral

Mistral architecture (7B, Codestral)

Llama

Llama architecture (1B-70B)

Phi

Phi architecture (1.5, 2)

Phi3

Phi-3 architecture (SuRoPE, SwiGLU, sliding window)

Qwen

Qwen architecture

Gemma

Gemma architecture (original)

Gemma2

Gemma-2 architecture (soft-capping, alternating local/global attention)

Implementations

impl ModelArchitecture

pub fn config_name(&self) -> &'static str

Get architecture name for HuggingFace model config

pub fn detect_from_model_id(model_id: &str) -> Option<Self>

Detect architecture from model ID string

pub fn uses_gqa(&self) -> bool

Check if this architecture uses GQA (Grouped Query Attention)

pub fn uses_sliding_window(&self) -> bool

Check if this architecture uses sliding window attention

pub fn default_sliding_window(&self) -> Option<usize>

Get default sliding window size for this architecture

Trait Implementations

impl Clone for ModelArchitecture

fn clone(&self) -> ModelArchitecture

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ModelArchitecture

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

Formats the value using the given formatter.

impl Default for ModelArchitecture

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for ModelArchitecture

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Hash for ModelArchitecture

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for ModelArchitecture

fn eq(&self, other: &ModelArchitecture) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for ModelArchitecture

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl Copy for ModelArchitecture

impl Eq for ModelArchitecture

impl StructuralPartialEq for ModelArchitecture