Struct TransformerBasedOptimizerConfig 

pub struct TransformerBasedOptimizerConfig<T: Float + Debug + Send + Sync + 'static> {

    pub model_dimension: usize,
    pub num_transformer_layers: usize,
    pub num_attention_heads: usize,
    pub attention_head_dimension: usize,
    pub feedforward_dimension: usize,
    pub sequence_length: usize,
    pub dropout_rate: f64,
    pub learning_rate: T,
    pub batch_size: usize,
    pub num_epochs: usize,
    pub activation_function: ActivationFunction,
    pub positional_encoding_type: PositionalEncodingType,
    pub memory_config: MemoryConfig,
    pub meta_learning_config: MetaLearningConfig<T>,
    pub performance_config: PerformanceConfig,
    pub enable_gradient_clipping: bool,
    pub gradient_clip_value: T,
    pub weight_decay: T,
    pub warmup_steps: usize,
    pub enable_layer_norm: bool,
    pub use_pre_norm: bool,
    pub enable_residual_connections: bool,
}

Configuration for transformer-based optimizer

Fields

model_dimension: usize

Model dimension (embedding size)

num_transformer_layers: usize

Number of transformer layers

num_attention_heads: usize

Number of attention heads

attention_head_dimension: usize

Dimension of each attention head

feedforward_dimension: usize

Feed-forward network dimension

sequence_length: usize

Maximum sequence length

dropout_rate: f64

Dropout rate

learning_rate: T

Learning rate

batch_size: usize

Batch size for training

num_epochs: usize

Number of training epochs

activation_function: ActivationFunction

Activation function type

positional_encoding_type: PositionalEncodingType

Positional encoding type

memory_config: MemoryConfig

Memory management configuration

meta_learning_config: MetaLearningConfig<T>

Meta-learning configuration

performance_config: PerformanceConfig

Performance tracking configuration

enable_gradient_clipping: bool

Enable gradient clipping

gradient_clip_value: T

Gradient clipping threshold

weight_decay: T

Weight decay factor

warmup_steps: usize

Warmup steps for learning rate schedule

enable_layer_norm: bool

Enable layer normalization

use_pre_norm: bool

Pre-norm vs post-norm

enable_residual_connections: bool

Enable residual connections

Implementations

impl<T: Float + Debug + Send + Sync + 'static> TransformerBasedOptimizerConfig<T>

pub fn small() -> Self

Create configuration for small models

pub fn large() -> Self

Create configuration for large models

pub fn for_training() -> Self

Create configuration optimized for training

pub fn for_inference() -> Self

Create configuration optimized for inference

pub fn validate(&self) -> Result<(), String>

Validate configuration

pub fn estimate_parameter_count(&self) -> usize

Calculate total parameters estimate

pub fn estimate_memory_usage(&self) -> f64

Calculate memory requirements (in MB)

Trait Implementations

impl<T: Clone + Float + Debug + Send + Sync + 'static> Clone for TransformerBasedOptimizerConfig<T>

fn clone(&self) -> TransformerBasedOptimizerConfig<T>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<T: Debug + Float + Debug + Send + Sync + 'static> Debug for TransformerBasedOptimizerConfig<T>

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

Formats the value using the given formatter.

impl<T: Float + Debug + Send + Sync + 'static> Default for TransformerBasedOptimizerConfig<T>

fn default() -> Self

Returns the “default value” for a type.

impl<'de, T> Deserialize<'de> for TransformerBasedOptimizerConfig<T>

where T: Deserialize<'de> + Float + Debug + Send + Sync + 'static,

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl<T> Serialize for TransformerBasedOptimizerConfig<T>

where T: Serialize + Float + Debug + Send + Sync + 'static,

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

where __S: Serializer,

Serialize this value into the given Serde serializer.